Split (1)

train · ~9.11M rows (showing the first 93.5k)

identifier stringlengths 11 32	pdf_url stringlengths 17 4.62k ⌀	lang stringclasses 120 values	error stringclasses 1 value	title stringlengths 2 500 ⌀	source_name stringlengths 1 435 ⌀	publication_year float64 1.9k 2.02k	license stringclasses 3 values	word_count int64 0 1.64M	text stringlengths 1 9.75M
https://openalex.org/W2140053715	https://www.scielo.br/j/mioc/a/HLYTk74sQGk9HTwTwVLHSHt/?lang=en&format=pdf	English	null	Mycobacterium leprae in six-banded (Euphractus sexcinctus) and nine-banded armadillos (Dasypus novemcinctus) in Northeast Brazil	Memórias do Instituto Oswaldo Cruz	2,012	cc-by	4,639	Cristiane Cunha Frota1/+, Luana Nepomuceno Costa Lima1, Adalgiza da Silva Rocha2, Philip Noel Suffys2, Benedito Neilson Rolim3, Laura Cunha Rodrigues4, Maurício Lima Barreto5, Carl Kendall6, Ligia Regina Sansigolo Kerr7 leprae was detected in 21% (6/29) of the animals, including five D. novemcinctus and one E. sexcinctus. This is the first Brazilian study to iden tify the presence of a biomarker of M. leprae in wild armadillos (D. novemcinctus and E. sexcinctus) in a leprosy hyperendemic area where there is continuous contact between humans and armadillos. Key words: Euphractus sexcinctus - Dasypus novemcinctus - Mycobacterium leprae - eco-epidemiology - leprosy M. leprae in the wild for the first time. The authors pro posed that armadillos might have acquired leprosy infec tions from untreated human patients in the USA. Mycobacterium leprae, the causative agent of leprosy, is not cultivable in vitro. The lack of growth on standard mycobacterial isolation media differentiates this organ ism from other mycobacterial pathogens. Human beings are the only known reservoir of infection, except in the southern United States of America (USA), where nine- banded armadillos (Dasypus novemcinctus) are believed to also provide a reservoir (Truman et al. 2011). The exact mode of transmission of leprosy between humans and armadillos is not known, though cross-re activity between IgM antibodies against phenolic glyco lipid-I of humans and armadillos has been reported (Tru man et al. 1991, Job et al. 1992). Infected nine-banded armadillos have been identified in the states of Texas and Louisiana and in Central and South America (Smith et al. 1983, Amezcua et al. 1984, Stallknecht et al. 1987, Zu marraga et al. 2001). Additionally, biomarkers of arma dillo infection have been detected in Colombia and Brazil (Deps et al. 2007, Cardona-Castro et al. 2009). Several studies have shown an association between armadillo exposure through hunting, cleaning and eating the meat and the development of leprosy (Clark et al. 2008, Deps et al. 2008, Truman 2008). More than half of the leprosy cases that have been reported in the southeastern USA have described some direct or indirect exposure to arma dillos (Bruce et al. 2000, Truman et al. 2011) and other studies have raised the hypothesis that exposure to these animals could be a significant risk factor for leprosy in Brazil (Kerr-Pontes et al. 2006, Deps et al. 2008). p ( ) In 1960, Shepard introduced the footpad mouse model to study experimental leprosy. In 1971, nine-banded ar madillos were successfully infected with the bacillus and developed clinical symptoms and pathologies similar to the human disease (Kirchheimer et al. 1972, Kirchheim er & Sanchez 1977). Cristiane Cunha Frota1/+, Luana Nepomuceno Costa Lima1, Adalgiza da Silva Rocha2, Philip Noel Suffys2, Benedito Neilson Rolim3, Laura Cunha Rodrigues4, Maurício Lima Barreto5, Carl Kendall6, Ligia Regina Sansigolo Kerr7 Cristiane Cunha Frota1/+, Luana Nepomuceno Costa Lima1, Adalgiza da Silva Rocha2, Philip Noel Suffys2, Benedito Neilson Rolim3, Laura Cunha Rodrigues4, Maurício Lima Barreto5, Carl Kendall6, Ligia Regina Sansigolo Kerr7 Cristiane Cunha Frota1/+, Luana Nepomuceno Costa Lima1, Adalgiza da Silva Rocha2, Philip Noel Suffys2, Benedito Neilson Rolim3, Laura Cunha Rodrigues4, Maurício Lima Barreto5, Carl Kendall6, Ligia Regina Sansigolo Kerr7 1Departamento de Patologia e Medicina Legal 7Departamento de Saúde Comunitária, Faculdade de Medicina, Universidade Federal do Ceará, Fortaleza, CE, Brasil 2Laboratório de Biologia Molecular Aplicada a Micobactérias, Instituto Oswaldo Cruz-Fiocruz, Rio de Janeiro, RJ, Brasil 3Controle de Zoonoses/Vetores, Secretaria de Saúde do Estado do Ceará, Fortaleza, CE, Brasil 4London School of Hygiene and Tropical Medicine, London, UK 5Instituto de Saúde Coletiva, Universidade Federal da Bahia, Salvador, BA, Brasil 6Department of Global Community Health and Behavioral Sciences, Center for Global Health Equity, Tulane University School of Public Health and Tropical Medicine, New Orleans, LA, USA 1Departamento de Patologia e Medicina Legal 7Departamento de Saúde Comunitária, Faculdade de Medicina, Universidade Federal do Ceará, Fortaleza, CE, Brasil 2Laboratório de Biologia Molecular Aplicada a Micobactérias, Instituto Oswaldo Cruz-Fiocruz, Rio de Janeiro, RJ, Brasil 3Controle de Zoonoses/Vetores, Secretaria de Saúde do Estado do Ceará, Fortaleza, CE, Brasil 4London School of Hygiene and Tropical Medicine, London, UK 5Instituto de Saúde Coletiva, Universidade Federal da Bahia, Salvador, BA, Brasil 6Department of Global Community Health and Behavioral Sciences, Center for Global Health Equity, Tulane University School of Public Health and Tropical Medicine, New Orleans, LA, USA Human beings are the main reservoir of the causative agent of leprosy, Mycobacterium leprae. In the Americas, nine-banded armadillos (Dasypus novemcinctus) also act as a reservoir for the bacillus. In the state of Ceará (CE), which is located in Northeast Brazil and is an endemic area of leprosy, there are several species of armadillos, in cluding D. novemcinctus and Euphractus sexcinctus (six-banded armadillo). Contact between humans and armadil los occur mainly through hunting, cleaning, preparing, cooking and eating. This study identified M. leprae DNA in the two main species of armadillos found in Northeast Brazil. A total of 29 wild armadillos (27 D. novemcinctus and 2 E. sexcinctus) were captured in different environments of CE countryside. Samples from the ear, nose, liver and spleen from each of these animals were tested by a nested M. leprae-specific repetitive element polymerase chain reaction assay. The samples that tested positive were confirmed by DNA sequencing. M. 209 209 Mem Inst Oswaldo Cruz, Rio de Janeiro, Vol. 107(Suppl. I): 209-213, 2012 online \| memorias.ioc.fiocruz.br Mycobacterium leprae in six-banded (Euphractus sexcinctus) and nine-banded armadillos (Dasypus novemcinctus) in Northeast Brazil Cristiane Cunha Frota1/+, Luana Nepomuceno Costa Lima1, Adalgiza da Silva Ro Philip Noel Suffys2, Benedito Neilson Rolim3, Laura Cunha Rodrigues4, Maurício Lima Barreto5, Carl Kendall6, Ligia Regina Sansigolo Kerr7 Financial support: CNPq (308539/2006-0), FUNCAP (303/05), CAPES (PROCAD 0204056) + Corresponding author: cristianefrota71@gmail.com Received 3 April 2012 Accepted 11 July 2012 Cristiane Cunha Frota1/+, Luana Nepomuceno Costa Lima1, Adalgiza da Silva Rocha2, Philip Noel Suffys2, Benedito Neilson Rolim3, Laura Cunha Rodrigues4, Maurício Lima Barreto5, Carl Kendall6, Ligia Regina Sansigolo Kerr7 Armadillos provide a much more effective animal model than the mouse footpad, produc ing 106 acid-fast bacilli/mL (Shepard 1985). Because ar madillos do not breed well in captivity, the animals to be used in the laboratory need to be captured in the wild. While searching for these animals in 1975, Walsh et al. (1977) found armadillos that were naturally infected with Unlike Brazil, where there are other species of arma dillos, only D. novemcinctus is found in the USA. Among the 21 species of armadillos found in Brazil, Euphractus sexcinctus (six-banded armadillo) is commonly found. This species is known to eat animal carcasses, justifying Financial support: CNPq (308539/2006-0), FUNCA CAPES (PROCAD 0204056) + Corresponding author: cristianefrota71@gmail.com Received 3 April 2012 Accepted 11 July 2012 M. leprae in wild armadillos • Cristiane Cunha Frota et al. 210 for 18 h at 60ºC. The digestion was conducted at 97ºC for 15 min (de Wit et al. 1991). The extract was purified once with phenol/chloroform/isoamyl alcohol (25:24:1) and then precipitated with ethanol. the local name of “vulture or gravedigger” and, because of this, hunters keep the animal alive and caged for sev eral days before eating (Dalponte & Tavares-Filho 2004, Armadillo Online! 2012). This type of “quarantine” is believed to “clean” the animal, but instead enhances its contact with humans. In the state of Ceará (CE), located in Northeast Brazil, armadillos are used as a source of meat and hunting is seen as a leisure pursuit. Polymerase chain reaction (PCR) and nucleotide sequencing - The M. leprae-specific repetitive element (RLEP) PCR was amplified in a nested PCR reaction. The primers for RLEP2-1 (5’-atatcgatgcaggcgt gag-3’) and RLEP2-2 (5’-ggatcatcgatgcact gttc-3’) amplified a 282-bp sequence of the RLEP element. The second set of inner primers, RLEP2-3 (5’- gggtaggggcgttttagtgt-3’) and RLEP2-2, am plified a 238-bp product. A 1 µL aliquot of the isolated DNA was added to 24 µL of PCR mix, which contained 15 mM Tris-HCl (pH 8.0), 50 mM KCl, 1.5 mM MgCl2, 0.2 mM dNTP, 5% DMSO, 1.25 units of Taq DNA Poly merase and 0.2 µM of each primer. The mixture was de natured at 94ºC for 4 min, followed by 35 PCR cycles (30 s at 94ºC, 30 s at 59.6ºC and 1 min at 72ºC), with a final extension at 72ºC for 10 min. Each run includ ed negative and positive controls. Cristiane Cunha Frota1/+, Luana Nepomuceno Costa Lima1, Adalgiza da Silva Rocha2, Philip Noel Suffys2, Benedito Neilson Rolim3, Laura Cunha Rodrigues4, Maurício Lima Barreto5, Carl Kendall6, Ligia Regina Sansigolo Kerr7 For the nested PCR, 0.5 µL of product was used as the DNA template. The amplification reactions were visualised on a 1.5% aga rose gel. Each PCR sample was double-blind tested by different researchers in Fortaleza and Rio de Janeiro. When the results for the same sample were different, a third PCR was performed for confirmation. Different amounts of purified DNA from M. leprae (kindly donat ed by Dr R Truman, Louisiana State University, USA) were added to all negative PCR samples to assess the presence of inhibitory substances. A standard curve was constructed by serial dilution of purified M. leprae DNA ranging from 10 fg-1 µg. Purified M. leprae DNA was also used as a positive control for the amplifications. g p Brazil is a high leprosy-burdened country, with 34,894 new cases detected in 2010 (WHO 2011). More than half (53.5%) of the cases are concentrated in areas where only 17.5% of the population live, which reflects that leprosy is a rural disease (MS/SVS 2008). A study conducted with cases reported to the Brazilian Ministry of Health between 1990-2007 shows that leprosy is geographically concentrated, and a spatial analysis shows 29 clusters of higher prevalence. In these clusters, the mean rate of de tection was more than the double the rate of the rest of the country (56.2 vs. 20.6 cases per 100,000 inhabitants). CE is one of the poorest states in Brazil and more than half of the municipalities in CE (46 out of 84) reported local transmission of new cases. CE was also included in one of the 29 Brazilian leprosy clusters (Penna et al. 2009). However, in Brazil, epidemiological studies are still con troversial in terms of whether armadillo meat intake and direct animal contact are associated with leprosy infec tion (Kerr-Pontes et al. 2006, Deps et al. 2008, Schmitt et al. 2010). Aiming to investigate the question further, we used molecular diagnostic tools to search for M. leprae DNA in the two main species of wild armadillos found in CE (D. novemcinctus and E. sexcinctus). MATERIALS AND METHODS A reference gyrA sequence (GenBank acces sion NC002677) was used to align the sequences. for 45 s and 72ºC for 90 s, followed by 35 cycles of 94ºC for 45 s, 62ºC for 45 s and 72ºC for 90 s. The final extension was for 10 min at 72ºC. The 187-bp PCR products were purified using the Invitrogen ChargeSwitch PCR Clean- Up kit prior to sequencing in an Applied Biosystems DNA sequencer (Perkin-Elmer Applied Biosystems) using a BigDye Terminator Cycle Sequencing kit. The sequences were identified using SecScape software v2.7 (Applied Biosystems). A reference gyrA sequence (GenBank acces sion NC002677) was used to align the sequences. in the Americas and in the USA (Louisiana), Mexico, Colombia and Brazil had only reported M. leprae infec tion in armadillos of the species D. novemcinctus (Mey ers et al. 1977, Truman et al. 1990, Deps et al. 2002). p ) Clinically, most of the animals exhibited nodule- like lesions indicative of leprosy or other degenerative diseases, as demonstrated by a clinical exam conducted by an expert veterinarian. However, no histopathologi cal study was conducted, as the tissue samples were not formalin-fixed for anatomopathological examination. Additionally, no blood samples were taken from the animals. Therefore, we cannot conclude that the PCR- positive armadillos had clinical leprosy. False-positive amplifications were addressed by us ing individual sterile section-cutting blades and sterile glassware for each biopsy sample. The armadillo biop sies and extracted DNA samples were carefully identi fied and kept in separate boxes. A simple DNA extrac tion protocol was established to minimise the risk of contamination. The extraction method used one purifi cation step to reduce the accumulation of impurities that would inhibit the polymerase reaction. The genome sequencing of the Tamil Nadu M. lep rae strain and other strains was conducted with a typing system based on single-nucleotide polymorphism (SNP) differences allowing continent distribution of the leprosy bacilli, which were classified as SNP types 1 to 4 (Monot et al. 2005). In addition, a correlation was observed be tween a mutation in the gyrA gene and the SNP types, which were clustered in gyrAC and gyrAT populations; the T SNP represents SNP type 3, while the C nucleotide represents the other three types. MATERIALS AND METHODS The results from the se quencing data demonstrated that the analysed samples from the armadillos in CE belong to the gyrAT (SNP type 3) population, which was also identified in samples from wild armadillos in Louisiana (Monot et al. 2005) and in humans in Brazil (Fontes et al. 2009). Although this is not yet definitive evidence that armadillos act as a source of Ethical considerations - This project was submitted to and approved by the Ethical Committee of the Fed eral University of Ceará. TABLE I p p y p ( ) The PCR amplification of M. leprae with the primers RLEP2-3 and RLEP2-2 was species-specific, targeting the same region of the primers developed by Donoghue et al. (2001). It generated a single band of 238 bp, a larger am plicon compared to the Donoghue primers that amplified a 99-bp product. In addition, the sequencing of the gyrA re gion (coding sequence location from 7515 to 11451 kb) was confirmatory for the presence of M. leprae DNA in the ar madillos analysed. The detection limit of the nested PCR for M. leprae was 1 pg/µL (data not shown). All negative samples were amplified after adding 1 pg of purified M. leprae DNA. The exception was for one sample that was only amplified after a 1000-fold dilution, therefore con firming the presence of substances that were inhibitory to the PCR. The assay was repeated for the few isolates for which there were discrepancies between the researchers. Polymerase chain reaction (PCR) results for the 29 wild armadillos analyzed according to the armadillo species in the state of Ceará, Brazil MATERIALS AND METHODS Armadillos - A total of 29 wild armadillos from two species (27 D. novemcinctus and 2 E. sexcinctus), consist ing of males and females weighing from 2.6-3.8 kg, were captured between July-August 2007. The animals were captured by local hunters under the supervision of a vet erinarian in rural sites of 12 selected endemic municipali ties from CE (Figure). Biopsy samples from the ear, nose, liver and spleen of each of these animals were studied. The M. leprae gyrA region was amplified using prim ers gyrAF (5’-CCCGGACCGTAGCCACGCTAAGTC-3’) and gyrAR (5’-CATCGCTGCCGGTGGGTCATTA-3’). The thermal profile involved an initial denaturation at 94ºC for 5 min and six cycles of 94ºC for 45 s, 68 to 63ºC Euthanasia - Animal captures were authorised by the Brazilian Institute of Environmental and Renewable Nat ural Resources. Before euthanasia, armadillos were anes thetised with tiletamine and zolazepam (5.0 mg/kg/I.M) (Virbac, Brazil). Euthanasia was conducted in the place of capture and the animals were kept in ice until they arrived at the laboratory in Fortaleza, the capital city of CE. Distribution map of the 12 places where armadillos were collected in the state of Ceará, Brazil. Sample biopsies and DNA extraction - Stringent pre cautions were necessary to avoid cross-contamination. Clean protective clothing was worn and gloves were changed frequently. Before the tissue dissection, the ani mals were carefully rinsed with distilled water. For each desired tissue, a sterile blade was used to cut. Ear, nose, liver and spleen samples were kept frozen at -20ºC in separate sterile plastic storage bags until DNA extraction was performed. The sampling extraction was conducted batch-wise, four samples at a time. The frozen sections were incubated with 50 µL of 100 mM Tris-HCl, pH 8.5, containing Tween-20 and 60 µg of proteinase K per mL Distribution map of the 12 places where armadillos were collected in the state of Ceará, Brazil. Mem Inst Oswaldo Cruz, Rio de Janeiro, Vol. 107(Suppl. I), 2012 211 for 45 s and 72ºC for 90 s, followed by 35 cycles of 94ºC for 45 s, 62ºC for 45 s and 72ºC for 90 s. The final extension was for 10 min at 72ºC. The 187-bp PCR products were purified using the Invitrogen ChargeSwitch PCR Clean- Up kit prior to sequencing in an Applied Biosystems DNA sequencer (Perkin-Elmer Applied Biosystems) using a BigDye Terminator Cycle Sequencing kit. The sequences were identified using SecScape software v2.7 (Applied Biosystems). RESULTS A total of 116 liver, spleen, ear and nose tissue sam ples from 29 armadillos were tested using the nested RLEP PCR assay. M. leprae was detected in six (21%) of the animals; five were from the species D. novemcinctus (samples 8, 21, 22, 23 and 25) and one was from the spe cies E. sexcinctus (sample 9) (Table I). M. leprae DNA was amplified in the ear biopsy samples of all six ani mals, but in only five of the liver or nose biopsy samples and three of the spleen biopsy samples (Table II). TABLE I Polymerase chain reaction (PCR) results for the 29 wild armadillos analyzed according to the armadillo species in the state of Ceará, Brazil Armadillo species Total (n) RLEP PCR positive n (%) Dasypus novemcinctus 27 5 (19) Euphractus sexcintus 2 1 (50) Total 29 6 (21) RLEP: Mycobacterium leprae-specific repetitive element. TABLE I DISCUSSION Distribution of positive polymerase chain reaction (PCR) results using the Mycobacterium leprae-specific repetitive element repetitive sequence primer pairs in different biopsy tissues from wild armadillos in the state of Ceará, Brazil This is the first study to identify M. leprae in two species of wild armadillos (D. novemcinctus and E. sex cinctus) in Brazil. The tested armadillos came from an endemic leprosy area where the prevalence rate was re ported to be 2.99/100,000 inhabitants in 2008 and where there is continuous contact between humans and these animals (MS/SVS 2011). In this region, the hunting and eating of armadillos is a popular and frequent practice, a situation similar to that described in Colombia (Cardona- Castro et al. 2009). It is noteworthy that previous studies Armadillo species Biopsy source Liver Spleen Ear Nose Dasypus novemcinctus (n = 5) 4 2 5 4 Euphractus sexcinctus (n = 1) 1 1 1 1 212 M. leprae in wild armadillos • Cristiane Cunha Frota et al. Dalponte JC, Tavares-Filho JA 2004. Diet of the yellow armadillo, Eu phractus sexcinctus, in south-central Brazil. Edentata 6: 37-41. infection for humans, this contributes to the evidence sup porting the hypothesis of zoonotic transmission, as sug gested by other authors (Job et al. 1986, Walsh et al. 1988, Cardona-Castro et al. 2009, Truman & Fine 2010). de Wit MY, Faber WR, Krieg SR, Douglas JT, Lucas SB, Montreewasu wat N, Pattyn SR, Hussain R, Ponnighaus JM, Hartskeerl RA 1991. Application of a polymerase chain reaction for the detection of My cobacterium leprae in skin tissues. J Clin Microbiol 29: 906-910. , ) Because M. leprae cannot be cultivated in vitro, its detection is based on the histopathological demonstra tion of the bacilli in dermal nerves, mouse footpad cul tivation and PCR assays of the selective amplification of M. leprae DNA (Truman & Fine 2010). This study used the RLEP repetitive element sequence of M. leprae, which is reported to be specific for M. leprae and is not present in other mycobacterial or bacterial species. In addition, the use of the repetitive sequence as the PCR target DNA provides the advantage of higher sensitivity over other targets because it is present at multiple sites in the genomic DNA (Donoghue et al. 2001, Truman et al. 2008). DISCUSSION It has been suggested that many homologous sequences of the RLEP may be present in other environ mental Mycobacterium species that have not been thor oughly investigated, which might generate false-positive results. Despite this fact, Martinez et al. (2011) found that the RLEP PCR assay can be used as a more specific and sensitive diagnostic test to detect M. leprae infec tion compared to the ones based on gene targets Ag 85B, sodA and 16S rRNA. Because we used several strategies to minimise false-positive amplifications and contami nation, we are confident that these are real infections. Deps PD, Alves BL, Gripp CG, Aragao RL, Guedes B, Filho JB, An dreatta MK, Marcari RS, Prates I, Rodrigues LC 2008. Contact with armadillos increases the risk of leprosy in Brazil: a case con trol study. Indian J Dermatol Venereol Leprol 74: 338-342. Deps PD, Antunes JM, Tomimori-Yamashita J 2007. Detection of Mycobacterium leprae infection in wild nine-banded armadillos (Dasypus novemcinctus) using the rapid ML Flow test. Rev Soc Bras Med Trop 40: 86-87. Deps PD, Santos AR, Yamashita-Tomimori J 2002. Detection of My cobacterium leprae DNA by PCR in blood sample from nine- banded armadillo: preliminary results. Int J Lepr Other Myco bact Dis 70: 34-35. Donoghue HD, Holton J, Spigelman M 2001. PCR primers that can detect low levels of Mycobacterium leprae DNA. J Med Micro biol 50: 177-182. Fontes AN, Sakamuri RM, Baptista IM, Ura S, Moraes MO, Martinez AN, Sarno EN, Brennan PJ, Vissa VD, Suffys PN 2009. Genetic diversity of Mycobacterium leprae isolates from Brazilian lep rosy patients. Lepr Rev 80: 302-315. Job CK, Drain V, Truman R, Deming AT, Sanchez RM, Hastings RC 1992. The pathogenesis of leprosy in the nine-banded armadillo and the significance of IgM antibodies to PGL-1. Indian J Lepr 64: 137-151. In conclusion, the presence of M. leprae DNA in wild armadillos (D. novemcinctus and E. sexcinctus) in a leprosy transmission area in Brazil provides additional evidence supporting the hypothesis that armadillos can play a role as an environmental reservoir for the bacillus in this area. Moreover, the finding supports the idea that intensive contact with these animals may increase the risk of infection in CE. The current global control strat egy depends on treating all human cases, but a definitive identification of an animal reservoir, as suggested by Tru man et al. DISCUSSION (2011), could challenge this strategy and may partially explain the continuing and growing presence of leprosy in the studied area. More detailed molecular studies will be useful in monitoring and confirming the transmission of M. leprae between wild armadillos and humans and in guiding new strategies for prevention. Job CK, Harris EB, Allen JL, Hastings RC 1986. A random survey of leprosy in wild nine-banded armadillos in Louisiana. Int J Lepr Other Mycobact Dis 54: 453-457. Kerr-Pontes LR, Barreto ML, Evangelista CM, Rodrigues LC, Heu kelbach J, Feldmeier H 2006. Socioeconomic, environmental and behavioural risk factors for leprosy in Northeast Brazil: results of a case-control study. Int J Epidemiol 35: 994-1000. Kirchheimer WF, Sanchez RM 1977. Quantitative aspects of leprosy in armadillos. Lepr India 49: 48-53. Kirchheimer WF, Storrs EE, Binford CH 1972. Attempts to establish the Armadillo (Dasypus novemcinctus linn.) as a model for the study of leprosy. II. Histopathologic and bacteriologic post-mor tem findings in lepromatoid leprosy in the Armadillo. Int J Lepr Other Mycobact Dis 40: 229-242. REFERENCES Martinez AN, Ribeiro-Alves M, Sarno EN, Moraes MO 2011. Evalua tion of qPCR-based assays for leprosy diagnosis directly in clini cal specimens. PLoS Negl Trop Dis 5: e1354. Amezcua ME, Escobar-Gutierrez A, Storrs EE, Dhople AM, Burch field HP 1984. Wild Mexican armadillo with leprosy-like infec tion. Int J Lepr Other Mycobact Dis 52: 254-255. Meyers WM, Walsh GP, Brown HL, Rees RJ, Convit J 1977. Natu rally acquired leprosy-like disease in the nine-banded armadillo (Dasypus novemcinctus): reactions in leprosy patients to lep romins prepared from naturally infected armadillos. J Reticu loendothel Soc 22: 369-375. Armadillo Online! 2012. [homepage on the Internet]. Minneapolis. Armadillos: Biology, Ecology and Images. Michigan State Uni versity Department of Zoology. [cited 2012 March 13]. Available from: msu.edu/~nixonjos/armadillo/species.html. Bruce S, Schroeder TL, Ellner K, Rubin H, Williams T, Wolf Jr JE 2000. Armadillo exposure and Hansen’s disease: an epidemiolog ic survey in southern Texas. J Am Acad Dermatol 43: 223-228. Monot M, Honore N, Garnier T, Araoz R, Coppee JY, Lacroix C, Sow S, Spencer JS, Truman RW, Williams DL, Gelber R, Virmond M, Flageul B, Cho SN, Ji B, Paniz-Mondolfi A, Convit J, Young S, Fine PE, Rasolofo V, Brennan PJ, Cole ST 2005. On the origin of leprosy. Science 308: 1040-1042. Cardona-Castro N, Beltran-Alzate JC, Romero-Montoya IM, Me lendez E, Torres F, Sakamuri RM, Li W, Vissa V 2009. Identifica tion and comparison of Mycobacterium leprae genotypes in two geographical regions of Colombia. Lepr Rev 80: 316-321. MS/SVC - Ministério da Saúde/Secretaria de Vigilância em Saúde 2008. Situação epidemiológica da hanseníase no Brasil. Pro grama Nacional de Controle de Hanseníase. [cited 2012 March 13]. Available from: portal.saude.gov.br/portal/arquivos/pdf/bo letim_novembro.pdf. Clark BM, Murray CK, Horvath LL, Deye GA, Rasnake MS, Long field RN 2008. Case-control study of armadillo contact and Han sen’s disease. Am J Trop Med Hyg 78: 962-967. Mem Inst Oswaldo Cruz, Rio de Janeiro, Vol. 107(Suppl. I), 2012 213 MS/SVC - Ministério da Saúde/Secretaria de Vigilância em Saúde 2011. Taxa de prevalencia de hanseníase 2011. [updated 2011 June; cited 2012 March 13]. Available from: tabnet.datasus.gov. br/cgi/idb2010/matriz.htm#morb. Truman RW, Fine PE 2010. “Environmental” sources of Mycobacte rium leprae: issues and evidence. Lepr Rev 81: 89-95. Truman RW, Job CK, Hastings RC 1990. Antibodies to the phenolic glycolipid-1 antigen for epidemiologic investigations of enzootic leprosy in armadillos (Dasypus novemcinctus). Lepr Rev 61: 19-24. Penna ML, de Oliveira ML, Penna GO 2009. REFERENCES The epidemiological behaviour of leprosy in Brazil. Lepr Rev 80: 332-344. Truman RW, Kumaresan JA, McDonough CM, Job CK, Hastings RC 1991. Seasonal and spatial trends in the detectability of leprosy in wild armadillos. Epidemiol Infect 106: 549-560. Schmitt JV, Dechandt IT, Dopke G, Ribas ML, Cerci FB, Viesi JMZ, Marchioro HZ, Zunino MMB, Miot HA 2010. pArmadillo meat intake was not associated with leprosy in a case control study, Curitiba (Brazil). Mem Inst Oswaldo Cruz 105: 857-862. Truman RW, Singh P, Sharma R, Busso P, Rougemont J, Paniz-Mon dolfi A, Kapopoulou A, Brisse S, Scollard DM, Gillis TP, Cole ST 2011. Probable zoonotic leprosy in the southern United States. N Engl J Med 364: 1626-1633. Sheppard CC 1960. The experimental disease that follows the injec tion of human leprosy bacilli into footpads of mice. J Expect Med 112: 445. Walsh GP, Meyers W, Binford CH 1977. Naturally acquired leprosy- like disease in the nine-banded armadillo (Dasypus novemcinc tus): recent epizootiologic findings. J Reticuloendothel Soc 22: 363-367. Sheppard CC 1985. Experimental leprosy. In RC Hastings (ed.), Lep rosy, Longman Group Limited, New York, p. 269-286. Smith JH, Long EG, Crouse DT, Christie JD, Folse DS, Imaeda T, Barksdale L 1983. Leprosy in wild armadillos (Dasypus novem cinctus) of the Texas Gulf Coast. Acta Leprol 2: 311-318. Walsh GP, Meyers WM, Binford CH, Gormus BJ, Baskin GB, Wolf RH, Gerone PJ 1988. Leprosy as a zoonosis: an update. Acta Lep rol 6: 51-60. Stallknecht DE, Truman RW, Hugh-Jones ME, Job CK 1987. Surveil lance for naturally acquired leprosy in a nine-banded armadillo population. J Wildl Dis 23: 308-310. WHO - World Health Organization 2011. Weekly epidemiological re cord. Available from: who.int/wer/2011/wer8636.pdf. Zumarraga MJ, Resoagli EH, Cicuta ME, Martinez AR, Oritiz de Rott MI, de Millan SG, Caimi K, Gioffre A, Alito A, Bigi F, Cataldi AA, Romano MI 2001. PCR-restriction fragment length poly morphism analysis (PRA) of Mycobacterium leprae from human lepromas and from a natural case of an armadillo of Corrientes, Argentina. Int J Lepr Other Mycobact Dis 69: 21-25. Truman R 2008. Armadillos as a source of infection for leprosy. South Med J 101: 581-582. Truman RW, Andrews PK, Robbins NY, Adams LB, Krahenbuhl JL, Gillis TP 2008. Enumeration of Mycobacterium leprae using real-time PCR. PLoS Negl Trop Dis 2: e328.
https://openalex.org/W2051177377	https://zenodo.org/records/2186258/files/article.pdf	German	null	Nierenkolik, Nierenblutung und Nephritis<sup>2</sup>)	Deutsche medizinische Wochenschrift/Deutsche Medizinische Wochenschrift	1,902	public-domain	4,392	') A. Kussmaul, Ein Dreigestirn grosser Naturforscher an der Heidelberger Universität im 19. Jahrhundert. Deutsche Revue, Januar und Februar 1902. l h l i i f i di i 2) l,Tortmg, gehalten am 13. Januar im Verein für innere Medizin. - Die ausführliche Entgegnung von J. Israel wird wegen Raum- mangels erst in der nächsten Nummer erscheinen. DEUTSCHE 1vfEfl1C1NISC1iI WOO NSO1fR1tP. Beide Erscheinungen sind Folgezustände der Nierencon- gestion". g ,,Eine grosse Zahl der bisher als Nephralgie, Nephralgie hematurique, angioneurotische Nierenblutung bezeichneten Krank- heitsbilder sind auf nephritische Prozesse zu beziehen". g g Es handelt sich, wie Sie sehen, um eine Frage, welche die innere Medizin sehr angeht, und es Ist wohl hier der Ort, diese Frage zur Besprechung zu bringen. p ,,Die Incision der Niere beeinflusst in vielen Fällen den nephritischen Prozess und seine Symptome gilnstig". p ,,Die Incision der Niere beeinflusst in vielen Fällen den nephritischen Prozess und seine Symptome gilnstig". D i k i f d K k d S h li h Damit kommen wir auf den Kernpunkt der Sache, nämlich, dass es sich in allen solchen Fällen um congestiv entziindliche Schwellungen der Niere mit Spannung der Kapsel handle, weiche durch die Spaltung gehoben werde. Ich werde mich dabei vorwiegend an die Mittheilunge von Israel halten, und zwar aus verschiedenen Grilnden. Erstens hat er die reichste Erfahrung auf dem betreffenden Gebiete, er hat die grundlegenden Fälle mitgetheilt, die mit der von ihm be- kannten Genauigkeit und Schärfe beobachtet und mit ausser- ordentlicher Treue und Sorgfalt beschrieben sind, sodass jeder selbst sich ein Urtheil bilden kann, was bei den Mittheilungen vieler seiner Nachfolger leider nicht möglich ist. Oft fehlt bei diesen der Urinbefund oder es ist nicht genau angegeben, was bei der operativen Autopsie oder bei der Sektion nach dem Tode gefunden wurde. p g g M. H.! Wir in der inneren Medizin haben doch recht häufig Gelegenheit, congestive Schwellungen mit Spannung der Kapsel zu beobachten. Ich erinnere Sie an die Schwellungen bei heftiger akuter Nephritis oder bei akuten Exazerbationeri einer chronischen Nephritis, an die Schwellung in Folge von venöser Stauung bei Compensationsstörungen u. s. w. Die Fälle verlaufen ja häufig tödtlich, und man hat deshalb Gelegenheit genug, auf dem Sektions- tische zu sehen, wie gewaltig die Nieren geschwollen und blau- roth sind, wie die Kapsel auf das äusserste gespannt und ver- dännt ist. Aber Nierenkoliken kommen dabei nicht vor, oder jedenfalls gehört das zu den unerhört seltensten Ausnahmen. Hier besteht also ein schreiender Widerspruch zwischen den Erfah- rungen der inneren Medizin und dem, was Israel und seine Nachfolger behaupten, von denen übrigens Viele weit über das hinausgegangen sind, was Israel selbst ausgesprochen oder ge- meint hat. Nun, dieser schreiende Widerspruch löst sich leicht an der Hand der Beobachtungen, namentlich der genauen von Israel mitgetheilten. DEUTSCHE 1vfEfl1C1NISC1iI WOO NSO1fR1tP. DEUTSCHE 1vfEfl1C1NISC1iI WOO NSO1fR1tP. 127 20. Februar. I. Nierenkolik, Nierenbiutung und Nephritis.2) Von Prof. Dr. III. Senator in Berlin. M. H.! Ich beabsichtige im Folgenden nicht, Ihnen neue Ent- deckungen oder merkwürdige Beobachtungen mitzutheilen, sondern nur Betrachtungen anzustellen über Nierenkolik, Nierenblutung und Nephritis und kritische Bemerkungen an die Beurtheilung zu knüpfen, die diesen Zuständen in neuerer Zeit zu Theil geworden ist, sowie über die zugehörigen therapeutischen Bestrebungen. Diese Bestrebungen sind ursprünglich ausgegangen von opera- tiven Eingriffen zu rein diagnostischen Zwecken, sind aber verhältnissmässig schnell schon angelangt bei der operativen Be- handlung des Morbus Brightii. g g Nierenkolik und Nierenbiutung, oder wenn sie zusammen auf- treten, Nephralgia haematurica, Symptome, die seit uralter Zeit den Aerzten bekannt sind, haben im Laufe der Zeit verschiedene Deutungen erfahren, Deutungen, die dem Entwickelungsgange der Medizin entsprachen. In alter Zeit, als Sektionen noch selten ge- macht wurden und das Mikroskop noch nicht eingeführt war, etwa bis zum ersten Drittel des vorigen Jahrhunderts, fand man sich mit solchen Fällen, wenn eine grob sichtbare Ursache nicht zu entdecken war, leicht ab. Man nannte sie ,,essentielle", und so wie essentielle Fieber und essentielle Wassersucht etc. spielen auch ,,essentielle" Nierenkolik und Nierenblutung in der alten Pathologie eine grosse Rolle. Als dann später die patho- logische Anatomie ihre grossen Fortschritte machte und ebenso die Diagnostik und man immer mehr dazu kam, in Fällen, die man vorher als essentielle bezeichnet hatte, greifbare Verände- rungen nachzuweisen, wurde man gegen alle essentiellen Krank- heitszustände und auch gegen die essentiellen Nierenkoliken und -Blutungen immer misstrauischer, und etwa um die Mitte oder im vorletzten Drittel des vergangenen Jahrhunderts sind diese essentiellen Krankheitszustände schon ganz aus der Pathologie verschwunden. DWTSOH M1mrcm1ScniE WOOHINSC}IR1FT. 128 Ño. Ño. In n&ierer Zeit aber, wo man wieder anfing, den funktionellen und vom Nervensystem abMngigen Störungen mehr Aufmerksam- keit zu schenken, gewann die Anschauung Boden, dass manche der früher sogenannten essentiellen Blutungen und Nephralgieen he- clingt seien durch rein nervöse Vorgänge, dass es wirklich neural- gische Nierenschmerzen gebe und ebenso ierenblutungen aus Nieren ohne anatomische Veränderung lediglich unter dem Ein- fluss von nervösen Störungen. Diese Auffassung ist vor langer Zeit von Lancereaux vertheidigt, aber, wie es scheint, in Ver- gessenheit gerathen, und erst in neuerer Zeit hat G. Kiemperer sie wieder aufgenommen. Aber schon beginnt auch gegen diese Auffassung eine Reaktion, und zwar knüpft sich diese Reaktion an einen Fortschritt, vielleicht einen der grössten in der Diagnostik der Nierenkrankheiten, den wir J. DEUTSCHE 1vfEfl1C1NISC1iI WOO NSO1fR1tP. Israel verdanken. Er hat be- kanntlich zu der schon früher, wenn auch seltener geübten Bloss- legung der Niere, mit der man auch wohl noch die Acupunktur des Organs verband, die Spaltung der Niere durch den Sektions- schnitt hinzugefügt und dadurch natürlich die Möglichkeit ge- geben, sich in viel ausgiebigerer Weise fiber den Zustand der Niere zu unterrichten. Bei diesem ursprünglich zu rein dia- gnostischen Zwecken geübten Verfahren kam Israel zu ganz unerwarteten Befunden und meistens auch zu ganz überraschen- den, wenn auch in der Mehrzahl nur vorübergehenden Erfolgen. Diese Befunde und Erfolge brachten Israel') dazu, eine von den bisherigen Ansichten ganz abweichende Anschauung in betreff der Ursachen und der Behandlung der Nierenkolik und Nieren- blutung auszusprechen. Er meint nämlich, dass es sich in solchen Fällen, natürlich wenn keine andere Ursache aufzufinden ist, um akut congestive Schwellung und dadurch bedingte Spannung der Nierenkapsel handle, welche eben durch die Spaltungbeseitigtwürden. Israel hat hierin lebhafte Zustimmung gefunden von Chirurgen, namentlich Spezialchirurgen der Harnorgane. Innere Kliniker und Aerzte haben sich bis jetzt nur ganz vereinzelt hören lassen. Zu- erst Naunyn,) welcher drei Fälle von Massenblutung bei chro- nischer interstitieller Nephritis mittheilt, die übrigens ohne opera- tiven Eingriff aufhörten. Dass es Blutungen aus gesunden Nieren giebt, lässt er gelten, bestätigt auch, dass es sich in vielen Fällen von Blutungen aus angeblich normalen Nieren eben um chro- nische Nephritis gehandelt habe, und meint, dass die Nephrotomie nach Israel noch zu einer weitergehenden Rolle in der Behand- lung des Morbus Brightii berufen sei. In ganz entgegengesetztem Sinne äusserte sich P. K. Pci3) in einem in der Versammlung niederländischer Aerzte und Naturforscher in Rotterdam gehal- tenen Vortrage, und i der Diskussion hat Rosenstejn sich ihm vollständig angeschlossen. anderen Nierensehriftstellern wiederholt bestätigt worden. Be- kannt ist die Beobachtung eines Falles von Schruinpfniere durch Bartels, wo wochenlang bis zum Tode kein Eiweiss im Ham gefunden wurde. Sogar bei ganz akuter Nephritis kann Albumin- uric fehlen, und erst im vorigen Jahre hat uns ja Herr Cassel liber sehr beweisende Fälle dieser Art berichtet.1) Auch der Satz 5: ,,Trotz grossen Reichthums an hya- linen, gekömnten und epithelialen Cylindemn kann der Urin eiweissfrei sein", ist nicht zu bestreiten, aber auch be- kannt. Wir nennen solche Zustände, wo nur Cylinder vorkommen. Cylindrurie. Vor vielen Jahren schon haben Rosenstein, Axel Key, Nothnagel u. a. solche Fälle veröffentlicht. DEUTSCHE 1vfEfl1C1NISC1iI WOO NSO1fR1tP. i b h i i f ll i y g Endlich Satz 6: ,,Es giebt Nephriten mit anfallsweise auftretenden profusen Blutungen", ist gleichfalls richtig, aber gleichfalls bekannt. Man spricht in solchen Fällen von hä- morrhagischer Nephritis, welche akut, subakut oder chronisch auftreten kann. Am seltensten sind solche Blutungen bei der sogenannten Schrumpfniere, aber doch auch schon beobachtet, nicht bloss bei akuten Exazerbationen oder Entzilndung, sondern auch ohne solche. Beiläufig will ich noch erwähnen, dass bei doppelseitiger Nephritis die Blutung garnicht selten nur aus einem Niere stamiut, was weiter nicht wunderbar ist, so wenig wie die Hämoptoö aus einer Lunge bei beiderseitiger Tuberkulose. E dli h k h S 1 E i h i i i Endlich kann man auch Satz 1: ,,Es gicht einseitige Nephriten" in dieser allgemein gehaltenen Fassung zustimmen. Denn unzweifelhaft kann eine Niere allein bei einem einseitigen Reiz erkranken. Wenn aber damit gemeint sein soll, dass es eine einseitige Nephritis gicht, die nicht auf örtlich begrenzte, nur einseitig wirkende Ursachen zurückzuführen ist, also Formen, wie man sie wohl als Morbus Brightii zusammenfasst, so muss ich dagegen entschiedenen Widerspruch erheben. Einen einsei- tigen Ivlorbus Brightii giebt es nicht, es müsste denn nur eine Niere vorhanden sein. Es bleiben nun noch die folgenden fünf Thesen übrig: Israel hat hierin lebhafte Zustimmung gefunden von Chirurgen, namentlich Spezialchirurgen der Harnorgane. Innere Kliniker und Aerzte haben sich bis jetzt nur ganz vereinzelt hören lassen. Zu- erst Naunyn,) welcher drei Fälle von Massenblutung bei chro- nischer interstitieller Nephritis mittheilt, die übrigens ohne opera- tiven Eingriff aufhörten. Dass es Blutungen aus gesunden Nieren giebt, lässt er gelten, bestätigt auch, dass es sich in vielen Fällen von Blutungen aus angeblich normalen Nieren eben um chro- nische Nephritis gehandelt habe, und meint, dass die Nephrotomie nach Israel noch zu einer weitergehenden Rolle in der Behand- lung des Morbus Brightii berufen sei. In ganz entgegengesetztem Sinne äusserte sich P. K. Pci3) in einem in der Versammlung niederländischer Aerzte und Naturforscher in Rotterdam gehal- tenen Vortrage, und i der Diskussion hat Rosenstejn sich ihm vollständig angeschlossen. g ,,Es giebt durch Nephritis erzeugte Nierenkoliken, welche völlig Nierensteinkoliken gleichen". ,,Es giebt durch Nephritis erzeugte Nierenkoliken, welche völlig Nierensteinkoliken gleichen". g g ,,Es giebt doppelseitige Nephriten. welche nur einseitige Koliken erzeugen". ,,Es giebt doppelseitige Nephriten. welche nur einseitige Koliken erzeugen". ,,Nephritische Blutungen können mit oder ohne Koliken eintreten oder verlaufen. Die Blutung ist nicht die Ursache der Kolik. ') J. Israel, Mittheilungen aus den G-renzgebieten der Medizin und Chirurgie 1899, Bd. V, S. 471 und Chir. Klinik der Nierenkrank- liciten, Berlin 1901, S. 403. Naunyn, Mittheilungen aus den Greuzgebieten der Medizin und Chirurgie 1900, Bd. V. P. K. Pel, Ebenda. VIII, 1901. l) Cassel, Berliner klinische Wochenschrift 1900, No. 10. ') J. Israel, Mittheilungen aus den G-renzgebieten der Medizin und Chirurgie 1899, Bd. V, S. 471 und Chir. Klinik der Nierenkrank- liciten, Berlin 1901, S. 403. Naunyn, Mittheilungen aus den Greuzgebieten der Medizin und Chirurgie 1900, Bd. V. P. K. Pel, Ebenda. VIII, 1901. DEUTSCHE 1vfEfl1C1NISC1iI WOO NSO1fR1tP. g Seine Ansichten hat Israel in 11 Sätzen niedergelegt, von denen die beiden letzten ausser Betracht bleiben können. Der eine (11) betrifft die Behandlung der Nierenwunde, der andere (10), welcher sagt, dass Anurie bei akuter aufsteigender Nephritis durch Nierenspaltung geheilt werden kann, wird wohl kaum be- stritten werden und liegt übrigens schon mehr auf chirurgischem Gebiete. Auch von den übrigen neun Sätzen kann man dreien vorweg unbedingt zustimmen. Nämlich: Satz 4: ,,Es giebt schwere Nephriten mit eiweissfreiem Urin und Ab- wesenheit von Cylindern." Das ist garnicht zu bestreiten, aber es ist nicht neu, es ist schon von Bright angegeben und von g Es sind 14 Fälle, und da heisst es bei Fall 12: die blossgelegte Niere auffallend klein und weich, Fall 11: die Niere auf- fallend schlaff und weich, Fall 10: die Niere von auffallend weicher Consistenz mit Ausnahme des unteren Endes, welches auffallend resistenter ist, Fall 8: Niere auffallend weiss und DEUTSCHE MEDICINTSCHE WOCHENSCI[R[FT. 20. Februar. 20. Februar. .129 ,,aüffallenderweise gar keine Bintung aus dem Parenehym", Fall 5: Niere nicht vergrössert, nicht empfindlich, Fall 4: an der Niere keinerlei Anomalie, Fall : Niere nicht ver- grössert, ohne besondere Härte. Fall 14: Niere zeigt keine Abweichung in Bezug auf Grösse nnd Consistenz. In drei Fällen (I, 6, 7) ist über Consistenz und Schwellung nichts angegeben, woraus wohl zu schliessen, dass sie in dieser Be- ziehung nichts Auffallendes darboten. Einen anderen Fall berichtet Albarran.') Hier fand sich nach Lösung der Niere aus einigen Verwachsungen ein kleiner grauer Kern an der Basis einer Pyramide, so klein, dass, wie Albarran sagt, er ihm entgangen wäre, wenn er den Schnitt ein paar Millimeter seitwärts gelegt hätte. Nun, wer da weiss, wie häufig solche Heerdchen alter, ab- gelaufener kleiner Entziindungsprozesse in den Leichen von Men- sehen gefunden werden, die kein Zeichen von Nephritis dargeboten hatten, eben weil der unbedeutende Prozess schnell zum Stillstand gekommen ist, für den hat ein solcher gelegentlich bei einer Ope- ration gemachter Befund nichts Ueberraschendes. Ueberraschend ist nur, dass man daraus ohne weiteres auf eine entzündliche Congestion der ganzen Niere schliesst. In 11 von 14 Föllen war also eine Spannung und congestive Schwellung nicht vorhanden, in mehreren wird sogar das Gegen- theil als vorhanden angegeben. In den übrigen drei Fällen (3, D und 13) ist es der Beschreibung nach mit der congestiven Schwellung auch nicht weit her, z. B. schwoll einmal die Niere nach der Dislokation etwas venös an. DEUTSCHE 1vfEfl1C1NISC1iI WOO NSO1fR1tP. In Fall 1 bei Harnsäurekrystallen im Una war eine Papille weiss gefärbt, im Fall 3, wo Erschei- nungen von Cysti tis bestanden, fand sich eine Perinephritis mit blasigen Abhebungen der Kapsel, im Fall 7, wo der Urin Blut und Cylinder enthielt, lag eine bewegliche Niere vor. In Fall 8 mit Erscheinungen von Pyclitis bekam Pat, nach zwei Jahren Oedem und vermuthlich Nephritis, endlich in Fall 10 war am unteren Pol der Niere ein blasser gelber Keil, vielleicht syphiliti- sehen Ursprungs. Das sind die fünf Fälle von ,,Nephritis", die zwei Fälle von Morbus Brightii mit schnell tödtlichem Ausgang und sieben Fälle, in denen eine Entzündung überhaupt nicht nachgewiesen ist. Freilich sagt Israel, dass, wo nur ein ausgeschnittenes Stück- ehen untersucht ist, ein negativer Befund nichts beweist, da in dem übrigen Organ doch eine Entzündung bestehen kann. Das ist richtig, wenn es auch eine sonderbare entzündliche Congestion sein müsste, die makroskopisch gar nicht zu erkennen ist und nur an einer zufällig nicht ausgeschnittenen Stelle mikroskopisch sich finden soll. Immerhin lässt sich der Einwand erheben, ja ich gehe noch weiter als Israel und behaupte, dass, wenn auch makroskopisch und an einem mikroskopischen Schnitt nichts ge- funden wird, oder sogar, wenn ein entzündliches Heerdchen ge- funden wird, noch ganz andere Dinge in der Niere vorhanden sein können, welche nach aller Erfahrung ganz sicher und unge- zwungen die Hämaturie und Kolik erklären, seien es Steinchen, oder Tuberkulose u. s. w. Das ist keine ausgedachte Vermuthung, die ja übrigens ebenso berechtigt ist, wie Israel's Einwand, son- dern dafür sprechen Thatsachen. Sehr schlagend ist eine bezug- liche Beobachtung von Braatz2), welcher wegen Nephralgie eine Niere blosslegte, spaltete und, da an der vollständig aufgeklappten Niere nichts Abnormes zu finden war, sie vernähte und reponirte. Der Erfolg war glänzend, bis nach drei Jahren die Beschwerden wiederkamen und zu nochmaliger Blosslegung und Exstirpation der Niere zwangen. Jetzt fand sich am oberen Pol der geschrumpften Niere ein käsig tuberkulöser Abszess nebst frischen tuberkulösen Eruptionen und am unteren Pol ein alter ausgeheilter Heerd, der offenbar die früheren Beschwerden veranlasst hatte. Braatz meint übrigens, dass die Spaltung der Niere eine schwere Schä- digung derselben darstellt. Doch wäre es möglich, dass, wenn die von Zondek empfohlene Schnittlegung gewählt wird, der Eingriff nicht so schlimme Folgen hat. t) Beilthifig sei hier initgetheilt, dass dic Dame jetzt, nach der im April 1890, also vor bald 12 Jahren, ausgeführten Neplitektuinic sich körperlich ganz wohl befindet. Niemals ist Häinaturie wieder auf- getreten. Ihr psychisches Verhalten hat insofern gelitten, als sie seit DEUTSCHE 1vfEfl1C1NISC1iI WOO NSO1fR1tP. Aber ich zähle sie doch nicht mit. Es giebt übrigens noch mehr Fälle von Kolik und Blutung, in denen die Niere sich als gesund erwies. \Venn ich nur die- jenigen berücksichtige, in denen eine mikroskopische Untersuchung des ganzen Organs oder eines ausgeschnittenen Stückchens ge- macht wurde, so finde ich deren sieben, nämlich je einen von Schede (1889), Klemperer (1897), Harris (1898), zwei Fälle von Rovsing (1898, Fall 2 und 4) und die beiden genannten von Israel (12 und 13). Neben den nur malrroskopisch untersuchten von Broca, Debaisieux, Durham, Israel u. a. sollten diese sieben doch zur Vorsicht mahnen in der Annahme, dass, wo die bekannten Ursachen fehlen, eine Entzündung die Schuld trägt. Wenn nun in jenen 11 von 14 Fällen, in denen eine conges- tive Schwellung und Spannung sicher nicht vorhanden war, nach der Operation eine Besserung eintrat, so kann diese unmög- lich von einer Entspannung herriihren. Dies allein geniigt eigentlich schon, mn der ganzen Theorie von der congestiven Schwellung als Ursache dec Koliken den Boden zu entziehen. Aber es lohnt sieh, noch etwas weiter auf die Sache einzugehen und zu prüfen, wie weit die angenommene Entztindung in solchen Fällen wirklich nachgewiesen ist. h h l i h h i d di 14 ll l' Ich halte mich zunächst wieder an die 14 Fälle Israel's. In zwei Fällen bestand eine ausgesprochene chronische Nephritis, wie durch die Sektion festgestellt ist ; der eine Fall starb am ersten, der andere am zweiten Tage nach der Operation, davon ist der eine Fall einer, in welchem die Niere auffallend schlaff und weich war. Von einem Erfolg kann ja bei dem schnell ein- getretenen Tode hier keine Rede sein, wir scheiden sie also aus der Betrachtung aus. Es bleiben 12 Fälle, von welchen in sieben fiber den Zustand der Niere berichtet ist. Fünf sind nur makro- skopisch untersucht, nämlich Fall 2, 4, 5, 6, 14, in zwei Fällen (12 und 113) ist auch ein ausgeschnittenes Stückchen mikroskopisch untersucht worden. Bei den makroskopischen Untersuchungen hat kein Fall etwas auffallendes gezeigt, bei den mikroskopisch untersuchten hat der Schnitt nichts von Entzündung nachgewiesen. In einigen dieser Fälle war der Urin sogar ohne Spur von ne- ph ritischen Erscheinungen", in anderen Fällen wird aus dem Blut- gehalt oder aus Cylindern auf Nephritis geschlossen, in zwei über- haupt nur aus Analogie. Und wie verhielt es sich mit der Nephritis in dem Rest von fünf Fällen? etwa sieben Jahren in Folge von Todesfällen in der Familie an perio- discher Melancholie leidet auf religiösem Hintergrunde. In den freien Zeiten tritt öfter, abet' nicht regelmässig, etwas Eiweiss ohne Cylinder zugleich mit Kopfdruck ein. Wie diese Albuminurie zu erklären sei, lasse ich dahingestellt, vielleicht hängt sie mit der Psychose zusammen, wie solches vielfach beobachtet ist, vielleicht besteht ein jedenfalls sehr geringfügiger Prozess in der einzigen noch vorhandenen Niere, welche trotzdem den Verlust ihrer Genossin vollständig durch Mehr- leistung ersetzt. In den ersten Tagen nach der Operation hatte sie den Verlust sogar tibercompensirt, indem mehr Wasser und Stickstoff im Ham ausgeschieden wurden, als der Nahrung entsprach und als später bei vollstäiidigem Wohlbefinden ausgeschieden wurde. g g ') Albarrau, Association française d'urologie, IV. session. Paris 1899. 8. 105. h di i i h li hi if 2) Braatz, Deutsche medizinische Woelienschi'ift 1900, No. 10. DEUTSCHE 1vfEfl1C1NISC1iI WOO NSO1fR1tP. In den meisten Fällen werden sie andere Ursachen haben, Steinchen oder Sand, die auf dem Schnitt nicht gefunden werden, Tuberkeln und andere Geschwulstbildungen, erkrankte aneurysmatische oder variköse Gefässchen, hämorrhagische Di- athese u. a. m. Aber es bleiben allerdings Fälle übrig, in denen sich keine Läsion nachweisen lässt, die in Heilung übergehen, auch unter Umständen, welche den Gedanken an eine Erkrankung der Niere so gut wie ausschliessen. Ich erinnere an Fälle, wo nach dem Blasenschnitt die Nierenbiutung aufhörte (Pas set) oder wo nach der Operation die Blutung öfters wiederkehrte, um schliesslich dauernd wegzubleiben (Andersen), und namentlich an die Fälle von Klemperer. Hier darf man wohl von Blutungen aus gesun- den Nieren oder von neuropathischen, durch vasomotorische Ein- flüsse bedingten Blutungen sprechen. g g Aber selbst davon abgesehen, wird doch durch die Spaltung und Vernähung des Schnittes im günstigsten Fall, wenn keine Eite- rung eintritt, zu der schon bestehenden Entzündung noch eine adhäsive Entztindung mit narbiger Verwachsung der Kapsel hinzugefügt. Und wenn nun in Folge der alten und neuen Entzilndung wirklich eine congestive Schwellung hinzuträte, müssten ja die wtithendsten Koliken und Blutungen die Folge sein, da die verwachsene Kapsel sich noch weniger als früher ausdehnen kann. Wenn die Theorie von der akut- entzündlichen Schwellung als lJrsache der Nephralgia haematurica richtig wäre, würde ja die Spaltung der Niere mit der nach- folgenden Entzilndung die Bedingungen für das Leiden geradezu schaffen! g g p M. H.! Ich fasse meine Ausführungen dahin zusammen, dass Entzündung der Niere als Ursache von Nephralgia haematurica nicht bewiesen ist, dass die Spaltung der Niere kein Mittel da- gegen ist und dass, wo die Blosslegung mit oder ohne Spaltung geholfen hat, der Erfolg auf andere Umstände (Lösung von Ver- wachsungen, Anheftung der beweglichen Niere u. s. w.) zurück- zuführen Ist. Die diagnostische Bedeutung der Nierenspaltung wird dadurch nicht beeinträchtigt. Die Spaltung der Niere soll eine Entspannung herbeiführen. Nun, wenn wirklich eine solche nöthig wäre (wir haben gesehen, dass in vielen Fällen die Niere gar nicht geschwollen war), so ist ja, wie Pel mit Recht bemerkt, die Blutung das beste Mittel zur Entspannung. g I. H.! Ich muss mit einer persönlichen Bemerkung schliessen. Meine Ausführungen waren grossen Theils kritischer Natur. DEUTSCHE 1vfEfl1C1NISC1iI WOO NSO1fR1tP. Kritik ist aber immer ein undankbares Geschäft, und ich habe mich, ob- wohl ich vor hhiigerer Zeit schon aufgefordert war, mich über die operative Behandlung des Morbus Brightii zu äussern, nur schwer dazu entschlossen, zumal meine Kritik sich so viel mit J. Israel beschäftigen musste, der von mir hochgeschätzt wird und mir persönlich befreundet ist. Aber ich habe mich endlich dazu ent- schlossen, weil ich mir sagte, dass Israel's Verdienste auf den verschiedenen Gebieten der Chirurgie so gross sind, dass meine Kritik ihnen keine Einbusse thun kann und wird, und ich halte mich vor Ihnen und vor mir selbst entschuldigt nach Art des alten Kritikers Aristoteles: Amicils Plato, amicus Israel - mugis arnica ventas. Ich kann nur einen Zustand von Spannung zugeben, welcher wirklich so hochgradig werden kann, dass Schmerzen bis zu wirk- licher Kolik dabei auftreten, das ist die durch Anurie hervor- ge ruf e n e Spannung. Durch Behinderung des Harnflusses kommt, wie Erfahrung und Experiment (C. Ludwig) seit lange gelehrt haben, der höchste Grad von Spannung zu Stande, und hier ist die Nephrotomie in der That wirksani und oft lebensrettend, gleichviel, wodurch die Anurie bedingt sein mag. Natürlich wird man sich nicht gleich bei stockender Harnabsonderung zur Operation entschliessen, sondern erst, wenn Gefahr drohende Symptome auftreten. Nicht Blut und Entzün- dungsprodukte zu beseitigen, handelt es sich hier, sondern dem Ham Abfluss zu verschaffen. M. H.! Ich habe mich bis jetzt mehr negativ verhalten, in- dem ich zeigte, dass von Spannung und Entzilndung als Ursache von Nierenkolik und -Blutung keine Rede sein kann; ich bin nun in der angenehmen Lage, doch auch eine positive und, wie ich glaube, besser begründete Erklärung zu geben, zunächst für die Kolikschmerzen in einem grossen, vielleicht grössten Theil der Fälle. Das sind die Verwachsungen der Niere. Daftirliefert uns wieder eine Durchsicht der Fälle von Israel eine gute Un- terlage. Dieses Dokument wurde zum persönlic g In seinen 14 Fällen war die Niere nicht weniger als neun Mal verwachsen, einige Male ausserordentlich stark. Und in den übrigen fünf Fällen braucht man um eine Ursache für die Schmerzen auch nicht verlegen zu sein. Denn da ist von Pyelitis (Fall 4), von Sand und Gries (9, 12), von Gonorrhoe und Cystitis (6) und von beweglicher Niere (7 und 12) die Rede. Aber auch andere Autoren fanden Verwachsungen, als sie wegen Nephralgie die Operation unternahmen. DEUTSCHE 1vfEfl1C1NISC1iI WOO NSO1fR1tP. g g Nach alle dem glaube ich mich wohl berechtigt zu sagen, dass, wenn selbst in der Niere ein Entzündungheerd be- steht, daraus noch nicht auf Entzündung als Ursache von Kolik und Blutung geschlossen werden darf, da da- neben andere Ursachen vorhanden sein können, die jene Symptome plausibler erklären. g Israel hat selbst bei einigen Fällen Bedenken über das Vor- handensein von ,,Nephritis" und bemüht sich, sie in seinem Sinne zu zerstreuen. Seine Nachfolger aber machen sich zumeist gar keine Skrupel mehr in dieser Beziehung, ja, es ist erstaunlich, wie Fälle mit winzigen alten, längst abgelaufenen Heerdchen in der Niere ohne Bedenken herangezogen werden als Beweis dafür, dass Kolik oder Blutungen oder beide zusammen durch eine entzünd- liche Congestion verursacht sind. So wird immer wieder der be- kannte Fall von Sabatier genannt, aus dem Jahre 1888, wo sich in der exstirpirten Niere ein kleiner sklerotischer Heerd fand, so geringfügig, dass der Anatom nicht anstand, die Integrität des Organs anzuerkennen. Ebenso wird der von mir (1891) berichtete Fall von Nierenexstirpation wegen lebensgefährlicher Blotting bei einer jungen Dame aus hämophiler Familie herangezogen. Ilier ergab die von Prof. O. Israel ausgeführte Untersuchung ver- einzelte sternförmig Narben, aber keine räumlich irgendwie aus- gedehnte Nephritis.1) p Uebrigens, wie soll man sich den günstigen Einfluss der Nierenspaltung und gar einen dauernden Erfolg bei akut entzünd- DEUTSCHE MEDIOINJSCHE WOCRENSCHEIFT. 130 No. S No. S licher Congestion in Folge eines alten Entzündungsheerdes denken? licher Congestion in Folge eines alten Entzündungsheerdes denken? dass bei Personen mit Hyperästhesie der Bauchganglien eine Ver- schieblichkeit der Niere, welche bei gesunden Personen ohne Be- deutung ist, zu nervösen Nierenkoliken Anlass geben kann. Schon das Herausholen der Niere, das doch ohne Druck und Quetschung des Organs nicht abgeht, noch dazu, wenn dasselbe mit Mühe, ja wie man oft liest, mit grosser Gewalt losgelöst werden muss , ist durchaus kein gleichgültiger Eingriff. Wie Menge gezeigt hat, kann man durch Druck auf die Niere, durch die Bauchdecken hindurch (bei Wanderniere) Hämaturie und Al- buminurie erzeugen, und dies ist doch ein verhältnissmässig harmloser Eingriff gegen die Manipulationen bei der Operation. Und dann kommt noch hinzu die Abklemmung der Gefässe, welche natürlich zu Entstehung von Nekrosen Anlass giebt. g g Was noch die Massenblutungen betrifft, so ist, wie ich aus- geführt habe, nicht bewiesen, dass gerade die Entzündungen sic verschulden. ') Poirier, Bulletin de lasoc. de chir. 1898, Mai10, Albarrani. c. 2) G-uyon, IX, Congrès de chir. 1895, S. 534. ) T aima, Deutsches Archiv für kiinisehe Medizin Bd. XLIX, S. 235. y g ) T aima, Deutsches Archiv für kiinisehe Medizin Bd. XLIX, S. 235. DEUTSCHE 1vfEfl1C1NISC1iI WOO NSO1fR1tP. Rovsing, dessen Krankenberichte an Genauigkeit auch nichts zu wünschen übrig lassen, fand unter seinen vier Fällen einmal Verwachsung und zweimal Verschie- bung der Niere, bei Poirier, Albarran') u. a. finden sieh gleich- falls Verwachsungen angegeben, ebenso bei Guyon2), der die im übrigen absolut normale Niere mit der Scheere loslösen musste. g Nun, dass Verwachsungen und Verschiebungen der Niere eine hinreichende Erklärung für Koliken sind und, wenigstens was die Verschiebungen betrifft, auch für Blutungen aus der Niere geben können, ist so bekannt, dass ich darauf nicht weiter ein- zugehen brauche. g Uebrigens hat vor Jahren schon Talma3) darauf hingewiesen,
https://openalex.org/W3093275894	https://repositorioinstitucional.ceu.es/bitstream/10637/12481/1/Intrafollicular_Satue_ANIMALS_2020.pdf	English	null	Intrafollicular and Systemic Dopamine, Noradrenaline and Adrenaline Concentrations in Cycling Mares	Animals	2,020	cc-by	5,815	animals animals animals animals Intrafollicular and Systemic Dopamine, Noradrenaline and Adrenaline Concentrations in Cycling Mares Katiuska Satué 1,* , Esterina Fazio 2 , Maria Dolores Rubio 3 , Cristina Cravana 2 and Pietro Medica 2 1 Department of Animal Medicine and Surgery, Faculty of Veterinary Medicine, CEU-Cardenal Herrera University, Tirant lo Blanc, 7, Alfara del Patriarca, 46115 Valencia, Spain 2 Department of Veterinary Sciences, Veterinary Physiology Unit, Messina University, Via Palatucci, 98168 Messina, Italy; fazio@unime.it (E.F.); ccravana@unime.it (C.C.); pmedica@unime.it (P.M.) 3 Department of Cellular Biology, Physiology and Immunology, Faculty of Veterinary, University of Córdoba, Campus of Rabanales, 14071 Córdoba, Spain; ba1rulum@uco.es * Correspondence: ksatue@uchceu.es; Tel.: +96-136-90-00-66013-66020 * Correspondence: ksatue@uchceu.es; Tel.: +96-136-90-00-66013-66020 1. Introduction In mares, the involvement of dopamine (DA), noradrenaline (NA) and adrenaline (AD) on reproductive physiology is documented, since the catecholamines regulate gonadotropin-releasing hormone (GnRH), follicle-stimulating hormone (FSH), luteinizing hormone (LH) and prolactin (PRL) secretion [1–3]. Furthermore, the administration of DA antagonists stimulates the follicular recrudescence in anestrus mares [4–7], although they are not always capable of maintaining a sustained increase of PRL [5,8]. This fact suggests that a local ovarian mechanism, involving PRL or DA, aﬀects the seasonal follicular growth. High concentrations of DA were found in the antral ﬂuid of preovulatory follicles in human [9–15] and rat [16]. In pig’s follicular ﬂuid (FF), NA concentrations increase signiﬁcantly during the follicular phase, with physiological implications in preovulatory events, as during luteinization [17]. Fernández-Pardal et al. [18] documented that small follicles contain greater concentrations of NA than medium follicles. However, signiﬁcantly greater concentrations of both AD and NA were found in the FF of large preovulatory follicles compared to medium-size follicles. In addition, Kozłowska et al. [19] showed that in the porcine ovaries’ cystic ﬂuid NA was lower as compared to medium follicles of the control group. The contents of DA and AD in the cystic ﬂuid were below the threshold of detection. After dexamethasone (DXM) injections, the concentration of NA in the FF from small follicles was greater than in the control group. The concentrations of DA in FF from small follicles was similar in the control and DXM groups. Moreover, the content of AD in the FF from small follicles in both groups, as well as the concentrations of DA and A in the FF from medium follicles of the control group, were below the threshold of detection. In humans, NA concentrations in FF were substantially higher than those in plasma samples, and a positive correlation between FF and plasma concentrations was found [13]. Although DA was detected in the FF of mare follicles of all sizes, NA has not been identiﬁed in medium or large follicles [20]. It was presumed, although not veriﬁed, that cortical ovarian samples contained primordial and/or primary follicles. The results indicated that both dopaminergic receptor type 2 (DA D2r) and follicle-stimulating hormone receptor (FSHr) mRNAs exist in similar quantities in the cortex of mare’s ovaries during winter anestrus and summer cyclicity. This suggests that DA does not inﬂuence follicular growth through FSHr production. 1. Introduction However, the amount of DA reaching the cortical DA D2r receptors could be rapidly modulated through dopaminergic innervation of these areas. From the clinical point of view, the study of FF has an immense value for better understanding the regulatory mechanisms of fertility in female reproduction. Considering that FF is a compartment with the ability to store and release catecholamines and that it could serve as a reserve to maintain a higher availability of these molecules, the hypothesis of this study was to verify the existence of catecholamines in FF for their possible contribution by systemic sources in mares. On this basis, the objective of the present study was to determine DA, NA and AD concentrations in FF and blood samples, by taking into account the correlations between molecules and the possible contribution of systemic catecholamine concentrations to FF contents. Received: 13 September 2020; Accepted: 13 October 2020; Published: 16 October 2020 Simple Summary: This study provides new evidence on the physiological changes of catecholamines in follicular ﬂuid during the follicular growth in the mare. Both dopamine and epinephrine increase in the follicular ﬂuid with the advance of follicular development, although norepinephrine decreases. These changes could be related to the existence of systemic, autocrine and/or paracrine mechanisms of synthesis, metabolism and interconversion of catecholamines for the regulation of follicular growth and development. Abstract: In some species, catecholamines in follicular ﬂuid (FF) are related to local physiological events responsible for the regulation of ovarian functions and oocyte maturation. The aim of the present study was to determine and compare intrafollicular and systemic concentrations of dopamine (DA), noradrenaline (NA) and adrenaline (AD) in cycling mares. Sixty ovaries were collected during breeding season from 30 mares raised for slaughterhouse meat production, with clinically normal reproductive tracts, were evaluated. Blood samples were collected prior to slaughter. Follicles were classiﬁed into three categories in relation to size: small (20–30 mm; n = 20), medium (≥31–40 mm; n = 20) and large (≥41 mm; n = 20). Follicular ﬂuid (FF) samples were extracted from each follicle. Intrafollicular DA, NA and AD concentrations were signiﬁcantly higher than the systemic concentrations (p < 0.05). Intrafollicular DA concentrations were higher in medium than small and large follicles (p < 0.05). Intrafollicular NA concentrations were higher in small than medium and large follicles (p < 0.05). Intrafollicular AD concentrations were higher in large than small and medium follicles (p < 0.05). Follicle diameter was signiﬁcantly and negatively correlated with NA and AD (p < 0.05). A signiﬁcant correlation of the same hormone concentration in FF and in systemic ﬂuid was observed (p < 0.05). In summary, the FF can serve as an intraovarian catecholamine-storing compartment, with the ability to release neurotransmitters in a regulated way. These results provide novel insights into the neuronal nature of the follicle, suggesting the involvement of catecholamines in normal ovarian functions in mares. Keywords: adrenaline; cycling mare; dopamine; follicular ﬂuid; noradrenaline Animals 2020, 10, 1896; doi:10.3390/ani10101896 www.mdpi.com/journal/animals 2 of 9 Animals 2020, 10, 1896 2.2. Collection of Blood and Ovaries From each animal blood samples (20 mL) were collected from the jugular vein 1 h before slaughtering (gunshot), while mares were in lairage, a procedure that took just a few seconds for each horse. Blood samples were collected using evacuated tubes (Venoject, Terumo®; Belgium) and were transferred into a polypropylene tube containing EDTA (1 mg/mL of blood). Samples were centrifuged at 1200× g for 10 min, and plasma was collected and stored at 4 ◦C in portable coolers for later transport to the laboratory. After the slaughter, only the ovaries of the normal reproductive tracts were collected. The period between the slaughter and collection of ovaries in no case exceeded 2 h, as proposed by Hinrichs et al. [22]. All ovaries were placed in containers with 0.9% physiologic saline plus penicillin (100 IU/mL) and streptomycin (50 mg/mL) and were transported to the laboratory in individually labeled plastic bags in thermal containers (at 25 ◦C) [23]. 2.3. Collection of Follicular Fluid Ovaries were washed three times with sterile saline and the follicles were direct measured with a digital Vernier caliper and categorized according to diameter as small (20–30 mm; n = 20), medium (≥31–40 mm; n = 20) or large (≥41 mm; n = 20). FF was aspirated using a diﬀerent sterile syringe and 22 G needle for each follicle. Following collection, the FF samples were centrifuged for 10 min at 1200× g to eliminate the cumulus oocyte complexes. Only the supernatant (pure FF) was removed and stored in 0.5 mL aliquots at −20 ◦C until subsequent analysis. 2.1. Animals All procedures involved in this study were approached by the CEU-Cardenal Herrera University Committee of Ethics in Animal Research and were carried out by considering the RD 37/2014 that regulates the protection of animals at the time of slaughter and the EU directive 2010/63/EU. The study was conducted in the northern hemisphere during months of the breeding season (April and May 2018). During this period, the ambient temperature ranged from 27–31 ◦C, with a relative humidity of 40–60%. A total of 30 clinically healthy mares (local autochthonous mares for meat production, which mainly include Draft, Hispano–Breton and related crosses) aged 6.6 ± 1.3 years were studied. According to system described by Henneke et al. [21], the animals had a body condition 3 of 9 Animals 2020, 10, 1896 score (BCS) from 7 to 8 out of 9, presenting a mean weight of 533 ± 7.3 kg. All animals were subjected to the same management and feeding conditions, represented by orchard grass–alfalfa mixed hay and had free access to mineral salt and fresh water in a sheltered area. The slaughter was localized in Valencia (Spain), with geographic coordinates of latitude: 39◦31′ 0.01” N and longitude: 0◦25′ 0.01” E. The oﬃcial veterinarians for each stockyard and slaughterhouse accepted responsible participation in the study, and only mares with a reproductive history of normality in their estrus cycles were included in the study. The veterinary examination of the animals prior to slaughter consisted of careful review of oﬃcial documentation, which included livestock of origin, sanitary registration number, suitable health status, deworming and vaccination plan, and clinical and reproductive history of the animals, along with clinically normal reproductive tracts after slaughter. The inclusion criteria for the animals were (1) absence of reproductive diseases in the clinical examination; (2) absence of inﬂammatory processes or infections that had required treatment or hospitalization during the month prior to the onset of the study; (3) to be vaccinated and dewormed correctly and (4) to be younger that 15-year-old, have no conformation defects that aﬀect the perineum and vulva; (5) normal involution of the uterus in previous births, and lack of previous history of reproductive diseases that aﬀect fertility. 3. Results 3. Results In all the samples, intrafollicular concentrations of DA, NA and AD were signiﬁcantly higher than the systemic ones (p < 0.05), and are presented as mean ± SD. In all the samples, intrafollicular concentrations of DA, NA and AD were significantly higher than the systemic ones (p < 0.05), and are presented as mean ± SD. Intrafollicular dopamine (DA) concentrations were higher in small (306 2 ± 113 8 ng/mL) Intrafollicular dopamine (DA) concentrations were higher in small (306.2 ± 113.8 ng/mL), medium (707.9 ± 360.6 ng/mL) and large (351.2 ± 132.5 ng/mL) follicles than systemic ﬂuid (36.65 ± 6.45 ng/mL; p < 0.05). Medium follicles showed higher DA concentrations than small and large follicles (p < 0.05; Figure 1). Intrafollicular dopamine (DA) concentrations were higher in small (306.2 ± 113.8 ng/mL), medium (707.9 ± 360.6 ng/mL) and large (351.2 ± 132.5 ng/mL) follicles than systemic fluid (36.65 ± 6.45 ng/mL; p < 0.05). Medium follicles showed higher DA concentrations than small and large follicles (p < 0.05; Figure 1). Systemic Small (20-30 mm) Medium (≥31-40 mm) Large (≥41 mm) -100 0 100 200 300 400 500 600 700 800 900 Concentrations (ng/mL) * DA NA AD * **** * * Figure 1. Concentrations of dopamine (DA), noradrenaline (NA) and adrenaline (AD) in systemic and follicular fluid (FF) of different follicles of mares (mean ± SD). Systemic fluid vs. small, medium and large follicles: * p < 0.05. Small vs. medium and large follicles: p < 0.05; medium vs. small and large: * p < 0.05; large vs. small and medium **** p < 0.05. Figure 1. Concentrations of dopamine (DA), noradrenaline (NA) and adrenaline (AD) in systemic and follicular ﬂuid (FF) of diﬀerent follicles of mares (mean ± SD). Systemic ﬂuid vs. small, medium and large follicles: * p < 0.05. Small vs. medium and large follicles: p < 0.05; medium vs. small and large: * p < 0.05; large vs. small and medium **** p < 0.05. Figure 1. Concentrations of dopamine (DA), noradrenaline (NA) and adrenaline (AD) in systemic and follicular fluid (FF) of different follicles of mares (mean ± SD). Systemic fluid vs. small, medium and large follicles: * p < 0.05. Small vs. medium and large follicles: p < 0.05; medium vs. small and large: * p < 0.05; large vs. small and medium **** p < 0.05. Figure 1. 2.4. Catecholamine Assay The systemic and intrafollicular DA (pg/mL), NA (ng/mL) and AD (ng/mL) concentrations were determined by competition EIA-Technical 3-CAt EIA (Demeditec Diagnostics GmbH, Germany) speciﬁcally validated for the equine species [24]. The percentages of recovery for DA, NA and AD in plasma were 90.0%, 97% and 92%, respectively, and in FF were 93.0%, 96% and 95%, respectively. The detection limits for DA, NA and AD concentrations were 5 ng/mL, 50 ng/mL and 10 ng/mL, respectively. Serial dilutions up to 1:64 of pooled plasma and FF samples showed ranges of 14.0–917 ng/mL, 1.3–81.4 ng/mL, and 4.9–339 ng/mL for DA, NA, and AD, respectively. The intra-analysis CVs for DA varied between 9.5% and 15.8%, for NA between 9.8% and 16.1%, and for AD between 6.9% and 15%. The inter-analysis CVs for DA ranged between 15.9% and 18.2%, for NA between 8.5% and 15%, and for AD between 13.2% and 15.4%. 4 of 9 13 of 9 Animals 2020, 10, 1896 Animals 2020, 10, x 2.5. Statistical Analyses 2.5. Statistical Analyses D i i i i Descriptive statistics mean ± standard deviation (SD) for DA, NA and AD concentrations in FF of small, medium and large follicles and in blood plasma were calculated. Normality was veriﬁed in all the data groups, using the Kolmogorov–Smirnov test. To determine the magnitude of variation in the concentrations of these constituents of FF and plasma in follicles of diﬀerent diameters, data were subjected to one-way ANOVA analysis. Post-hoc comparisons were performed using Tukey’s test. The relationship between FF and systemic DA, NA and AD concentrations was examined by linear regression analysis, and the correlation was expressed by Pearson’s correlation coeﬃcient. Diﬀerences were considered statistically signiﬁcant when p < 0.05. Descriptive statistics mean ± standard deviation (SD) for DA, NA and AD concentrations in FF of small, medium and large follicles and in blood plasma were calculated. Normality was verified in all the data groups, using the Kolmogorov–Smirnov test. To determine the magnitude of variation in the concentrations of these constituents of FF and plasma in follicles of different diameters, data were subjected to one-way ANOVA analysis. Post-hoc comparisons were performed using Tukey’s test. The relationship between FF and systemic DA, NA and AD concentrations was examined by linear regression analysis, and the correlation was expressed by Pearson’s correlation coefficient. Differences were considered statistically significant when p < 0.05. 4. Discussion 4.1. Intrafollicular Catecholamine Concentrations in Mares/Humans/Other Species The intrafollicular concentration of catecholamines in this study completely diﬀered from those previously reported in the same species [20], despite the similarity in categorized follicle sizes (small: <25 mm; medium: 26–35 mm and large: >35 mm). Indeed, King et al. [20] observed no noradrenergic activity in the FF of medium and large follicles, showing an inverse relationship between follicular size and intrafollicular DA concentrations. Our results may indicate that DA and NA exercise a role in early follicular recruitment, but not in late antral follicular development; nevertheless, intrafollicular AD is present during the development of the dominant follicle until it reaches preovulatory size. Despite these diﬀerences, the dynamic shown by DA during the follicular development period was similar in both the studies. In fact, DA concentrations were signiﬁcantly higher in large and medium than small follicles. The variability between these results may be due to the wide standard deviations, as well as the small number of follicles considered in these studies, manipulation of the samples, laboratory methods employed and diﬀerences in the metabolic activities within the follicles [25]. Indeed, diverse methods used to determination of catecholamines in FF as high performance liquid chromatography (HPLC) [13] or kits ELISA [14] diﬀerent from the one used in this study could be related with the diﬀerences among results of diverse researchers. Substantial intrafollicular catecholaminergic activity in mares suggests an extra- or intra-ovarian production or synthesis. In women, the FF of large follicles accumulates NA in concentrations higher than systemic ones and on similar form to mare, signiﬁcant correlations between both ﬂuids were obtained [25]. Although the mechanisms for accumulation of catecholamines in the FF remain unknown, in women undergoing treatment for in vitro fertilization, NA can be released from sympathetic ﬁbers innervating the interiors of the follicles, interstitial gland and ovarian vasculature [25–27]. Contrary to the case in the mare, the NA in the FF in women [10,13,25] and cows [28] markedly increased in preovulatory follicles. These increases are produced in synergy with increased preovulatory release of NA from the nerve terminals in pigs [18] and humans [29] at the time of preovulatory gonadotropin surge. The synergistic action of catecholamines with gonadotrophins could favor the myocontractility necessary for expulsion of the oocyte at ovulation, as observed in cows [30] and women [29]. 3. Results 3. Results Parameters N (90) r Equation of Regression Line DA (ng/mL) 0.64 Systemic/FF = 1.3557 + 0.99 * NA (ng/mL) 0.67 Systemic/FF = 1.3410 + 0.00154 * AD (ng/mL) 0.93 Systemic/FF = 0.90257 + 0.01575 * Parameters N (90) r Equation of Regression Line DA (ng/mL) 0.64 Systemic/FF = 1.3557 + 0.99 * NA (ng/mL) 0.67 Systemic/FF = 1.3410 + 0.00154 * AD (ng/mL) 0.93 Systemic/FF = 0.90257 + 0.01575 * 3. Results 3. Results Concentrations of dopamine (DA), noradrenaline (NA) and adrenaline (AD) in systemic and follicular ﬂuid (FF) of diﬀerent follicles of mares (mean ± SD). Systemic ﬂuid vs. small, medium and large follicles: * p < 0.05. Small vs. medium and large follicles: p < 0.05; medium vs. small and large: * p < 0.05; large vs. small and medium **** p < 0.05. Intrafollicular noradrenaline (NA) concentrations were higher in small (472.2 ± 227.7 ng/mL), medium (244.7 ± 93.9 ng/mL) and large (210.1 ± 81.6 ng/mL) follicles than systemic fluid (17.12 ± 2.35 ng/mL; p < 0.05). Large follicles showed higher NA concentrations than small and medium follicles (p < 0 05 Figure 1) Intrafollicular noradrenaline (NA) concentrations were higher in small (472.2 ± 227.7 ng/mL), medium (244.7 ± 93.9 ng/mL) and large (210.1 ± 81.6 ng/mL) follicles than systemic fluid (17.12 ± 2.35 ng/mL; p < 0.05). Large follicles showed higher NA concentrations than small and medium follicles (p < 0.05; Figure 1). (p < 0.05; Figure 1). Intrafollicular adrenaline (AD) concentrations were higher in small (48.9–83.2 ng/mL), medium (53.1–71.8 ng/mL) and large (47.7–93.3 ng/mL) follicles than systemic fluid (6.59–19.6 ng/mL; p < 0.05). Large follicles showed higher AD concentrations than small and medium follicles (p < 0.05; Fi 1) Intrafollicular adrenaline (AD) concentrations were higher in small (48.9–83.2 ng/mL), medium (53.1–71.8 ng/mL) and large (47.7–93.3 ng/mL) follicles than systemic ﬂuid (6.59–19.6 ng/mL; p < 0.05). Large follicles showed higher AD concentrations than small and medium follicles (p < 0.05; Figure 1). 5 of 9 Animals 2020, 10, 1896 Follicular diameter increased signiﬁcantly in medium and large respect to small follicle sizes (p < 0.05) and was signiﬁcantly and negatively correlated with NA and AD (p < 0.05). A signiﬁcant correlation of the same hormone concentration in FF and in systemic ﬂuid was observed (p < 0.05; Table 1). Table 1. Systemic and intrafollicular dopamine (DA), noradrenaline (NA) and adrenaline (AD correlation coeﬃcients in follicles of mares. Results are presented as mean ± SD; * p < 0.05. Table 1. Systemic and intrafollicular dopamine (DA), noradrenaline (NA) and adrenaline (AD) correlation coeﬃcients in follicles of mares. Results are presented as mean ± SD; * p < 0.05. correlation coeﬃcients in follicles of mares. Results are presented as mean ± SD; p < 0.05. 4. Discussion Although in the mare, the sources of catecholamines in the FF are unknown, experimental studies in women and in laboratory animals have shown diﬀerent origins of these neurotransmitters. In women, the access of NA to the granulosa cells of larger follicles depends on diﬀusion through the basal lamina and granulosa cells layers during follicular growth. Once NA is taken up by granulosa cells, it is metabolized via mitochondrial monoamine oxidase A (MAO-A) within these cells. Furthermore, granulosa cells also take up and metabolize NA in humans [25] and rats [31]. 6 of 9 Animals 2020, 10, 1896 4.2. Ovarian Synthesis, Uptake, or Interconversion Mechanisms of Catecholamines in Mares/Primates/Other Species According to Receptors’ Expression 4.2. Ovarian Synthesis, Uptake, or Interconversion Mechanisms of Catecholamines in Mares/Primates/Oth Species According to Receptors’ Expression Another point is that DA antagonist treatment of acyclic mares provides controversial results. Hence, some studies reported that treatment with sulpiride, domperidone or perphenazine stimulated ovarian recrudescence, advancing the ﬁrst ovulation of the year in mares maintained outside [4,5,7]; other authors reported a stimulatory eﬀect of sulpiride, but not of domperidone, in mares housed indoors only [32], or a stimulatory eﬀect in photo-stimulated mares housed indoors [2,33]; ﬁnally, others authors have not observed an increase in ovulation [34]. Factors such as environment, photoperiod, temperature, or stress may inﬂuence the eﬃcacy of these treatments [2,34,35]. The exact mechanism of action of catecholamines, DA in particular, on follicular dynamics is not known, but is believed to involve the regulation of prolactin as its pituitary production, which primarily is regulated through inhibition by the neurotransmitter DA. The DA antagonist increases the circulating prolactin and estradiol concentrations after the ﬁrst week of treatment, indicating a follicular development as a result of inhibition of DA and increased stimulation of gonadotrophin receptors on the ovary [36]. However, the DA antagonists allow follicular development only in the presence of FSH secretion and the ability of the ovary to respond to them in the mare. Indeed, the mRNA expression for FSH receptors in the ovarian cortex was minimal during anestrus and increased approximately ﬁve-fold during the breeding season [37]. Moreover, the presence of catecholamines in the FF is intimately related to the expression of receptors in the ovarian follicle [25]. 4. Discussion In mares, DA D2r receptors are highly expressed in the luteal tissue and ovarian cortex, but they are not highly expressed in granulosa and theca cells, while D1r receptors are only expressed in corpus luteum [20]. The low presence of DA receptors in granulosa cells makes the synthesis of this catecholamine decrease in the FF of larger-sized follicles. However, these ﬁndings do not support the idea that DA synthesis is exclusively mediated by the dopaminergic receptors, since in our study, both DA and NA considerably rise in the follicles of small and medium size, although they decrease later in the preovulatory follicles. The substantial variation observed in the mare could be related to the complex metabolism of catecholamines, since DA is the precursor of follicular NA and both can be metabolized. These ﬁndings suggest that the depletion of endogenous DA and NA in late follicular development stages could be a consequence of reduced synthesis and uptake, or of interconversion mechanisms of DA for the ﬁnal synthesis of AD, with the aim of reaching the preovulatory stage. Furthermore, NA targets β-adrenergic receptors of granulosa cells and elevates cAMP, allowing them to serve as catecholamine-storing cells within the ovarian follicle and ensuring that DA and/or NA are present in the granulosa cell compartment [31]. In addition, adrenergic nerves supply the theca and the evidence for DA secretion by nerve cells to the Graaﬁan follicle has been shown in cows [28]. Furthermore, it has also been reported that DA-containing nerve terminals exist in the thecal cell layer and around the walls of mature follicles in guinea pigs, but not in the granulosa cell layer. However, granulosa cells also express transporters for NA and DA, favoring their intracellular storage [14,25,31,38]. Metanephrine, a metabolite of DA and NA in FF and granulosa cells, also provides clear evidence of the follicular metabolism of catecholamines [25]. In addition, a ligand (NA in FF) and its receptor (ADRB-2 in granulosa cells) co-exist in the large antral follicles of primates [39]. Likewise, NA can be synthesized by the neuron-like cells in the ovary and oocytes. Indeed, in non-human primates, oocytes, but no other ovarian cell types, express the enzyme DA hydroxylase (DBH), and under experimental conditions can take up its and convert it to NA [40]. Hence, a complex ovarian DA system, which includes receptor-mediated roles for DA and DA metabolism, may be assumed [14]. References 1. Melrose, P.A.; Walker, R.F.; Douglas, R.H. Dopamine in the cerebrospinal ﬂuid of prepubertal and adult horses. Brain Behav. Evol. 1990, 35, 98–106. [CrossRef] [PubMed] 1. Melrose, P.A.; Walker, R.F.; Douglas, R.H. Dopamine in the cerebrospinal ﬂuid of prepubertal and adult horses. Brain Behav. Evol. 1990, 35, 98–106. [CrossRef] [PubMed] 2. Daels, P.F.; Fatone, S.; Hansen, B.S.; Concannon, P.W. Dopamine antagonist-induced reproductive function in anoestrous mares: Gonadotrophin secretion and eﬀects of environmental cues. J. Reprod. Fertil. 2000, 56, 173–183. 2. Daels, P.F.; Fatone, S.; Hansen, B.S.; Concannon, P.W. Dopamine antagonist-induced reproductive function in anoestrous mares: Gonadotrophin secretion and eﬀects of environmental cues. J. Reprod. Fertil. 2000, 56, 173–183. 3. Satué, K.; Gardon, J.C.; Marcilla, M. Physiology and Metabolic Anomalies of Dopamine in Horses: A Review; Open Access Peer-Reviewed Chapter; InTechOpen: London, UK, 2018; pp. 85–109. [CrossRef] 4. Besognet, B.; Hansen, B.H.; Daels, P.F. Dopaminergic regulation of gonadotrophin secretion in seasonally anoestrous mares. J. Reprod. Fertil. 1996, 108, 55–61. [CrossRef] [PubMed] Besognet, B.; Hansen, B.H.; Daels, P.F. Induction of reproductive function in anestrous mares using a dopamine antagonist. Theriogenology 1997, 47, 467–480. [CrossRef] . Bennett-Wimbush, K.; Loch, W.E.; Plata-Madrid, H.; Evans, T. The eﬀect of perphenazine and bromocryp on follicular dynamics and endocrine proﬁles in anestrous mares. Theriogenology 1998, 49, 717–733. [Cross 7. Panzani, D.; Zicchino, I.; Taras, A.; Marmorini, P.; Crisci, A.; Rota, A.; Camillo, F. Clinical use of dopamine antagonist sulpiride to advance ﬁrst ovulation in transitional mares. Theriogenology 2011, 75, 138–143. [CrossRef] 8. Thompson, D.L., Jr.; DePew, C.L. Prolactin, gonadotropin, and hair shedding responses to daily sulpiride administration in geldings in winter. J. Anim. Sci. 1997, 75, 1087–1091. [CrossRef] [PubMed] 9. Bódis, J.; Bognár, Z.; Hartmann, G.; Török, A.; Csaba, I.; Halvax, L. Analysis of noradrenaline, dopamine and serotonin levels in follicular ﬂuid following superovulatory treatment. Orv. Hetil. 1991, 132, 2475–2477. [PubMed] 10. Bódis, J.; Török, A.; Tinneberg, H.R.; Hanf, V.; Hamori, M.; Cledon, P. Inﬂuence of serotonin on progesterone and estradiol secretion of cultured human granulosa cells. Fertil. Steril. 1992, 57, 1008–1011. [CrossRef] 11. Bódis, J.; Hartmann, G.; Török, A.; Bognár, Z.; Tinneberg, H.R.; Cledon, P.; Hanf, V. Relationship between the monoamine and gonadotropin content in follicular ﬂuid of preovulatory graaﬁan follicles after superovulation treatment. Exp. Clin. Endocrinol. 1993, 101, 178–182. [CrossRef] 12. Bódis, J.; Hartmann, G.; Tinneberg, H.R.; Török, A.; Hanf, V.; Papenfuss, F.; Schwarz, H. 4. Discussion Since NA is the most abundant neurotransmitter released by the sympathetic nerves in most mammalian species [27], the granulosa cells take up and store it, releasing it upon depolarization. Although ovarian denervation inhibits the follicular growth, it reduces, but does not eliminate, NA in the gland [41], implying the existence of an additional source of catecholamine synthesis or participation of intraovarian cells in ovarian NA homeostasis. 7 of 9 Animals 2020, 10, 1896 4.3. Eﬀects of Systemic and Follicular Fluid Catecholamines on the Reproductive Physiology in Mares 4.3. Eﬀects of Systemic and Follicular Fluid Catecholamines on the Reproductive Physiology in Mares In summary, the presence of DA, NA and AD in the FF of the mare indicates the existence of interspeciﬁc diﬀerences regarding the content of catecholamines with respect to other species. The great amount of DA, NA and AD in FF could indicate the existence of systemic and autocrine and/or paracrine mechanisms related to catecholamine synthesis, metabolism and subsequent use, guaranteeing the successful growth and development of ovarian follicles. This physiological response could have pivotal applications in the reproductive clinic for the diagnosis of pathological processes that occur with infertility in the mare. However, future investigations would be required to clarify these complex mechanisms of interconversion and sources of catecholamines in the FF of mares. Author Contributions: K.S. and P.M. conceived and designed the experiment; K.S. performed the experiment; K.S., E.F., C.C. analyzed the data; K.S. wrote the paper. M.D.R. reviewed the paper. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Funding: This research received no external funding. Acknowledgments: The authors express their gratitude to the Laboratory of Physiology of University Complutense of Madrid (Spain) for their technical assistance. Acknowledgments: The authors express their gratitude to the Laboratory of Physiology of University Complutense of Madrid (Spain) for their technical assistance. Conﬂicts of Interest: The authors declare no conﬂict of interest. Conﬂicts of Interest: The authors declare no conﬂict of interest. References Relationship between the monoamine, progesterone and estradiol content in follicular ﬂuid of preovulatory graaﬁan follicles after superovulation treatment. Gynecol. Obstet. Investig. 1993, 35, 232–235. [CrossRef] [PubMed] 13. Itoh, M.T.; Ishizuka, B.; Kuribayashi, Y.; Abe, Y.; Sumi, Y. Noradrenaline concentrations in human preovulatory follicular ﬂuid exceed those in peripheral plasma. Exp. Clin. Endocrinol. Diabetes 2000, 108, 506–509. [CrossRef] 8 of 9 Animals 2020, 10, 1896 14. Saller, S.; Kunz, L.; Berg, D.; Berg, U.; Lara, H.; Urra, J.; Hecht, S.; Pavlik, R.; Thaler, C.J.; Mayerhofer, A. Dopamine in human follicular ﬂuid is associated with cellular uptake and metabolism-dependent generation of reactive oxygen species in granulosa cells: Implications for physiology and pathology. Hum. Reprod. 2014, 29, 555–567. [CrossRef] [PubMed] 15. Musalı, N.; Özmen, B.; ¸Sükür, Y.E.; Ergüder, B.˙I.; Atabeko˘glu, C.S.; Sönmezer, M.; Berker, B.; Aytaç, R. Follicular ﬂuid norepinephrine and dopamine concentrations are higher in polycystic ovary syndrome. Gynecol. Endocrinol. 2016, 32, 460–463. [CrossRef] [PubMed] 16. Ben-Jonathan, N.; Arbogast, L.A.; Rhoades, T.A.; Bahr, J.M. Norepinephrine in the rat ovary: Ontogeny and de novo synthesis. Endocrinology 1984, 115, 1426–1431. [CrossRef] 17. Bahr, J.M.; Ben-Jonathan, N. Elevated catecholamines in porcine follicular ﬂuid before ovulation. Endocrinology 1985, 117, 620–623. [CrossRef] 18. Fernández-Pardal, J.; Gimeno, M.F.; Gimeno, A.L. Catecholamines in sow graaﬁan follicles at proestrus and at diestrus. Biol. Reprod. 1986, 34, 439–445. [CrossRef] 19. Kozłowska, A.; Majewski, M.; Jana, B. Changes in the cholinergic innervation pattern of porcine ovaries with cysts induced by dexamethasone administration. J. Mol. Neurosci. 2014, 54, 10–19. [CrossRef] 20. King, S.S.; Jones, K.L.; Mullenix, B.A.; Heath, D.T.; Everson, K.A.; Arbogast, L.A. Evidence for dopaminergic activity in equine follicles and ovarian germinal epithelium. Anim. Reprod. Sci. 2006, 94, 175–178. 21. Henneke,D.R.; Potter,G.D.; Kreider,J.L.; Yeates,B.F.Relationshipbetweenconditionscoring, physicalmeasurement and body fat percentage in mares. Equine Vet. J. 1983, 15, 371–372. [CrossRef] 22. Hinrichs, K. Assisted reproduction techniques in the horse. Reprod. Fertil. Dev. 2012, 25, 80–93. [CrossRef] [PubMed] 23. Foss, R.; Ortis, H.; Hinrichs, K. Eﬀect of potential oocyte transport protocols on blastocyst rates after intracytoplasmic sperm injection in the horse. Equine Vet. J. 2013, 45, 39–43. [CrossRef] [PubMed] 24. Marcilla, M.; Muñoz, A.; Satué, K. Longitudinal changes in serum catecholamines, dopamine, serotonin, ACTH and cortisol in pregnant Spanish mares. Res. Vet. Sci. 2017, 115, 29–33. [CrossRef] [PubMed] 25. Saller, S.; Merz-Lange, J.; Raﬀael, S.; Hecht, S.; Pavlik, R.; Thaler, C.; Berg, D.; Berg, U.; Kunz, L.; Mayerhofer, A. Nagy, P.; Guillaume, D.; Daels, P. Seasonality in mares. Anim. Reprod. Sci. 2000, 60–61, 245–262. [CrossRef] References Norepinephrine, active norepinephrine transporter, and norepinephrine-metabolism are involved in the generation of reactive oxygen species in human ovarian granulosa cells. Endocrinology 2012, 153, 1472–1483. [CrossRef] [PubMed] 26. Aguado, L.I. Role of the central and peripheral nervous system in the ovarian function. Microsc. Res. Tech. 2002, 59, 462–473. [CrossRef] 27. Mayerhofer, A.; Smith, G.D.; Danilchik, M.; Levine, J.E.; Wolf, D.P.; Dissen, G.A.; Ojeda, S.R. Oocytes are a source of catecholamines in the primate ovary: Evidence for a cell-cell regulatory loop. Proc. Natl. Acad. Sci. USA 1998, 95, 10990–10995. [CrossRef] 28. Paredes, A.H.; Salvetti, N.R.; Diaz, A.E.; Dallard, B.E.; Ortega, H.H.; Lara, H.E. Sympathetic nerve activity in normal and cystic follicles from isolated bovine ovary: Local eﬀect of beta-adrenergic stimulation on steroid secretion. Reprod. Biol. Endocrinol. 2011, 9, 66. [CrossRef] 29. Christian, C.A.; Moenter, S.M. The neurobiology of preovulatory and estradiol-induced gonadotropin-releasing hormone surges. Endocr. Rev. 2010, 31, 544–577. [CrossRef] 30. Kotwica, G.; Kurowicka, B.; Franczak, A.; Grzegorzewski, W.; Wrobel, M.; Mlynarczuk, J.; Kotwica, J. The concentrations of catecholamines and oxytocin receptors in the oviduct and its contractile activity in cows during the estrous cycle. Theriogenology 2003, 60, 953–964. [CrossRef] 31. Greiner, M.; Paredes, A.; Rey-Ares, V.; Saller, S.; Mayerhofer, A.; Lara, H.E. Catecholamine uptake, storage, and regulated release by ovarian granulosa cells. Endocrinology 2008, 149, 4988–4996. [CrossRef] 32. Mari, G.; Morganti, M.; Merlo, B.; Castagnetti, C.; Parmeggiani, F.; Govoni, N.; Galeati, G.; Tamanini, C. Administration of sulpiride or domperidone for advancing the ﬁrst ovulation in deep anestrous mares. Theriogenology 2009, 71, 959–965. [CrossRef] [PubMed] 33. Duchamp, J.; Daels, P.F. Combined eﬀect of sulpiride and light treatment on the onset of cyclicity in anestrous mares. Theriogenology 2002, 58, 599–602. 34. McCue, P.M.; Buchanan, B.R.; Farquhar, V.J.; Squires, E.L.; Cross, D.L. Eﬃcacy of domperidone on induction of ovulation in anestrous and transitional mares. Proc. Annu. Conv. Am. Assoc. Equine Pract. 1999, 45, 217–218. 35. Nagy, P.; Guillaume, D.; Daels, P. Seasonality in mares. Anim. Reprod. Sci. 2000, 60–61, 245–262. [CrossRef] 9 of 9 Animals 2020, 10, 1896 36. McCue, P.M. Management of seasonal anestrus: Hormone therapy. In Equine Reproductive Procedures; Dascanio, J.J., McCue, P.M., Eds.; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2014; pp. 146–148. 37. King, S.S.; Jones, K.L.; Mullenix, B.A.; Heath, D.T. Seasonal relationships between dopamine D1 and D2 receptor and equine FSH receptor mRNA in equine ovarian epithelium. Anim. Reprod. Sci. 2008, 108, 259–266. [CrossRef] 38. References Rey-Ares, V.; Lazarov, N.; Berg, D.; Berg, U.; Kunz, L.; Mayerhofer, A. Dopamine receptor repertoire of human granulosa cells. Reprod. Biol. Endocrinol. 2007, 5, 40. [CrossRef] 39. Merz, C.; Saller, S.; Kunz, L.; Xu, J.; Yeoman, R.R.; Ting, A.Y.; Lawson, M.S.; Stouﬀer, R.L.; Hennebold, J.D.; Pau, F.; et al. Expression of the beta-2 adrenergic receptor (ADRB-2) in human and monkey ovarian follicles: A marker of growing follicles? J. Ovarian Res. 2015, 8, 8. [CrossRef] 40. Lara, H.E.; Porcile, A.; Espinoza, J.; Romero, C.; Luza, S.M.; Fuhrer, J.; Miranda, C.; Roblero, L. Release of norepinephrine from human ovary: Coupling to steroidogenic response. Endocrine 2001, 15, 187–192. [CrossRef] 41. Dees, W.L.; Hiney, J.K.; Schultea, T.D.; Mayerhofer, A.; Danilchik, M.; Dissen, G.A.; Ojeda, S.R. The primate ovary contains a population of catecholaminergic neuron-like cells expressing nerve growth factor receptors. Endocrinology 1995, 136, 5760–5768. [CrossRef] Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional aﬃliations. Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional aﬃliations. © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
https://openalex.org/W4389430153	https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0295037&type=printable	English	null	Robustness of performance during domain change in an esport: A study of within-expertise transfer	PloS one	2,023	cc-by	11,236	Robustness of performance during domain change in an esport: A study of within- expertise transfer Joe ThompsonID1, Justin W. O’Camb2, Robin C. A. BarrettID2, Scott Harrison3, Mark R. Blair1,2 1 Department of Psychology, Douglas College, New Westminster, Canada, 2 Department of Psychology, Simon Fraser University, Burnaby, Canada, 3 Cognitive Science Program, Simon Fraser University, Burnaby, Canada a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 thompsonj14@dougalscollege.ca Editor: Stergios Makris, Edge Hill University, UNITED KINGDOM Received: September 19, 2022 Accepted: November 13, 2023 Published: December 7, 2023 Copyright: © 2023 Thompson et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: Data for the present analyses and associated analysis code can be found here: https://github.com/SFU-Cognitive- Science-Lab/StarTransfer. Raw data (including raw replay files) will not be released publicly as it is possible due to concerns that such data could compromise participant anonymity. Science-Lab/StarTransfer. Raw data (including raw replay files) will not be released publicly as it is possible due to concerns that such data could compromise participant anonymity. PLOS ONE RESEARCH ARTICLE OPEN ACCESS Citation: Thompson J, O’Camb JW, Barrett RCA, Harrison S, Blair MR (2023) Robustness of performance during domain change in an esport: A study of within-expertise transfer. PLoS ONE 18(12): e0295037. https://doi.org/10.1371/journal. pone.0295037 Abstract Research on the transfer of skill from the circumstances in which it was learned to partially or completely novel tasks or situations is a foundational topic in the study of learning, mem- ory, education, and expertise. A long history of transfer research has led to the conclusion that skill learning is generally domain specific. One important transfer problem occurs when a domain of expertise undergoes a fundamental shift, as when experts must adapt to changes in technology, rules, or professional practice. Here we examine skill maintenance in StarCraft 2, a video game of skills which undergoes frequent changes due to updates and includes a variety of gameplay options. Of particular interest are two competing predictions about how transfer will interact with expertise in this domain. The first approach emphasizes perceived similarity of the domains and predicts that skilled individuals will exhibit more favourable transfer than novices as these people will know enough to avoid processes, methods, and strategies which no longer apply after a domain change. The second empha- sizes maximal adaptation to task constraints and predicts that experts will suffer the most during a domain change because of the loss of exploitable affordances. Neither approach did a good job explaining behaviour after the major game update called ‘StarCraft 2: Heart of the Swarm,’ perhaps because transfer was generally strong across all players. However, when examining transfer in the context of larger changes to gameplay, transfer seemed slightly better in more experienced players. The theoretical implications of this apparent interaction effect, and of the apparent resilience of more experienced StarCraft 2 players to transfer costs, are discussed. PLOS ONE PLOS ONE Introduction Funding: The authors received no specific funding for this work. An understanding of transfer is the holy grail of learning research. This is because, insofar as prior experience can be construed as a training task and future behaviour is construed as a transfer task, learning itself is a transfer effect [1]. A deep understanding of the mechanisms Competing interests: The authors have declared that no competing interests exist. PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 1 / 17 PLOS ONE Robustness of performance during domain change in an esport underlying transfer, would presumably allow for the prediction and control of learning, skill, and expertise. A good theory of learning, therefore, should be expected to predict and explain transfer effects (e.g., Thorndike & Woodworth [2]; for a review, see Adams [3]). One of the early lessons from transfer research is that skill is domain-specific (e.g., Chase & Simon [4]). There are many examples where training tasks do not improve performance on a related task as much as one might like. Transfer costs can be found in sports [5, 6], infant loco- motion [7], mathematics [8], luggage security screening tasks [9], spatial ability [10, 11], and education [12, 13]. Domain specificity has also been observed in research on transfer in video games. Tetris experts showed increased mental rotation ability, but only for Tetris or Tetris like shapes [10]. They didn’t show improvement in spatial ability tasks. When participants new to Tetris were given a 12-hour Tetris training program the domain specific improvements in mental rotation observed in experts were not observed in new trainees. Programs aimed at improving cogni- tive skills through digital games have repeatedly failed to show any real transfer effects [14– 18]. In 2016, Lumos Labs, the company behind Lumosity, was made to pay $2 million [19] after their claims of far transfer between their games and other aspects of daily living were found to be unsupported by evidence. There are even occasions where transfer performance is worse after training [20]. Indeed, the conventional wisdom in skill research seems to be that far transfer, where training improves performance on a wide range of very different tasks, is rarer than expected and should never be assumed without empirical investigation [1]. Of course, real-world transfer effects must exist, or else learning would be impossible. Introduction For example, elite players of ball sports tend to have diverse sporting experiences [21], and some study techniques facilitate transfer better than others [22]. Nevertheless, predicting transfer remains highly challenging. While researchers can make real-world predictions by relying on the conventional wisdom that skill is generally domain specific, the conventional wisdom has little predictive value without an understanding of the moderators of transfer. The present study examines the utility of two different theoretical frameworks which make competing predictions about transfer in the context of competitive esport video game play. The maximal adaptation approach construes expertise development as a process of increasing specialization, suggesting that transfer should become less effective with skill [23]. The similar- ity-based approach, in contrast, suggests that large transfer costs will be ameliorated in experts due to their educated perceptions about the similarity between test and transfer tasks [1]. We first test the prediction, common to all theories of transfer, that learning is generally domain- specific [1, 23]. We then examine competing predictions about how transfer effects will be impacted by expertise. PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 Theoretical background The present work defines transfer costs as a decrease in performance as participants transition from a training task to a test task [1]. The definition of ‘training’ and ‘test’ are intentionally left arbitrary, allowing transfer to be examined across many learning domains and across various theoretical frameworks. The definition also allows us to distinguish good transfer (where per- formance is maintained or improves from training tasks to test tasks), partial transfer (where performance is degraded but not to novice levels), non-transfer (where performance is degraded to novice levels), and negative transfer (where performance is degraded to below novice levels). A unified theory of transfer, therefore, would provide principles that could be used to predict how an animal will perform in a given situation (i.e., the transfer task) based on its set of learned experiences (i.e., the training task). Given the grandeur of this theoretical goal, it should be no surprise that no complete and unified theory of transfer is yet available. PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 2 / 17 PLOS ONE Robustness of performance during domain change in an esport However, researchers have proposed theoretical frameworks and approaches to transfer which do nevertheless make predictions about when and where transfer will be observed. The present study will compare the predictions of two such approaches. However, researchers have proposed theoretical frameworks and approaches to transfer which do nevertheless make predictions about when and where transfer will be observed. The present study will compare the predictions of two such approaches. The similarity-based approach to transfer can be found in Kimball & Holyoak [1]. This approach maintains that successful transfer requires tasks which are objectively similar—in the sense that both tasks involve similar abilities—but also perceived as similar by participants. This recognition of similarity is thought to inspire the cognitive system to leverage skills from the training task in execution of the transfer task. We consider this to be a theoretical approach, rather than a well-specified theory, as researchers might disagree about how similarity should be defined [24]. However, this approach will lead to the prediction that transfer effects should be more favourable as participants become more experienced in the training task. This predic- tion stems from the fact that skilled participants, being more knowledgeable about the similarity between training and test tasks, will be able to best exploit whatever prior knowledge is relevant to the transfer task. Theoretical background In the context of expertise, this means that whatever opportunities of trans- fer are afforded by the training task will be identified and exploited by experts. While Kimball & Holyoak [1] cache the similarity-based framework in theoretically neutral terms, more developed versions of the framework do exist. In Gick’s and Holyoak’s [25] theo- retical framework, for example, transfer problems are put in terms of mental representations and rules for transforming these representations into procedures that produce behaviour. This framework follows Kimball’s & Holyoak’s [1] prediction that transfer is dependent on per- ceived similarity. In Gick’s and Holyoak’s framework, tasks would be perceived as similar when a common set of rules or representations would be deployed across a variety of different situations. In such cases, the cognitive system will attempt to apply its prior knowledge to a new situation, and this attempt will fail miserably if, despite perceived similarity, the transfer tasks has important structural differences. The predictions of Gick’s & Holyoak’s [25] framework align with Kimball & Holyoak [1]. The cognitive systems of experts are likely, in the language of Gick & Holyoak to have rules that better match the structural regularities of the world making them less likely to attempt transfer when they should not, and this would lead to less negative transfer. We might also expect experts to have a great deal of positive transfer insofar as they have a highly developed network of many rules such that a solution to novel problems can be identified through the application of old knowledge. For example, if expert knowledge in physics is construed in terms of larger rule sets, then transfer of knowledge to novel physics problems might be expected insofar as a circuitous application of existing rules arrive at a solution. Similarly, experts may have more nuanced perceptions that represent a larger number of problem-relevant features, allowing them to use more resources in solving novel problems. So while the theories developed within Gick’s & Holyoak’s [25] framework may employ different theoretical language, the predictions of these theories align with the similarity-based approach. These approaches predict that trans- fer costs, when they can be found, should be ameliorated by expertise. Not all frameworks would predict that the most experienced individuals will enjoy the most favourable transfer. Indeed, expertise is often construed as a maximal adaptation to task con- straints [23]. PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 Theoretical background In this case, it is reasonable to expect that changes to a domain of expertise will be especially damaging to experts, as their optimized performance is based on subtle task affor- dances that are overlooked by novices. This framework too, has sometimes been cached out in both representational and less-traditional terms. For example, representational descriptions of the maximal adaptation approach have been provided in the education literature (see, e.g., Gegnefurtner & Seppa¨nen [13]). Under this conceptualization, the representations and sche- mas tied to expert performance are optimized to specific task conditions and cannot trivially be assimilated into other tasks. 3 / 17 PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 PLOS ONE Robustness of performance during domain change in an esport Of course, the maximal adaptation approach does not need to be construed in such repre- sentational terms. For example, consider Adolph’s [26] finding that the avoidance capacities of young infants do not transfer from crawling to walking. Crawling-skilled but walking-novici- ate babies tend to plunge over impossible gaps that they would not attempt to cross when crawling. Adolph views such transfer costs as due to the organization of control systems in infants. These motor systems are tied to posture in such a way that crawling skill cannot hope to transfer perfectly to walking. Adolph argues that highly general representations and rules, of the kind hypothesized in Gick’s [25] account of expert transfer, are not practical in the context of locomotor development, in part, because infant bodies change so rapidly. Adults are, of course, not babies. Nevertheless, the learning of experts may well be tied to domain specifics that could interfere with some transfer performance. For example, while expert chess players have superior memory for chess positions, their ability is usually restricted to realistic chess positions and not truly random chess positions [4, 27]. We might expect, therefore, that a change to the rules of chess would seriously compromise chess memory per- formance if the rule change were to allow for new board positions that were previously impos- sible in a standard game of chess. From a perspective of maximal adaptation to task constraints, therefore, we might predict that novices will have better transfer than experts. Theoretical background Regardless of how it is expressed, the maximal adaptation approach supposes that (a) exper- tise is a process of increasing optimization to the subtle particularities of a task environment and (b) transfer of skill to tasks outside of the learning environment will be laden with transfer costs. The additional optimization of an expert to a single learning environment is not likely to reduce transfer costs and, on the contrary, may make such costs worse. At the very least, the maximal adaptationist perspective does not hold out much hope for superior transfer of experts [13]. To test the predictive utility of these approaches one needs rich experience data comprising a large swath of the expertise continuum. The maximal adaptation framework, in particular, is designed for domains where learning occurs over hundreds or thousands of hours. Such task environments will be hard to construct in laboratory conditions as expertise takes thousands of hours to develop, and because cross-sectional comparisons of experts and novices can be misleading by only comparing snapshots of each player’s current level of expertise without showing each person’s individual development from novice to expert [28]. Here we examine transfer using real-world telemetry from skilled electronic sport (‘esport’) performance. First, StarCraft 2 is a domain of expertise, with full time professional players who have thousands of hours of practice. Furthermore, esports data collection can be performed automatically by participant computers, producing detailed measures of skilled performance across large sample sizes. This allows us to examine the impacts of domain change longitudi- nally and in the context of a player’s entire history of gameplay. The competitive esport StarCraft 2 StarCraft 2 is a real-time strategy game where the player manages economic and military resources of a civilization, while at the same time attempting to destroy the civilization of their opponent. The player must focus both on macro-level tasks, like managing their economy and expanding their bases, while also focusing on micro-level tasks like commanding groups of military units into battle. Given the central role of Chess in the history of expertise research [4], it is useful to com- pare Chess and StarCraft 2. Both are strategy games which involve the competition between artificial armies following stable starting conditions. It is useful, however, to highlight two important differences between the two games. PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 4 / 17 PLOS ONE Robustness of performance during domain change in an esport First, StarCraft 2 requires a great deal of task switching. Unlike Chess, where players begin with a complete army, in StarCraft 2 players must switch between the collection of resources, the production of an army, the control of one’s own army, and discovering hidden informa- tion about the opponent. These shifts require directing attention to various areas of the inter- face, and switching the viewscreen to look at various places on the large scale map upon which the game is played (for an investigation of attention in StarCraft 2 see McColeman et al. [29]). Second, StarCraft 2 takes place in real-time rather than in turns like Chess, so there is a large incentive to issue commands as quickly as possible. Previous research has found that one of the most robust predictors of skill in StarCraft 2 is the speed at which players can act after allo- cating attention to a new area within the game [28]. In short, the game demands speedy perfor- mance while under high cognitive load. Our research focuses on two transfer problems found in the video game StarCraft 2. First, on March 12, 2013, StarCraft 2 went through the ‘Heart of the Swarm’ expansion. The Heart of the Swarm expansion was a particularly large change as it introduced seven new usable game units, each with unique appearances and abilities that had important gameplay implications. This change enabled new kinds of tactics and made many strategies less effective. Any player who purchased the Heart of the Swarm expansion had to integrate these units into their play. The competitive esport StarCraft 2 Even if they themselves were not using the new units, they would have to learn to counter the new dynamics imposed by the new units if their opponents opted to use them. Different theoretical frameworks disagree as to how the impact of the expansion on perfor- mance should interact with prior expertise. Kimball’s & Holyoak’s [1] similarity-based frame- work would predict that more experienced participants will have more detailed and accurate knowledge of the differences between pre and post-expansion gameplay, and therefore also expect the magnitude of transfer effects to become more favourable with prior experience [1]. However, a maximal-adaptation approach [23] would lead one to expect less favourable trans- fer with prior experience insofar as only experts are skilled enough to rely on subtle contingen- cies that were disturbed by the release of Heart of the Swarm. A second transfer problem that emerges in the context of StarCraft 2 comes from choices in gameplay. Players are given a wide range of latitude in how they approach the game. Unlike in Chess, where the available chess pieces remain the same for all players, StarCraft 2 players must choose a ‘Race’ at the beginning of each match, each of which is associated with an entirely new set of pieces. Each piece is associated with its own advantages and disadvantages, and many pieces are associated with unique special abilities that alter the basic mechanics of gameplay. There are also more fundamental differences between the StarCraft 2 races. Unlike Chess, where pieces are granted at the beginning of the game, StarCraft 2 players must manage economic considerations while developing their civilization, first collecting resources and building prerequisite buildings prior to being able to produce any military units. Some races, for example, may produce their units through a series of disjointed structures, while another must ‘grow’ their pieces through a central structure called a hive. This requires that every action of a player is then coloured in some way by their choice of race, meaning that the perfect transfer of skill between races seems unlikely given the domain specificity of expertise. Given the large change associated with changing race and the general finding that transfer is difficult, we once again predict a general loss in performance. As with the prior analysis, experienced players who switch Races enjoy a superior knowledge of the game in general, sug- gesting more favourable transfer from a similarity-based framework. PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 Definition of experience The definition of experience is challenging in the complex environments such as StarCraft 2 [30]. This is because StarCraft 2 has several different game modes, including team games and games with opponents selected by the player rather than opponents selected by the match- making system. This issue is especially challenging when one is investigating transfer between game modes because, if transfer were poor, then experience would be better measured by more restrictive definitions that are specific to the number of games a player has played in a particular game mode. If transfer effects are strong, then more permissive definitions, which include a wider range of game modes, might be better. Given that transfer effects can be unpre- dictable, we begin with a more permissive definition but plan to rerun analyses with more restrictive definitions as necessary. We initially measure a player’s current experience, there- fore, against the number of verified StarCraft 2 games in our database (117,978 such games are in our database) which they have participated in up until that date. To ensure results are not artifacts of our definition of experience, we sometimes rerun analyses with a stricter definition of experience as one-versus-one games with an opponent found using the match-making sys- tem (81,655 such games are in our database; (see S1 Text for details about the verification process). The competitive esport StarCraft 2 In contrast, an emphasis on maximal-adaptation to task constraints suggests that the major change in gameplay associ- ated with a change of race would be especially punishing to those who were highly adapted to the task in their preferred race. 5 / 17 PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 PLOS ONE Robustness of performance during domain change in an esport Data collection Participants were recruited from internet gaming communities, and 124 StarCraft 2 players filled out a survey and, collectively, submitted 164,001 game replay files for analysis. The col- lected self-report survey was, for the present project, consulted only to determine each player’s years of experience with the original StarCraft game, on which StarCraft 2 was based (see Anal- ysis 3 for details). The primary measures are performance-based measures found in replay files collected from each player. The collected replay files are automatically generated by the game itself, and are encoded such that they are difficult to modify. They include the information the game engine requires to replay game in full. This file, once parsed, provides us with time- stamped lists of every action players took during each game, producing the files from which our performance data is calculated. The average game lasted roughly 15 minutes. To be included in our study, games needed to be verifiable StarCraft 2 games played by the survey respondent (see S1 Text for details about the verification process). After this initial exclusion, 109 players and 117,978 StarCraft 2 games remained for analysis. StarCraft 2 is a complex game with several different game modes. Some of these game modes, such as games between two teams of players, are relatively rare. We further restricted our analysis to games that contained two human players that were randomly matched against each other using the game developers automated matchmaking system. This helps to ensure that players were matched against motivated opponents of roughly equal skill. This left us with 107 players (103 males; 2 females; 1 other; 1 unknown) and 81,655 replays, which is equivalent to roughly 20,000 hours of second-by-second performance data. Ages of these players range from 16–41, with a mean age of 24.65, and a standard deviation of 5.28 years. Dependent variable: League-equivalent performance Our primary performance measure for our analysis is League-equivalent Performance and is derived from looking-doing latency, which is defined as the mean latency of actions that immediately follow a shift in attention within StarCraft 2. Looking-doing latencies are the first component of Perception-Action Cycles of a StarCraft 2 game, a cycle which includes fixating PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 6 / 17 PLOS ONE Robustness of performance during domain change in an esport the game screen on a new location, acting, and shifting the screen away [31]. This measure is of interest to psychologists as it is inspired by the perception-action cycles that make up real-world tasks [32]. Prior work has found that looking-doing latency is a robust predictor of StarCraft 2 skill [28], exhibits patterns of learning-related changes in information-access behaviours consis- tent with laboratory eye-tracking research [29], and shows age-related changes [33]. There are two factors that make using raw looking-doing latency less desirable for the pres- ent study. First, looking doing latency is associated with a player’s game ‘race’. Given that the mechanics of gameplay differ significantly between races, it is not surprising that previous cross-sectional datasets revealed differences in performance speed by race [34]. Second, look- ing doing latency numbers are a timing measure and so interpreting transfer results in terms of the total continuum of skilled performance is more difficult than necessary. In the present work, we rescale looking-doing latencies into a measure of performance with the aim of addressing these two problems. Using the cross-sectional dataset of 3,307 games from Thompson, Blair, & Henrey [28], we fit the relationship between Looking-Doing Latency and League, taking into consideration game race. This analysis revealed that Zerg players have lower Looking-Doing-Latencies for any given league than the other two races, and that the relationship between Looking-Doing Latency and League was clearly linear. There was no apparent interaction between league, race, and looking doing latency. We therefore refit the models, one for Zerg players, and the other combining Protoss and Terran players, and used these models to transform Looking- Doing Latency into a league-based continuous scale for the present data that we call League- equivalent Performance. A score of 1 equates to a Looking-Doing Latency of the weakest level of online play: Bronze-League. A League-equivalent Performance score of 8 equates to profes- sional level play for that given race. Dependent variable: League-equivalent performance Performance data spanned a wide range of skill levels. Roughly 25% of our games included performance comparable to the lowest levels of skill (League-equivalent performance <1.5 = 7.4%; <2.5 = 13.5%; <3.5 = 23.3%), and a little more than a third of our games came from the highest levels (League-equivalent performance >5.5 = 37.3%). PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 Analysis 1: Transfer of performance between two major versions of StarCraft 2 Data in the shaded region show loss of performance when transferring. https://doi.org/10.1371/journal.pone.0295037.g001 Results from expansion analysis. (A) League-equivalent Performance for the games of participant #78. Green dots are games i l SC2 Wi f lib (W L) d i k d f h H f h S i (H TS) Li f h Fig 1. Results from expansion analysis. (A) League-equivalent Performance for the games of participant #78. Green dots are games of the original SC2 Wings of liberty (WoL), and pink dots are games of the Heart of the Swarm expansion (HoTS). Lines for each game type represent best fitting model of performance by game number. (B) Estimated performance for the next game using the original SC2 (WoL) model plotted against the estimated performance for the initial game using the HoTS expansion model. The size of each point reflects the total number of games used to estimate that player’s performance. The highlighted point is the estimate for the same player as panel A. Data in the shaded region show loss of performance when players transferred expansions. (C) Estimated transfer effect of the switch to the expansion (estimated performance WoL—HotS) plotted against total experience at the time of transfer. The highlighted point is the estimate for player #78, from panel A. Data in the shaded region show loss of performance when transferring. https://doi.org/10.1371/journal.pone.0295037.g001 https://doi.org/10.1371/journal.pone.0295037.g001 experienced players. This was examined with a significance test of the Pearson correlation coefficient between the number of games played before the expansion and the prediction-dif- ference transfer estimate. Analysis 1 revolves around the comparison of linear models before and after the release of the Heart of the Swarm expansion. Players need to have played at least 20 games before and 20 games after the expansion to be included in this analysis, leaving us with a sample size of 32 players and 19,312 games. The data of a single participant (#78 in our dataset) is shown in Fig 1A, with league-equiva- lent performance plotted for each successive game. The change from the original version of SC2 to the expansion version is indicated by a change in the dot colour. The best fitting linear models for each expansion are highlighted in black for visibility. A descriptive comparison of the model predictions is summarized in Fig 1B, where the proximity of data points to the refer- ence line suggests strong transfer. Analysis 1: Transfer of performance between two major versions of StarCraft 2 In analysis 1 we compared player performance (using the league-equivalent performance mea- sure for every game for every player) from before and after the release of the StarCraft 2: Heart of the Swarm expansion and examine whether transfer effects might vary by prior skill. To make full use of the dataset, we constructed two linear models for each player that predicted League-equivalent performance based on the number of games played—the first model used performance data from games prior to the expansion, and the second used data from games after the expansion. We then compared the model predictions for the first game played after the expansion. This allowed us to generate a prediction-difference transfer estimate for each player and to conduct one-sample t-test to establish whether the earliest StarCraft 2: Heart of the Swarm performance was in keeping with the skilled performance associated with StarCraft 2: Wings of Liberty. We hypothesized a non-zero prediction-difference transfer estimate, i.e., we expected the performance estimate from before the expansion to exceed the performance estimate from after, as the domain specificity of expertise implies that performance should suffer after such a major change. We then tested the predictions of the similarity-based and the maximal-adapta- tion based frameworks by examining whether transfer would be favourable for more 7 / 17 PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 PLOS ONE Robustness of performance during domain change in an esport Fig 1. Results from expansion analysis. (A) League-equivalent Performance for the games of participant #78. Green dots are games of the original SC2 Wings of liberty (WoL), and pink dots are games of the Heart of the Swarm expansion (HoTS). Lines for each game type represent best fitting model of performance by game number. (B) Estimated performance for the next game using the original SC2 (WoL) model plotted against the estimated performance for the initial game using the HoTS expansion model. The size of each point reflects the total number of games used to estimate that player’s performance. The highlighted point is the estimate for the same player as panel A. Data in the shaded region show loss of performance when players transferred expansions. (C) Estimated transfer effect of the switch to the expansion (estimated performance WoL—HotS) plotted against total experience at the time of transfer. The highlighted point is the estimate for player #78, from panel A. Analysis 1: Transfer of performance between two major versions of StarCraft 2 Predictions of early Heart of the Swarm Performance were similar regardless of whether predictions were generated by pre-expansion or post-expansion data. Fig 1C, which depicts the relationship between experience at the time of transfer and pre- diction-transfer-estimates, reveals little evidence of a transfer-experience interaction. We began the formal analysis with a one-sample t-test on prediction-difference transfer estimates. Sample standard deviations were low. This is not surprising given that the typical model prediction was based on 1,081 of games per player, the majority of which contain 253 first-action-latencies. Tight standard deviations lead to a respectable power to detect 1 league- equivalent performance difference (α = .05, 1- β ~99%, N = 32, δ = 1). Nevertheless, we failed to find significant prediction-differences between the models (t(31) = -1.193, p = 0.242, 95% CI: -0.36, 0.095). In short, performance in the first games of Heart of the Swarm performance did not seem to differ markedly from Wings of Liberty performance. We also conducted a correlation test on the Pearson correlation between the number of games played before the Heart of the Swarm expansion and the prediction-difference transfer estimate. We failed to find significant prediction-differences between the models (r = -.038, t (30) = -.209, p = 0.835, 95% CI: -0.38, 0.314), though power was only sufficient to detect a large effect size (α = .05, 1- β = ~85%, N = 32, δ = 0.5). Two points were influential, but excluding 8 / 17 PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 PLOS ONE Robustness of performance during domain change in an esport these points had no effect on the results. To ensure that these results were not due to an inclu- sive definition of experience (which includes many game modes), we also reran the aforemen- tioned analyses using a more restrictive definition of experience that included only 1v1 games created using the match-making system. The choice of definition had no discernable impact on the pattern of results. In summary, analysis 1 revealed evidence of strong transfer of perfor- mance following a major game update, and little evidence that transfer effects differ by prior experience. Analysis 2: Transfer between two different ‘races’ in StarCraft 2 Analysis 1 discovered strong transfer between the two different expansions of StarCraft 2. This is surprising given the domain specificity of expertise. In Analysis 2 we investigate even more substantive changes to gameplay. Unlike chess, where both players always control pieces that are equivalent sets of units, play- ers in StarCraft 2 often have their own units corresponding to which “race” a player chooses to command at the beginning of each match. Each race is equipped with distinct buildings, units, and game play mechanics, all working together to enable a diverse set of unique tactical possi- bilities which must be learned to gain proficiency in that race. Due to the significant differ- ences in the tactics, strategies, and timings that result from differences in how each race functions, players often choose one race to play as their as their dominant race, as playing with another race would require significant additional learning. As such, if a player chooses to use a race other than their dominant race in any given match, the difference in required skills for that race creates a transfer problem for that individual game. Given that a change of race implies changes to so many aspects to gameplay, we expect weaker transfer of performance than in Analysis 1. than in Analysis 1. Analysis 2 brought new data analytic challenges. Unlike Analysis 1, which focused on a sin- gle point in time (i.e. the release of StarCraft 2: Heart of the Swarm), players choose their race at the beginning of each match, and could therefore change what race they play as at any point in time. Consequently, a more sophisticated analysis using mixed effects models was required. Analysis 2 focuses on two independent variables, a training/transfer variable that differentiates the games using the players dominant race (the one they use most often) from the transfer games where players are using a different race (often called “playing off-race”), and an experi- ence variable that quantifies that players amount of experience at the time of play. A model without the training/transfer variable, which makes predictions solely based on general experi- ence, effectively assumes perfect transfer: predictions are not adjusted if a player chooses to play their non-dominant race. Our first research question asks whether a model that includes the training/transfer variable predict performance (once again using our league-equivalent performance measure derived from looking-doing-latency) better than a model without it. Analysis 2: Transfer between two different ‘races’ in StarCraft 2 Our second, and more central research question, is investigated by adding an interaction term to our model to determine if players with more experience have more favourable transfer per- formance than players with less experience. Interestingly, players exhibited a strong preference for a dominant race, and many players avoid off-race games. To be included in analysis 2, play- ers needed to play at least 30 off-race games and at least 50 games overall. One player was dropped for having an even distribution of races. This left us with 18 players and 14,625 games for our analysis. The fact that so few players qualified for this analysis is evidence of participant reluctance to play off-race. Analysis 2 brought new data analytic challenges. Unlike Analysis 1, which focused on a sin- gle point in time (i.e. the release of StarCraft 2: Heart of the Swarm), players choose their race at the beginning of each match, and could therefore change what race they play as at any point in time. Consequently, a more sophisticated analysis using mixed effects models was required. Analysis 2 focuses on two independent variables, a training/transfer variable that differentiates the games using the players dominant race (the one they use most often) from the transfer games where players are using a different race (often called “playing off-race”), and an experi- ence variable that quantifies that players amount of experience at the time of play. A model without the training/transfer variable, which makes predictions solely based on general experi- ence, effectively assumes perfect transfer: predictions are not adjusted if a player chooses to play their non-dominant race. Our first research question asks whether a model that includes the training/transfer variable predict performance (once again using our league-equivalent performance measure derived from looking-doing-latency) better than a model without it. Our second, and more central research question, is investigated by adding an interaction term to our model to determine if players with more experience have more favourable transfer per- formance than players with less experience. Interestingly, players exhibited a strong preference for a dominant race, and many players avoid off-race games. To be included in analysis 2, play- ers needed to play at least 30 off-race games and at least 50 games overall. One player was dropped for having an even distribution of races. This left us with 18 players and 14,625 games for our analysis. Analysis 2: Transfer between two different ‘races’ in StarCraft 2 (C) Estimated transfer effect of the switch to Off Race (Dominant Race—Off Race) plotted against total experience. The highlighted point is the estimate for player #78, from panel A. Data in the shaded region show loss of performance when playing Off Race. https://doi.org/10.1371/journal.pone.0295037.g002 in black for visibility. A descriptive summary of our results can be found in Fig 2B. The prox- imity of dominant race performance and off race performance to the reference line, is again suggestive of strong transfer. Fig 2C shows what may be a weak transfer-experience interac- tion, as less experienced players seem to be suffering greater drops in performance when they play off-race. Although Analysis 1 only had power to detect a large interaction effect between training and experience, Analysis 2 has power to detect considerably smaller effects. To assess the power to detect an interaction effect, we used the simr package in R [35]. A range of effect sizes were selected a priori. However, simulating power for mixed effects models requires spec- ification of a covariance structure, which is not supplied by psychological theory or the Star- Craft 2 community. We therefore calculated power in a post-hoc fashion, by using the observed covariance structures derived from Analysis 2 below. We defined a small interaction effect as a reduced transfer cost of one tenth of a league of skill after 1,000 games of experience. Simulation results suggested that we have respectable power even for detection of small effects (α = 0.05; 1- β = 94.5%, 95% CI [92.9, 95.83], δ = .01). We regressed the number of competitive games experienced onto League-equivalent Per- formance in a mixed effect model, using a random intercept for player using R [36], with the lme4 package by Bates, Maechler & Bolker [37]. We compared this model to more complex models built with all prior significant effects; each of the nested models were compared using likelihood ratio tests. No games were found to be overly influential (Di<1). The first model was League-equivalent Performance predicted by number of competitive games with a random intercept for each player. Analysis 2: Transfer between two different ‘races’ in StarCraft 2 The fact that so few players qualified for this analysis is evidence of participant reluctance to play off-race. dropped for having an even distribution of races. This left us with 18 players and 14,625 games for our analysis. The fact that so few players qualified for this analysis is evidence of participant reluctance to play off-race. The data of a single participant (#78 in our dataset) is shown in Fig 2A, with league-equiva- lent performance plotted for each successive game. Dominant race games and off race games are in differing dot colours. The best fitting linear models for each expansion are highlighted PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 9 / 17 PLOS ONE Robustness of performance during domain change in an esport Fig 2. Results for race analysis. (A) League-equivalent Performance for the games of participant #78. Blue dots are games of the players main race, and orange dots are off race games. Lines for each game type represent best fitting model of performance by game number. (B) Estimated performance for the model of Dominant Race games plotted against the estimated performance using the model of Off Race games. The size of each point reflects the total number of games used to estimate that player’s performance. The highlighted point is the estimate for player #78, from panel A. Data in the shaded region show loss of performance when playing Off Race. (C) Estimated transfer effect of the switch to Off Race (Dominant Race—Off Race) plotted against total experience. The highlighted point is the estimate for player #78, from panel A. Data in the shaded region show loss of performance when playing Off Race. https://doi.org/10.1371/journal.pone.0295037.g002 Fig 2. Results for race analysis. (A) League-equivalent Performance for the games of participant #78. Blue dots are games of the players main race, and orange dots are off race games. Lines for each game type represent best fitting model of performance by game number. (B) Estimated performance for the model of Dominant Race games plotted against the estimated performance using the model of Off Race games. The size of each point reflects the total number of games used to estimate that player’s performance. The highlighted point is the estimate for player #78, from panel A. Data in the shaded region show loss of performance when playing Off Race. Analysis 2: Transfer between two different ‘races’ in StarCraft 2 Each subsequently added effect was found to significantly increase the fit of the model: a random slope for each player (χ2 = 2201.9, p<0.001), our dichotomous training/transfer variable (χ2 = 24.911, p<0.001) (Beta = 0.26, 95% CI [0.179, 0.339]), and an interaction effect between number of competitive games and dominant or off-race games (χ2 = 16.666, p<0.001) (Beta = -0.012, 95% CI [-0.017, -0.006]). Our observed effect size was for this interaction was that those playing off-race preserved an PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 10 / 17 PLOS ONE Robustness of performance during domain change in an esport additional 0.12 leagues for every 1,000 games of experience possessed. Given that our best esti- mate of the transfer cost of playing off-race is about 0.26 leagues of skill, we expect it would typically take about 2,000 games of experience to eliminate the transfer costs associated with playing off race. This interaction effect is small given that 1,000 games seems to have a main effect of roughly 1.3 leagues of overall improvement. One challenge with this analysis was the definition of player experience. StarCraft 2 has a variety of game modes, and so it is sometimes unclear what sort of experience should be counted as prior experience. Following Thompson [30] experience was defined to include both 1 versus 1 and team games. In other words, we define experience as any game of StarCraft 2 that passed our initial exclusion criteria (~118,000 games), regardless of game mode. Given that the topic of the present study was transfer, we also reran analysis 2 with a more restrictive definition. In this case, we restricted experience to games with exactly one participant and one opponent using the games match-making system. Using this alternate definition of experience led to worse model fits overall. This fact is consistent with the generally strong transfer we see across variations in the game, that is, team games seem like valid examples of experience and excluding them weakens the model fits. Regardless, in these new analyses, models were improved by the addition of a random slope (χ2 = 1896.5, p<0.001) and a dichotomous train- ing/transfer variable (χ2 = 24.4, p<0.001), but, unlike the initial version, the interaction term did not significantly improve models (χ2 = 0.282, p = 0.596). Analysis 2: Transfer between two different ‘races’ in StarCraft 2 Our best interpretation of this sit- uation is that, while the superior models support the predicted finding that experience modu- lates transfer, this finding is neither strong, nor especially robust. Analysis 3: Transfer between StarCraft 1 and StarCraft 2 In Analysis 2 we again found strong transfer, defying the general expectation of domain spe- cific performance. Given the presence of strong transfer across the three StarCraft 2 races, a transition involving entirely different units and mechanics, one might wonder what sort of intra-domain transfer could have a meaningful impact on performance. Any player of Star- Craft 2 would recognize that it was modelled after its predecessor. Both games have dedicated full-time professional players and both are recognized as Esports in the scientific literature [38]. The games are also similar enough to enjoy some overlap in their communities, as evi- denced by the fact that Esport websites covering StarCraft 2 often report on StarCraft 1 as well. Furthermore, there are individuals capable of professional play in both games [39, 40]. In short, to two games are different enough to be considered separate domains but similar enough for some transfer to be possible. To gauge farther transfer, our final analysis examined whether the first fifty games of Star- Craft 2 performance can be predicted by self-reported experience in StarCraft 1, the game to which StarCraft 2 is a sequel. Self-reports of experience were obtained by asking participants the year in which they began and ended playing StarCraft 1 and calculating the difference. Our change from the preferred direct measure of performance in Analyses 1 and 2, to a measure of self-reported experience was out of necessity. Not only are StarCraft 1 replays unavailable for most players, but even if the records could be collected, the data contained in StarCraft 1 replays is more limited and does not allow for the calculation of looking-doing latency or lea- gue-equivalent performance. While the previous two analyses involved a comparison of performance based on game recordings, the analogous data are not available for StarCraft 1. In this final analysis we look at whether experience in StarCraft 1 is predictive of skill in StarCraft 2 by adding self-reported years of experience with StarCraft 1 to our best model of the first fifty games of StarCraft 2 per- formance and evaluate this updated model with a likelihood ratio test. 11 / 17 PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 PLOS ONE Robustness of performance during domain change in an esport Fig 3. Results for StarCraft 1 analysis. Mean League-equivalent performance by self-reported years of SC1 experience for each player’s first 50 SC2 games. Analysis 3: Transfer between StarCraft 1 and StarCraft 2 Statistical inference was based on mixed effects models that are not depicted in this figure. See text for details. https://doi.org/10.1371/journal.pone.0295037.g003 Fig 3. Results for StarCraft 1 analysis. Mean League-equivalent performance by self-reported years of SC1 experience for each player’s first 50 SC2 games. Statistical inference was based on mixed effects models that are not depicted in this figure. See text for details. Fig 3. Results for StarCraft 1 analysis. Mean League-equivalent performance by self-reported years of SC1 experience for each player’s first 50 SC2 games. Statistical inference was based on mixed effects models that are not depicted in this figure. See text for details. https://doi.org/10.1371/journal.pone.0295037.g003 https://doi.org/10.1371/journal.pone.0295037.g003 Participants in Analysis 3 satisfied our general exclusion criteria and played at least 3 com- petitive 1-versus-1 matches in their first 25 games. A total of 97 players satisfied these criteria, and 93 also reported when they began and ended their play of StarCraft 1. A depiction of the observed relationship between self-reported StarCraft experience and StarCraft 2 performance are summarized in Fig 3. We found no significant relationship between the number of years of StarCraft 1 experience and the first three competitive games of StarCraft 2 performance (r = 0.157, t(30) = 1.51, df = 91, p = 0.136, 95% CI: -0.05, 0.350), despite respectable power (α = .05, 1- β = ~99%, N = 93, δ = 0.5). There is little evidence of an association between StarCraft 1 experience and early StarCraft 2 performance. PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 Discussion This chunk-centric perspec- tive suggests that expert performance is due to highly specific, and non-transferrable action sequences that become mastered throughout experience. However, despite the efforts of prior work [34] to identify chunks and measure their impact on performance, better StarCraft 2 players were faster regardless of the action sequences being performed. The present findings are more general and suggest that the players who are expected to be the most specialized (i.e., experts) were, contrary to the maximal adaptation framework, the most resilient to transfer costs. One of the puzzles of the present data are the surprisingly small transfer costs observed across all participants. Analysis 1 showed remarkable transfer. Predictions of post-expansion performance were virtually the same, regardless of whether the model was built from pre or post-expansion performance. There were consequently no transfer costs for experience to ameliorate. The stronger transfer costs for non-dominant race play, as described in Analysis 2, were not unexpected given that changing race impacts more pieces and gameplay mechanics. Neverthe- less, effect sizes were surprisingly weak. Playing a non-dominant race was associated with about a quarter of a league in terms of performance. This is quite small relative to the typical level of league-equivalent performance in our sample, which was about 5. When taken into consideration alongside Analysis 3, our results suggest impressive transfer within the domain of StarCraft 2, but not between StarCraft 2 and its predecessor. One explanation for generally impressive transfer is that the novices in our sample are already too knowledgeable about StarCraft 2 to suffer a within-domain transfer cost. This explanation, while technically consistent with the similarity-based approach, would undercut its utility as an approach to predicting transfer. The below-average players in our sample each have around 250 hours of experience and includes players with a league equivalent perfor- mance of 3.8, placing them close to what is known as ‘platinum league’ on a scale that, includ- ing professional play, contains 8 levels of skill in total. Even this level of performance would not be considered exceptional, or even intermediate, in the StarCraft 2 community. Impor- tantly, we do not claim our results are inconsistent with this possibility. Instead, given the ubiquity of transfer costs in the learning literature, we think it implausible that the skill of our novices alone could explain the weak or absent transfer costs that we observe here. Discussion Transfer effects are hard to predict. Perhaps not surprisingly, observations of transfer have played an important role in falsifying theories of learning [2, 3]. The topic of transfer continues to be a central to theories of learning, expertise, motor skills and categorization. In the present study we investigated transfer effects in a real time strategy game using replay files that docu- ment performance, often over hundreds of hours of experience. This research method has the advantage of allowing us to assess transfer using identical performance measures during train- ing and transfer. Further, we have a sense of how observed performance levels fit within the spectrum of expertise on a given task, and can thus assess transfer losses in relation to total skill in the domain. This allows us a unique opportunity to evaluate predictions about transfer in complex task domains. The specific predictions we focused on here are in relation to one potential moderating factor of transfer costs: prior experience. Frameworks predicting transfer based on perceived task-similarity [1] have suggested that more experienced individuals PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 12 / 17 PLOS ONE Robustness of performance during domain change in an esport should achieve better transfer, while those frameworks that construe experts as maximally adapted to a single task [23] suggest the opposite. We documented transfer costs when players played a race other than their dominant one (Analysis 2), and these costs seemed slightly larger for novices. An additional 1,000 games of experience (which corresponds to roughly 250 hours of play) was associated with a reduced transfer cost of 0.12 leagues when players changed gameplay race, and our best models suggest that 2,000 games would be enough to ameliorate most of the transfer costs of playing off-race. Our general pattern of results therefore fit best with the similarity-based framework described in [1]. Of course, this apparent theoretical victory is diminished by a weak, and not particularly robust, interaction effect between transfer cost and experience. In contrast, it seems like StarCraft 2 performance does not fit well with the concept of maxi- mal adaptation. Previous research provided evidence against one application of the maximal adaptation framework to motor chunking in StarCraft 2. Specifically, prior research examined the idea that the impressive speeds of StarCraft 2 players should be attributable to a small num- ber of extremely specialized, and overlearned, motor-chunks [27]. PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 Discussion If novitiate levels of skill were sufficient to ameliorate such costs, domain specificity should be very rare in the natural world. Instead, we find it more likely that, if our StarCraft 2 novices are somehow using their knowledge to circumvent transfer costs, then there must be something unique to this domain which makes such transfer more impressive than usual. PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 13 / 17 PLOS ONE Robustness of performance during domain change in an esport A second, more plausible, explanation for our results is that there is some feature of Star- Craft 2 that instills resilience in player performance. One such feature may be the changing task environment of esports. Most domains are stable, like chess, with changes being both rare, and minor. In esports, software developers are constantly changing the parameters and rules of the game to ensure competitive play that is entertaining for viewers. Indeed, esports com- munities have even developed a common language to describe these changes. For example, when companies ‘balance’ the game, some methods of play receive an advantage or ‘buff,’ while others become less effective, or are ‘nerfed’ [41], and the over pattern of strong strategies is called the ‘meta’. It is plausible that playing against a backdrop of constant change in the task has yielded impressive protection against the sort of transfer costs examined in Analysis 1. A final, and related, possibility is that player decision making is impacting their resilience to transfer costs. StarCraft 2 players have always been able to choose their own race and therefore have control over this aspect of their task environment. Furthermore, earlier studies of age- related change in StarCraft 2 suggested that older players, while slower, can adapt their play style to remain competitive [33]. In Analysis 2 of the present study, we noticed that many play- ers were clearly avoiding certain races (see exclusion criteria for analysis 2) and those who did play off-race exhibited transfer costs that were ameliorated by experience. Our evidence there- fore fits with the picture of a cognitive system that, in its attempt to win games and improve, is attempting to dynamically manage task difficulty and transfer effects by exerting control over the task environment (e.g., by adjusting race, strategy, or play-style). There are some important limitations of our work. Discussion First, since data was collected based on files donated by players, it is not possible to verify that players submitted all of the digital rec- ords associated with their gameplay. Given that records were collected automatically by the game, it seems unlikely that players would go to the effort selectively donate unrepresentative games, such as games where they performed well. However, it is possible that a player could, for example, have lost a large batch of their noviciate performance due to a hard drive failure. This sort of data, which we would consider missing at random, could yield misleading learning curves for some players. Secondly, it is important to recognize that the study of transfer needs to be separated from the question of generalizability. While we study transfer between different kinds of gameplay within the game of StarCraft 2, we make no claims about whether similarly powerful transfer effects would be observed within other games or domains. It is possible that there is something special about StarCraft 2 task environment that produces strong transfer effects. It is also possible that strong within-domain transfer, is common but undetected by older research methods. In 1973, Newell [42] advocated for the analysis of a complex task as a way forward for cognitive psychology. Such an endeavour requires integrating the diaspora of findings in cognitive science into a larger theoretical network that can be evaluated for its capacity to predict and explain behaviour. Our results confirm that such a unified framework is not yet available. However, recent advances in our ability to collect data on complex tasks through telemetry [31, 43–45] provides cognitive science with the opportunity to identify new phenomena to be explained. We hope this incentivizes comprehensive theoretical approaches with the capacity to make clear and nuanced predictions about the presence and magnitude of transfer costs. Supporting information S1 Text. Supplementary definitions. (PDF) Supporting information g S1 Text. Supplementary definitions. (PDF) 14 / 17 PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 PLOS ONE Robustness of performance during domain change in an esport Author Contributions Conceptualization: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Scott Harrison, Mark R. Blair. Conceptualization: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Scott Harrison, Mark R. Blair. Data curation: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Scott Harrison, Mark R. Blair. Formal analysis: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Scott Harrison, Mark R. Blair. Funding acquisition: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Mark R. Blair. Investigation: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Scott Harrison, Mark R. Blair. Methodology: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Scott Harrison, Mark R. Blair. Project administration: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Scott Harrison, Mark R. Blair. Resources: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Scott Harrison, Mark R. Blair. Software: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Scott Harrison, Mark R. Blair. Supervision: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Mark R. Blair. Validation: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Scott Harrison, Mark R. Blair. Visualization: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Scott Harrison, Mark R. Blair. Writing – original draft: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Scott Harrison, Mark R. Blair. Writing – review & editing: Joe Thompson, Justin W. O’Camb, Robin C. A. Barrett, Scott Harrison, Mark R. Blair. Acknowledgments We would like to acknowledge the support of the Cognitive Science Laboratory at SFU, which was crucial for the present project. We have especial gratitude to Cal Woodruff for their sup- port in data processing and comments on the manuscript. We would also like to thank Nick Li and Jagpreet Kaur for their feedback in the creation of the manuscript. References 1. Kimball DR, Holyoak KJ. Transfer and Expertise. In: Tulving E, editor. The Oxford Handbook of Mem- ory. Oxford University Press; 2000. pp. 109–122. 2. Woodworth RS, Thorndike EL. The influence of improvement in one mental function upon the efficiency of other functions. (I). Psychol Rev. 1901; 8: 247–261. https://doi.org/10.1037/h0074898 3. Adams J. Historical review and appraisal of research on the learning, retention, and transfer of human motor skills. 1987. https://doi.org/10.1037/0033-2909.101.1.41 4. Chase WG, Simon HA. Perception in chess. Cognit Psychol. 1973; 4: 55–81. https://doi.org/10.1016/ 0010-0285(73)90004-2 4. Chase WG, Simon HA. Perception in chess. Cognit Psychol. 1973; 4: 55–81. https://doi.org/10.1016/ 0010-0285(73)90004-2 PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 15 / 17 PLOS ONE Robustness of performance during domain change in an esport 5. Keetch KM, Lee TD, Schmidt R. Especial skills: specificity embedded within generality. J Sport Exerc Psychol. 2008. https://doi.org/10.1123/jsep.30.6.723 PMID: 19164838 6. Williams AM, Ward P, Ward JD, Smeeton N. Domain specificity, task specificity, and expert perfor- mance. Res Q Exerc Sport. 2008; 79: 428–433. https://doi.org/10.1080/02701367.2008.10599509 PMID: 18816957 7. Kretch KS, Adolph KE. Cliff or Step? Posture-Specific Learning at the Edge of a Drop-Off. Child Dev. 2013; 84: 226–240. https://doi.org/10.1111/j.1467-8624.2012.01842.x PMID: 22906143 8. Bassok M, Holyoak K. Interdomain Transfer Between Isomorphic Topics in Algebra and Physics. 1989. https://doi.org/10.1037/0278-7393.15.1.153 9. Smith JD, Redford JS, Washburn DA, Taglialatela LA. Specific-Token Effects in Screening Tasks: Pos- sible Implications for Aviation Security. J Exp Psychol Learn Mem Cogn. 2005; 31: 1171–1185. https:// doi.org/10.1037/0278-7393.31.6.1171 PMID: 16393038 10. Sims V, Mayer R. Domain specificity of spatial expertise: The case of video game players. 2002. https:// doi.org/10.1002/ACP.759 11. Meneghetti C, Borella E, Pazzaglia F. Mental rotation training: transfer and maintenance effects on spa- tial abilities. Psychol Res. 2016; 80: 113–127. https://doi.org/10.1007/s00426-014-0644-7 PMID: 25575755 12. Bransford JD, Schwartz DL. Rethinking Transfer: A Simple Proposal with Multiple Implications. Rev Res Educ. 1999; 24: 61–100. https://doi.org/10.2307/1167267 13. Gegnefurter A, Seppa¨nen M. Transfer of expertise: An eye tracking and think aloud study using dynamic medical visualizations. Comput Educ. 2013; 63: 393–403. https://doi.org/10.1016/j.compedu. 2012.12.021 14. Bainbridge K, Mayer RE. Shining the Light of Research on Lumosity. J Cogn Enhanc. 2018; 2: 43–62. https://doi.org/10.1007/s41465-017-0040-5 15. Shute V, Ventura M, Ke F. The power of play: The effects of Portal 2 and Lumosity on cognitive and non- cognitive skills. Comput Educ. 2015. https://doi.org/10.1016/j.compedu.2014.08.013 16. Sala G, Tatlidil KS, Gobet F. Video game training does not enhance cognitive ability: A comprehensive meta-analytic investigation. References Psychol Bull. 2018; 144: 111–139. https://doi.org/10.1037/bul0000139 PMID: 29239631 17. Nguyen L, Murphy K, Andrews G. A Game a Day Keeps Cognitive Decline Away? A Systematic Review and Meta-Analysis of Commercially-Available Brain Training Programs in Healthy and Cognitively Impaired Older Adults. Neuropsychol Rev. 2021. https://doi.org/10.1007/s11065-021-09515-2 PMID: 34251578 18. Lassonde KA, Osborn RM. Lumosity Does Not Best Classroom Memory Improvement Strategies. Scholarsh Teach Learn Psychol. 2019. https://doi.org/10.1037/stl0000125 19. Lumosity to Pay $2 Million to Settle FTC Deceptive Advertising Charges for Its “Brain Training” Pro- gram. In: Federal Trade Commission [Internet]. 4 Jan 2016 [cited 3 Jun 2022]. Available: http://www.ftc. gov/news-events/news/press-releases/2016/01/lumosity-pay-2-million-settle-ftc-deceptive-advertising- charges-its-brain-training-program 20. Lewis D, Miles GH. Retroactive Interference in Performance on the Star Discrimeter as a Function of Amount of Interpolated Learning. Percept Mot Skills. 1956; 6: 295–298. https://doi.org/10.2466/pms. 1956.6.3.295 21. Abernethy B, Baker J, Cȏte´ J. Transfer of pattern recall skills may contribute to the development of sport expertise. 2005. https://doi.org/10.1002/ACP.1102 22. Dunlosky J, Rawson KA, Marsh EJ, Nathan MJ, Willingham DT. Improving Students’ Learning With Effective Learning Techniques: Promising Directions From Cognitive and Educational Psychology. Psy- chol Sci Public Interest J Am Psychol Soc. 2013; 14: 4–58. https://doi.org/10.1177/1529100612453266 PMID: 26173288 23. Ericsson KA, Lehmann A. Expert and exceptional performance: evidence of maximal adaptation to task constraints. Annu Rev Psychol. 1996. https://doi.org/10.1146/annurev.psych.47.1.273 PMID: 15012483 24. Medin DL, Goldstone RL, Gentner D. Respects for Similarity. Psychol Rev. 1993; 100: 254–278. https:// doi.org/10.1037/0033-295X.100.2.254 25. Gick ML, Holyoak KJ. CHAPTER 2—The Cognitive Basis of Knowledge Transfer. In: Cormier SM, Hag- man JD, editors. Transfer of Learning. San Diego: Academic Press; 1987. pp. 9–46. https://doi.org/10. 1016/B978-0-12-188950-0.50008–4 26. Adolph KE. Learning to Move. Curr Dir Psychol Sci. 2008; 17: 213–218. https://doi.org/10.1111/j.1467- 8721.2008.00577.x PMID: 19305638 16 / 17 PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 PLOS ONE Robustness of performance during domain change in an esport 27. De Groot AD. Thought and choice in chess. Thought and Choice in Chess. 2014. p. 464. 28. Thompson JJ, Blair MR, Chen L, Henrey AJ. Video Game Telemetry as a Critical Tool in the Study of Complex Skill Learning. de Beeck HPOp, editor. PLoS ONE. 2013; 8: e75129. https://doi.org/10.1371/ journal.pone.0075129 PMID: 24058656 29. McColeman C, Thompson J, Anvari N, Azmand SJ, Barnes J, Barrett RCA, et al. Digit eyes: Learning- related changes in information access in a computer game parallel those of oculomotor attention in lab- oratory studies. Atten Percept Psychophys. 2020; 82: 2434–2447. https://doi.org/10.3758/s13414-020- 02019-w PMID: 32333371 30. References Thompson JJC. Use of digital records for studying skill learning. Thesis, Arts & Social Sciences: Depart- ment of Psychology. 2018. Available: https://summit.sfu.ca/item/18032 31. Thompson JJ, McColeman CM, Stepanova ER, Blair MR. Using Video Game Telemetry Data to Research Motor Chunking, Action Latencies, and Complex Cognitive-Motor Skill Learning. Top Cogn Sci. 2017; 9: 467–484. https://doi.org/10.1111/tops.12254 PMID: 28176483 32. Land M, Hayhoe M. In what ways do eye movements contribute to everyday activities? Vision Res. 2001. https://doi.org/10.1016/s0042-6989(01)00102-x PMID: 11718795 33. Thompson JJ, Blair MR, Henrey AJ. Over the Hill at 24: Persistent Age-Related Cognitive-Motor Decline in Reaction Times in an Ecologically Valid Video Game Task Begins in Early Adulthood. PLOS ONE. 2014; 9: e94215. https://doi.org/10.1371/journal.pone.0094215 PMID: 24718593 34. Thompson JJ, McColeman CM, Blair MR, Henrey AJ. Classic motor chunking theory fails to account for behavioural diversity and speed in a complex naturalistic task. PloS One. 2019; 14: e0218251. https:// doi.org/10.1371/journal.pone.0218251 PMID: 31194810 35. Green P, MacLeod CJ. SIMR: an R package for power analysis of generalized linear mixed models by simulation. Methods Ecol Evol. 2016; 7: 493–498. https://doi.org/10.1111/2041-210X.12504 36. R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Founda- tion for Statistical Computing; 2022. Available: https://www.R-project.org/ 37. Bates D, Ma¨chler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using lme4. J Stat Softw. 2015; 67: 1–48. https://doi.org/10.18637/jss.v067.i01 38. Pedraza-Ramirez I, Musculus L, Raab M, Laborde S. Setting the scientific stage for esports psychology: a systematic review. Int Rev Sport Exerc Psychol. 2020; 13: 319–352. 39. Bigfan. Jaedong: Never Give Up! In: Team Liquid [Internet]. 26 Feb 2018 [cited 3 Jun 2022]. Available: https://tl.net/forum/brood-war/531589-jaedong-never-give-up 40. Favie III C. The best female esports players, ranked. In: Clutchpoints [Internet]. 14 Nov 2020 [cited 3 Jun 2022]. Available: https://clutchpoints.com/the-best-female-esports-players-ranked/ 41. Seo Y. Professionalized consumption and identity transformations in the field of eSports. J Bus Res. 2016; 69: 264–272. https://doi.org/10.1016/j.jbusres.2015.07.039 42. Newell A. You Can’t Play 20 Questions with Nature and Win: Productive Comments on the Papers of this Symposium. Visual Information Processing. Elsevier; 1973. pp. 283–308. https://doi.org/10.1016/ B978-0-12-170150-5.50012–3 43. Goldstone RL, Lupyan G. Discovering Psychological Principles by Mining Naturally Occurring Data Sets. Top Cogn Sci. 2016; 8: 548–568. https://doi.org/10.1111/tops.12212 PMID: 27404718 44. Stafford T, Vaci N. Maximizing the Potential of Digital Games for Understanding Skill Acquisition. Curr Dir Psychol Sci. 2022. https://doi.org/10.1177/09637214211057841 45. Huang J, Yan E, Cheung G, Nagappan N, Zimmermann T. PLOS ONE \| https://doi.org/10.1371/journal.pone.0295037 December 7, 2023 References Master Maker: Understanding Gaming Skill Through Practice and Habit From Gameplay Behavior. Top Cogn Sci. 2017; 9: 437–466. https://doi.org/ 10.1111/tops.12251 PMID: 28198102 17 / 17 17 / 17
https://openalex.org/W2580489284	https://bmcbioinformatics.biomedcentral.com/track/pdf/10.1186/s12859-017-1478-2	English	null	Dynamic substrate preferences predict metabolic properties of a simple microbial consortium	BMC bioinformatics	2,017	cc-by	9,212	© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Erbilgin et al. BMC Bioinformatics (2017) 18:57 DOI 10.1186/s12859-017-1478-2 Erbilgin et al. BMC Bioinformatics (2017) 18:57 DOI 10.1186/s12859-017-1478-2 Open Access * Correspondence: trnorthen@lbl.gov 1Environmental Genomics and Systems Biology Division, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720, USA 2Joint Genome Institute, 2800 Mitchell Dr, Walnut Creek, CA 94598, USA Full list of author information is available at the end of the article Dynamic substrate preferences predict metabolic properties of a simple microbial consortium Onur Erbilgin1, Benjamin P. Bowen1,2, Suzanne M. Kosina1, Stefan Jenkins1,3, Rebecca K. Lau1 and Trent R. Northen1,2* Abstract Background: Mixed cultures of different microbial species are increasingly being used to carry out a specific biochemical function in lieu of engineering a single microbe to do the same task. However, knowing how different species’ metabolisms will integrate to reach a desired outcome is a difficult problem that has been studied in great detail using steady-state models. However, many biotechnological processes, as well as natural habitats, represent a more dynamic system. Examining how individual species use resources in their growth medium or environment (exometabolomics) over time in batch culture conditions can provide rich phenotypic data that encompasses regulation and transporters, creating an opportunity to integrate the data into a predictive model of resource use by a mixed community. Results: Here we use exometabolomic profiling to examine the time-varying substrate depletion from a mixture of 19 amino acids and glucose by two Pseudomonas and one Bacillus species isolated from ground water. Contrary to studies in model organisms, we found surprisingly few correlations between resource preferences and maximal growth rate or biomass composition. We then modeled patterns of substrate depletion, and used these models to examine if substrate usage preferences and substrate depletion kinetics of individual isolates can be used to predict the metabolism of a co-culture of the isolates. We found that most of the substrates fit the model predictions, except for glucose and histidine, which were depleted more slowly than predicted, and proline, glycine, glutamate, lysine and arginine, which were all consumed significantly faster. Conclusions: Our results indicate that a significant portion of a model community’s overall metabolism can be predicted based on the metabolism of the individuals. Based on the nature of our model, the resources that significantly deviate from the prediction highlight potential metabolic pathways affected by species-species interactions, which when further studied can potentially be used to modulate microbial community structure and/or function. Keywords: Microbiology, Quantitative metabolomics, Substrate preferences, Predicting community function Background following rainfall, light–dark cycles, digestion in animals, etc. Additionally, some biotechnologies that use microor- ganisms are also batch processes, such as the large-scale fermentations of microbe-processed foods (e.g. cheese, wine, etc.). Most of these processes use mixed microbial cultures, including one-pot processes of biomass conver- sion to biofuels and other biosynthetic products [2–4]. Studying the temporal substrate utilization by individuals is an important first step in developing approaches to better model these biochemical processes. While some work on mixed-substrate growth has been performed in continuous culture at steady state [1], under- standing substrate usage and competition in batch cultures may have both ecological and practical applica- tions. Many environmental processes happen with pulsed inputs: for example the release of substrates into the soil * Correspondence: trnorthen@lbl.gov 1Environmental Genomics and Systems Biology Division, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720, USA 2Joint Genome Institute, 2800 Mitchell Dr, Walnut Creek, CA 94598, USA Full list of author information is available at the end of the article Erbilgin et al. BMC Bioinformatics (2017) 18:57 Page 2 of 12 Page 2 of 12 Exometabolomics, also known as metabolic footprinting, is a powerful platform for studying how microbes and their consortia modify substrate pools, as analysis is only of the extracellular metabolites [5]. With the development of exometabolomics pipelines, the metabolic connections be- tween microbes have begun to be studied at a large scale and have allowed for a more comprehensive approach to monitoring the dynamic transformations of relatively complex mixtures of substrates [5]. Some key examples in- clude optimizing multiple steps of lignocellulose degrad- ation [6, 7], understanding metabolic interactions between species in mixed communities [8], and determining the ecological role of individuals within a mixed commu- nity [9–11]. We have recently found exometabolite niche partitioning in two soil environments where sympatric microbes were found to target largely non- overlapping portions of the available substrates, thus min- imizing substrate competition [10]. These experiments were focused on the endpoint depletion of substrates by isolates, not the temporal sequence of utilization. However, the order of substrate utilization (i.e. substrate preferences) may further discriminate the adaptive strategies of individ- ual organisms for common substrates. substrates in an environment, these parameters have great potential in providing a direct measure of an organism’s substrate preferences within that environ- ment, effectively creating a resource usage model for the organism. Background When taken into consideration with other spe- cies’ models, they may enable the prediction of the overall net metabolism of microbial consortia by aggregating indi- vidual contributions to environmental substrate usage. Observed deviations from these predictions could help identify interspecies interactions that modulate an organ- ism’s metabolism, e.g. communication and antagonism between microbes within communities. Here we compare the temporal depletion of 20 sub- strates by three isolates and fit these data to the Behrends model (Eq. 1), describing their substrate preferences within this ‘environment’. We then examine if the first substrates depleted result in maximal growth rate, or re- late to growth medium or biomass composition. Finally, we developed a model that simply combines the usage profiles of individual species to test if a consortium ini- tially composed of an equal mixture of each of the three isolates consumes substrates in an identical manner to when they are grown individually, i.e. the presence of other microbes does not affect their substrate usage. Any deviations from this model may indicate compounds that are actively regulated. For example, if a compound is con- sumed significantly faster or earlier than predicted by the model, this would indicate an additional interaction between species such as synergistic or competitive growth. g In addition to exometabolomics, several genomics- enabled analyses have been used to model cellular metabolism and metabolic interactions between species in mixed communities [12]. However, this type of analysis relies on the availability of a sequenced genome for each organism, and for the genome to be properly annotated. No genome annotation is perfect; there are both false positives (presence of a function when there is none) and negatives (assertion that a function is missing when it is not). There is also the issue of genomes that harbor a sig- nificant number of genes of unknown or hypothetical function. These genomes may very well harbor full bio- chemical pathways that cannot be predicted based on our current databases. New biochemical pathways are con- stantly being discovered and characterized, and the pres- ence of these novel pathways would drastically alter the reconstructed metabolic network of a species. Further- more, while these models can be used to investigate the space of potential interspecies interactions without opti- mizing an objective function (e.g. Background biomass or ATP produc- tion), predicting which interactions actually happen in the environment require optimization to predict what the in- and out-fluxes would be, and thus how the metabolisms of different species would network together. Here we have focused on developing a model based on experimental data of how microbial species deplete resources over time, with no assumptions made based on genomic data. This modeling approach is a first step to uncovering the fundamental metabolic interactions within microbial communities. It serves as a test: if a resource behaves as modeled (passes the test), its use is not affected by the presence of other species. On the other hand, if a resource fails the test, this indicates that that particular resource may be influencing a phenotypic change in at least one species to gain a competitive advantage, or involved in a larger exometabolic network that connects different species. When this test is applied to a well- defined ecosystem, it will highlight the “important” resources in that environment, narrowing down the number of metabolic interactions to study in an environ- ment. Furthermore, new data on how the resources are used can be incorporated as parameters into this model, improving its ability to accurately model how all of the resources are used by a community. Results and Discussion In order to determine the substrate usage profiles of in- dividuals, we designed a defined medium composed of sufficient levels of standard vitamins, minerals, phos- phate, and ammonium and limiting levels of carbon (glucose and nineteen amino acids (see Methods). This medium was designed such that the species would reach As recently shown in the pioneering work by Behrends et al., the kinetics of substrate depletion from a mixture of substrates can be effectively fit using a few parameters [13]: see Eq. (1) in Methods. When compared across all Erbilgin et al. BMC Bioinformatics (2017) 18:57 Page 3 of 12 Page 3 of 12 stationary phase within 12 h and every substrate could be detected in a single LC-MS run. necessarily mean that a compound is utilized by an or- ganism. The compound may be enzymatically trans- formed to a different compound outside of the cell and then utilized, or it may be simply be imported into the cell and not participate in any metabolism. While strange, the latter scenario has been reported to occur in Cyanobacteria [16]. Bacilli and pseudomonads represent some of the most ubiquitous soil bacteria, and we selected the common soil bacterium Bacillus cereus for comparison with two closely related Pseudomonas species, Pseudomonas lini and Pseudomonas baetica (Additional file 1: Figure S1) that were isolated from groundwater; taxonomic assertions were confirmed by BLAST search results on the sequenced 16S rRNA gene. For simplicity, we will refer to the species as Bc (Bacillus cereus), Pl, (Pseudomonas lini) and Pb (Pseudo- monas baetica). Each species was grown individually in the defined medium, with supernatant samples collected every hour for 12 h, and one final time point at 26 h. y To examine the sequence of substrate depletion in finer detail, we used the model to calculate the time at which each species depleted half of the total amount of each compound (Th), and when the compound was depleted from 90% to 10% of the total amount available to the spe- cies (usage window) (Fig. 1), and mapped them onto the growth curve of each species (Fig. 2a–c). For Bc, we ob- served that compounds were half-depleted in three distinct groups (Figs. 2a and d, dotted boxes). Results and Discussion Bc initially utilized glucose, then a cluster of 13 amino acids that all had Th values within 0.25 h of each other during early logarithmic growth, and finally half-depleted the remaining six sub- strates in late exponential and stationary phases. Neither of the pseudomonads appeared to utilize substrates in these types of groups, but instead had a more even distri- bution throughout their growth curve (Figs. 2b–d). How- ever, the growth curve of Pb did show multiple growth phases (Fig. 2c), and so compounds can be mapped to the growth phase in which they are half-depleted (Fig. 2d). This observation is more in line with the traditional view of catabolite repression and multi-auxic growth, where a lag phase will be observed each time the organism reorga- nizes its metabolism to utilize different substrates [17]. The absolute concentrations of the 20 growth sub- strates were quantified at each time point, and the data were fit to a previously described model for compound depletion during microbial batch culture [13] (Fig. 1, Algorithm 1). We observed that all compounds followed the Behrends model over the course of growth for each species, with the exception of two compounds: glycine increased over the first 5 h of culture from all three spe- cies and then decreased logarithmically, and the methio- nine depletion profile for Bc was indeterminable due to both variance in the data and a lack of time points from 12 to 24 h (Additional file 2). Overall, these observations corroborate previous assertions that substrate utilization by microbes in batch culture follow the shape of a logis- tic growth type curve [13–15]. It is important to note that in our assay, the disappearance of signal does not Fig. 1 Modeling usage parameters. Example curve fitting to Behrends model (cyan). Blue square indicates the modeled T50 parameter of the Behrends model, or inflection point of the curve, and the width parameter of the model is depicted by the green bar centered at T50. The orange square represents the calculated Th value, or when half of the total amount of compound has been depleted, and the red bar depicts the calculated usage window, or time when the compound is depleted from 90 to 10% of the total amount used by the species Fig. 1 Modeling usage parameters. Example curve fitting to Behrends model (cyan). Results and Discussion Blue square indicates the modeled T50 parameter of the Behrends model, or inflection point of the curve, and the width parameter of the model is depicted by the green bar centered at T50. The orange square represents the calculated Th value, or when half of the total amount of compound has been depleted, and the red bar depicts the calculated usage window, or time when the compound is depleted from 90 to 10% of the total amount used by the species Erbilgin et al. BMC Bioinformatics (2017) 18:57 Page 4 of 12 Fig. 2 Th and width values for the strains. a–c Th and width for each compound mapped onto the growth curve of each strain. Colored circles represent average Th and colored horizontal lines represent the average usage window (time of depletion from 90 to 10% of total resource used by the strain). Solid black line is the average OD600 of each strain measured over time (n = 3), with shading representing standard deviation. d Comparison of Th values between strains, of all compounds, with error bars representing standard error. Dashed boxes in a and d indicate the grouping of compounds utilized by Bc, and dashed brackets in c and d indicate the different growth phases observed for Pb Fig. 2 Th and width values for the strains. a–c Th and width for each compound mapped onto the growth curve of each strain. Colored circles represent average Th and colored horizontal lines represent the average usage window (time of depletion from 90 to 10% of total resource used by the strain). Solid black line is the average OD600 of each strain measured over time (n = 3), with shading representing standard deviation. d Comparison of Th values between strains, of all compounds, with error bars representing standard error. Dashed boxes in a and d indicate the grouping of compounds utilized by Bc, and dashed brackets in c and d indicate the different growth phases observed for Pb other. Additionally, the Th values across all substrates for the two Pseudomonas species were close, but not identical, consistent with their short phylogenetic distance but different species identity (Fig. 2d); a similar observation has been described previously [15]. Results and Discussion Considering the differ- ences in growth curves between the two species, this is quite intriguing, as the general order in which the species consume the metabolites is not different, but there is this difference in growth profiles, supporting the hypothesis that there could be significant physiological differences between such closely related species. It is surprising that for these three species we observed three different combinations of growth curve and substrate utilization profile: a temporally distinct group- ing of compound utilization with only one observed growth phase (Fig. 2a), an even distribution of substrate utilization with only one growth phase (Fig. 2b) and an even distribution over multiple growth phases (Fig. 2c). This is quite significant given that two of the species belong to the same genus (Pl and Pb). This suggests that the metabolic regulatory systems between the two species are different: while Pb slows down its growth, presumably because it is undergoing a large-scale “switch” of meta- bolic systems, Pl does not, which may indicate that either all its metabolic systems are constitutively active, or the regulation of the systems is so perfectly timed that the organism can seamlessly switch from one metabolic re- gime to another. Bc may also have an efficient metabolic regulatory system, as even though we observe distinct temporal gaps between groups of compounds, we did not observe multiple growth phases. Bc was markedly different from the two pseudomonads, differing greatly in the amount of time it depleted 8 of the compounds (Fig. 2d and Additional file 3: Table S1). Of these, the utilization of glucose was particularly interesting, as it was predominantly depleted before there was any appreciable increase in biomass (Fig. 2a). This may indicate that there is a significant delay in substrate conversion to biomass in this species, or that Bc rapidly transforms glucose into some other compound, for example glycogen. Bc was markedly different from the two pseudomonads, differing greatly in the amount of time it depleted 8 of the compounds (Fig. 2d and Additional file 3: Table S1). Of these, the utilization of glucose was particularly interesting, as it was predominantly depleted before there was any appreciable increase in biomass (Fig. 2a). This may indicate that there is a significant delay in substrate conversion to biomass in this species, or that Bc rapidly transforms glucose into some other compound, for example glycogen. Results and Discussion We next wondered if the preferred substrates offer some physiological benefit over less preferable sub- strates. It is a general assumption in microbiology that substrates consumed first may be more advantageous than those consumed later [18], and that this would We next wondered if the preferred substrates offer some physiological benefit over less preferable sub- strates. It is a general assumption in microbiology that substrates consumed first may be more advantageous than those consumed later [18], and that this would To compare the differences in substrate depletion be- tween species, we compared Th across the three species (Fig. 2d and Additional file 3: Table S1). Across all three species, glutamine, glutamate, alanine, arginine, proline and asparagine, were half-depleted within 1 h of each Page 5 of 12 Erbilgin et al. BMC Bioinformatics (2017) 18:57 Page 5 of 12 depend on the competitive ‘strategy’ of the organism. Major strategies suggested include maximal growth rate and maximal biomass yield. Generally, copiotrophs (or- ganisms that grow in nutrient-rich conditions) are thought of as r-strategists (maximal growth rate) and oligotrophs (organisms that can only grow in low- nutrient conditions) as K-strategists (maximum yield) [19, 20]. The strains used in this study are copiotrophs, and we would expect that their order of substrate con- sumption would be related to maximal growth rate [18]. greatest physiological advantage. Bc could possess a metabolic strategy that does not perfectly follow the well-established paradigm of catabolite repression. Ul- timately, it is clear that bacteria dramatically differ in regulation of catabolite uptake, and it is not prudent to make general assumptions on microbial metabolism based solely on observations from a few model organ- isms and/or the energetic potential of substrates. Our experiments to test these correlations yielded a number of interesting results in addition to those de- scribed above. First, all three species grew on glucose as the sole carbon source without added amino acids, which was not predicted based on genomic functional predictions. The genomes of these organisms were avail- able in the Integrated Microbial Genomes (IMG) data- base (img.jgi.doe.gov), where functional predictions are made by associating annotated genes with KEGG Orthology terms and KEGG pathways and MetaCyc reactions [22]. Results and Discussion These analyses indicated auxotrophy for lysine, phenylalanine, tyrosine, histidine and serine in the case of Bc, and for lysine, histidine, leucine and co- enzyme A for Pl and Pb, meaning these organisms could not grow with a single carbon source, as they would be unable to synthesize those amino acids or cofactors. This observation highlights that all computational predictions should be treated as only suggestions, and should always be tested experimentally before making any assertions. Additionally, there were a number of compounds that did not support growth as sole carbon sources, but were depleted throughout the growth of the species in our complete defined medium (Fig. 3, lightly shaded com- pounds). This finding indicates that caution should be employed when making physiological assertions based on single-substrate studies, for example those that have individual substrates arrayed in multi-well plates; many microbes can only utilize certain compounds when other substrates are present, the phenomenon of co-metabolism [23]. We should note, however, that we do not know the details of how these compounds are depleted in the rich defined medium, only that they are depleted from the medium; they may simply be exogenously transformed. Finally, we determined the maximum depletion rate of all the substrates by the three species and normalized to grams cell dry weight (gCDW). We observed these rates to be less than 400 mMol/hour/except for glucose deple- tion by Bc, which we calculated to be about 5079 mMol/ hour/gCDW (Additional file 3: Table S1). In comparison, various studies of different organisms have measured the glucose uptake rate to range from 2 to 60 mMol/hour/ gCDW [24–26]. The depletion of glucose corresponds to rapid loss of signal representing glucose roughly 2.4 h into the growth curve (see Additional file 2), when hardly any biomass has been made. This is likely an artifact of our targeted analysis, as we are not directly observing what is sumption would be related to maximal growth rate [18]. We tested some of these general assumptions by com- paring the calculated Th values and maximum usage rate of each compound to the specific growth rate, starting molarity of the compound, and predicted total protein composition of each species, in order to determine what the substrate preference order might be correlated with (Fig. 3 and Additional file 1: Figure S3). Results and Discussion The specific growth rate of a species on a compound was determined by growing the species on that compound as a sole car- bon source (see Methods). Due to the excess nitrogen added to the medium (see Methods), we do not expect the C:N ratio of a given carbon source to have a signifi- cant impact on the growth rate of the organism. Surpris- ingly, the only significant (p < 0.05) correlations between all of these tests were that the specific growth rate of Pl on a given compound was weakly correlated with the Th of that compound (r = −0.652, p = 0.030) and with the maximum depletion rate of that compound relative to biomass (r = 0.656, p = 0.028) (Fig. 3c and d). These cor- relations support the common assumptions listed above, especially those rationalizing catabolite repression, as the compound that provides the higher rate of growth is depleted earlier and more rapidly than others. It is inter- esting that glucose did not confer the fastest specific growth rate for any of the strains, despite glucose gener- ally being considered a superior source of energy. This is not surprising, however, as it is known that pseudomo- nads preferentially use amino acids over glucose [21]. The rationalization of this phenotype is that in the soil environments where many pseudomonads (and B. ce- reus) live, decomposition products such as amino acids and organic acids are more readily available than sugars [21]. However, the lack of any strong or significant correlations in the bacillus and the other pseudomonad indicates that there are other factors at play that deter- mine an organism’s preferred substrate usage. It is apparent that not all microbes prefer to use substrates sequentially at all; the grouping of substrate utilization by Bc is a striking example of this. The resources within the second utilization group (Fig. 2a) conferred a wide range of specific growth rates, from zero to the highest observed for all substrates, and all were utilized within 2 h of each other (Fig. 3a). It is likely the case that the simultaneous usage of these substrates confers the Erbilgin et al. BMC Bioinformatics (2017) 18:57 Page 6 of 12 Fig. 3 Physiological Correlations. Results and Discussion Correlations between specific growth rate on a compound as a sole carbon source, and Th (a, c, e) or maximum compound depletion rate relative to biomass (grams cell dry weight (gCDW)) (b, d, f) in complete defined medium for species Bc (a, b), Pl (c, d) and Pb (e, f). Compounds that did not support growth as a sole carbon source (specific growth rate of zero) are shaded lighter at the bottom of each plot. Pearson correlation coefficients (r) and p-values (p) for the set of compounds for which the specific growth rate was nonzero are depicted in the upper-right of each plot. Correlations that had a p-value less than 0.05 were colored red. Error bars depict standard error Fig. 3 Physiological Correlations. Correlations between specific growth rate on a compound as a sole carbon source, and Th (a, c, e) or maximum compound depletion rate relative to biomass (grams cell dry weight (gCDW)) (b, d, f) in complete defined medium for species Bc (a, b), Pl (c, d) and Pb (e, f). Compounds that did not support growth as a sole carbon source (specific growth rate of zero) are shaded lighter at the bottom of each plot. Pearson correlation coefficients (r) and p-values (p) for the set of compounds for which the specific growth rate was nonzero are depicted in the upper-right of each plot. Correlations that had a p-value less than 0.05 were colored red. Error bars depict standard error Predicting consortium metabolism based on models of individual isolates happening to the glucose; extracellular enzymes may con- vert it to another molecule that is then imported into the cell at a different rate as opposed to the cell transporting glucose directly. This raises the question of why Bc would expend extra energy to synthesize these enzymes, when it presumably can use glucose as is. Perhaps it converts glu- cose to a molecule that is not usable by other species, thus sequestering a valuable energy source and gaining a com- petitive advantage. Having modeled the substrate usage of each species for each compound, we hypothesized that these models could be combined to predict how a consortium composed of the three species might utilize the substrates. We simu- lated the time-dependent depletion of each compound by a consortium composed of the bacillus and two pseudo- monads (see Methods, Eq. 2 and Algorithm 2). Briefly, the Page 7 of 12 Page 7 of 12 Erbilgin et al. BMC Bioinformatics (2017) 18:57 functions describing the compound usage by each species were summed (Additional file 1: Figure S2A), and the time at which this summed use curve reached the total avail- able compound was determined. This time of depletion was then used to predict how much of a given metabolite each species would have utilized when grown in co- culture, and the compound usage by each species was re- modeled (Additional file 1: Figure S2B colored dashed lines) and added together to form the co-culture predic- tion (Additional file 1: Figure S2B solid black line). These predictive models allowed us to make several hypotheses that are relatively simple to test. First is the usage curve of each metabolite by the co-culture. Related to this, we can predict the time at which all of a given metabolite will be depleted, and when all metabolites will be depleted. From this we predict that 14 compounds will be nearly depleted (less than 10% of starting concentration) by 6 h, and all but methionine will be completely consumed by 9 h (Fig. 4). Based on this, one could reasonably argue that a consortium composed of these three species would reach stationary phase sometime between 6 and 9 h, in contrast to the individual species, which all reached stationary phase after 9 h. Predicting consortium metabolism based on models of individual isolates To test our predictions, we inoculated a 3-member co- culture at equal optical density in the defined medium (see Methods), collected supernatant time points every hour, and measured the concentrations of all 20 sub- strates as described for monocultures. We found that many of our predictions were valid: nearly all com- pounds (17) were depleted to below 10% of starting con- centration by 6 h (Fig. 4, gold) and the co-culture accordingly reached stationary phase at this time as well (Additional file 1: Figure S4), presumably because all available substrates were consumed. Fig. 4 Co-culture observations compared to predictions, normalized to t0 concentration of each metabolite. Blue, green and red dashed lines represent the observed depletion of each compound by Bc, Pl and Pb, respectively, when grown in isolation. The solid black line is the predicted depletion of a co-culture of all three strains. The golden circles represent the measured compound concentration in the co-culture medium. Error bars and/or shading represent standard error (n = 3). Glycine at time point 4 could not be calculated because the measurement was outside the dynamic range of the calibration curve, and the r2 was not determined (n.d.) for glycine. Non-normalized figure is shown as Additional file 1: Figure S5 Fig. 4 Co-culture observations compared to predictions, normalized to t0 concentration of each metabolite. Blue, green and red dashed lines represent the observed depletion of each compound by Bc, Pl and Pb, respectively, when grown in isolation. The solid black line is the predicted depletion of a co-culture of all three strains. The golden circles represent the measured compound concentration in the co-culture medium. Error bars and/or shading represent standard error (n = 3). Glycine at time point 4 could not be calculated because the measurement was outside the dynamic range of the calibration curve, and the r2 was not determined (n.d.) for glycine. Non-normalized figure is shown as Additional file 1: Figure S5 Page 8 of 12 Page 8 of 12 Erbilgin et al. BMC Bioinformatics (2017) 18:57 Compounds that follow the model are evenly shared When analyzing the kinetics of depletion of the com- pounds, we observed that many (13) compounds agreed very well with the prediction, having R2 values greater than 0.9 (Fig. 4). Predicting consortium metabolism based on models of individual isolates Most of the compounds with high R2 values began to decrease slightly earlier or at a slightly faster rate than predicted, which could be attributed to experimental error in initial culture density. However, the depletion of most compounds were still very close to the predicted model, indicating that the shared usage be- tween the species could be very close to “blind” condi- tions, where the presence of other species does not affect the substrate usage decisions of each individual species. It is important to note that the high substrate concentrations likely explain the successful predictions using this simple modeling approach. Specifically, the substrate concentrations, initially at high micromolar concentrations, are likely well above the Km for the transporters and rate-limiting enzymes. For example many bacterial amino acid transporters have Km values in the low micromolar range [27, 28], such that the transporters and enzymes are saturated. We anticipate that much more detailed models accounting for sub- strate concentration would be required at soil- and groundwater-relevant substrate concentrations, which can be as low as 0.5–10% of the concentrations used in this study ([29] and Jenkins et al., in preparation). more difficult to explain and suggests at least one microbe has altered its phenotype due to the presence of other microbes, or that other exometabolites are influen- cing consortial behavior. For example, one species may have up-regulated metabolic pathways involving these compounds in an effort to outcompete others, either for the purpose of direct competition for the substrate, or in order to synthesize antibiotic compounds [30]. Alterna- tively, co-culturing of these microbes has resulted in an emergent function of increased flux of the substrate (s) through the system. This could be due to a cross-feeding effect where one microbe depletes an inhibitory com- pound of another microbe or one microbe’s products induce the co-metabolism of that product and one of these substrates. Testing these hypotheses would require an extensive untargeted metabolomics study, an extremely interesting direction for future studies. Compounds that deviate from the model Compounds that deviate from the model The remaining 7 compounds (glucose, histidine, glutam- ate, lysine, arginine, proline and glycine) deviated signifi- cantly from our predictions (R2 < 0.9) (Fig. 4, red text), suggesting some sort of interspecies interaction (s) is/are present that affect the depletion of those compounds. These interactions could include both direct interactions (e.g. signaling molecules) or indirect (e.g. the effect one species has on the medium). Among indirect interac- tions, metabolites secreted by one species that were not measured in this study (e.g. overflow metabolites such as acetate) could be consumed by another species, thus altering its resource usage, or could inhibit a certain metabolic pathway or even enhance the degradation of a metabolite due to co-metabolism. Glucose and histidine were both depleted more slowly than predicted. The simplest explanations for this are that the metabolic systems that deplete these compounds are indeed concentration dependent, or that these compounds are secreted by at least one member in the co-culture, resulting in an apparent slowdown of net depletion. Another possibility for this would be that there is a buildup of product in the co-culture that exerts feedback inhibition on the metabolism of these two compounds. Conclusions This study examining substrate competition for 20 abun- dant substrates by 3 species demonstrates that at least some portion of the metabolic behavior of a microbial consortium can be predicted by measuring the metabol- ism of microbes grown in monoculture. This likely can also apply to more complex situations, for example sep- arately measuring the metabolism of an existing micro- bial community and a foreign isolate, and predicting what the metabolic function might be if the isolate were introduced into the community. In any system, com- pounds that do not fit the predictions indicate emergent functions of the coculture and may highlight substrates that are somehow affected by species-species interac- tions. These may be occurring passively in the cases of feedback inhibition and co-metabolism, or actively in the case of one species altering its phenotype in order to outcompete others. Further studying these outlier sub- strates can shed light on metabolic interactions between microbes within a community. Finally, by studying the growth kinetics on varying levels of substrates, growth models based on the Monod equation can be generated and used to predict relative species abundance in these co-cultures. Ultimately, incorporating this predictive strategy when studying community metabolisms can help pinpoint interesting biological questions, as well as aid in the design of synthetic consortia. Metabolomics sample extraction All bacterial species were initially inoculated from frozen glycerol stocks onto an R2A agar plate prepared using Difco R2A Agar (BD, Franklin Lakes, NJ) and incubated overnight at 30 °C. The medium used for metabolomics experiments consisted of 1× Wolfe’s vitamins and 1× Wolfe’s minerals solutions [37], 1.5 mg/mL ammonium chloride, 0.6 mg/mL potassium phosphate, and 0.1 mg/ mL each of D-glucose and the following L-amino acids: alanine, aspartate, glutamate, phenylalanine, glycine, histidine, isoleucine, lysine, leucine, methionine, aspara- gine, proline, glutamine, arginine, serine, valine, threonine and tryptophan. Tyrosine was additionally supplied at 0.01 mg/mL. Species were individually cultured in 5 mL of this medium overnight at 30 °C from the R2A plate, then washed 3× by centrifugation at 5000 xg and resuspending in fresh medium. Washed cells were used to inoculate 50 mL of the medium in 250 mL Erlenmeyer flasks, at an initial optical density (OD600) of 0.012–0.017 as measured by a SpectraMax Plus 384 plate reader. These cultures were incubated at 30 °C, shaking at 200 rpm. Biomass was monitored by OD600 measurements, with grams cell dry weight (gCDW) inferred by a predetermined correlation factor, k with units gCDW/L/OD. To determine k, gCDW was measured from four cell suspensions of each species that were harvested by centrifugation and washed in phosphate buffered saline solution (Sigma-Aldrich). An example calculation is as follows: 1 mL of OD600 0.496 corresponded to 0.0005 gCDW. 0.0005 gCDW/.001 L/ 0.496 OD600 = 0.9 gCDW/L/OD. gCDW for each time point was calculated by multiplying k by the measured OD600 and the volume of the culture (50 mL subtracted by the volume of culture removed for sampling). Hourly time points of 1 mL of cell culture and controls (see above) were aspirated and centrifuged at 5000 xg to pellet the cells. 800 μL was aspirated from the top, taking care not to disturb the cell pellet, and split into two 400 μL aliquots, which were immediately frozen at −80 °C. A calibration curve was created with the medium used for culturing: 1× culture medium, 1/2×, 1/10×, 1/100×, 1/ 1000× and 1/10000× dilutions were prepared using culture medium without any carbon sources as the diluent. All experimental, control and calibration curve samples were lyophilized overnight, and metabolites were extracted in 300 μL methanol with 25 μM 13C-phenylalanine for use as an internal standard. Isolates and identification Isolates and identification The 16S rRNA gene for each isolate was amplified using primers 27 F (AGAGTTTGATCMTGGCTCAG) and 14 92R (CGGTTACCTTGTTACGACTT), and sequenced at the Eurofins sequencing facility (Eurofins MWG Operon LLC, Louisville, KY). Forward and reverse sequences were In contrast, glycine, proline, lysine, arginine and glu- tamate were all depleted faster than predicted. This is Page 9 of 12 Erbilgin et al. BMC Bioinformatics (2017) 18:57 Page 9 of 12 Erbilgin et al. BMC Bioinformatics (2017) 18:57 manually merged and used as queries using nucleotide BLAST against the 16S rRNA sequence database at NCBI. Growth assays of species on individual carbon sources were performed in 96-well Falcon tissue culture plates with flat bottom and low evaporation lid, in a total volume of 200 μL. The medium consisted of the same concentrations of Wolfe’s vitamins and minerals, ammo- nium chloride and potassium phosphate. Individual carbon sources were added at a concentration of 0.5 mg/ mL. Species were pre-cultured and washed as before, and wells were inoculated at an OD600 of 0.05. The plates were incubated at 30 °C, shaking at “medium” speed in BioTek Synergy HT and Tecan Infinite F200 Pro plate readers, for 48 h. Phylogenetic tree construction 16S rRNA gene sequences were obtained from IMG (img.jgi.doe.gov), except for B. cereus, P. lini and P. bae- tica, which were directly sequenced (see above). Gene sequences were aligned using MUSCLE [31, 32], curated using GBlocks [33], and the tree was constructed using PhyML [34] with 100 bootstraps, using the phylogeny.fr web server [35, 36]. The final tree was rendered using FigTree (http://tree.bio.ed.ac.uk/software/figtree/). Metabolomics sample extraction Final extracted samples were stored in Agilent 96-well sample plates and immediately analyzed via LCMS or stored at −80 °C. Metabolomics data acquisition and quantification An Agilent 1290 LC system equipped with a ZIC-pHILIC column (150 mm × 2.1 mm, 5 μm 100 Å, Merck SeQuant) was used for metabolite separation with the following LC conditions: solvent A, 5 mM ammonium acetate; solvent B, 9:1 acetonitrile:H2O with 5 mM ammonium acetate; flowrate: 0.25 mL/min; timetable: 0 min at 100% B, 1.5 min at 100% B, 25 min at 50% B, 26 min at 35% B, 32 min at 35% B, 33 min at 100% B, and 40 min at 100% B; column compartment temperature of 40 °C. Mass spec- trometry analyses were performed using Agilent 6460 triple quadrupole mass spectrometer. Agilent software (Santa Clara, CA): Optimizer was used for establishing fragmentor and collision cell voltages as well as precursor and product ion transitions while Mass Hunter QQQ Quantitative Analysis (version 6.0) was used for compound quantification. Retention times, collision energies and tran- sitions for each compound are listed in Additional file 1: Table S2. For co-culture experiments, 50 mL cultures were inoc- ulated with an OD600 of 0.012 of each species, resulting in an initial co-culture density of 0.036. 200 μL of cell culture was aspirated for OD600 measurements taken in a 96-well Falcon tissue culture plate with flat bottom. For all growth experiments, the water used to prepare the medium and uninoculated medium were incubated alongside the experimental flasks, as controls. Substrate depletion modeling The exact steps are shown in Algorithm 1: C− X species i C− aij0 1 þ e t−t50ij 0 wij0 þ oij 0 ð2Þ ð2Þ where C is the total amount of substrate j that is available to the mixed culture of set species. This is defined as the starting concentration of j minus the smallest oj in species.aij ', oij ′, t50ij , and wij ', are parameters that describe the depletion of j by species i in the co-culture of the individual in the set species, shown in Algorithm 2: where C is the total amount of substrate j that is available to the mixed culture of set species. This is defined as the starting concentration of j minus the smallest oj in species.aij ', oij ′, t50ij , and wij ', are parameters that describe the depletion of j by species i in the co-culture of the individual in the set species, shown in Algorithm 2: Th and usage window values were calculated from the Behrends model. All correlation coefficients and p-values were calculated using the pearsonr function in the stats package of scipy. Substrate depletion modeling The Anaconda package and custom IPython notebooks were used for all computational tasks [38], which are Page 10 of 12 Page 10 of 12 Erbilgin et al. BMC Bioinformatics (2017) 18:57 publicly available at https://github.com/biorack in the “Predicting metabolic properties of a microbial co- culture” repository. Data were stored and organized using Pandas [39] and NumPy [40], and graphs created using Matplotlib [41]. Metabolite depletion was modeled using leastsq from scipy.optimize [42], fitting the data to the Behrends model (Eq. 1): expressions were summed to generate an approximate total usage curve, and the time at which this curve crossed the total amount of available compound was de- termined: the time when all available compound has been used tdj . The amount of available compound was defined to be the starting concentration of a compound minus the lowest offset parameter between the three species, as the species with the lowest offset parameter for a substrate will presumably deplete the substrate to that level, but not more, even in a co-culture. The tdj was used to approximate the amount of compound that each species would have consumed by that time. The in- dividual usage curves were capped at this compound level at this time and subtracted from the starting concentration of compound to revert the curves back to depletion curves. These new curves were used to solve new parameters for the Eq. 1, generating new models of compound depletion in mixed conditions. These new models were then summed, producing the predicted total co-culture usage of each compound. This can be summarized by the general Eq. 2: ξ ¼ a 1 þ e t−t50 w þ o ð1Þ ξ ¼ a 1 þ e t−t50 w þ o ð1Þ Where a is amplitude and o is offset (see Fig. 1). These two parameters were defined from the data: amplitude was defined to be the average of the t = 0 data point and the maximum value data point in the data set of each compound, and offset was defined as the lowest value in the data set. All other parameters were solved using leastsq, with the criteria that they had to be positive values. Consent for publication 16. Baran R, Bowen BP, Northen TR. Untargeted metabolic footprinting reveals a surprising breadth of metabolite uptake and release by Synechococcus sp. PCC 7002. Mol Biosyst. 2011;7(12):3200–6. 17. Gorke B, Stulke J. Carbon catabolite repression in bacteria: many ways to make the most out of nutrients. Nat Rev Microbiol. 2008;6(8):613–24. 17. Gorke B, Stulke J. Carbon catabolite repression in bacteria: many ways to make the most out of nutrients. Nat Rev Microbiol. 2008;6(8):613–24. Funding h k 10. Baran R, Brodie EL, Mayberry-Lewis J, Hummel E, Da Rocha UN, Chakraborty R, Bowen BP, Karaoz U, Cadillo-Quiroz H, Garcia-Pichel F, et al. Exometabolite niche partitioning among sympatric soil bacteria. Nat Commun. 2015;6:8289. 10. Baran R, Brodie EL, Mayberry-Lewis J, Hummel E, Da Rocha UN, Chakraborty R, Bowen BP, Karaoz U, Cadillo-Quiroz H, Garcia-Pichel F, et al. Exometabolite niche partitioning among sympatric soil bacteria. Nat Commun. 2015;6:8289. This work has been funded by the Lawrence Berkeley National Lab under Contract No. DE-AC02-05CH11231 with the U.S. Department of Energy. Co-culture predictions The equations representing the depletion of a compound (depletion curve) by a species were subtracted from the initial starting concentration of the compound, creating an expression that represented the amount of compound used by each species over time (usage curve); these are the curves shown in Additional file 1: Figure S2A. These Page 11 of 12 Page 11 of 12 Page 11 of 12 Erbilgin et al. BMC Bioinformatics (2017) 18:57 References Additional file 1: Supplemental Table 2 and Supplemental Figures 1-5. (PDF 781 kb) Additional file 2: Raw data and fitted curves for levels of each metabolite during isolate growth curves. (PDF 2162 kb) Additional file 3: Supplemental Table 1. (XLSX 55 kb) Additional file 1: Supplemental Table 2 and Supplemental Figures 1-5. (PDF 781 kb) Additional file 2: Raw data and fitted curves for levels of each metabolite during isolate growth curves. (PDF 2162 kb) Additional file 2: Raw data and fitted curves for levels of each metabolite during isolate growth curves. (PDF 2162 kb) Additional file 3: Supplemental Table 1. (XLSX 55 kb) Abbreviations Bc: Bacillus cereus; gCDW: Grams cell dry weight; IMG: Integrated microbial genomics website (img.jgi.doe.gov); Km: Michaelis constant; LC-MS: Liquid chromatography coupled with mass spectrometry; OD600: Optical density at 600 nm; Pb: Pseudomonas baetica; Pl: Pseudomonas lini; Th: Time of half depletion of a resource 6. Zha Y, Westerhuis JA, Muilwijk B, Overkamp KM, Nijmeijer BM, Coulier L, Smilde AK, Punt PJ. Identifying inhibitory compounds in lignocellulosic biomass hydrolysates using an exometabolomics approach. BMC Biotechnol. 2014;14:22. 6. Zha Y, Westerhuis JA, Muilwijk B, Overkamp KM, Nijmeijer BM, Coulier L, Smilde AK, Punt PJ. Identifying inhibitory compounds in lignocellulosic biomass hydrolysates using an exometabolomics approach. BMC Biotechnol. 2014;14:22. 7. Casey E, Mosier NS, Adamec J, Stockdale Z, Ho N, Sedlak M. Effect of salts on the Co-fermentation of glucose and xylose by a genetically engineered strain of Saccharomyces cerevisiae. Biotechnol Biofuels. 2013;6(1):83. 7. Casey E, Mosier NS, Adamec J, Stockdale Z, Ho N, Sedlak M. Effect of salts on the Co-fermentation of glucose and xylose by a genetically engineered strain of Saccharomyces cerevisiae. Biotechnol Biofuels. 2013;6(1):83. Ethics approval and consent to participate Not applicable. 18. Kremling A, Geiselmann J, Ropers D, de Jong H. Understanding carbon catabolite repression in Escherichia coli using quantitative models. Trends Microbiol. 2015;23(2):99–109. 18. Kremling A, Geiselmann J, Ropers D, de Jong H. Understanding carbon catabolite repression in Escherichia coli using quantitative models. Trends Microbiol. 2015;23(2):99–109. Availability of data and materials 11. Wilmes P, Bowen BP, Thomas BC, Mueller RS, Denef VJ, VerBerkmoes NC, Hettich RL, Northen TR, Banfield JF. Metabolome-proteome differentiation coupled to microbial divergence. MBio. 2010;1(5):e00246–10. The IPython notebooks and data used in this study will be publicly available at https://github.com/biorack/bmc_bioinf_2016_erbilgin upon publication of this manuscript. 12. Perez-Garcia O, Lear G, Singhal N. Metabolic Network Modeling of Microbial Interactions in Natural and Engineered Environmental Systems. Front Microbiol. 2016;7:673. Declarations 19. Fierer N, Bradford MA, Jackson RB. Toward an ecological classification of soil bacteria. Ecology. 2007;88(6):1354–64. 19. Fierer N, Bradford MA, Jackson RB. Toward an ecological classification of soil bacteria. Ecology. 2007;88(6):1354–64. This manuscript has been authored by Lawrence Berkeley National Lab under Contract No. DE-AC02-05CH11231 with the U.S. Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. gy 20. Pianka ER. R-Selection and K-Selection. Am Nat. 1970;104(940):592. 20. Pianka ER. R-Selection and K-Selection. Am Nat. 1970;104(940):592. 21. Rojo F. Carbon catabolite repression in Pseudomonas : optimizing metabolic versatility and interactions with the environment. FEMS Microbiol Rev. 2010;34(5):658–84. 22. Markowitz VM, Chen IM, Palaniappan K, Chu K, Szeto E, Pillay M, Ratner A, Huang J, Woyke T, Huntemann M, et al. IMG 4 version of the integrated microbial genomes comparative analysis system. Nucleic Acids Res. 2014; 42(Database issue):D560–567. Acknowledgements The strains used in this study were a generous gift from Romy Chakraborty at Lawrence Berkeley National Laboratory. at Lawrence Berkeley National Laboratory. This material by ENIGMA- Ecosystems and Networks Integrated with Genes and Molecular Assemblies (http://enigma.lbl.gov), a Scientific Focus Area Program at Lawrence Berkeley National Laboratory is based upon work supported by the U.S. Department of Energy, Office of Science, Office of Biological & Environmental Research under contract number DE-AC02-05CH11231 8. Henriques ID, Aga DS, Mendes P, O’Connor SK, Love NG. Metabolic footprinting: a new approach to identify physiological changes in complex microbial communities upon exposure to toxic chemicals. Environ Sci Technol. 2007;41(11):3945–51. y y This material by ENIGMA- Ecosystems and Networks Integrated with Genes and Molecular Assemblies (http://enigma.lbl.gov), a Scientific Focus Area Program at Lawrence Berkeley National Laboratory is based upon work supported by the S D f Off f S Off f B l l & 9. Halter D, Goulhen-Chollet F, Gallien S, Casiot C, Hamelin J, Gilard F, Heintz D, Schaeffer C, Carapito C, Van Dorsselaer A, et al. In situ proteo- metabolomics reveals metabolite secretion by the acid mine drainage bio- indicator, Euglena mutabilis. ISME J. 2012;6(7):1391–402. Authors’ contributions 13. Behrends V, Ebbels TM, Williams HD, Bundy JG. Time-resolved metabolic footprinting for nonlinear modeling of bacterial substrate utilization. Appl Environ Microbiol. 2009;75(8):2453–63. OE, TRN conceived and designed the experiments. OE, BPB, SMK, SJ, RL performed the experiments. OE, BPB, SJ, TRN analyzed the data. OE, TRN wrote the manuscript. TRN contributed materials and analysis tools. All authors read and approved the final manuscript. 14. Behrends V, Geier B, Williams HD, Bundy JG. Direct assessment of metabolite utilization by Pseudomonas aeruginosa during growth on artificial sputum medium. Appl Environ Microbiol. 2013;79(7):2467–70. Competing interests 15. La Rosa R, Behrends V, Williams HD, Bundy JG, Rojo F. Influence of the Crc regulator on the hierarchical use of carbon sources from a complete medium in Pseudomonas. Environ Microbiol. 2016;18(3):807–18. The authors declare they have no competing interests. Consent for publication Not applicable. Consent for publication Not applicable. Received: 2 July 2016 Accepted: 7 January 2017 Author details 1 1Environmental Genomics and Systems Biology Division, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720, USA. 2Joint Genome Institute, 2800 Mitchell Dr, Walnut Creek, CA 94598, USA. 3Present Address: Intrexon Corporation, 1750 Kraft Dr, Blacksburg, VA 24060, USA. 23. Nzila A. Update on the cometabolism of organic pollutants by bacteria. Environ Pollut. 2013;178:474–82. 24. Fuhrer T, Fischer E, Sauer U. Experimental identification and quantification of glucose metabolism in seven bacterial species. J Bacteriol. 2005;187(5):1581–90. Received: 2 July 2016 Accepted: 7 January 2017 Received: 2 July 2016 Accepted: 7 January 2017 25. Dauner M, Storni T, Sauer U. Bacillus subtilis metabolism and energetics in carbon- limited and excess-carbon chemostat culture. J Bacteriol. 2001;183(24):7308–17. 25. Dauner M, Storni T, Sauer U. Bacillus subtilis metabolism and energetics in carbon- limited and excess-carbon chemostat culture. J Bacteriol. 2001;183(24):7308–17. Page 12 of 12 Erbilgin et al. BMC Bioinformatics (2017) 18:57 Accessed Nov 2015. • We accept pre-submission inquiries • Our selector tool helps you to ﬁnd the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and we will help you at every step: • We accept pre-submission inquiries • Our selector tool helps you to ﬁnd the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research Submit your manuscript at www.biomedcentral.com/submit Submit your next manuscript to BioMed Central and we will help you at every step: Erbilgin et al. BMC Bioinformatics (2017) 18:57 Erbilgin et al. BMC Bioinformatics (2017) 18:57 26. Blank LM, Sauer U. TCA cycle activity in Saccharomyces cerevisiae is a function of the environmentally determined specific growth and glucose uptake rates. Microbiology. 2004;150:1085–93. 27. Short SA, White DC, Kaback HR. Mechanisms of active transport in isolated bacterial membrane vesicles. IX. The kinetics and specificity of amino acid transport in Staphylococcus aureus membrane vesicles. J Biol Chem. 1972; 247(23):7452–8. 28. Lombardi FJ, Kaback HR. Mechanisms of active transport in isolated bacterial membrane vesicles. 8. The transport of amino acids by membranes prepared from Escherichia coli. J Biol Chem. 1972;247(24):7844–57. 29. Warren CR. Rapid and sensitive quantification of amino acids in soil extracts by capillary electrophoresis with laser-induced fluorescence. Soil Biol Biochem. 2008;40(4):916–23. 30. Garbeva P, Silby MW, Raaijmakers JM, Levy SB, Boer W. Transcriptional and antagonistic responses of Pseudomonas fluorescens Pf0-1 to phylogenetically different bacterial competitors. ISME J. 2011;5(6):973–85. 31. Edgar RC. MUSCLE: a multiple sequence alignment method with reduced time and space complexity. BMC Bioinf. 2004;5:113. 32. Edgar RC. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 2004;32(5):1792–7. 33. Castresana J. Selection of conserved blocks from multiple alignments for their use in phylogenetic analysis. Mol Biol Evol. 2000;17(4):540–52. 34. Guindon S, Dufayard JF, Lefort V, Anisimova M, Hordijk W, Gascuel O. New algorithms and methods to estimate maximum-likelihood phylogenies: assessing the performance of PhyML 3.0. Syst Biol. 2010;59(3):307–21. 35. Dereeper A, Guignon V, Blanc G, Audic S, Buffet S, Chevenet F, Dufayard JF, Guindon S, Lefort V, Lescot M, et al. Phylogeny.fr: robust phylogenetic analysis for the non-specialist. Nucleic Acids Res. 2008;36(Web Server issue): W465–469. 36. Dereeper A, Audic S, Claverie JM, Blanc G. BLAST-EXPLORER helps you building datasets for phylogenetic analysis. BMC Evol Biol. 2010;10:8. 37. Atlas R. Handbook of Microbiological Media, 4 ed. Boca Raton: CRC Press; 2010. 38. Anaconda Software Distribution Vers 2–2.4.0. Computer Software. https:// continuum.io. Accessed Nov 2015. 39. McKinney W. Data Structures for Statistical Computing in Python. In: Proceedings of the 9th Python in Science. 2010. p. 51–6. 40. Walt S, Colbert C, Varoquaux G. The NumPy Array: A Structure for Efficient Numerical Computation. Comput Sci Eng. 2011;13:22–30. Numerical Computation. Comput Sci Eng. 2011;13:22–30. 41. Hunter JD. Matplotlib: A 2D Graphics Environment. Comput Sci Eng. 2007;9:90–5. 42. SciPy: Open Source Scientific Tools for Python. http://www.scipy.org/. Accessed Nov 2015. 42. SciPy: Open Source Scientific Tools for Python. http://www.scipy.org/.
https://openalex.org/W3100640624	https://zenodo.org/records/4022293/files/universe-06-00098.pdf	English	null	A Plausible Model of Inflation Driven by Strong Gravitational Wave Turbulence	Universe	2,020	cc-by	10,213	Received: 10 June 2020; Accepted: 14 July 2020; Published: 16 July 2020 Abstract: It is widely accepted that the primordial universe experienced a brief period of accelerated expansion called inﬂation. This scenario provides a plausible solution to the horizon and ﬂatness problems. However, the particle physics mechanism responsible for inﬂation remains speculative with, in particular, the assumption of a scalar ﬁeld called inﬂaton. Furthermore, the comparison with the most recent data raises new questions that encourage the consideration of alternative hypotheses. Here, we propose a completely different scenario based on a mechanism whose origins lie in the nonlinearities of the Einstein ﬁeld equations. We use the analytical results of weak gravitational wave turbulence to develop a phenomenological theory of strong gravitational wave turbulence where the inverse cascade of wave action plays a key role. In this scenario, the space-time metric excitation triggers an explosive inverse cascade followed by the formation of a condensate in Fourier space whose growth is interpreted as an expansion of the universe. Contrary to the idea that gravitation can only produce a decelerating expansion, our study reveals that strong gravitational wave turbulence could be a source of inﬂation. The fossil spectrum that emerges from this scenario is shown to be in agreement with the cosmic microwave background radiation measured by the Planck mission. Direct numerical simulations can be used to check our predictions and to investigate the question of non-Gaussianity through the measure of intermittency. Keywords: general relativity; gravitational waves; turbulence A Plausible Model of Inﬂation Driven by Strong Gravitational Wave Turbulence Sébastien Galtier 1,* , Jason Laurie 2 and Sergey V. Nazarenko 3 Sébastien Galtier 1,* , Jason Laurie 2 and Sergey V. Nazarenko 3 Sébastien Galtier 1,* , Jason Laurie 2 and Sergey V. Nazarenko 3 1 Laboratoire de Physique des Plasmas, École polytechnique, Université Paris-Saclay, Institut Universitaire de France, CEDEX, F-91128 Palaiseau, France 2 Mathematics Group, School of Engineering and Applied Science, Aston University, Birmingham B4 7ET, UK; j.laurie@aston.ac.uk 2 Mathematics Group, School of Engineering and Applied Science, Aston University, Birmingham B4 7ET j.laurie@aston.ac.uk 3 3 Institut de Physique de Nice, Université Nice-Sophia Antipolis, Parc Valrose, 06108 Nice, France; sergey.nazarenko@inphyni.cnrs.fr 3 Institut de Physique de Nice, Université Nice-Sophia Antipolis, Parc Valrose, 06108 Nice, France; sergey.nazarenko@inphyni.cnrs.fr * Correspondence: sebastien.galtier@lpp.polytechnique.fr * Correspondence: sebastien.galtier@lpp.polytechnique.fr universe universe Universe 2020, 6, 98; doi:10.3390/universe6070098 universe universe 1. Introduction Since the problem is highly non-trivial, we will examine a simpliﬁed theoretical framework from which analytical results were recently derived for the regime of weak GW turbulence [14]. We will use these results to develop a theory of strong GW turbulence which is phenomenological by nature because, unlike for weak turbulence, the problem of strong turbulence is unsolvable perturbatively. In this way we will follow a very classical approach of turbulence based on the idea of critical balance (see, e.g., [15–20]). The mechanism of cascade in GW turbulence requires an initial excitation of the space-time metric denoted hi. We will assume that it happens at a wavenumber ki (or in a wavenumber window localized around ki). Since the state of the universe before inﬂation is inherently unknown, the description of a source of excitation remains speculative and subject to criticism. However, it is likely that the primordial universe was in a tumultuous state—as a heritage of the quantum foam [21]—with for example the creation of primordial black holes (PBH) due to space-time ﬂuctuations [22] (see Ref. [23] for the PBH formation by bubble collisions). They are expected to disappear quickly by radiation [24], however, the merger of PBH is possible before and could actually be a potent source of GWs. Our scenario of GW turbulence may start around 10−36 s or later, which is far enough from the Planck time to consider the general relativity model as applicable [14] (although inﬂaton models also use Einstein’s equations). Therefore, in the context of PBH mergers, the GWs produced will be modeled in Einstein’s equations by a forcing at ki, however, PBH which require a quantum description, are not modeled. In doing so, we follow the methodology used for inﬂaton, the effects of which are introduced into the energy-stress tensor. The case hi ≪1 is favourable to the development of weak GW turbulence for which a theory has been recently derived [14]. The theory describes the nonlinear evolution of weak ripples on the Poincaré–Minkowski ﬂat space-time metric. In this theory, like in the rest of the present paper, the cosmological constant is neglected (Λ = 0) and a pure vacuum space is considered. Therefore, the vacuum Einstein model is used, which in terms of the Ricci tensor reads Rµν = 0. The theory of Ref. 1. Introduction Understanding the origin of the universe—before or around the Planck time τP ∼10−43 s—is currently out of reach because it would require using a quantum theory of gravity that remains to be built. For a time signiﬁcantly greater than τP, the situation is different because the general relativity theory [1] provides a theoretical framework for describing the evolution of the universe as a whole [2]. It is believed that around 10−36 s the primordial universe experienced an accelerated expansion called inﬂation which led to an increase of the size of the universe by a factor of at least 1028 [3,4]. This superluminal expansion is supposed to have eventually stopped around 10−32 s. The inﬂation scenario has met a growing success (see, however, Ref. [5] for an alternative) because it can explain why the cosmic microwave background (CMB) radiation appears so uniform at large scales and why the universe is ﬂat [6]. While the inﬂationary paradigm is widely accepted, the detailed particle physics mechanism responsible for inﬂation—like the existence of a scalar ﬁeld called inﬂaton—remains unknown [2]. Furthermore, current experiments (in LHC) can only provide limited information with conditions corresponding to the age of ∼10−12 s [7]. www.mdpi.com/journal/universe www.mdpi.com/journal/universe Universe 2020, 6, 98; doi:10.3390/universe6070098 Universe 2020, 6, 98 2 of 16 Universe 2020, 6, 98 Inﬂation ﬁnds its energy from the vacuum and a phase transition associated with the grand-uniﬁed-theory symmetry breaking [3,8]. The inﬂationary scenario has, however, several tuning parameters leading to a wide spectrum of speculative models. The most recent data from the Planck satellite shows, with a high precision, that we live in a remarkably simple universe with, e.g., a small spatial curvature and a nearly scale-invariant density ﬂuctuation spectrum with nearly Gaussian statistics [6]. These observations have made it possible to exclude several models, while raising new questions that could weaken the inﬂationary paradigm and encourage consideration of alternatives [9–12]. Here, we propose a plausible alternative based on the nonlinearities of the (non-modiﬁed) general relativity equations which have been neglected so far when considering the primordial universe. Our approach is, therefore, different from the Starobinsky’s model where an extra R2 term was introduced in the Hilbert–Einstein action [13]. As the fundamental hypothesis of our study we will neglect the role of inﬂaton in the mechanism of inﬂation and we will focus our attention only on gravitational wave (GW) turbulence. 1. Introduction [14] is restricted to a 2.5 + 1 diagonal metric tensor which includes only one type (+) of GW (the × GW being excluded). However, the weak GW turbulence theory is limited because: (i) Initially weak GW turbulence quickly leads to strongly nonlinear turbulence at large scales (see below) and (ii) the GW dilution due to the universe global expansion will overpower the nonlinearity of the GW interactions [25]. The expansion arises because of the space-time ripples which contribute to the coarse-grained Einstein model through nonlinear wave interactions leading to an effective energy-momentum tensor typical for usual radiation, e.g., the electromagnetic waves. As we show in Appendix A, the rate of such an expansion for the statistically homogeneous and isotropic GW ﬁelds is sufﬁcient to produce the GW dilution which overpowers the wave–wave interactions if the wave phases are random (as is the case for weak GW turbulence theory). However, if the phases are not random, which usually occurs when the nonlinearities are not weak, then the Universe 2020, 6, 98 3 of 16 wave–wave interactions are typically as fast as the dilution process and a strong GW turbulence theory can be applicable. Thus, for the nonlinear GW interactions to be effective in the statistically homogeneous and isotropic GW ﬁelds these ﬁelds must be strongly nonlinear. Note that under some special metrics the expansion effect may be naturally slow or even suppressed, e.g., if the universe is forced to be ﬁnite as in the anti-de Sitter geometry. However, even taken alone, the fact that the initially weak GW turbulence quickly becomes strong (as can be theoretically predicted) calls for studying the case of strong GW turbulence, and this is the main objective of this paper. The paper is structured as follows: Section 2 is devoted to the Friedmann equations in order to recall the assumptions of the basic model and notations; in Section 3 we brieﬂy present the analytical results of weak GW turbulence published recently and then deduce the phenomenological theory of strong GW turbulence; the formation of a condensate and its interpretation in terms of inﬂation is discussed in Section 4; in Section 5 we show that our prediction is in good agreement with the CMB measured by the Planck mission; in the last section we conclude with a summary and a discussion. 1. Introduction A consideration of the GW dilution in an expanding universe, where the expansion is caused by the GW themselves is presented in Appendix A. 1 One should not be confused with the reference to quantum mechanics, which is made purely for a simple illustration of the relation between the energy and the wave action spectra. The system we are considering is purely classical, in a sense that the occupation numbers at all the momentum states are large. It would be possible to extend our consideration to the cases of small or moderate occupation numbers via writing a quantum kinetic equation based on, e.g., the Fermi golden rule. However, these cases can only be relevant to the direct cascade at very high k which is beyond the scope of the present paper. 2. Friedmann Equations In this section, we recall the Friedmann equations in the simpliﬁed case where the cosmological constant is neglected (Λ = 0) and the metric is ﬂat; they read H2 = 8πG 3 ρ (1) ˙ρ + 3H ρ + P c2 = 0 (2) (1) (2) with H ≡ ˙a/a the Hubble parameter, ˙ the time derivative, a(t) the cosmic scale factor, c the speed of light, ρ(t) the density and P(t) the pressure. These equations are derived from the Friedmann–Lemaître–Robertson–Walker (FLRW) diagonal metric with interval ds2 = gµνdxµdxν = −c2dt2 + a2(t)dx2 (3) (3) where x are the co-moving coordinates. The basic assumption behind the FLRW metric is that the universe is homogeneous and isotropic (assumption only valid at large scale) and thus the density and pressure are uniform functions (in space) which depend only on t. The Friedmann equations describe, therefore, the large-scale evolution of the universe. In a vacuum space (ρ = P = 0) the solution of the Friedmann Equations (1) and (2) is trivially a static universe. On the other hand, in presence of weak GW turbulence (isotropic and homogeneous but in the statistical sense) the vacuum universe exhibits rich dynamics with a dual cascade. Note that inﬂation models use the Friedmann equations where the density and pressure are substituted with new quantities arising from assuming existence of an inﬂaton ﬁeld [26]. It is known [27], and explicitly shown in Appendix A, that even in vacuum GW ﬁelds (metric disturbances) produce effective density and pressure terms in the Friedmann equations which have a form typical for radiation. The respective radiative density and pressure terms correspond to the wave–mean interactions, i.e., the effect of the small-scale GWs onto the coarse-grained metric evolution. However, the difference with a usual (e.g., electromagnetic) radiation is that the GW ﬁeld is nonlinear, which can affect in a substantial way the character of energy dilution due to the universe expansion. Namely, as we show in Appendix A, the wave–wave interaction process is typically as fast as the wave–mean interaction if their phases are not random. Universe 2020, 6, 98 4 of 16 3. Metric Cascades in GW Turbulence Let us brieﬂy recap the results concerning weak GW turbulence. It is characterized by a direct cascade of energy E and an inverse cascade of wave action (or particle number) N [14]. The respective turbulent spectra are deﬁned from the space-time ﬂuctuations hµν which are introduced as perturbations of the Minkowski metric ηµν, gµν = ηµν + hµν (4) (4) with gµν being the metric and \|hµν\| ≪1. We shall give a phenomenological argument to explain why such a double cascade exists for weak GW turbulence. This argument is similar to the famous Fjørtoft argument predicting the dual cascade behavior in two-dimensional hydrodynamic turbulence [28]. Hereafter, we assume statistical homogeneity and isotropy of the turbulent state. We deﬁne ˆEk to be the one-dimensional (1D) spectrum of the total energy density E by the deﬁnition E = R ∞ 0 ˆEkdk. (A similar deﬁnition holds for the 1D wave action density spectrum ˆNk.) Assume that at wavenumber ki ∼1/λi we have an injection of wave action ﬂux ζi and energy ﬂux εi. For the demonstration, we will assume the presence of sinks at small and large wavenumbers, k0 and k∞respectively, such that 0 < k0 < ki < k∞< ∞. We also deﬁne ζ0, ε0, ζ∞and ε∞as the ﬂux values at the corresponding sinks. By virtue of the conservation of the wave action and the energy (in absence of sources and sinks), the relations (5) ζi = ζ0 + ζ∞, εi = ε0 + ε∞ (5) should be satisﬁed in the steady state, i.e., the injected ﬂuxes are in balanced with the removal at the sinks. The energy and the wave action spectral densities are linked through the relation ˆEk = ω ˆNk (with ω = ck for GWs), which could be interpreted as the usual quantum-mechanical relation, between the energy and the occupation number k-space density of quasi-particles–in our case gravitons1. The relationship between the energy and wave action leads to the complementary relationship between the respective ﬂuxes, ε = ωζ, and to three equations involving k0, ki and k∞. 3. Metric Cascades in GW Turbulence By using the scaling relation at scale ℓ[32], we can show that E> ℓ= R k<2π/ℓˆEk dk, the total energy density contained in scales greater than ℓ, scales as E> ℓ∼ c4 32πG h2 ℓ ℓ2 (9) (9) and E> ℓ∼ωN> ℓ∼k2 ˆNk ∼k4/3 ∼ℓ−4/3, we ﬁnd (with the notation introduced above) and E> ℓ∼ωN> ℓ∼k2 ˆNk ∼k4/3 ∼ℓ−4/3, we ﬁnd (with the notation introduced above) and E> ℓ∼ωN> ℓ∼k2 ˆNk ∼k4/3 ∼ℓ−4/3, we ﬁnd (with the notation introduced above) hℓ∼hi ℓ λi 1/3 (10) (10) The GW turbulence becomes strong when hℓ= hs ∼1; with an initial excitation hi ∼10−1 it leads to ℓ= λs ∼103λi. However and as explained above, for statistically isotropic and homogeneous weak GW turbulence, the wave–wave interaction is overpowered by the wave–mean interaction leading to the mean expansion and GW dilution. Thus, for the GW interactions to have a signiﬁcant contribution, the initial excitation has to be close to ks. Therefore, the weak wave turbulence regime in Figure 1 is illustrated mainly for pedagogical reasons to emphasize the overall wave–kinetic scenario. k ˆHk 0 ks ki kP Planck Scale Strong wave turbulence Weak wave turbulence k−1 k−5/3 k−2 ζ ǫ Figure 1. 1D metric spectrum ˆHk produced by an injection of wave action and energy ﬂuxes at wavenumber ki. The weak GW turbulence regime is localized in the interval ks < k ≪kP, where kP is the Planck wavenumber and ks determines the wavenumber below which GW turbulence is strong. In this scenario, the inverse cascade leads to the formation of a condensate at k = 0. The growth of the condensate corresponds to an increase of the cosmic scale factor. k ˆHk 0 ks ki kP Planck Scale Strong wave turbulence Weak wave turbulence k−1 k−5/3 k−2 ζ ǫ kP 0 Figure 1. 1D metric spectrum ˆHk produced by an injection of wave action and energy ﬂuxes at wavenumber ki. The weak GW turbulence regime is localized in the interval ks < k ≪kP, where kP is the Planck wavenumber and ks determines the wavenumber below which GW turbulence is strong. In this scenario, the inverse cascade leads to the formation of a condensate at k = 0. The growth of the condensate corresponds to an increase of the cosmic scale factor. 3. Metric Cascades in GW Turbulence Solving these equations, for a sufﬁciently large inertial range (k0 ≪ki ≪k∞), we obtain in the limit ε0 ε∞ = k0 ki 1 −ki/k∞ 1 −k0/ki →0 (6) ζ∞ ζ0 = ki k∞ 1 −k0/ki 1 −ki/k∞ →0 (7) (6) (7) which means that the energy and the wave action ﬂuxes are opposite to each other in the k-space. Note that this is nothing but a standard argument about the spectral ﬂux directions in weak turbulence theory with two positive quadratic integrals of motion, e.g., see Ref. [29]. which means that the energy and the wave action ﬂuxes are opposite to each other in the k-space. Note that this is nothing but a standard argument about the spectral ﬂux directions in weak turbulence theory with two positive quadratic integrals of motion, e.g., see Ref. [29]. As explained in Ref. [14] and shown numerically in Ref. [30], the inverse cascade is exp with in principle the possibility for the wave action spectrum, excited at ki, to reach the waven As explained in Ref. [14] and shown numerically in Ref. [30], the inverse cascade is explosive with in principle the possibility for the wave action spectrum, excited at ki, to reach the wavenumber k = 0 in a ﬁnite time. However, the description fails at scale ks (or λs ∼1/ks) where the turbulence becomes strongly nonlinear. We may evaluate λs by using the weak GW turbulence theory for which we know the exact power law solutions [31]. We have the following 1D isotropic constant-ﬂux Universe 2020, 6, 98 5 of 16 stationary solution corresponding to the inverse cascade (see Figure 1 where the metric spectrum is schematically reported), ˆNk ∼ζ1/3k−2/3 (8) (8) Let us denote by hℓthe typical value of the metric disturbance at length scale ℓ. Under the assumptions of statistical homogeneity and isotropy we deﬁne hℓ= p ⟨(h(x + ℓ) −h(x))2⟩where ⟨·⟩ represents the ensemble and spatial average over x, and x and x + ℓare two positions separated by an increment ℓ. Due to statistical homogeneity and isotropy hℓonly depends on the magnitude of the increment ℓ= \|ℓ\|. 3. Metric Cascades in GW Turbulence Weak GW turbulence can be seen as a local description for which the assumption of a perturbed Minkowski space-time metric (4) applies well. For larger scales, however, the metric could be different with, for example, the presence of a non-zero large-scale curvature. The phenomenological scenario described hereafter applies to this situation as well. The presence of an inverse cascade in the regime of weak GW turbulence is an indication that at wavenumber k < ks the inverse cascade continues in the regime of strong turbulence. Following the classical theory of strong wave turbulence (for different 6 of 16 Universe 2020, 6, 98 applications see, e.g., [15–20]), we conjecture that the turbulence saturation state is determined by a critical balance (CB) in which the linear wave period τGW = 2π/ω and the nonlinear time τNL ∼ℓ/(hℓc) are approximately the same over a wide range of scales. The scale-by-scale balance relation τGW ∼τNL leads to statistical ﬂuctuations with hℓ∼1. This result means that strong turbulence may develop on the background of the Minkowski space-time keeping the ﬂuctuations ﬁnite. Note that the presence of such strong ﬂuctuations may be accompanied by the creation of structures like PBH, which do not contradict our statistical prediction. The 1D metric spectrum ˆHk = 4πk2 Z ⟨h(x + ℓ)h(x)⟩e−ik·ℓdℓ= 2πk2 Z 2 D h2E −h2 ℓ e−ik·ℓdℓ (11) (11) (where the factor 4πk2 arises from angle integration in Fourier space) for the CB regime can be determined dimensionally with the relation h2 ℓ∼k ˆHk, leading to the scaling law ˆHk ∼k−1 (see Figure 1), and thus the wave action spectrum is ˆNk ∼k0. Then, the CB phenomenology of strong GW turbulence tells us that the spectrum can reach the mode k = 0 in ﬁnite time because it is a ﬁnite-capacity turbulence system: namely, the integral R ki 0 ˆNk dk converges, which means that the CB spectrum holds only a ﬁnite amount of wave action over the range [0, ki]. This implies that it takes only a ﬁnite amount of time to form the CB spectrum everywhere in the range [0, ki] given that the forcing pumps wave action at a ﬁnite rate and that most of the wave action is transferred upscale of the forcing scale (which follows from the dual cascade argument given above). This scenario is supported by the numerical simulations in Galtier et al. [30]. 3. Metric Cascades in GW Turbulence It is important to note that the excitation of the metric from ki > 0 to k = 0 in a ﬁnite time does not violate the causality principle because it does not correspond to the propagation of information in physical space from a given position to inﬁnity. Instead, the inverse cascade means a continuous increase of the wavelength of a ﬂuctuation as a consequence of its interaction with other ﬂuctuations of predominately similar wavelengths. Clearly, this can be done locally in physical space. Then, the mode k = 0 corresponds to the level of the background over which there are ﬂuctuations of different wavelengths (see, e.g., Refs. [33–35] where the zero mode plays an important role). Note also that this description does not require to ﬁx the size (ﬁnite or inﬁnite) of the system. 4. Condensate and Inﬂation When the spectral front reaches the mode k = 0 a condensate emerges in Fourier space (see Figure 1). This situation is similar to the formation of non-equilibrium Bose–Einstein condensation generated by an inverse cascade (see, e.g., Refs. [33,35–39]). In our case, the growth in time of the condensate is at the expense of the ﬂuctuations which can be maintained as long as the wave action ﬂux ζ is ﬁnite (possibly constant). For the sake of simplicity, let us consider a perturbed FLRW metric with interval (initially a(0) = 1 ds2 = −c2dt2 + a2(t)dx2 + hµνdxµdxν (12) (12) with ⟨hµν⟩= 0 where the angle brackets denote the physical space average. We can formally write ds2 = −c2dt2 + dx2 + h′ µνdxµdxν (13) (13) where h′ ij ≡(a2 −1)δij + hij, i, j = 1, 2, 3, with h′ 0,µ = h0,µ, and h′ µ,0 = hµ,0 for µ = 0, 1, 2, 3. Therefore, we see that the level of the background “condensate” metric h′ ij(k = 0) ≡(a2 −1)δij provides a measure of the cosmic scale factor. In particular, a growth of the condensate means an expansion of the universe (and a ﬂattening of the space-time). Since this growth occurs via draining the wave action from the of the cosmic scale factor. In particular, a growth of the condensate means an expansion of the universe (and a ﬂattening of the space-time). Since this growth occurs via draining the wave action from the ﬂuctuations, this mechanism is contributing to “ironing out" of small-scale inhomogeneities. In previous works on the dynamics of the Bose–Einstein condensation, it was shown that the condensate growth accelerates [35,36] and an explosive evolution (as (tc −t)x with x < 0 and t < tc) 7 of 16 Universe 2020, 6, 98 was not excluded; in the weak turbulence regime a power-law in time was predicted [40]. An analogous accelerated condensate growth scenario for GW turbulence would mean inﬂation. Inﬂation would not be in contradiction with the causality principle because it would be the result of an ampliﬁcation of the background metric. This mechanism would be limited in time because an expansion of the universe leads also to a dilution of the GW ﬁeld (see Appendix A and also Ref. [14]). 4. Condensate and Inﬂation We may expect, however, that it is only when the source of the wave action ﬂux is depleted that the GW ﬂuctuations start to decay. This eventually leads to a natural saturation of the condensate and the end of inﬂation (because of the weakening of the GW ﬁeld and hence the nonlinear transfer). We may estimate the time-scale necessary for the formation of a condensate by using the turbulence phenomenology, which gives several predictions. First, we can say that the development of weak GW turbulence happens at the typical time-scale of the four-wave kinetic equation [14] τWT ∼τGW ϵ4 (14) (14) where the GW time is that of the linear wave evolution τGW ∼1/(kic) and the small parameter ϵ is deﬁned as the ratio between the linear and nonlinear time-scales ϵ ∼τGW/τNL ∼hi ∼10−1. We ﬁnd τWT ∼104λi/c. But, as we have already discussed, the GW turbulence quickly ceases to be weak (because the initial excitation is close to ks), and the characteristic time becomes the one of the dynamical (rather than the weak turbulence kinetic) equation, namely τdyn ∼τGW ϵ2 (15) (15) which is shown in Appendix A to be the same time as for the dilution. Note that expression (15) is also valid for weak GWs with h ≪1 if their phases are not random. Non-random phases could appear, e.g., due to presence of coherent structures such as solitons, PBH, or wormholes. With the parameters given above, we ﬁnd that τdyn ∼100λi/c. In the CB state the nonlinearity parameter ϵ is of order one, so which is shown in Appendix A to be the same time as for the dilution. Note that expression (15) is also valid for weak GWs with h ≪1 if their phases are not random. Non-random phases could appear, e.g., due to presence of coherent structures such as solitons, PBH, or wormholes. With the parameters given above, we ﬁnd that τdyn ∼100λi/c. In the CB state the nonlinearity parameter ϵ is of order one, so τdyn ∼τGW ∼τNL ∼λi c (16) (16) It is important to realize that expressions (14)–(16) arise from the Einstein vacuum model, which is free of tuning parameters, and their form is dictated solely by the nonlinear properties of this model. 6. Conclusions In summary, we propose an alternative scenario for inﬂation which is driven by strong GW turbulence. We show that a small-scale excitation of the metric leads to a self-accelerating inverse cascade that could lead to the formation of a condensate whose growth is interpreted as an expansion of the universe. The condensate growth is expected to accelerate—possibly explosively—leading to a phase of inﬂation. It is shown that the scalar spectrum obtained with this scenario is compatible with the CMB measured by the Planck mission (without introducing tuning parameters [45]), however, it is not possible to quantify the expansion (number of e-folds). This new scenario does not preclude the appearance of a reheating phase of the universe and particle creation after inﬂation [10]. The nonlinear mechanism described here does not require the introduction of the cosmological constant and has no tuning parameters. It also supports the idea of inﬂation which was recently questioned [11,12]. Note that this scenario is a priori not applicable to the recent cosmic acceleration of the universe for which the conditions are not favorable for the development of GW turbulence. The turbulent inﬂation introduced here is mainly a phenomenological theory, inspired by the analytical results obtained in weak GW turbulence. At present, an essential part of it remains in conjecture, speciﬁcally the view that the inverse cascade will continue through the strongly turbulent stage. Indeed, strictly speaking the dual cascade behavior relies on the conservation of the wave action, which is a property of the four-wave kinetic equation and therefore breaks down when this equation is no longer applicable. The situation here is similar to the behavior described by the Gross–Pitaevskii model: when the inverse cascade becomes strong, the energy invariant ceases to be quadratic, and the dual cascade argument becomes, technically, invalid. However, it is known from numerical simulations of the Gross–Pitaevskii model [46,47] that the condensation process started at the weakly turbulent regime as an inverse cascade, continues at the strongly turbulent stage with the appearance of strongly nonlinear defects which move like hydrodynamic vortices. These tend to continuously annihilate, so that no defects remain after a ﬁnite time, with the correlation length becoming inﬁnite. 5. Fossil Spectrum and CMB According to our scenario, phase-transitions in the early universe give rise to GW turbulence which is fuelling inﬂation. This is followed by GW dilution and the termination of inﬂation when the source of GW is depleted. Then, the fossil gravitational spectrum should be the one we have obtained for strong GW turbulence but with a much smaller amplitude due to the effects of GW dilution (it is safe to assume that the nonlinear interactions of the waves are negligible during the majority of the dilution stage). We may try to compare our prediction with the CMB radiation measured by the Planck mission [6]. After inﬂation we are left with a Minkowski metric plus very small ﬂuctuations, therefore we can ply use the Newtonian law (17) ∇2φ = 4πGρ, (17) with φ the gravitational potential, to derive the corresponding 1D scalar spectrum ˆΦk = 2πk2 Z 2 D φ2E −φ2 ℓ e−ik·ℓdℓ. (18) (18) Universe 2020 6 98 8 of 16 Universe 2020, 6, 98 Universe 2020, 6, 98 8 of 16 Here, φℓ = p ⟨(φ(x + ℓ) −φ(x))2⟩is linked to the potential by the relation φ2 ℓ ∼k ˆΦk. Expression (17) leads to the scaling relation φℓ∼ρℓℓ2. Connexion with our prediction can be made through the relation ρℓφℓ∼E> ℓ ∼ℓ−2 (relation (9) is used) which is compatible with the CB phenomenology. Then, we ﬁnd ˆΦk ∼k−1 (19) (19) which is the Harrison–Zeldovich spectrum (with the classical notation ˆΦk ∼kns−2, therefore ns = 1). This solution is scale-invariant in the sense that the ﬂuctuations in the gravitational potential are independent of length scale [41]. The Planck data are actually compatible with ns ≃0.967 which corresponds to a 1D metric spectrum ˆHk ∼kns−2 ∼k−1.033 only slightly steeper than our prediction. Note that ﬁnite-capacity turbulence systems, which are characterized by an explosive cascade, exhibit anomalous scalings with power laws slightly different from the steady-state predictions [35,36,42] (for other physical examples see Refs. [31,34,43,44] and for weak GW turbulence see Ref. [30]). Therefore, the slight discrepancy between our prediction and the Planck data could be the signature of an anomalous scaling. Finally, we can show that the tensor-to-scalar ratio r (ratio between the metric and scalar spectra) is independent of k. Its amplitude is, however, difﬁcult to estimate without numerical simulation because it requires the knowledge of the Kolmogorov constants. 6. Conclusions By this analogy, we conjecture that in the vacuum Einstein model, the condensation process will also continue through the strongly turbulent stage, possibly with some singular coherent objects, such as wormholes, PBH, or solitons appearing in the system at a transient stage (similar to the appearance of the vortices in the Gross–Pitaevskii model). Obviously substantial work remains to be performed for such a scenario Universe 2020, 6, 98 9 of 16 to be conﬁrmed by direct numerical simulations (this issue is left for future work) and if possible by analytical calculations. To describe strong GW turbulence, we have used the CB hypothesis stating that turbulence saturates in a state where the linear and the nonlinear time-scales balance each other across a wide range of spatial scales. The CB theory was originally introduced in the ﬁeld of astrophysical MHD turbulence in Ref. [15]. It remains phenomenological in nature due to the fact that strongly nonlinear turbulence is a notoriously difﬁcult subject, and to date there exist no exact theory of strong turbulence even in such well-studied systems as the classical Navier–Stokes ﬂuids. However, we should recall that the CB theory has played an enormously positive role in the ﬁeld of turbulence, becoming a central conjecture which has attracted a signiﬁcant number of authors who are aiming to check its validity, either theoretically or numerically. In the twenty-ﬁve years since its introduction, the CB theory has received a substantial theoretical and numerical support, and yet it is far from being fully conﬁrmed. This experience leads us to predict that thorough study of strong GW turbulence will be equally difﬁcult and time consuming. However, we believe that the CB approach introduced in the present paper will serve its positive role as a guide for future studies and act as a reference theory for analysing numerical and observational data, in the same way as it has been serving a positive role in the ﬁeld of MHD turbulence. We should note that there has been several other explanations and interpretations of the measured CMB spectrum, and the fact that the CB turbulence theory predictions are compatible with it does not, of course, mean that it is the ultimate valid explanation of it. One particularly interesting approach was put forward in Ref. 6. Conclusions [48] who suggested that the scaling of the CMB spectrum should be related to conformal invariance of the correlation functions in a dark energy dominated universe. In fact, their suggestion is not contradictory with the turbulent scenario, and one cannot rule out that full conformal invariance does indeed arise in turbulence of such type of systems. However, invoking dark energy without clarifying its nature has the same detriment as introducing a hypothetical inﬂaton ﬁeld in the inﬂation theories. Secondly, assuming the full conformal invariance may be natural based on the symmetries of the system, but it is not obvious if it should naturally arise dynamically during the evolution. To this effect, it is worth reminding the reader about the theory of 2D hydrodynamic turbulence based on a conformal invariance conjecture which was suggested by Polyakov [49]. In spite of the theory’s immense mathematical beauty, it predicted a turbulence spectrum exponent that has not as yet been conﬁrmed by numerical simulations. One of the hottest question in cosmology is about the non-Gaussianity of the CMB [6]. In the framework of turbulence, non-Gaussianity is natural and associated in particular with intermittency [50], which is traditionally measured via structure functions (two-point measurements). In our case, we can deﬁne the following set of structure functions of order p (20) Sp ij(ℓ) ≡⟨[hij(x + ℓ) −hij(x)]p⟩ (20) Under the assumption of statistical homogeneity and isotropy, they are expected to have scalings Sp ij(ℓ) ∼ℓζp with ζp a function that depends on p in a non-trivial way. With direct numerical simulations of strong wave turbulence it will be possible to measure ζp (this issue is left for future work) and therefore have an empirical prediction that can be compared directly with the Planck data (it should be viewed as a new way to analyze and interpret the data). This new prediction can be used in ﬁne to make a distinction between the different scenarios of inﬂation, e.g., those based on turbulence or inﬂaton. Finally, it is interesting to note that the effects of small scale ﬂuctuations on the large-scale dynamics has been studied by [51]: it was shown analytically that the back reaction is much stronger for GWs than for matter density ﬂuctuations. While, in another study [52], it was suggested that solitonic GWs of cosmological origin can contribute to the expansion of the universe. Acknowledgments: We thank K. Clough for useful discussions. Acknowledgments: We thank K. Clough for useful discussions. Conﬂicts of Interest: The authors declare no conﬂict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. Appendix A. Universe Expansion Driven by Nonlinear Gravitational Wave Interactions We systematically derive a set of mean and ﬂuctuation Equations (A10)–(A12) for nonlinear gravitational wave evolution in a ﬂat empty universe based on the Einstein vacuum equations. We show how, for non-weak gravitational wave amplitudes or non-random phases, the leading nonlinear wave contribution is of the same order as the term corresponding to universe dilation, implying that consideration of gravitational wave–wave interactions is important for describing the expansion of the universe. This motivates the use of the critical balance argument for strong gravitational wave turbulence of the main text. g Consider the Einstein vacuum equations Consider the Einstein vacuum equations Rµκ = 0 (A1) (A1) restricted to small perturbations hµκ of metric gµκ about the Friedmann-Lemaître-Robertson-Walker (FLRW) metric ¯gµκ: restricted to small perturbations hµκ of metric gµκ about the Friedmann-Lemaître-Robertson-Walker (FLRW) metric ¯gµκ: gµκ = ¯gµκ + hµκ, \|hµκ\| ≪\| ¯gµκ\| (A2) ¯gµκ =   −c2 0 0 0 0 a2(t) 0 0 0 0 a2(t) 0 0 0 0 a2(t)   gµκ = ¯gµκ + hµκ, \|hµκ\| ≪\| ¯gµκ\| (A2) ¯gµκ =   −c2 0 0 0 0 a2(t) 0 0 0 0 a2(t) 0 0 0 0 a2(t)   gµκ = ¯gµκ + hµκ, \|hµκ\| ≪\| ¯gµκ\| (A2) (A2) where a(t) is the scale factor for the spatial expansion of the universe and c is the speed of light. It is worth remarking that in [53] and subsequently in [54], the authors considered weak gravitational waves in FLRW space without assuming slowness of a(t). This makes sense if matter or (non-gravitational) radiation produces a fast expansion with ˙a/a = O(1). However, in the vacuum case considered in this article, the temporal evolution of the scale factor is slow. We will show a posteriori that ˙a is of the same order as the weak gravitational wave perturbation hµκ, i.e., ˙a ∼ √ ¨a ∼h ∼ϵ ≪1 and thus nonlinear gravitational wave interactions cannot be neglected when describing the expansion of the universe. In our analysis, we will utilize these scalings and perform a formal expansion on Equation (A1) in ϵ. 6. Conclusions Clearly, these few examples underline the need to better understand the role of nonlinearities in cosmology that have been underestimated so far. Universe 2020, 6, 98 Universe 2020, 6, 98 10 of 16 10 of 16 Author Contributions: All authors have contributed equally to the different parts of the paper. All authors have read and agreed to the published version of the manuscript. Funding: Sergey Nazarenko is supported by the Chaire D’Excellence IDEX (Initiative of Excellence) awarded by Université de la Côte d’Azur, France and by the Simons Foundation Collaboration grant ‘Wave Turbulence’ (Award ID 651471). This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 823937 (Hydrodynamic approach to Light Turbulence). Appendix A. Universe Expansion Driven by Nonlinear Gravitational Wave Interactions The Ricci tensor Rµκ can be expressed in terms of Christoffel symbols Γλ µκ via Rµκ = ∂κΓν µν −∂νΓν µκ + Γα µνΓν κα −Γα µκΓν να (A3) Rµκ = ∂κΓν µν −∂νΓν µκ + Γα µνΓν κα −Γα µκΓν να (A3) where the Christoffel symbol is deﬁned in terms of the metric gµκ and co-metric gµκ as Γσ µκ = 1 2 gνσ ∂µgκν + ∂κgµν −∂νgκµ (A4) (A4) Universe 2020, 6, 98 Universe 2020, 6, 98 11 of 16 By substituting (A2) into the expressions (A4) and (A3) we produce a formal ϵ-expansion of the vacuum equations, with each order in ϵ denoted as follows, Rµκ = R(0) µκ + ϵR(1) µκ + ϵ2R(2) µκ + · · · = 0 As the Einstein vacuum equations lead to an over-determined system of equations for the perturbation hµκ, we can ﬁx the coordinate system without loss of generality. We apply the four harmonic gauge conditions Hλ = gµκΓλ µκ = 0 (A5) (A5) deﬁned up to the linear perturbation of hµκ. The harmonic gauge conditions (A5) still leaves coordinate freedom with respect to any additional harmonic perturbation. Therefore, we can additionally choose the space-time coordinates in such a way that h00 = 0, and hl0 = 0, for l = 1, 2, 3 We ﬁnd that at order O(ϵ0), R(0) µκ = 0, i.e., a pure Minkowski space with hµν = 0 and a = constant. At order O(ϵ), we ﬁnd We ﬁnd that at order O(ϵ0), R(0) µκ = 0, i.e., a pure Minkowski space with hµν = 0 and a = constant. At order O(ϵ), we ﬁnd R(1) 00 = 1 2a2 ¨hjj R(1) ll = −1 2c2 ¨hll + 1 2a2 ∂2 llhjj −1 a2 ∂l∂jhlj + 1 2a2 ∂2 jjhll R(1) l0 = 1 2a2 ∂l ˙hjj −1 2a2 ∂j ˙hlj R(1) lm = 1 2a2 ∂l∂mhjj −1 2c2 ¨hlm −1 2a2 ∂l∂jhmj −1 2a2 ∂m∂jhlj + 1 2a2 ∂2 jjhml where l, m = 1, 2, 3 are ﬁxed and j = 1, 2, 3 is explicitly summed over such that j ̸= l ̸= m. Here ˙ denotes the temporal derivative. The harmonic gauge conditions (A5) give where l, m = 1, 2, 3 are ﬁxed and j = 1, 2, 3 is explicitly summed over such that j ̸= l ̸= m. Here ˙ denotes the temporal derivative. Appendix A. Universe Expansion Driven by Nonlinear Gravitational Wave Interactions The harmonic gauge conditions (A5) give H0(1) = 1 2c2a2 ˙hjj = 0 Hl(1) = 1 2a4 ∂lhll −1 2a4 ∂lhjj + 1 a4 ∂jhjl = 0 H0(1) = 1 2c2a2 ˙hjj = 0 Hl(1) = 1 2a4 ∂lhll −1 2a4 ∂lhjj + 1 a4 ∂jhjl = 0 where l = 1, 2, 3 is ﬁxed and j = 1, 2, 3 is summed over such that j ̸= l. where l = 1, 2, 3 is ﬁxed and j = 1, 2, 3 is summed over such that j ̸= l. where l = 1, 2, 3 is ﬁxed and j = 1, 2, 3 is summed over such that j ̸= l. Using the harmonic gauge conditions, we can simplify the Einstein vacuum equations at O(ϵ) to wave equations of the form: R(1) lm = −1 2c2 ¨hlm + 1 2a2 ∇2hlm = 0 (A6) (A6) for l, m = 1, 2, 3. It is important to note that Equation (A6) depends on a time dependent scale factor a(t) whose evolution can only be described by considering the subsequent order in ϵ. The term R(2) µκ involves a mix of contributions arising from terms quadratic in h (denote them hhR(2) µκ ) and linear in h contributions ∼˙ah (denote them ahR(2) µκ ), and ﬁnally contributions of type ( ˙a/a)2 and ¨a/a (denote them aR(2)). Then for l, m = 1, 2, 3. It is important to note that Equation (A6) depends on a time dependent scale factor a(t) whose evolution can only be described by considering the subsequent order in ϵ. The term R(2) µκ involves a mix of contributions arising from terms quadratic in h (denote them hhR(2) µκ ) and linear in h contributions ∼˙ah (denote them ahR(2) µκ ), and ﬁnally contributions of type ( ˙a/a)2 and ¨a/a (denote them aR(2)). Then R(2) µκ =hhR(2) µκ + ahR(2) µκ + aR(2) µκ R(2) µκ =hhR(2) µκ + ahR(2) µκ + aR(2) µκ Universe 2020, 6, 98 Universe 2020, 6, 98 12 of 16 Using the harmonic gauge, the order O(ϵ2) contribution to the Einstein vacuum equations simplify to aR(2) 00 =3 ˙a2 a2 aR(2) ll = −a¨a + 2 ˙a2 c2 where l = 1, 2, 3 with no summation, with the rest of the components of aR(2) µκ being zero. Appendix A. Universe Expansion Driven by Nonlinear Gravitational Wave Interactions I t f ith + d l i ti h th t µ 0 In terms of waves with + and × polarizations, we have that 0 with + and × polarizations, we have that 0 In terms of waves with + and × polarizations, we have that hµκ =(2π)−3 Z ˆλk ˆeλ + ˆµk ˆeµ µκ exp(ik · x) dk where ˆeλ = Rˆe+RT and ˆeµ = Rˆe×RT are deﬁned through rotations R from z-direction to the direction of k of the elementary perturbations where ˆeλ = Rˆe+RT and ˆeµ = Rˆe×RT are deﬁned through rotations R from z-direction to the direction of k of the elementary perturbations ˆe+ =   0 0 0 0 0 1 0 0 0 0 −1 0 0 0 0 0  , and ˆe× =   0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0   ˆe+ =   0 0 0 0 0 1 0 0 0 0 −1 0 0 0 0 0  , and ˆe× =   0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0   We consider the monochromatic wave amplitudes ˆλk and ˆµk of the form Appendix A. Universe Expansion Driven by Nonlinear Gravitational Wave Interactions We also determine that where l = 1, 2, 3 with no summation, with the rest of the components of aR(2) µκ being zero. We also determine that ahR(2) lm = ˙a ˙hlm 2c2a (A7) (A7) and ahR(2) 0m = ahR(2) l0 = 0. The form of (A7) is precisely what results from the second term of Equation (8) in Ref. [54]. The complete expression for hhR(2) µκ is lengthy and will not be reproduced here. It can be determined through the use of computational algebra software such as Mathematica. Note that by considering the Einstein vacuum equations at order O(ϵ2), it gives rise to two equations: one equation for the mean and one equation for the ﬂuctuations. Let us introduce the spatial average ⟨·⟩:= limV→∞[(1/V) R V · dx], then up to order O(ϵ2) we have ⟨Rµκ⟩=⟨R(0) µκ + R(1) µκ + R(2) µκ ⟩= ⟨R(2) µκ ⟩= 0 (A8) (A8) where we have taken into account that R(0) µκ = 0, while R(1) µκ is linear in terms of the perturbation hµκ and vanishes upon the spatial average. (Actually, this implies that the averaged metric remains of FLRW form which will be seen from the structure of the ﬁnal averaged equations.) Now note that ahR(2) µκ has zero mean (being linear in hµκ) and that ⟨aR(2) µκ ⟩= aR(2) µκ . We evaluate terms ⟨hhR(2) µκ ⟩assuming a distribution of gravitational waves with amplitudes hµκ = (2π)−3 Z ˆhµκ(k, t) exp(ik · x) dk hµκ = (2π)−3 Z ˆhµκ(k, t) exp(ik · x) dk where ˆhµκ(k, t) are time-dependent wave amplitudes for a wave with wave vector k = (k1, k2, k3). For the calculation below, it sufﬁces to take the leading order time dependence of ˆhk(t) which follows from the wave Equation (A6), namely ˆhµκ(k, t) ∼exp(−i R t 0 ωk dt′) with ωk = c\|k\|/a(t′). where ˆhµκ(k, t) are time-dependent wave amplitudes for a wave with wave vector k = (k1, k2, k3). For the calculation below, it sufﬁces to take the leading order time dependence of ˆhk(t) which follows from the wave Equation (A6), namely ˆhµκ(k, t) ∼exp(−i R t 0 ωk dt′) with ωk = c\|k\|/a(t′). We consider the monochromatic wave amplitudes ˆλk and ˆµk of the form ˆλk =(2π)3[Λ0δ(k −k0) + Λ∗ 0δ(k + k0)] ˆµk =(2π)3[M0δ(k −k0) + M∗ 0δ(k + k0)] Universe 2020, 6, 98 13 of 16 which we substitute into the expression for hhR(2) µκ and apply space averaging, with the aid of the Mathematica software, to get a rather simple result: which we substitute into the expression for hhR(2) µκ and apply space averaging, with the aid of the Mathematica software, to get a rather simple result: ⟨hhR(2) 00 ⟩=c2\|k0\|2 a6 (\|Λ0\|2 + \|M0\|2) ⟨hhR(2) 0l ⟩=c\|k0\|k0l a5 (\|Λ0\|2 + \|M0\|2) ⟨hhR(2) lm ⟩=k0lk0m a4 (\|Λ0\|2 + \|M0\|2) l, m = 1, 2, 3 One can further consider the case of an homogeneous and isotropic distribution of gravitational waves by angle-averaging the above formulas. This gives One can further consider the case of an homogeneous and isotropic distribution of gravitational waves by angle-averaging the above formulas. This gives ⟨hhR(2) 00 ⟩=2πc2 a6 Z ∞ 0 k4(nλ k + nµ k ) dk ⟨hhR(2) 0l ⟩=⟨hhR(2) lm ⟩= 0 for l ̸= m ⟨hhR(2) ll ⟩= 2π 3a4 Z ∞ 0 k4(nλ k + nµ k ) dk where spectra nλ k and nµ k are deﬁned via where spectra nλ k and nµ k are deﬁned via ⟨ˆλk ˆλk′⟩=(2π)3nλ k δ(k −k′) ⟨ˆµk ˆµk′⟩=(2π)3nµ k δ(k −k′) where averaging now is over many periods of wave oscillations or, equivalently, wave phases. By deﬁning the gravitational wave energy density ρ [55] by where averaging now is over many periods of wave oscillations or, equivalently, wave phases. By deﬁning the gravitational wave energy density ρ [55] by ρ = c2 4Ga6 Z ∞ 0 k4(nλ k + nµ k ) dk, (A9) (A9) that acts on the 00 component, the pressure P = c2ρ/3 as the contribution acting on the jj components, and G as the Einstein constant. We consider the monochromatic wave amplitudes ˆλk and ˆµk of the form Equation (A8) for the 00 and the jj components (j = 1, 2 or 3) become respectively 3¨a a + 8πGρ =0 −(a¨a + 2 ˙a2) c2 + 8πGa2P c4 =0 By rearranging and using the fact that P = c2ρ/3 one get the usual expressions of the Friedmann equations By rearranging and using the fact that P = c2ρ/3 one get the usual expressions of the Friedmann equations ¨a a = −8πGρ 3 = −4πG 3 ρ + 3P c2 (A10) ˙a2 a2 =4πG 3 ρ + 3P c2 = 8πGρ 3 (A11) (A10) (A11) that can be rearranged further into the Friedmann equations of the form (1) and (2). These equations show that our initial scaling ˙a ∼h ∼ √ ¨a ∼ϵ is justiﬁed. Also, Equation (A8) for the off-diagonal components become 0 = 0 identities which mean that the averaged metric remains of FLRW type. Keeping in mind the relation P = c2ρ/3, which is typical for radiation or ultra-relativistic matter, we see that gravitational waves make the universe expand in a way that usual radiation would. However, it remains to be seen how the expansion itself affects the gravitational waves. The crucial 14 of 16 Universe 2020, 6, 98 difference with usual (e.g., electromagnetic) radiation is that the waves are nonlinear and, therefore, may interact with each other and thus modify the usual energy dilution process. To describe the evolution of the gravitational waves we need an equation for the ﬂuctuating ﬁeld up to order O(ϵ2), i.e., R(0) µκ + R(1) µκ + R(2) µκ −⟨R(0) µκ + R(1) µκ + R(2) µκ ⟩= 0 Recall that R(0) µκ = ⟨R(1) µκ ⟩= ⟨ahR(2) lm ⟩= 0, ⟨aR(2) µκ ⟩= aR(2) µκ , and that R(1) lm and ahR(2) lm are given by Equations (A6) and (A7) respectively. Then, this enables us to write the ﬂuctuation equation as Recall that R(0) µκ = ⟨R(1) µκ ⟩= ⟨ahR(2) lm ⟩= 0, ⟨aR(2) µκ ⟩= aR(2) µκ , and that R(1) lm and ahR(2) lm are given by Equations (A6) and (A7) respectively. Then, this enables us to write the ﬂuctuation equation as 1 c2 ¨hlm −1 a2 ∇2hlm −˙a ˙hlm c2a = 2 ⟨hhR(2) lm ⟩−hhR(2) lm (A12) (A12) for l, m = 1, 2, 3. We consider the monochromatic wave amplitudes ˆλk and ˆµk of the form If we neglect the right-hand side of (A12), we would arrive at the familiar equation for the gravitational wave dilution found in [53,54], and by neglecting the sub-leading order in ϵ, its solution is hlm ∼a(t) exp −ic\|k\| Z t 0 1 a(t′) dt′ + ik · x which leads to the typical dilution law of radiation ρ ∼1/a4. The right-hand side of Equation (A12) describes nonlinear wave–wave interactions. Neglecting this term would be justiﬁed if the universe would be ﬁlled with a substance causing it to expand fast, so that ˙a = O(1). The key point is that in vacuum space, considered here, ˙a and h are of the same order of magnitude and, therefore, in general the right-hand side of Equation (A12) is of the same order as the dilution term ∼˙a ˙hlm, i.e., the timescales of the dilution and the wave–wave interaction are comparable. Note that randomness of phases may weaken the wave–wave interactions, as is the case for weak gravitational turbulence [14]. However, when the wave amplitudes are not weak and/or the phases are not random, the wave–wave interaction term becomes important and one has to seek a reasonable model closure for its description. This is precisely the critical balance approach suggested in the main text. References Turbulence of weak gravitational waves in the early Universe. Phys. Rev. 2017, 119, 221101. [CrossRef] [PubMed] 15. Goldreich, P.; Sridhar, S. Toward a theory of interstellar turbulence. 2: Strong alfvenic turbulence. Astrophys. J. 1995, 438, 763–775. [CrossRef] 16. Nazarenko, S.V.; Schekochihin, A.A. Critical balance in magnetohydrodynamic, rotating and stratiﬁed turbulence: Towards a universal scaling conjecture. J. Fluid Mech. 2011, 677, 134–153. [CrossRef] 7. Newell, A.C.; Rumpf, B. Wave Turbulence. Ann. R 17. Newell, A.C.; Rumpf, B. Wave Turbulence. Ann. Rev. Fluid Mech. 2011, 43, 59–78. [CrossRef] 18. Passot, T.; Sulem, P.L. A Model for the Non-universal Power Law of the Solar Wind Sub-ion-scale Magnetic Spectrum. Astrophys. J. 2015, 812, L37. [CrossRef] 19. Meyrand, R.; Galtier, S.; Kiyani, K.H. Direct Evidence of the Transition from Weak to Strong Magnetohydrodynamic Turbulence. Phys. Rev. Lett. 2016, 116, 105002. [CrossRef] 0. Alexakis, A.; Biferale, L. Cascades and transitions in turbulent ﬂows. Phys. Rep. 2018, 767, 1–101. [Cross 1 Wh l J A G Ph R 1955 97 511 536 [C R f] 0. Alexakis, A.; Biferale, L. Cascades and transitions in turbulent ﬂows. Phys. Rep. 2018, 767, 1–101. [Cross 1. Wheeler, J.A. Geons. Phys. Rev. 1955, 97, 511–536. [CrossRef] 21. Wheeler, J.A. Geons. Phys. Rev. 1955, 97, 511–536. [CrossRef] 21. Wheeler, J.A. Geons. Phys. Rev. 1955, 97, 511–536. [CrossRef] 2. Carr, B.; Kühnel, F.; Sandstad, M. Primordial black holes as dark matter. Phys. Rev. D 2016, 94, 083 [CrossRef] 23. Hawking, S.W.; Moss, I.G.; Stewart, J.M. Bubble collisions in the very early universe. Phys. Rev. D 1982, 26, 2681–2693. [CrossRef] 24. Hawking, S. Particle creation by black-holes. Commun. Math. Phys. 1975, 43, 199–220. [CrossRef] 25. Clough, K.; Niemeyer, J.C. On the difﬁculty of generating gravitational wave turbulence in the early universe. Class. Quant. Grav. 2018, 35, 187001. [CrossRef] 26. Peter, P.; Uzan, J.P. Primordial Cosmology; Oxford University Press: Oxford, UK, 2013. 26. Peter, P.; Uzan, J.P. Primordial Cosmology; Oxford 27. Isaacson, R.A. Gravitational Radiation in the Limit of High Frequency. II. Nonlinear Terms and the Effective Stress Tensor. Phys. Rev. 1968, 166, 1272–1279. [CrossRef] 28. Fjørtoft, R. On the Changes in the Spectral Distribution of Kinetic Energy for Two-dimensional, Non-divergent Flow. Tellus 1953, 5, 225. [CrossRef] 29. Newell, A.C.; Nazarenko, S.; Biven, L. Wave turbulence and intermittency. Phys. D Nonlinear Phenom. 2001, 152–153, 520–550. [CrossRef] 30. Galtier, S.; Nazarenko, S.V.; Buchlin, E.; Thalabard, S. Nonlinear diffusion models for gravitational wave turbulence. Phys. References 1. Einstein, A. Die Feldgleichungen der Gravitation. In Sitzungsberichte der Königlich Preußischen Akademie der Wissenschaften (Berlin); Springer: Berlin, Germany,1915; pp. 844–847. 2. Weinberg, S. Cosmology; Oxford University Press: New York, NY, USA, 2008. 3. Guth, A.H. Inﬂationary universe: A possible solution to the horizon and ﬂatness problems. Phys. Rev. D 1981, 23, 347–356. [CrossRef] 4. Linde, A.D. A new inﬂationary universe scenario: A possible solution of the horizon, ﬂatness, homogeneity, isotropy and primordial monopole problems. Phys. Lett. B 1982, 108, 389–393. [CrossRef] 5. Peter, P.; Pinto-Neto, N. Cosmology without inﬂation. Phys. Rev. D 2008, 78, 063506. [CrossRef] 5. Peter, P.; Pinto-Neto, N. Cosmology without inﬂation. Phys. Rev. D 2008, 78, 063506. [CrossRef 6. Planck Collaboration; Ade, P.A.R.; Aghanim, A.; Arnaud, M.; Arroja, F.; Ashdown, M.; Aumont, J.; Baccigalupi, C.; Ballardini, M.; Banday, A.J.; et al. Planck 2015 results. XX. Constraints on inﬂation. Astron. Astrophys. 2016, 594, A20. 6. Planck Collaboration; Ade, P.A.R.; Aghanim, A.; Arnaud, M.; Arroja, F.; Ashdown, M.; Aumont, J.; Baccigalupi, C.; Ballardini, M.; Banday, A.J.; et al. Planck 2015 results. XX. Constraints on inﬂation. Astron. Astrophys. 2016, 594, A20. 7. Sirunyan, A.; CMS Collaboration. Observation of t¯tH production. Phys. Rev. Lett. 2018, 120, 231801. [CrossRef] [PubMed] 7. Sirunyan, A.; CMS Collaboration. Observation of t¯tH production. Phys. Rev. Lett. 2018, 120, 231801. [CrossRef] [PubMed] 8. Binétruy, P.; Bohé, A.; Caprini, C.; Dufaux, J.F. Cosmological backgrounds of gravitational waves and eLISA/NGO: Phase transitions, cosmic strings and other sources. J. Cosm. Astrop. Phys. 2012, 6, 027. [CrossRef] 9. Goldwirth, D.; Piran, T. Initial conditions for inﬂation. Phys. Rep. 1992, 214, 223–292. [CrossRef] 9. Goldwirth, D.; Piran, T. Initial conditions for inﬂation. Phys. Rep. 1992, 214, 223 292. [CrossRef] 10. Hollands, S.; Wald, R. Essay: An Alternative to Inﬂation. Gen. Relat. Grav. 2002, 34, 2043–2055. [CrossRef] 11. Ijjas, A.; Steinhardt, P.; Loeb, A. Inﬂationary paradigm in trouble after Planck. Phys. Lett. B 2013, 723, 261–266. [CrossRef] 10. Hollands, S.; Wald, R. Essay: An Alternative to Inﬂation. Gen. Relat. Grav. 2002, 34, 2043–2055. 11. Ijjas, A.; Steinhardt, P.; Loeb, A. Inﬂationary paradigm in trouble after Planck. Phys. Lett. B 2013, 723, 261–266. [CrossRef] 15 of 16 Universe 2020, 6, 98 12. Ijjas, A.; Steinhardt, P.; Loeb, A. Inﬂationary schism. Phys. Lett. B 2014, 736, 142–146. [CrossRef] 13. Starobinsky, A.A. A new type of isotropic cosmological models without singularity. Phys. Lett. B 1980, 91, 99–102. [CrossRef] 4. Galtier, S.; Nazarenko, S.V. References D Nonlinear Phenom. 2019, 390, 84–88. [CrossRef] 1. Nazarenko, S.V. Wave Turbulence. In Lecture Notes in Physics; Springer: Berlin, Germany, 2011; Volume 32. Maggiore, M. Gravitational Waves; Oxford University Press: Oxford, UK, 2008; Volume 1. 33. Dyachenko, S.; Newell, A.C.; Pushkarev, A.; Zakharov, V.E. Optical turbulence: Weak turbulence, condensates and collapsing ﬁlaments in the nonlinear Schrödinger equation. Phys. D Nonlinear Phenom. 1992, 57, 96–160. [CrossRef] 34. Galtier, S.; Nazarenko, S.V.; Newell, A.C.; Pouquet, A. A weak turbulence theory for incompressible magnetohydrodynamics. J. Plasma Phys. 2000, 63, 447–488. [CrossRef] 35. Lacaze, R.; Lallemand, P.; Pomeau, Y.; Rica, S. Dynamical formation of a Bose-Einstein condensate. Phys. D Nonlinear Phenom. 2001, 152, 779–786. [CrossRef] 36. Semikoz, D.V.; Tkachev, I.I. Kinetics of Bose Condensation. Phys. Rev. Lett. 1995, 74, 3093–3097. [CrossRef] [PubMed] 37. Zhao, D.; Yu, M. Generalized nonlinear Schrödinger equation as a model for turbulence, collapse, and inverse cascade. Phys. Rev. E 2011, 83, 036405. [CrossRef] 38. Miller, P.; Vladimirova, N.; Falkovich, G. Oscillations in a turbulence-condensate system. Phys. Rev. E 2013, 87, 065202. [CrossRef] [PubMed] 39. Reeves, M.; Billam, T.; Anderson, B.; Bradley, A. Inverse Energy Cascade in Forced Two-Dimensional Quantum Turbulence. Phys. Rev. Lett. 2013, 110, 104501. [CrossRef] [PubMed] 40. Zakharov, V.; Nazarenko, S. Dynamics of the Bose-Einstein condensation. Phys. D Nonlinear Phenom. 2005, 201, 203–211. [CrossRef] 41. Dodelson, S.; Gates, E.; Turner, M. Cold Dark Matter. Science 1996, 274, 69–75. [CrossRef] Universe 2020, 6, 98 16 of 16 16 of 16 42. Semikoz, D.V.; Tkachev, I.I. Condensation of bosons in the kinetic regime. Phys. Rev. D 1997, 55, 489–502. [CrossRef] 43. Connaughton, C.; Nazarenko, S. Warm Cascades and Anomalous Scaling in a Diffusion Model of Turbulence. Phys. Rev. Lett. 2004, 92, 044501. [CrossRef] [PubMed] 44. Thalabard, S.; Nazarenko, S.; Galtier, S.; Medvedev, S. Anomalous spectral laws in differential models of turbulence. J. Phys. A Math. Theor. 2015, 48, 285501. [CrossRef] 45. Boyle, L.; Steinhardt, P.; Turok, N. Inﬂationary Predictions for Scalar and Tensor Fluctuations Reconsidered. Phys. Rev. Lett. 2006, 96, 111301. [CrossRef] 46. Nazarenko, S.; Onorato, M. Wave turbulence and vortices in Bose Einstein condensation. Phys. D Nonlinear Phenom. 2006, 219, 1–12. [CrossRef] 47. Nazarenko, S.; Onorato, M. Freely decaying Turbulence and Bose–Einstein Condensation in Gross–Pitaevski Model. J. Low Temp. Phys. 2007, 146, 31–46. [CrossRef] 48. Antoniadis, I.; Mazur, P.O.; Mottola, E. Conformal invariance, dark energy, and CMB non-gaussianity. J. Cosmol. Astropart. Phys. 2012, 2012, 024. [CrossRef] 49. Polyakov, A.M. c⃝2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). References The theory of turbulence in two dimensions. Nuclear Phys. B 1993, 396, 367–385. [Cr 50. Frisch, U. Turbulence; Cambridge University Press: Cambridge, UK, 1995. Frisch, U. Turbulence; Cambridge University Press: Cambridge, UK, 1995. Chevalier, C.; Debbasch, F.; Ollivier, Y. Multiscale cosmological dynamics. Phys. A Statis. Mech. Appl 2009, 388, 5029–5035. [CrossRef] 52. Belinski, V.A.; Vereshchagin, G.V. On the cosmological gravitational waves and cosmological distances. Phys. Lett. B 2018, 778, 332–338. [CrossRef] 53. Lifshitz, E. On the Gravitational Stability of the Expanding Universe. J. Phys. (USSR) 1946, 10, 116. 54. Grishchuk, L. Ampliﬁcation of Gravitational Waves in an Isotropic Universe. JETP 1974, 40, 409. 55. Guzzetti, M.; Bartolo, N.; Liguori, M.; Matarrese, S. Gravitational Waves From Inﬂation. Rivista Del Nuovo Cimento 2016, 39, 339. c⃝2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
https://openalex.org/W2734668696	https://europepmc.org/articles/pmc5532622?pdf=render	English	null	Phosphoinositide 3-Kinase-Dependent Signalling Pathways in Cutaneous Squamous Cell Carcinomas	Cancers	2,017	cc-by	13,275	Received: 2 May 2017; Accepted: 3 July 2017; Published: 11 July 2017 Received: 2 May 2017; Accepted: 3 July 2017; Published: 11 July 2017 Abstract: Cutaneous squamous cell carcinoma (cSCC) derives from keratinocytes in the epidermis and accounts for 15–20% of all cutaneous malignancies. Although it is usually curable by surgery, 5% of these tumours metastasise leading to poor prognosis mostly because of a lack of therapies and validated biomarkers. As the incidence rate is rising worldwide it has become increasingly important to better understand the mechanisms involved in cSCC development and progression in order to develop therapeutic strategies. Here we discuss some of the evidence indicating that activation of phosphoinositide 3-kinases (PI3Ks)-dependent signalling pathways (in particular the PI3Ks targets Akt and mTOR) has a key role in cSCC. We further discuss available data suggesting that inhibition of these pathways can be beneﬁcial to counteract the disease. With the growing number of different inhibitors currently available, it would be important to further investigate the speciﬁc contribution of distinct components of the PI3Ks/Akt/mTOR pathways in order to identify the most promising molecular targets and the best strategy to inhibit cSCC. Keywords: Akt; cutaneous squamous cell carcinoma; mTOR; PI3K Review Review Phosphoinositide 3-Kinase-Dependent Signalling Pathways in Cutaneous Squamous Cell Carcinomas Joanna M. Janus, Ryan F. L. O’Shaughnessy, Catherine A. Harwood and Tania Maffucci * Centre for Cell Biology and Cutaneous Research, Blizard Institute, Barts and The London School of Medicine and Dentistry, Queen Mary University of London, London E1 2AT, UK; j.m.janus@qmul.ac.uk (J.M.J.); r.f.l.oshaughnessy@qmul.ac.uk (R.F.L.O.); caharwood@doctors.org.uk (C.A.H.) * Correspondence: t.maffucci@qmul.ac.uk; Tel.: +44-020-7882-8423 cancers cancers cancers Cancers 2017, 9, 86; doi:10.3390/cancers9070086 www.mdpi.com/journal/cancers 2. The Epidermis 2. The Epidermis The epidermis of the skin contains stratiﬁed layers of squamous epithelium (Figure 1), mostly consisting of keratinocytes [23]. Keratinocytes are specialised cells named after their ability to produce keratin, a protein essential in the formation of intermediate ﬁlaments and in maintaining the barrier function of the skin. Keratinocytes continuously divide in the basal layer of the epidermis, and then differentiate as they migrate upwards through the spinous and granular layers towards the surface of the skin to ultimately form a layer of anucleate corniﬁed cells called the stratum corneum [24–26]. As the cells migrate upwards they become more ﬂattened and synthesise a number of different proteins (including different keratins) and lipids from specialised organelles, such as lamellar bodies and keratohyalin granules [27]. Intercellular junctions, such as desmosomes, are crucial to maintain the barrier function and modulate cell signalling [28]. The different desmosomal components have speciﬁc expression patterns within the epidermis and this is important to control not only the structure, but also the speciﬁc function of each stratum [29]. By the time they reach the surface the keratinocytes have become denucleated and form the tough keratinised layer of the stratum corneum, allowing the skin to remain waterproof and resistant to external stresses [23]. The epidermis of the skin contains stratified layers of squamous epithelium (Figure 1), mostly consisting of keratinocytes [23]. Keratinocytes are specialised cells named after their ability to produce keratin, a protein essential in the formation of intermediate filaments and in maintaining the barrier function of the skin. Keratinocytes continuously divide in the basal layer of the epidermis, and then differentiate as they migrate upwards through the spinous and granular layers towards the surface of the skin to ultimately form a layer of anucleate cornified cells called the stratum corneum [24–26]. As the cells migrate upwards they become more flattened and synthesise a number of different proteins (including different keratins) and lipids from specialised organelles, such as lamellar bodies and keratohyalin granules [27]. Intercellular junctions, such as desmosomes, are crucial to maintain the barrier function and modulate cell signalling [28]. The different desmosomal components have specific expression patterns within the epidermis and this is important to control not only the structure, but also the specific function of each stratum [29]. 1. Introduction Keratinocyte carcinomas (KC), comprising basal cell carcinoma (BCC) and cutaneous squamous cell carcinoma (cSCC), are the main forms of non-melanoma skin cancers (NMSC). They represent one third of all malignancies [1,2] and are the most common malignancy in the UK [3]. In 2014, there were 131,772 cases of NMSC registered in the UK, although this is a signiﬁcant underestimation as there are acknowledged problems of under-recording [4]. The crude incidence rate indicates 233 new NMSC cases for every 100,000 males and 176 for every 100,000 females [5]. A recent study reported that approximately 3.3 million people were treated for NMSC in USA in 2012 [6]. More worryingly, the incidence of NMSC has risen over the years and it is still rising worldwide [7–12]. For instance one study estimated that on average the incidence of NMSC has increased by 3–8% yearly among white populations in Australia, Canada, Europe, and the USA in the last 30 years [12]. Morbidity associated with NMSC is high and available treatments can be disﬁguring and expensive. One study estimated that in 2008 the cost due to skin cancer was in the range of £106–112 million in England, with expected cost per case estimated at £889–1226 for NMSC (bottom-up and top-down approaches) [13]. p p p pp Approximately 75–80% of KC are BCC and 18–20% are cSCC [7,14]. While BCC is usually a localised cancer, approximately 5% of cSCC are able to metastasise, usually to lymph nodes [2,15]. As a consequence of this, although 95% of cSCC are curable with surgical resection, it has been estimated that 20% of skin cancer deaths are attributable to cSCC [16]. Indeed the ability of cSCC to metastasise leads to a 3-year disease-free survival rate of 56% [17] and a ﬁve-year survival rate of 25% to 35% [18–21]. Such a poor prognosis is due to a lack of therapies for this subset of patients Cancers 2017, 9, 86; doi:10.3390/cancers9070086 www.mdpi.com/journal/cancers 2 of 18 2 f 17 Cancers 2017, 9, 86 as currently there is no FDA-approved therapy with a speciﬁc indication for metastatic cSCC [22]. The development of therapies is further complicated by the fact that no molecular biomarkers that can predict disease behaviour or treatment response have been validated [22]. 1. Introduction With the rising incidence of this disease, a better understanding of the biochemical pathways involved in cSCC development and progression is urgently needed in order to identify molecular targets and design drugs that can be beneﬁcial to patients. currently there is no FDA-approved therapy with a specific indication for metastatic cSCC [22]. The development of therapies is further complicated by the fact that no molecular biomarkers that can predict disease behaviour or treatment response have been validated [22]. With the rising incidence of this disease, a better understanding of the biochemical pathways involved in cSCC development and progression is urgently needed in order to identify molecular targets and design drugs that can be beneficial to patients. 3. Overview of cSCC Carcinogenesis Chronic exposure to UV radiation has been described as the most important environmental risk factor for cSCC development, with other factors, including exposure to ionising agents and chemical carcinogens, also identiﬁed [31]. Indeed the majority of cSCC occurs on sun-exposed areas of the body and has been strongly associated with chronic sun exposure [32]. Approximately 65% of cSCC arise from dysplastic regions in the epidermis known as actinic keratoses (AK), which occur as a result of increased UV exposure [33]. The factors responsible for this progression are, however, still largely unknown and indeed not all AK progress to cSCC [34,35]. Nevertheless, AK are an important clinical risk factor for cSCC [36]. Genetically, cSCC is a very heterogeneous disease. Chromosomal changes have been identiﬁed by genome-wide studies, and mainly comprise loss of heterozygosity due to allelic loss and uniparental disomy at 3p, 9p, 2q, 8p and 13, and allelic gain on 3q and 8q [37,38]. Mutations in the Notch gene family and many other key genes, including TP53, have also been reported [39]. In fact, because of the complex mutational patterns, it is very difﬁcult to identify driver genes in cSCC and this has strongly limited the translation from genomics to the clinic [39]. Indeed while identiﬁcation of mutations in BRAF for advanced melanoma and Hedgehog signalling for BCC has paved the road to clinical use of BRAF and smoothened inhibitors respectively, a similar direct translation has not occurred in cSCC [39]. Nevertheless accumulating evidence from clinical use of epidermal growth factor receptor inhibitors or immune modulatory drugs suggests that targeted therapies may be beneﬁcial [39,40]. There is, therefore, an urgent need to deﬁne the critical molecular mechanisms and key signalling pathways involved in cSCC carcinogenesis in order to identify new molecular targets. y g g p y g y g It is now well documented that alteration of speciﬁc signalling pathways occurs during cSCC carcinogenesis. For instance reverse phase protein microarray analysis revealed speciﬁc activation of the mitogen-activated protein kinase (MAPK) pathway in cSCC compared to AK and normal skin [41]. Similarly, a core set of 196 genes was found to be differentially expressed between AK and cSCC and gene set enrichment analysis indicated a key role for MAPK pathway in cSCC compared to AK [42]. 2. The Epidermis 2. The Epidermis By the time they reach the surface the keratinocytes have become denucleated and form the tough keratinised layer of the stratum corneum, allowing the skin to remain waterproof and resistant to external stresses [23]. Figure 1. Representative diagram of the epidermis. The basement membrane, separating the dermis from the epidermis, and the distinct strata are indicated. Keratinocytes that have left the basal layer of skin are squamous in morphology therefore th Figure 1. Representative diagram of the epidermis. The basement membrane, separating the dermis from the epidermis, and the distinct strata are indicated. Figure 1. Representative diagram of the epidermis. The basement membrane, separating the dermis from the epidermis, and the distinct strata are indicated. Figure 1. Representative diagram of the epidermis. The basement membrane, separating the dermis from the epidermis, and the distinct strata are indicated. 3 of 18 Cancers 2017, 9, 86 Cancers 2017, 9, 86 Keratinocytes that have left the basal layer of skin are squamous in morphology therefore they are generally referred to as squamous cells and are the most abundant cell type within the epidermis. KC are classiﬁed as BCC or cSCC depending on their histopathological characteristics; BCC cells tend to resemble those from the basal layer of the epidermis whilst cSCC tend to resemble the squamous cells [30]. 3. Overview of cSCC Carcinogenesis Consistent with this, more recently it has been shown that inhibition of MEK causes senescence, but not apoptosis, in cSCC cell lines and reduces tumour growth in vivo [43]. Several lines of evidence also indicate that activation of the enzymes belonging to the phosphoinositide 3-kinase (PI3K) family is involved in cSCC carcinogenesis (as discussed in more detail below). 4. The PI3K Pathway in Epidermal Homeostasis Finally, data also indicate a role for PI3K in regulation of keratinocyte survival [62]. Transgenic mouse models have further supported a key role for PI3K-dependent pathways in epidermis. Mice bearing a keratinocyte-speciﬁc PTEN null mutation developed epidermal hyperplasia and hyperkeratosis [63]. A negative role for PTEN in regulation of skin growth was also conﬁrmed in another study describing the phenotype of mice carrying a speciﬁc deletion of PTEN in the skin [64]. Additional evidence includes characterisation of a conditional PDK1 knockout model (with PDK1 ablated in activated CD4 T cells, regulatory T cells and mature keratinocytes) that revealed a central role for this enzyme in keratinocytes homeostasis [65]. Similarly, another study reported that epidermis-speciﬁc PDK1 knockout mice displayed a thin and shiny epidermis and impaired barrier function and pointed to a role for this enzyme in asymmetric cell division in the epithelium [66]. Finally, the Akt1/Akt2 null mouse lacks the stratum corneum and dies neonatally, possibly because of defects in the skin barrier [67]. Possibly the most compelling evidence of a key role for PI3K-dependent pathways in skin derives from the observation that germline mutations of PTEN lead to a number of severe disorders known as PTEN hamartoma tumour syndromes (PHTS) which are characterised by hyperplastic changes in the skin [68]. A typical example of PHTS is Cowden Syndrome, where most patients develop skin hamartomas and various skin lesions [68,69]. While evidence in literature has demonstrated the importance of the family of PI3Ks and corresponding PI3Ks-dependent pathways, less attention has been paid to the fact that eight distinct PI3K isoforms exist which are grouped into three classes according to their structures and substrate speciﬁcity [45,46,70,71], as depicted in Figure 2. Class I PI3Ks are dimers comprising a catalytic and a regulatory subunit and they catalyse the synthesis of PIP3 in vivo. Class II PI3Ks are monomers that mainly catalyse the synthesis of phosphatidylinositol 3-phosphate (PI3P) in vivo although evidence also indicates that they can catalyse the synthesis of phosphatidylinositol 3,4-bisphosphate (PI(3,4)P2). Class III PI3K only catalyses the synthesis of PI3P [46,71]. Isoform speciﬁc knock-out and knock-in mice and the investigation of the effects of isoform-speciﬁc inhibitors have shed much light on our knowledge of the physiological roles and the cellular functions that are regulated by each PI3K. 4. The PI3K Pathway in Epidermal Homeostasis PI3Ks catalyse the phosphorylation of position 3 within the inositol ring of speciﬁc phosphoinositides leading to the synthesis of lipid products that can then bind and mediate the activation of many signalling molecules [44–47]. Due to the ability of their products to activate many downstream effectors, PI3Ks have a well-established role in regulation of several cellular processes, including cell proliferation, growth, survival, migration, and metabolism [44–47]. Amongst the many enzymes that are regulated by PI3Ks, 3-phosphoinositide-dependent protein kinase 1 (PDK1) and protein kinase B/Akt are by far the most studied and well-characterised. Upon activation, binding of the PI3K product phosphatidylinositol 3,4,5-trisphosphate (PIP3) to Akt induces translocation of this enzyme to the plasma membrane where it can be activated through phosphorylation at its residue Thr308 by PDK1 and at residue Ser473 by additional kinases, including the complex 2 of mechanistic target of rapamycin (mTORC2) [48,49]. Activated Akt in turn regulates a plethora of signalling molecules, ultimately controlling cell proliferation, cell cycle, survival, and migration [50,51]. Three Akt isoforms exist, with data pointing to speciﬁc, non-redundant roles for each of them, in particular in Cancers 2017, 9, 86 4 of 18 cancer [52]. One of the key enzymes regulated by Akt is mTOR, a master kinase involved in protein synthesis, ribosome biogenesis, autophagy and several other cellular functions [53,54]. Activation of PI3K is normally tightly regulated and activation of PI3K-dependent pathways is also controlled by speciﬁc phosphatases, including the tumour suppressor phosphatase and tensin homolog (PTEN) which dephosphorylates PIP3 and switches off the signals [55]. PI3K-dependent pathways are crucial for regulation of epidermal homeostasis [56–58]. Data obtained through overexpression of constitutively active and dominant negative PI3K indicated a role of this pathway in the early phases of keratinocytes differentiation [56]. Consistent with this, it was reported that pan-PI3K inhibition induced premature differentiation of keratinocytes [57]. Activation of PI3K was indeed detected in mouse primary keratinocytes upon induction of differentiation and this was mirrored by activation of Akt [58], also conﬁrmed by analysis of three-day old mouse skin that revealed increased active Akt in differentiating layers [58]. Activation of Akt has been associated with epidermal terminal differentiation with Akt1 in particular shown to be important for control of the barrier function of the corniﬁed layer [59,60]. In this respect recent data have pointed to a role for Akt1 on nuclear degradation and differentiation through lamin A/C degradation [61]. 4. The PI3K Pathway in Epidermal Homeostasis It is now well established that these enzymes are not redundant and play distinct roles [72–74], but few studies have investigated the potential contribution of each of the eight PI3K isoforms to normal skin homeostasis. 5 of 18 5 f 17 Cancers 2017, 9, 86 Figure 2. List of the eight mammalian PI3K isoforms and their classification into three distinct classes. For class I PI3Ks only the four catalytic subunits are shown. Their main lipid products are also indicated. Figure 2. List of the eight mammalian PI3K isoforms and their classiﬁcation into three distinct classes. For class I PI3Ks only the four catalytic subunits are shown. Their main lipid products are also indicated. Figure 2. List of the eight mammalian PI3K isoforms and their classification into three distinct classes. For class I PI3Ks only the four catalytic subunits are shown. Their main lipid products are also indicated. Figure 2. List of the eight mammalian PI3K isoforms and their classiﬁcation into three distinct classes. For class I PI3Ks only the four catalytic subunits are shown. Their main lipid products are also indicated. Expression of the class I PI3K catalytic subunits p110α and p110β was detected in mouse epidermis and in cultured murine keratinocytes [57]. Ribonucleotide protection assays also revealed the presence of a transcript encoding the class I isoform p110γ in murine skin although the protein could not be detected [57]. Interestingly, upregulation of p110γ both at the mRNA and protein levels was observed during wound repair, in particular during the inflammatory phase [57]. Analysis of three-day old mouse skin revealed a specific localisation of the class I regulatory subunit p85α at cell- cell contacts of suprabasal differentiating keratinocytes [58]. Expression of two members of the class II subfamily of PI3Ks has also been reported in human epidermis, with PI3K-C2α found to be expressed throughout the epidermis and PI3K-C2β mainly restricted to suprabasal layers [75]. To the best of our knowledge no study so far has specifically investigated the expression levels and localisation of the class III PI3K hVps34 in the epidermis. In this respect it is worth mentioning that a recent study reported that autophagy is important during epidermal development and differentiation [76]. Due to the role of hVps34 in regulation of autophagy [77] it would be important to investigate the potential contribution of this PI3K isoform to skin homeostasis. 4. The PI3K Pathway in Epidermal Homeostasis Expression of the class I PI3K catalytic subunits p110α and p110β was detected in mouse epidermis and in cultured murine keratinocytes [57]. Ribonucleotide protection assays also revealed the presence of a transcript encoding the class I isoform p110γ in murine skin although the protein could not be detected [57]. Interestingly, upregulation of p110γ both at the mRNA and protein levels was observed during wound repair, in particular during the inﬂammatory phase [57]. Analysis of three-day old mouse skin revealed a speciﬁc localisation of the class I regulatory subunit p85α at cell-cell contacts of suprabasal differentiating keratinocytes [58]. Expression of two members of the class II subfamily of PI3Ks has also been reported in human epidermis, with PI3K-C2α found to be expressed throughout the epidermis and PI3K-C2β mainly restricted to suprabasal layers [75]. To the best of our knowledge no study so far has speciﬁcally investigated the expression levels and localisation of the class III PI3K hVps34 in the epidermis. In this respect it is worth mentioning that a recent study reported that autophagy is important during epidermal development and differentiation [76]. Due to the role of hVps34 in regulation of autophagy [77] it would be important to investigate the potential contribution of this PI3K isoform to skin homeostasis. A transient upregulation of p110α and p110β was detected in differentiating primary human keratinocytes in vitro [57]. Similarly, treatment of cultured human keratinocytes with calcium induced phosphorylation of p85α as well as activation of all class I PI3K isoforms, as assessed by in vitro assays [78]. Another study however showed that overexpression of either dominant negative p85 mutant (Δp85) or constitutively active p110α (p110α CAAX) did not induce differentiation of primary human keratinocytes, as assessed by Western blotting analysis of involucrin expression levels [75]. These authors further showed that overexpression of the class II PI3K-C2β, but not PI3K- C2α, was able to induce differentiation of primary human keratinocytes in vitro, although downregulation of these enzymes, either alone or in combination, did not appear to affect their calcium-induced differentiation [75]. Importantly, no difference in epidermal differentiation was dete ted i t a e i i e ith eithe i ea ed o ab e t PI3K C2β e e io uli out a ajo A transient upregulation of p110α and p110β was detected in differentiating primary human keratinocytes in vitro [57]. 4. The PI3K Pathway in Epidermal Homeostasis Similarly, treatment of cultured human keratinocytes with calcium induced phosphorylation of p85α as well as activation of all class I PI3K isoforms, as assessed by in vitro assays [78]. Another study however showed that overexpression of either dominant negative p85 mutant (∆p85) or constitutively active p110α (p110α CAAX) did not induce differentiation of primary human keratinocytes, as assessed by Western blotting analysis of involucrin expression levels [75]. These authors further showed that overexpression of the class II PI3K-C2β, but not PI3K-C2α, was able to induce differentiation of primary human keratinocytes in vitro, although downregulation of these enzymes, either alone or in combination, did not appear to affect their calcium-induced differentiation [75]. Importantly, no difference in epidermal differentiation was detected in transgenic Cancers 2017, 9, 86 6 of 18 mice with either increased or absent PI3K-C2β expression, ruling out a major role for this enzyme in this process in vivo [75]. e with either increased or absent PI3K-C2β expression, ruling out a major role for this enzyme in process in vivo [75]. Evidence suggests that deregulation of PI3Ks-dependent pathways (possibly of speciﬁc PI3Ks-dependent pathways) can lead to alteration of the normal differentiation pattern and normal skin organisation. For instance, it was shown that stable overexpression of an inducible, constitutively-active mutant of p110α enhanced keratinocyte proliferation and migration, delayed differentiation in human keratinocytes and induced formation of disorganised, hyperplastic epithelium in organotypic skin cultures [57]. Selective roles for p110α or p110β were also reported in a transgenic mouse model which develops dermal lesions resembling PHTS [69]. By using mice lacking PTEN in epidermal keratinocytes (PTEN∆) and mice with concurrent ablation of either p110α or p110β or both PI3K isoforms, the authors showed that p110α mainly regulated survival of suprabasal keratinocytes while p110β mainly regulated proliferation of basal keratinocytes in such a context of PTEN loss. A similar distinct regulation of Akt activation in the two layers was also observed in these transgenic mice [69]. Importantly, while PTEN∆mice developed multiple cutaneous hamartomas, concurrent ablation of either p110α or p110β signiﬁcantly delayed both the development and severity of these skin lesions and simultaneous ablation of both PI3K isoforms completely prevented their development [69]. 4. The PI3K Pathway in Epidermal Homeostasis Relative mRNA levels of p110α and p110β were higher in cells from suprabasal and basal layers, respectively, and this was observed in cells from ear epidermis of both PTEN∆and wild-type mice, possibly suggesting a different role of the two isoforms also in normal skin epidermis. Further studies are required to better deﬁne the contribution of each PI3K isoform in normal skin homeostasis and whether selective deregulation of some of them is associated with skin diseases. Improved understanding of the speciﬁc signalling pathways regulated by the distinct enzymes would also provide important information. For instance, although Akt undoubtedly plays a crucial role, it is very likely that PI3Ks mediate epidermal homeostasis via a number of different signalling pathways. Induction of PI3K signalling in the epidermis led to changes in expression of over 100 genes, with many associated with cell motility and adhesion as well as cell cycle control and DNA repair [57]. PI3K signalling has also been shown to inhibit the activity of the integrin-regulated YAP1 protein which is involved in epithelial cell proliferation [79]. Deﬁning the contribution of the distinct isoforms could shed new light into the speciﬁc signalling pathways that these enzymes can control in epidermis. 5. PI3Ks-Dependent Pathways upon UV Irradiation UV radiation causes DNA damage, for instance through generation of cyclobutane pyrimidine dimers (CPD) [80,81]. CPD have been associated with initiation of UVB-induced skin carcinogenesis [82] and repair or reduction of CPD in UVB-exposed murine skin reduces the risk of tumour development [83]. The nucleotide excision repair (NER) pathway is one of the mechanisms involved in the repair of UV-induced DNA damage [84]. It has been demonstrated that PTEN is necessary for efﬁcient NER through regulation of the xeroderma pigmentosum proteins [85] and, therefore, alteration of its expression levels and/or function (and consequent deregulation of PI3Ks-dependent pathways) can lead to impaired DNA repair upon UV exposure. Indeed mice lacking PTEN in their epidermis are predisposed to skin tumourigenesis upon exposure to low sub-erythemal UV radiation [86]. UV radiation can induce alteration of PTEN levels/function through genetic alteration of the gene [87] or possibly through inactivation of the enzyme by UV-induced reactive oxygen species [88]. Indeed, reduced expression levels of PTEN were detected in transformed human keratinocytes upon chronic exposure to UVA radiation [89]. Similarly, it was shown that UVB radiation reduced PTEN levels in primary human keratinocytes, HaCaT keratinocytes and in mouse skin and this was associated with increased survival [90]. These authors further showed that downregulation of PTEN occurred at the transcriptional level and it was mediated by UVB-dependent activation of ERK and Akt [90]. Alteration of the PI3K pathway can also occur as consequence of alteration in the microRNA proﬁle upon exposure to UV as observed in a study on SKH-1 hairless mice [91]. Consistent with the detected alteration of PTEN, several lines of evidence indicate that 7 of 18 Cancers 2017, 9, 86 the PI3Ks/Akt/mTOR pathway is activated upon exposure to UV radiation. Phosphorylation of Akt [92,93] and mTOR [93] was reported in HaCaT cells treated with low doses of UVB as well as in SKH-1 mice treated with an acute dose of solar-simulated light (SSL) [94]. Moreover activation of Akt and mTOR was detected in sun-protected human skin after acute doses of physiologically-relevant SSL exposure [95]. Interestingly, one study reported differential regulation of Akt phosphorylation by UV, with phosphorylation of Ser473 mainly mediated by UVB and phosphorylation of Thr308 mediated by UVA in normal human epidermal keratinocytes [96]. On the other hand, both UV types were able to activate mTOR, as assessed by phosphorylation of S6K [96]. 6. PI3Ks-Dependent Pathways in cSCC Deregulation of the PI3Ks/Akt/mTOR pathway is one of the most common mechanisms responsible for development and progression of many cancer types [97–100]. Reverse phase protein microarray analysis revealed activation of a number of key proteins involved in this pathway in advanced and non-advanced human cSCC compared to AK [41]. Constitutive activation of the Akt/mTOR pathway in epidermal tumours was also reported in another study, with levels of phosphorylated Akt and mTOR shown to be much higher in 15 samples of SCC than in the same number of normal or AK skin samples [101]. Moderate/strong phosphorylation of Akt at Ser473 was also detected in 10 out of 15 cSCC and in eight out of 10 metastatic cSCC [101]. A speciﬁc role for distinct Akt isoforms has also been suggested by the observation that down-regulation of Akt1 and upregulation of Akt2 occur commonly in cSCC [102]. In addition, activation of upregulated Akt2 is associated with high-grade tumours [102]. Some studies have investigated the mechanisms responsible for activation of PI3Ks-dependent pathways in cSCC. Activating mutations of PIK3CA, a common characteristic of many cancer types, including lung SCC and head and neck SCC (HNSCC), have been reported but do not appear to occur at high frequency in cSCC [103]. For instance whole exome sequencing on DNA from 39 patients reported that PIK3CA was mutated only ﬁve times in four patients and, importantly, none of these mutations were the “classical” hotspot mutations observed in other tumour types [104]. On the other hand, a more recent study of 122 recurrent, metastatic cSCC identiﬁed clinically-relevant genomic alterations of PIK3CA in 6% of the cases [105]. This was consistent with data from a cohort of metastatic cSCC (29 cSCC lymph node metastases) that identiﬁed a PIK3CA P471L mutation in some of these tumours [22]. Importantly, a sustained clinical response was observed in one patient with metastatic cSCC harbouring mutations (including the PIK3CA P471L mutation) upon treatment with the mTOR inhibitor temsirolimus [105]. It remains to be established whether this mutation is indeed associated with hyperactivation of PI3K-dependent pathways. It is worth mentioning that this speciﬁc mutation was also detected in one primary cSCC sample [106], possibly suggesting that this event might not be speciﬁcally associated with metastatic cSCC although additional studies would be required to conﬁrm this observation. 5. PI3Ks-Dependent Pathways upon UV Irradiation As UV represents the most important environmental risk factor for cSCC [39], it would be important to deﬁne the speciﬁc contribution of PI3Ks-dependent pathways, and in particular of the selective PI3K isoforms, on UV-driven cSCC carcinogenesis. 6. PI3Ks-Dependent Pathways in cSCC Analysis of transgenic mice bearing a PTEN null mutation speciﬁcally in the keratinocytes revealed that 100% of these mice developed spontaneous tumours within 8.5 months of birth, mostly squamous papillomas [63]. Importantly, many of these papillomas further developed into SCCs which were able to invade the dermis. In addition, the keratinocyte-speciﬁc PTEN ablation resulted in accelerated tumourigenesis upon chemical treatment [63]. Analysis of mouse skin tumours showed that PTEN was detectable in differentiating areas of the papilloma and in the most differentiating areas of cSCC whereas it was undetectable in non-differentiating inﬁltrative areas of cSCC [114]. Models of mouse skin tumourigenesis further demonstrated the central role for PI3K/Akt during both tumour formation and progression stages. Evidence includes demonstration of the critical role for Akt in insulin like growth factor-1 (IGF-1)-mediated mouse skin tumour promotion [115,116]. An increase in Akt activity was also detected throughout the entire process in the two-stage model of mouse skin carcinogenesis [114] and overexpression of Akt in mouse primary basal keratinocytes accelerated tumourigenesis upon injection into mice [114]. Furthermore transgenic mice expressing increased levels of Akt or constitutively-active Akt in the basal layer of stratiﬁed epithelia displayed higher sensitivity to the tumour promoter 12-O-tetradecanoylphorbol-13-acetate and increased sensitivity to two-stage skin carcinogenesis [117]. The speciﬁc mechanisms by which PI3Ks/Akt regulates cSCC promotion involve both increased cell proliferation and resistance to apoptosis, as detected in PTEN-deﬁcient keratinocytes [63]. Similarly, the pathway has been implicated in resistance to apoptosis mediated by the receptor tyrosine kinase Axl in cSCC [118]. Interestingly, it has been recently demonstrated that Axl is involved in development of resistance to a class I PI3K p110α inhibitor in HNSCC and in oesophageal SCC (OSCC) [119], suggesting a complex interplay between the Axl-dependent and PI3Ks-dependent signalling pathways in SCC. The speciﬁc mechanisms by which PI3Ks/Akt regulates cSCC promotion involve both increased cell proliferation and resistance to apoptosis, as detected in PTEN-deﬁcient keratinocytes [63]. Similarly, the pathway has been implicated in resistance to apoptosis mediated by the receptor tyrosine kinase Axl in cSCC [118]. Interestingly, it has been recently demonstrated that Axl is involved in development of resistance to a class I PI3K p110α inhibitor in HNSCC and in oesophageal SCC (OSCC) [119], suggesting a complex interplay between the Axl-dependent and PI3Ks-dependent signalling pathways in SCC. 6. PI3Ks-Dependent Pathways in cSCC A few additional mutations in other PI3K isoforms were observed in this same study [106], although the limited number of specimens does not allow the drawing of any conclusions about their importance and relevance. Loss of PTEN function is a common mechanism responsible for hyperactivation of PI3Ks-dependent pathways in many cancer types. Although somatic mutations of PTEN are rare in skin lesions, reduced levels of PTEN have been detected in human AK and cSCC, indicating that either epigenetic modiﬁcations or post-transcriptional downregulation of PTEN might be involved in the progression of the disease. Indeed, while initial studies did not detect any deletion (47 cSCC) [107] or somatic mutations (21 cSCC) [108] or hypermethylation of the promoter (20 cSCC) of PTEN [109], a more recent study showed that loss of protein expression of PTEN was observed in 15 out of 16 cSCC and this was associated with an increase in ﬁbroblast growth factor 10, which in turn plays a 8 of 18 Cancers 2017, 9, 86 central role in cSCC promotion [110]. Some mechanisms that can lead to inactivation/loss of PTEN have been observed in animal models. For instance it has been shown that loss of protein expression of PTEN can occur upon genetic ablation of the developmental transcription factor grainy head-like 3. This is associated with activation of the PI3K pathway and formation of aggressive cSCC which are completely inhibited by restoration of PTEN [111]. Finally, PTEN alteration can occur as a result of UV exposure, as discussed above. Alternative mechanisms to PTEN alteration, ultimately leading to hyperactivation of PI3Ks-dependent pathways, might also exist in the context of cSCC. For instance increased formation of spontaneous precancerous lesions and cSCC was reported in transgenic mice expressing the tyrosine kinase Fyn (K14-Fyn Y528F mice) together with increased activation of several signalling pathways, including increased phosphorylation of PDK1 [112]. PI3K/Akt activation has also been detected downstream of the basement membrane proteins laminin-332/collagen VII and proved to be crucial in mediating their contribution to cSCC tumourigenesis and invasion [113]. The impact of activation of PI3Ks-dependent pathways on cSCC development and progression has been demonstrated in many studies using transgenic animal models. Conditional knockout of PTEN in skin induces neoplasia and is critical for skin cancer development [64,88]. 7. Targeting PI3Ks-Dependent Pathways in cSCC The PI3Ks/Akt/mTOR pathway is a well-established target for anti-cancer drugs development [97,98,120–125] and several inhibitors have been developed, targeting PI3K, Akt, mTOR, as represented very schematically in Figure 3. As for the class I subfamily, several inhibitors are currently available, including inhibitors that target all isoforms with similar IC50 (pan-PI3K) or mainly one/more-than-one selective isoforms (isoform-speciﬁc, i.e., with a much lower IC50 towards one/more-than-one isoforms compared to the others) [125]. Isoform-sparing PI3K inhibitors have also been developed, as is the case of GDC-0032, an inhibitor showing much less potency towards p110β (β-sparing) [126]. Finally, dual PI3K/mTOR inhibitors have also been developed [125]. 9 of 18 f Cancers 2017, 9, 86 Figure 3. Schematic and simplified representation of the class I PI3Ks/Akt/mTOR pathway, some of the main cellular functions regulated by it, and the main family of inhibitors targeting it. Figure 3. Schematic and simpliﬁed representation of the class I PI3Ks/Akt/mTOR pathway, some of the main cellular functions regulated by it, and the main family of inhibitors targeting it. Figure 3. Schematic and simplified representation of the class I PI3Ks/Akt/mTOR pathway, some of the main cellular functions regulated by it, and the main family of inhibitors targeting it. Figure 3. Schematic and simpliﬁed representation of the class I PI3Ks/Akt/mTOR pathway, some of the main cellular functions regulated by it, and the main family of inhibitors targeting it. Isoform-specific PI3K inhibitors were developed with the aim of reducing side-effects and increasing potency, by specifically targeting the main isoform(s) involved in the development/progression of each specific cancer type [124,125]. For instance this led to trials of p110α inhibitors in cancers harbouring activating PIK3CA mutations or p110β inhibitors in tumours driven by PTEN loss, as this specific isoform was reported to be critical in this context [127–131]. Similarly, due to their high expression in immune cells, inhibitors of p110δ and p110γ (or targeting both isoforms) have been tested in many haematological malignancies, with a selective p110δ inhibitor (Idelalisib) approved for use in chronic lymphocytic leukemia and follicular B-cell non-Hodgkin lymphoma [125]. With the increasing evidence suggesting the importance of the microenvironment for tumour development/progression, the potential beneficial effects of p110δ and p110γ inhibitors in other cancer settings are also being tested. 7. Targeting PI3Ks-Dependent Pathways in cSCC T th b t f k l d t di f h t d lt f li i l t i l i d t Isoform-speciﬁc PI3K inhibitors were developed with the aim of reducing side-effects and increasing potency, by speciﬁcally targeting the main isoform(s) involved in the development/progression of each speciﬁc cancer type [124,125]. For instance this led to trials of p110α inhibitors in cancers harbouring activating PIK3CA mutations or p110β inhibitors in tumours driven by PTEN loss, as this speciﬁc isoform was reported to be critical in this context [127–131]. Similarly, due to their high expression in immune cells, inhibitors of p110δ and p110γ (or targeting both isoforms) have been tested in many haematological malignancies, with a selective p110δ inhibitor (Idelalisib) approved for use in chronic lymphocytic leukemia and follicular B-cell non-Hodgkin lymphoma [125]. With the increasing evidence suggesting the importance of the microenvironment for tumour development/progression, the potential beneﬁcial effects of p110δ and p110γ inhibitors in other cancer settings are also being tested. To the best of our knowledge no studies so far have reported results from clinical trials aimed to assess the effect of pan-PI3Ks or isoform-specific inhibitors in cSCC. On the other hand, these inhibitors have been tested or are being tested in other SCC [125]. For instance, as the PI3K pathway is the most frequently mutated pathway in HNSCC, several inhibitors have been or are being tested in this context, either alone or in combination with other interventions [132]. These include pan-PI3K inhibitors (Buparlisib (BMK120), PX-866, Copanlisib (BAY 80-6946), SF1126) and isoform-specific inhibitors (Alpelisib (BYL-719, NVP-BYL719) or the p110δ inhibitor AMG319), as well as Akt and mTOR inhibitors [132]. According to the clinicaltrials.gov website, at the time of writing this review, other trials are ongoing or are recruiting participants to test PI3K inhibitors in different SCC, including OSCC and squamous non-small cell lung cancer either alone or in combination with other drugs. Overall data in literature indicate that targeting the PI3Ks/Akt/mTOR pathway could be To the best of our knowledge no studies so far have reported results from clinical trials aimed to assess the effect of pan-PI3Ks or isoform-speciﬁc inhibitors in cSCC. On the other hand, these inhibitors have been tested or are being tested in other SCC [125]. 7. Targeting PI3Ks-Dependent Pathways in cSCC For instance, as the PI3K pathway is the most frequently mutated pathway in HNSCC, several inhibitors have been or are being tested in this context, either alone or in combination with other interventions [132]. These include pan-PI3K inhibitors (Buparlisib (BMK120), PX-866, Copanlisib (BAY 80-6946), SF1126) and isoform-speciﬁc inhibitors (Alpelisib (BYL-719, NVP-BYL719) or the p110δ inhibitor AMG319), as well as Akt and mTOR inhibitors [132]. According to the clinicaltrials.gov website, at the time of writing this review, other trials are ongoing or are recruiting participants to test PI3K inhibitors in different SCC, including OSCC and squamous non-small cell lung cancer either alone or in combination with other drugs. Overall data in literature indicate that targeting the PI3Ks/Akt/mTOR pathway could be beneficial in cSCC [133]. For instance studies have demonstrated the beneficial effects of the mTOR inhibitor rapamycin in animal models, such as in mice receiving chronic sub-erythrogenic doses of UVB and UVA, where rapamycin increased latency of large tumours and reduced their multiplicity [134]. Decreased tumour multiplicity, size, and progression were also detected in hairless mice exposed to UVB upon treatment with rapamycin alone or in combination with cyclosporine [135]. Rapamycin also reduced tumour incidence and multiplicity in a chemically-induced mouse model [136]. Another study further reported that rapamycin reduced not only the tumour burden of mice harbouring early and advanced tumour lesions but also recurrent skin SCCs in a chemically-induced cancer model, basically resulting in regression of carcinogen-induced skin SCC [137]. More importantly, the beneficial effects of mTOR inhibitors towards cutaneous carcinogenesis have been observed in specific subsets of patients Prolonged immunosuppression strongly increases the risk of Overall data in literature indicate that targeting the PI3Ks/Akt/mTOR pathway could be beneﬁcial in cSCC [133]. For instance studies have demonstrated the beneﬁcial effects of the mTOR inhibitor rapamycin in animal models, such as in mice receiving chronic sub-erythrogenic doses of UVB and UVA, where rapamycin increased latency of large tumours and reduced their multiplicity [134]. Decreased tumour multiplicity, size, and progression were also detected in hairless mice exposed to UVB upon treatment with rapamycin alone or in combination with cyclosporine [135]. Rapamycin also reduced tumour incidence and multiplicity in a chemically-induced mouse model [136]. 7. Targeting PI3Ks-Dependent Pathways in cSCC These cutaneous malignancies are also generally more aggressive and numerous than those seen in the general population [138,139]. A signiﬁcantly reduced risk of developing post-transplant de novo malignancies and non-skin solid malignancy was observed in patients receiving mTOR inhibitors (sirolimus/everolimus) as immunosuppressants compared to patients receiving calcineurin inhibitors (CNI) [141]. Switching renal transplant recipients receiving CNI-based therapies to sirolimus resulted in reduced incidence of de novo KC formation [142–145] and even regression of pre-existing premalignant lesions [144]. While these data suggest a potential beneﬁcial role for mTOR inhibitors, it is important to mention that in many cancer settings the use of some inhibitors of the PI3Ks-dependent pathways has unfortunately led to the discovery of compensatory mechanisms that reduce their therapeutic efﬁciency [146,147]. One of the most characterised mechanisms of resistance was identiﬁed through the use of mTOR inhibitors that were reported to induce hyperactivation of Akt through removal of a negative feedback loop [148–150]. Increased Akt phosphorylation upon treatment with rapamycin has also been observed in keratinocytes, conﬁrming the existence of such a feedback loop in these cells [60]. Possibly consistent with this, a study in SKH-1 mice reported that while rapamycin indeed reduced tumourigenesis when it was applied topically after mice were exposed for 15 weeks to SSL, tumourigenesis was actually increased if rapamycin was applied during SSL exposure and for an additional 10 weeks [94]. Importantly this study further showed that the selective PDK1/Akt inhibitor PHT-427 was able to prevent this latter effect, indicating that combination of drugs targeting distinct components of the PI3Ks-dependent pathways could prevent or oppose potential compensatory mechanisms [94]. The question remains as to whether targeting PI3Ks directly using either pan-PI3Ks or isoform-speciﬁc inhibitors would represent a valid therapeutic option in cSCC. It was previously shown that inhibition of PI3Ks with the pan inhibitor LY294002 reduced chemically-induced skin tumour promotion in a mouse model overexpressing IGF1 [116]. Additionally, selective simultaneous inhibition of p110α and p110β not only prevented the development of PHTS in mice lacking PTEN in epidermal keratinocytes (PTEN∆) but it was also able to reverse advanced skin hamartomas [69]. With the increasing number of PI3Ks inhibitors currently available, an improved understanding of the relative contribution of each isoform in cSCC carcinogenesis, in particular in the context of metastatic cSCC, would be useful to ascertain the potential impact of these drugs. 7. Targeting PI3Ks-Dependent Pathways in cSCC Another study further reported that rapamycin reduced not only the tumour burden of mice harbouring early and advanced tumour lesions but also recurrent skin SCCs in a chemically-induced cancer model, basically resulting in regression of carcinogen-induced skin SCC [137]. More importantly, the beneﬁcial effects of mTOR inhibitors towards cutaneous carcinogenesis have been observed in speciﬁc subsets of patients. 10 of 18 Cancers 2017, 9, 86 Prolonged immunosuppression strongly increases the risk of cSCC in organ transplant recipients, with a 65–100 fold increased incidence observed in transplant recipients compared to the general population [138–140]. These cutaneous malignancies are also generally more aggressive and numerous than those seen in the general population [138,139]. A signiﬁcantly reduced risk of developing post-transplant de novo malignancies and non-skin solid malignancy was observed in patients receiving mTOR inhibitors (sirolimus/everolimus) as immunosuppressants compared to patients receiving calcineurin inhibitors (CNI) [141]. Switching renal transplant recipients receiving CNI-based therapies to sirolimus resulted in reduced incidence of de novo KC formation [142–145] and even regression of pre-existing premalignant lesions [144]. While these data suggest a potential beneﬁcial role for mTOR inhibitors, it is important to mention that in many cancer settings the use of some inhibitors of the PI3Ks-dependent pathways has unfortunately led to the discovery of compensatory mechanisms that reduce their therapeutic efﬁciency [146,147]. One of the most characterised mechanisms of resistance was identiﬁed through the use of mTOR inhibitors that were reported to induce hyperactivation of Akt through removal of a negative feedback loop [148–150]. Increased Akt phosphorylation upon treatment with rapamycin has also been observed in keratinocytes, conﬁrming the existence of such a feedback loop in these cells [60]. Possibly consistent with this, a study in SKH-1 mice reported that while rapamycin indeed reduced tumourigenesis when it was applied topically after mice were exposed for 15 weeks to SSL, tumourigenesis was actually increased if rapamycin was applied during SSL exposure and for an additional 10 weeks [94]. Importantly this study further showed that the selective PDK1/Akt inhibitor PHT-427 was able to prevent this latter effect, indicating that combination of drugs targeting distinct components of the PI3Ks-dependent pathways could prevent or oppose potential compensatory mechanisms [94]. Prolonged immunosuppression strongly increases the risk of cSCC in organ transplant recipients, with a 65–100 fold increased incidence observed in transplant recipients compared to the general population [138–140]. References 2017, 7, 5–19. [CrossRef] [PubMed] 10. Hannuksela-Svahn, A.; Pukkala, E.; Karvonen, J. Basal cell skin carcinoma and other nonmelanom cancers in Finland from 1956 through 1995. Arch. Dermatol. 1999, 135, 781–786. [CrossRef] [PubMed 1. Wassberg, C.; Thorn, M.; Johansson, A.; Bergstrom, R.; Berne, B.; Ringborg, U. Increasing incidence rate squamous cell carcinoma of the skin in Sweden. Acta Derm. Venereol. 2001, 81, 268–272. [CrossRef] [PubM 12. John, S.M.; Trakatelli, M.; Gehring, R.; Finlay, K.; Fionda, C.; Wittlich, M.; Augustin, M.; Hilpert, G.; Barroso Dias, J.M.; Ulrich, C.; et al. CONSENSUS REPORT: Recognizing non-melanoma skin cancer, including actinic keratosis, as an occupational disease—A call to action. J. Eur. Acad. Dermatol. Venereol. 2016, 30, 38–45. [CrossRef] [PubMed] 13. Vallejo-Torres, L.; Morris, S.; Kinge, J.M.; Poirier, V.; Verne, J. Measuring current and future cost of skin cancer in England. J. Public Health 2014, 36, 140–148. [CrossRef] [PubMed] 14. Katalinic, A.; Kunze, U.; Schäfer, T. Epidemiology of cutaneous melanoma and non-melanoma skin cancer in Schleswig-Holstein, Germany: Incidence, clinical subtypes, tumour stages and localization (epidemiology of skin cancer). Br. J. Dermatol. 2003, 149, 1200–1206. [CrossRef] [PubMed] 5. Fu, T.; Aasi, S.Z.; Hollmig, S.T. Management of high-risk squamous cell carcinoma of the skin. Curr. T Options Oncol. 2016, 17, 34. [CrossRef] [PubMed] 16. Gurney, B.; Newlands, C. Management of regional metastatic disease in head and neck cutaneous malignancy. 1 cutaneous squamous cell carcinoma. Br. J. Oral Maxillofac. Surg. 2014, 52, 294–300. [CrossRef] [PubMed] 17. Martinez, J.C.; Otley, C.C.; Stasko, T.; Euvrard, S.; Brown, C.; Schanbacher, C.F.; Weaver, A.L.; Transplant-Skin Cancer Collaborative. Deﬁning the clinical course of metastatic skin cancer in organ transplant recipients: A multicenter collaborative study. Arch. Dermatol. 2003, 139, 301–306. [CrossRef] [PubMed] 18. Carucci, J.A. Press for an underestimated nemesis. JAMA Dermatol. 2013, 149, 1147–1148. [CrossRef] [PubMed] 19. Karia, P.S.; Han, J.; Schmults, C.D. Cutaneous squamous cell carcinoma: Estimated incidence of disease, nodal metastasis, and deaths from disease in the United States, 2012. J. Am. Acad. Dermatol. 2013, 68, 957–966. [CrossRef] [PubMed] 20. Jambusaria-Pahlajani, A.; Kanetsky, P.; Karia, P.S.; Hwang, W.; Gelfand, J.M.; Whalen, F.M.; Elenitsas, R.; Xu, X.; Schmults, C.D. Evaluation of AJCC tumor (T) staging for cutaneous squamous cell carcinoma and a proposed alternative tumor staging system. JAMA Dermatol. 2013, 16, 1–9. [CrossRef] [PubMed] 21. Thompson, A.K.; Kelley, B.F.; Prokop, L.J.; Murad, M.H.; Baum, C.L. Risk factors for cutaneous squamous cell carcinoma recurrence, metastasis, and disease-speciﬁc death a systematic review and meta-analysis. 8. Conclusions Despite several data indicating that PI3Ks-dependent signalling pathways are important in cSCC much still needs to be understood about the contribution of these enzymes and, in particular, the selective contribution of each of the distinct PI3K isoforms to the disease. Currently, the lack of strong evidence indicating either speciﬁc mutations or selective activation of speciﬁc PI3K isoform(s) during cSCC carcinogenesis, in particular during progression to metastatic cSCC, makes it difﬁcult to envisage which selective PI3K inhibitor(s) or which speciﬁc drugs combination(s) could be beneﬁcial in this context. Additional investigations, including a better characterisation of the role of distinct PI3Ks, are needed to determine whether targeting selective PI3Ks could represent a useful strategy to counteract this disease, in particular for metastatic cSCC. Acknowledgments: Work in our laboratory is supported by British Skin Foundation. Author Contributions: Joanna M. Janus and Tania Maffucci wrote the paper, and Ryan F. L. O’Shaughnessy and Catherine A. Harwood contributed to the manuscript. Conﬂicts of Interest: The authors declare no conﬂict of interest. The funding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results. 11 of 18 11 of 18 Cancers 2017, 9, 86 References 1. Surdu, S. Non-melanoma skin cancer: Occupational risk from UV light and arsenic exposure. Rev. Environ. Health 2014, 29, 255–264. [CrossRef] [PubMed] 1. Surdu, S. Non-melanoma skin cancer: Occupational risk from UV light and arsenic exposure. Rev. Environ. Health 2014, 29, 255–264. [CrossRef] [PubMed] 2. Burton, K.A.; Ashack, K.A.; Khachemoune, A. Cutaneous squamous cell carcinoma: A review of high-risk and metastatic disease. Am. J. Clin. Dermatol. 2016, 17, 491–508. [CrossRef] [PubMed] 2. Burton, K.A.; Ashack, K.A.; Khachemoune, A. Cutaneous squamous cell carcinoma: A review of high-risk and metastatic disease. Am. J. Clin. Dermatol. 2016, 17, 491–508. [CrossRef] [PubMed] 3. Madan, V.; Lear, J.T.; Szeimies, R.M. Non-melanoma skin cancer. Lancet 2010, 375, 673–685. [CrossR 3. Madan, V.; Lear, J.T.; Szeimies, R.M. Non-melanoma skin cancer. Lancet 2010, 375, 673–685. [CrossRef] 4. Leigh, I.M. Progress in skin cancer: The U.K. experience. Br. J. Dermatol. 2014, 171, 443–445. [CrossRef] [PubMed] 4. Leigh, I.M. Progress in skin cancer: The U.K. experience. Br. J. Dermatol. 2014, 171, 443–445. [CrossRef] [PubMed] Skin Cancer Incidence Statistics. Available online: http://www.cancerresearchuk.org/health-professional 5. Skin Cancer Incidence Statistics. Available online: http://www.cancerresearchuk.org/health-professional/ cancer-statistics/statistics-by-cancer-type/skin-cancer/incidence#ref-10 (accessed on 4 July 2017). Skin Cancer Incidence Statistics. Available online: http://www.cancerresearchuk.org/health profess cancer-statistics/statistics-by-cancer-type/skin-cancer/incidence#ref-10 (accessed on 4 July 2017). 6. Rogers, H.W.; Weinstock, M.A.; Feldman, S.R.; Coldiron, B.M. Incidence estimate of nonmelanoma skin cancer (keratinocyte carcinomas) in the U.S. population, 2012. JAMA Dermatol. 2015, 151, 1081–1086. [CrossRef] [PubMed] 6. Rogers, H.W.; Weinstock, M.A.; Feldman, S.R.; Coldiron, B.M. Incidence estimate of nonmelanoma skin cancer (keratinocyte carcinomas) in the U.S. population, 2012. JAMA Dermatol. 2015, 151, 1081–1086. [CrossRef] [PubMed] 7. Athas, W.F.; Hunt, W.C.; Key, C.R. Changes in nonmelanoma skin cancer incidence between 1977–1978 and 1998–1999 in northcentral New Mexico. Cancer Epidemiol. Biomark. Prev. 2003, 12, 1105–1108. 8. Gray, D.T.; Suman, V.J.; Su, W.P.D.; Clay, R.P.; Harmsen, W.S.; Roenigh, R.K. Trends in the population-based incidence of squamous cell carcinoma of the skin ﬁrst diagnosed between 1984 and 1992. Arch. Dermatol. 1997, 133, 735–740. [CrossRef] [PubMed] 9. Apalla, Z.; Nashan, D.; Weller, R.B.; Castellsagué, X. Skin cancer: Epidemiology, disease burden, pathophysiology, diagnosis, and therapeutic approaches. Dermatol. Ther. 2017, 7, 5–19. [CrossRef] [PubMed] 10. Hannuksela-Svahn, A.; Pukkala, E.; Karvonen, J. Basal cell skin carcinoma and other nonmelanoma skin cancers in Finland from 1956 through 1995 Arch Dermatol 1999 135 781–786 [CrossRef] [PubMed] 9. Apalla, Z.; Nashan, D.; Weller, R.B.; Castellsagué, X. Skin cancer: Epidemiology, disease burden, pathophysiology, diagnosis, and therapeutic approaches. Dermatol. Ther. References JAMA Dermatol. 2016, 152, 419–428. [CrossRef] [PubMed] 12 of 18 12 of 18 Cancers 2017, 9, 86 22. Li, Y.Y.; Hanna, G.J.; Laga, A.C.; Haddad, R.I.; Lorch, J.H.; Hammerman, P.S. Genomic analysis of metastatic cutaneous squamous cell carcinoma. Clin. Cancer Res. 2015, 21, 1447–1456. [CrossRef] [PubMed] 23. Baroni, A.; Buommino, E.; De Gregorio, V.; Ruocco, E.; Ruocco, V.; Wolf, R. Structure and function of the epidermis related to barrier properties. Clin. Dermatol. 2012, 30, 257–262. [CrossRef] [PubMed] 24. Eckert, R.L.; Rorke, E.A. Molecular biology of keratinocyte differentiation. Environ. Health Perspect. 1989, 80, 109–116. [CrossRef] [PubMed] 25. Blanpain, C.; Fuchs, E. Epidermal homeostasis: A balancing act of stem cells in the skin. Nat. Rev. Mol. Cell Biol. 2009, 10, 207–217. [CrossRef] [PubMed] 26. Radoja, N.; Gazel, A.; Banno, T.; Yano, S.; Blumenberg, M. Transcriptional proﬁling of epidermal differentiation. Physiol. Genom. 2006, 27, 65–78. [CrossRef] [PubMed] 27. Raymond, A.A.; Gonzalez de Peredo, A.; Stella, A.; Ishida-Yamamoto, A.; Bouyssie, D.; Serre, G.; Monsarrat, B.; Simon, M. Lamellar bodies of human epidermis: Proteomics characterization by high throughput mass spectrometry and possible involvement of CLIP-170 in their trafﬁcking/secretion. Mol. Cell Proteomics 2008, 7, 2151–2175. [CrossRef] [PubMed] 28. Brooke, M.A.; Nitoiu, D.; Kelsell, D.P. Cell-cell connectivity: Desmosomes and disease. J. Pathol. 2012, 226, 158–171. [CrossRef] [PubMed] 29 Johnson J L ; Najor N A ; Green K J Desmosomes: Regulators of cellular signalling and adhesion in 28. Brooke, M.A.; Nitoiu, D.; Kelsell, D.P. Cell-cell connectivity: Desmosomes and disease. J. Pathol. 2012, 226, 158–171. [CrossRef] [PubMed] 29. Johnson, J.L.; Najor, N.A.; Green, K.J. Desmosomes: Regulators of cellular signalling and adhesion in epidermal health and disease Cold Spring Harb Perspect Med 2014 4 a015297 [CrossRef] [PubMed] 9. Johnson, J.L.; Najor, N.A.; Green, K.J. Desmosomes: Regulators of cellular signalling and adhesio epidermal health and disease. Cold Spring Harb. Perspect. Med. 2014, 4, a015297. [CrossRef] [PubMed] 30. Neel, V.A.; Sober, A.J. Tumors arising from the epidermis. In Holland-Frei Cancer Medicine, 6th ed.; Kufe, D.W., Pollock, R.E., Weichselbaum, R.R., Bast, R.C., Gansler, T.S., Holland, J.F., Frei, E., Eds.; BC Decker: Hamilton, ON, Canada, 2003. 31. Alam, M.; Ratner, D. Cutaneous squamous-cell carcinoma. N. Engl. J. Med. 2001, 344, 975–983. [CrossRef] [PubMed] 32. Harris, R.B.; Grifﬁth, K.; Moon, T.E. Trends in the incidence of nonmelanoma skin cancers in south eastern Arizona, 1985–1996. J. Am. Acad. Dermatol. 2001, 45, 528–536. [CrossRef] [PubMed] 33. References Criscione, V.D.; Weinstock, M.A.; Naylor, M.F.; Luque, C.; Eide, M.J.; Bingham, S.F.; Department of Veteran Affairs Topical Tretinoin Chemoprevention Trial Group. Actinic keratoses: Natural history and risk of malignant transformation in the veterans affairs topical tretinoin chemoprevention trial. Cancer 2009, 115, 2523–2530. [CrossRef] [PubMed] 34. Harvey, I.; Frankel, S.; Marks, R.; Shalom, D.; Nolan-Farrell, M. Non-melanoma skin cancer and solar keratoses II analytical results of the south wales skin cancer study. Br. J. Cancer 1996, 74, 1308–1312. [CrossRef] [PubMed] 35. Werner, R.N.; Sammain, A.; Erdmann, R.; Hartmann, V.; Stockﬂeth, E.; Nast, A. The natural history of actinic keratosis: A systematic review. Br. J. Dermatol. 2013, 169, 502–518. [CrossRef] [PubMed] 36. Salasche, S.J. Epidemiology of actinic keratoses and squamous cell carcinoma. J. Am. Acad. Dermatol. 2000, 42, 4–7. [CrossRef] [PubMed] 37. Purdie, K.J.; Harwood, C.A.; Gulati, A.; Chaplin, T.; Lambert, S.R.; Cerio, R.; Kelly, G.P.; Cazier, J.B.; Young, B.D.; Leigh, I.M.; et al. Single nucleotide polymorphism array analysis deﬁnes a speciﬁc genetic ﬁngerprint for well-differentiated cutaneous SCCs. J. Investig. Dermatol. 2009, 129, 1562–1568. [CrossRef] [PubMed] 38. Purdie, K.J.; Lambert, S.R.; Teh, M.T.; Chaplin, T.; Molloy, G.; Raghavan, M.; Kelsell, D.P.; Leigh, I.M.; Harwood, C.A.; Proby, C.M.; et al. Allelic imbalances and microdeletions affecting the PTPRD gene in cutaneous squamous cell carcinomas detected using single nucleotide polymorphism microarray analysis. Genes Chromosom. Cancer 2007, 46, 661–669. [CrossRef] [PubMed] 39. Harwood, C.A.; Proby, C.M.; Inman, G.J.; Leigh, I.M. The promise of genomics and the development of targeted therapies for cutaneous squamous cell carcinoma. Acta Derm. Venereol. 2016, 96, 3–16. [CrossRef] [PubMed] 40. Cranmer, L.D.; Engelhardt, C.; Morgan, S.S. Treatment of unresectable and metastatic cutaneous squamous cell carcinoma. Oncologist 2010, 15, 1320–1328. [CrossRef] [PubMed] 41. Einspahr, J.G.; Calvert, V.; Alberts, D.S.; Curiel-Lewandrowski, C.; Warneke, J.; Krouse, R.; Stratton, S.P.; Liotta, L.; Longo, C.; Pellacani, G.; et al. Functional protein pathway activation mapping of the progression of normal skin to squamous cell carcinoma. Cancer Prev. Res. 2012, 5, 403–413. [CrossRef] [PubMed] 13 of 18 13 of 18 Cancers 2017, 9, 86 42. Lambert, S.R.; Mladkova, N.; Gulati, A.; Hamoudi, R.; Purdie, K.; Cerio, R.; Leigh, I.; Proby, C.; Harwood, C.A. Key differences identiﬁed between actinic keratosis and cutaneous squamous cell carcinoma by transcriptome proﬁling. Br. J. Cancer 2014, 110, 520–529. [CrossRef] [PubMed] 43. Adelmann, C.H.; Truong, K.A.; Liang, R.J.; Bansal, V.; Gandee, L.; Saporito, R.C.; Lee, W.; Du, L.; Nicholas, C.; Napoli, M.; et al. References MEK is a therapeutic and chemopreventative target in squamous cell carcinoma. J. Investig. Dermatol. 2016, 136, 1920–1924. [CrossRef] [PubMed] 4. Cantley, L.C. The phosphoinositide 3-kinase pathway. Science 2002, 296, 1655–1657. [CrossRef] [PubMe 44. Cantley, L.C. The phosphoinositide 3-kinase pathway. Science 2002, 296, 1655–1657. [CrossRef] [PubMed] 45. Maffucci, T. An introduction to phosphoinositides. Curr. Top. Microbiol. Immunol. 2012, 362, 1–42. [PubMed] 46. Falasca, M.; Maffucci, T. Regulation and cellular functions of class II phosphoinositide 3-kinases. Biochem. J. 44. Cantley, L.C. The phosphoinositide 3-kinase pathway. Science 2002, 296, 1655–1657. [CrossRef] [PubMed] 45. Maffucci, T. An introduction to phosphoinositides. Curr. Top. Microbiol. Immunol. 2012, 362, 1–42. [PubMed] y p p p y 45. Maffucci, T. An introduction to phosphoinositides. Curr. Top. Microbiol. Immunol. 2012, 362, 1–42. [PubMed] p p p 46. Falasca, M.; Maffucci, T. Regulation and cellular functions of class II phosphoinositide 3-kinases. Biochem. J. 2012, 443, 587–601. [CrossRef] [PubMed] 47. Engelman, J.A.; Luo, J.; Cantley, L.C. The evolution of phosphatidylinositol 3-kinases as regulators of growth and metabolism. Nat. Rev. Genet. 2006, 7, 606–619. [CrossRef] [PubMed] 48. Pearce, L.R.; Komander, D.; Alessi, D.R. The nuts and bolts of AGC protein kinases. Nat. Rev. Mol. Cell Biol. 2010, 11, 9–22. [CrossRef] [PubMed] 49. Huang, J.; Manning, B.D. A complex interplay between Akt, TSC2 and the two mTOR complexes. Biochem. Soc. Trans. 2009, 37, 217–222. [CrossRef] [PubMed] 50. Franke, T.F.; Kaplan, D.R.; Cantley, L.C. PI3K: Downstream AKTion blocks apoptosis. Cell 1997, 88, 4 [CrossRef] 51. Manning, B.D.; Cantley, L.C. AKT/PKB signaling: Navigating downstream. Cell 2007, 129, 1261–1274. [CrossRef] [PubMed] 52. Clark, A.R.; Toker, A. Signalling speciﬁcity in the Akt pathway in breast cancer. Biochem. Soc. Trans. 2014, 42, 1349–1355. [CrossRef] [PubMed] 53. Howell, J.J.; Manning, B.D. mTOR couples cellular nutrient sensing to organismal metabolic homeostasis. Trends Endocrinol. Metab. 2011, 22, 94–102. [CrossRef] [PubMed] 54. Laplante, M.; Sabatini, D.M. mTOR signaling in growth control and disease. Cell 2012, 149, 274–293. [CrossRef] [PubMed] 55. Worby, C.A.; Dixon, J.E. PTEN. Annu. Rev. Biochem. 2014, 83, 641–669. [CrossRef] [PubMed] 56. Sayama, K.; Yamasaki, K.; Hanakawa, Y.; Shirakata, Y.; Tokumaru, S.; Ijuin, T.; Takenawa, T.; Hashimoto, K. Phosphatidylinositol 3-kinase is a key regulator of early phase differentiation in keratinocytes. J. Biol. Chem. 2002, 277, 40390–40396. [CrossRef] [PubMed] 57. Pankow, S.; Bamberger, C.; Klippel, A.; Werner, S. Regulation of epidermal homeostasis and repair by phosphoinositide 3-kinase. J. Cell Sci. 2006, 119, 4033–4046. [CrossRef] [PubMed] 58. References Wang, Q.; Von, T.; Bronson, R.; Ruan, M.; Mu, W.; Huang, A.; Maira, S.M.; Zhao, J.J. Spatially distinct role of class Ia PI3K isoforms in the development and maintenance of PTEN hamartoma tumor syndrome. Genes Dev. 2013, 27, 1568–1580. [CrossRef] [PubMed] 70. Vanhaesebroeck, B.; Leevers, S.J.; Ahmadi, K.; Timms, J.; Katso, R.; Driscoll, P.C.; Woscholski, R.; Parker, P.J.; Waterﬁeld, M.D. Synthesis and function of 3-phosphorylated inositol lipids. Annu. Rev. Biochem. 2001, 70, 535–602. [CrossRef] [PubMed] 71. Falasca, M.; Maffucci, T. Role of class II phosphoinositide 3-kinase in cell signalling. Biochem. Soc. Trans. 2007, 35, 211–214. [CrossRef] [PubMed] 72. Vanhaesebroeck, B.; Guillermet-Guibert, J.; Graupera, M.; Bilanges, B. The emerging mechanisms of isoform-speciﬁc PI3K signalling. Nat. Rev. Mol. Cell Biol. 2010, 11, 329–341. [CrossRef] [PubMed] 73. Vanhaesebroeck, B.; Ali, K.; Bilancio, A.; Geering, B.; Foukas, L.C. Signalling by PI3K isoforms: Insights from gene-targeted mice. Trends Biochem. Sci. 2005, 30, 194–204. [CrossRef] [PubMed] 74. Maffucci, T.; Falasca, M. New insight into the intracellular roles of class II phosphoinositide 3-kinases. Biochem. Soc. Trans. 2014, 42, 1378–1382. [CrossRef] [PubMed] 74. Maffucci, T.; Falasca, M. New insight into the intracellular ro Biochem. Soc. Trans. 2014, 42, 1378–1382. [CrossRef] [PubMed] 75. Harada, K.; Truong, A.B.; Cai, T.; Khavari, P.A. The class II phosphoinositide 3-kinase C2beta is not essential for epidermal differentiation. Mol. Cell. Biol. 2005, 25, 11122–11130. [CrossRef] [PubMed] 76. Akinduro, O.; Sully, K.; Patel, A.; Robinson, D.J.; Chikh, A.; McPhail, G.; Braun, K.M.; Philpott, M.P.; Harwood, C.A.; Byrne, C.; et al. Constitutive autophagy and nucleophagy during epidermal differentiation. J. Investig. Dermatol. 2016, 136, 1460–1470. [CrossRef] [PubMed] 77. Backer, J.M. The regulation and function of class III PI3Ks: Novel roles for Vps34. Biochem. J. 2008, 410, 1–17. [CrossRef] [PubMed] 78. Xie, Z.; Singleton, P.A.; Bourguignon, L.Y.; Bikle, D.D. Calcium-induced human keratinocyte differentiation requires src- and fyn-mediated phosphatidylinositol 3-kinase-dependent activation of phospholipase C-gamma1. Mol. Biol. Cell 2005, 16, 3236–3346. [CrossRef] [PubMed] 79. Elbediwy, A.; Vincent-Mistiaen, Z.I.; Spencer-Dene, B.; Stone, R.K.; Boeing, S.; Wculek, S.K.; Cordero, J.; Tan, E.H.; Ridgway, R.; Brunton, V.G.; et al. Integrin signalling regulates YAP and TAZ to control skin homeostasis. Development 2016, 143, 1674–1687. [CrossRef] [PubMed] 80. Mouret, S.; Baudouin, C.; Charveron, M.; Favier, A.; Cadet, J.; Douki, T. Cyclobutane pyrimidine dimers are predominant DNA lesions in whole human skin exposed to UVA radiation. Proc. Natl. Acad. Sci. USA 2006, 103, 13765–13770. [CrossRef] [PubMed] 81. References Calautti, E.; Li, J.; Saoncella, S.; Brissette, J.L.; Goetinck, P.F. Phosphoinositide 3-kinase signaling to Akt promotes keratinocyte differentiation versus death. J. Biol. Chem. 2005, 280, 32856–32865. [CrossRef] [PubMed] 59. O’Shaughnessy, R.F.; Welti, J.C.; Cooke, J.C.; Avilion, A.A.; Monks, B.; Birnbaum, M.J.; Byrne, C. AKT-dependent HspB1 (Hsp27) activity in epidermal differentiation. J. Biol. Chem. 2007, 282, 17297–17305. [CrossRef] [PubMed] 60. Sully, K.; Akinduro, O.; Philpott, M.P.; Naeem, A.S.; Harwood, C.A.; Reeve, V.E.; O’Shaughnessy, R.F.; Byrne, C. The mTOR inhibitor rapamycin opposes carcinogenic changes to epidermal Akt1/PKBα isoform signaling. Oncogene 2013, 32, 3254–3262. [CrossRef] [PubMed] 61. Naeem, A.S.; Zhu, Y.; Di, W.L.; Marmiroli, S.; O’Shaughnessy, R.F. AKT1-mediated Lamin A/C degradation is required for nuclear degradation and normal epidermal terminal differentiation. Cell Death Differ. 2015, 22, 2123–2132. [CrossRef] [PubMed] 62. Madonna, S.; Scarponi, C.; Pallotta, S.; Cavani, A.; Albanesi, C. Anti-apoptotic effects of suppressor of cytokine signaling 3 and 1 in psoriasis. Cell Death Dis. 2012, 3, e334. [CrossRef] [PubMed] 63. Suzuki, A.; Itami, S.; Ohishi, M.; Hamada, K.; Inoue, T.; Komazawa, N.; Senoo, H.; Sasaki, T.; Takeda, J.; Manabe, M.; et al. Keratinocyte-speciﬁc Pten deﬁciency results in epidermal hyperplasia, accelerated hair follicle morphogenesis and tumor formation. Cancer Res. 2003, 63, 674–681. [PubMed] 64. Backman, S.A.; Ghazarian, D.; So, K.; Sanchez, O.; Wagner, K.U.; Hennighausen, L.; Suzuki, A.; Tsao, M.S.; Chapman, W.B.; Stambolic, V.; et al. Early onset of neoplasia in the prostate and skin of mice with tissue-speciﬁc deletion of Pten. Proc. Natl. Acad. Sci. USA 2004, 101, 1725–1730. [CrossRef] [PubMed] 14 of 18 Cancers 2017, 9, 86 65. Yu, M.; Owens, D.M.; Ghosh, S.; Farber, D.L. Conditional PDK1 ablation promotes epidermal and T-Cell-mediated dysfunctions leading to inﬂammatory skin disease. J. Investig. Dermatol. 2015, 135, 2688–2696. [CrossRef] [PubMed] 66. Dainichi, T.; Haydenm, M.S.; Park, S.G.; Oh, H.; Seeley, J.J.; Grinberg-Bleyer, Y.; Beck, K.M.; Miyachi, Y.; Kabashima, K.; Hashimoto, T.; et al. PDK1 is a regulator of epidermal differentiation that activates and organizes asymmetric cell division. Cell Rep. 2016, 15, 1615–1623. [CrossRef] [PubMed] 67. Peng, X.D.; Xu, P.Z.; Chen, M.L.; Hahn-Windgassen, A.; Skeen, J.; Jacobs, J.; Sundararajan, D.; Chen, W.S.; Crawford, S.E.; Coleman, K.G.; et al. Dwarﬁsm, impaired skin development, skeletal muscle atrophy, delayed bone development, and impeded adipogenesis in mice lacking Akt1 and Akt2. Genes Dev. 2003, 17, 1352–1365. [CrossRef] [PubMed] 68. Hobert, J.A.; Eng, C. PTEN hamartoma tumor syndrome: An overview. Genet. Med. 2009, 11, 687–694. [CrossRef] [PubMed] 69. References [CrossRef] [PubMed] 92. Han, W.; He, Y.Y. Requirement for metalloproteinase-dependent ERK and AKT activation in UVB-induced G1-S cell cycle progression of human keratinocytes. Photochem. Photobiol. 2009, 85, 997–1003. [CrossRef] [PubMed] 93. Carr, T.D.; DiGiovanni, J.; Lynch, C.J.; Shantz, L.M. Inhibition of mTOR suppresses UVB-induced keratinocyte proliferation and survival. Cancer Prev. Res. 2012, 5, 1394–1404. [CrossRef] [PubMed] 94. Dickinson, S.E.; Janda, J.; Criswell, J.; Blohm-Mangone, K.; Olson, E.R.; Liu, Z.; Barber, C.; Petricoin, E.F.; Calvert, V.S.; Einspahr, J.; et al. Inhibition of Akt enhances the chemopreventive effects of topical rapamycin in mouse skin. Cancer Prev. Res. 2016, 9, 215–224. [CrossRef] [PubMed] 95. Bermudez, Y.; Stratton, S.P.; Curiel-Lewandrowski, C.; Warneke, J.; Hu, C.; Bowden, G.T.; Dickinson, S.E.; Dong, Z.; Bode, A.M.; Saboda, K.; et al. Activation of the PI3K/Akt/mTOR and MAPK signaling pathways in response to acute solar-simulated light exposure of human skin. Cancer Prev. Res. 2015, 8, 720–728. [CrossRef] [PubMed] 96. Syed, D.N.; Afaq, F.; Mukhtar, H. Differential activation of signaling pathways by UVA and UVB radiation in normal human epidermal keratinocytes. Photochem. Photobiol. 2012, 88, 1184–1190. [CrossRef] [PubMed] 97. Wong, K.K.; Engelman, J.A.; Cantley, L.C. Targeting the PI3K signaling pathway in cancer. Curr. Opin. Genet. D 20 87 90 [C R f] [P bM d] 96. Syed, D.N.; Afaq, F.; Mukhtar, H. Differential activation of signaling pathways by UVA and UVB radiation in normal human epidermal keratinocytes. Photochem. Photobiol. 2012, 88, 1184–1190. [CrossRef] [PubMed] in normal human epidermal keratinocytes. Photochem. Photobiol. 2012, 88, 1184–1190. [CrossRef] [PubMed] 97. Wong, K.K.; Engelman, J.A.; Cantley, L.C. Targeting the PI3K signaling pathway in cancer. Curr. Opin. Genet. Dev. 2010, 20, 87–90. [CrossRef] [PubMed] 97. Wong, K.K.; Engelman, J.A.; Cantley, L.C. Targeting the PI3K signaling pathway in cancer. Curr. Opin. Genet. Dev. 2010, 20, 87–90. [CrossRef] [PubMed] 98. Vanhaesebroeck, B.; Vogt, P.K.; Rommel, C. PI3K: From the bench to the clinic and back. Curr. Top. Microbiol. Immunol. 2010, 347, 1–19. [PubMed] 99. Zhao, L.; Vogt, P.K. Class I PI3K in oncogenic cellular transformation. Oncogene 2008, 27, 5486–5496. [CrossRef] [PubMed] 100. Vivanco, I.; Sawyers, C.L. The phosphatidylinositol 3-Kinase AKT pathway in human cancer. Nat. Rev. Cancer. 2002, 2, 489–501. [CrossRef] [PubMed] 101. Chen, S.J.; Nakahara, T.; Takahara, M.; Kido, M.; Dugu, L.; Uchi, H.; Takeuchi, S.; Tu, Y.T.; Moroi, Y.; Furue, M. Activation of the mammalian target of rapamycin signalling pathway in epidermal tumours and its correlation with cyclin-dependent kinase 2. Br. J. Dermatol. References Ikehata, H.; Kawai, K.; Komura, J.; Sakatsume, K.; Wang, L.; Imai, M.; Higashi, S.; Nikaido, O.; Yamamoto, K.; Hieda, K.; et al. UVA1 genotoxicity is mediated not by oxidative damage but by cyclobutane pyrimidine dimers in normal mouse skin. J. Investig. Dermatol. 2008, 128, 2289–2296. [CrossRef] [PubMed] 82. De Gruijl, F.R.; Rebel, H. Early events in UV carcinogenesis-DNA damage, target cells and mutant p53 foci. Photochem. Photobiol. 2008, 84, 382–387. [CrossRef] [PubMed] 83. Yarosh, D.; Alas, L.G.; Yee, V.; Oberyszyn, A.; Kibitel, J.T.; Mitchell, D.; Rosenstein, R.; Spinowitz, A.; Citron, M. Pyrimidine dimer removal enhanced by DNA repair liposomes reduces the incidence of UV skin cancer in mice. Cancer Res. 1992, 52, 4227–4231. [PubMed] 84. Marteijn, J.A.; Lans, H.; Vermeulen, W.; Hoeijmakers, J.H. Understanding nucleotide excision repair and its roles in cancer and ageing. Nat. Rev. Mol. Cell Biol. 2014, 15, 465–481. [CrossRef] [PubMed] 85. Ming, M.; He, Y.Y. PTEN in DNA damage repair. Cancer Lett. 2012, 319, 125–129. [CrossRef] [PubMed] 84. Marteijn, J.A.; Lans, H.; Vermeulen, W.; Hoeijmakers, J.H. Understanding nucleotide excision repair and its roles in cancer and ageing. Nat. Rev. Mol. Cell Biol. 2014, 15, 465–481. [CrossRef] [PubMed] 85. Ming, M.; He, Y.Y. PTEN in DNA damage repair. Cancer Lett. 2012, 319, 125–129. [CrossRef] [PubMed] 15 of 18 15 of 18 Cancers 2017, 9, 86 86. Ming, M.; Feng, L.; Shea, C.R.; Soltani, K.; Zhao, B.; Han, W.; Smart, R.C.; Trempus, C.S.; He, Y.Y. PTEN positively regulates UVB-induced DNA damage repair. Cancer Res. 2011, 71, 5287–5295. [CrossRef] [PubMed] 7. Hocker, T.; Tsao, H. Ultraviolet radiation and melanoma: A systematic review and analysis of repo sequence variants. Hum. Mutat. 2007, 28, 578–588. [CrossRef] [PubMed] 88. Ming, M.; He, Y.Y. PTEN: New insights into its regulation and function in skin cancer. J. Investig. Dermatol. 2009, 129, 2109–2112. [CrossRef] [PubMed] 89. He, Y.Y.; Pi, J.; Huang, J.L.; Diwan, B.A.; Waalkes, M.P.; Chignell, C.F. Chronic UVA irradiation of human HaCaT keratinocytes induces malignant transformation associated with acquired apoptotic resistance. Oncogene 2006, 25, 3680–3688. [CrossRef] [PubMed] 90. Ming, M.; Han, W.; Maddox, J.; Soltani, K.; Shea, C.R.; Freeman, D.M.; He, Y.Y. UVB-induced ERK/AKT-dependent PTEN suppression promotes survival of epidermal keratinocytes. Oncogene 2010, 29, 492–502. [CrossRef] [PubMed] 91. Singh, A.; Willems, E.; Singh, A.; Ong, I.M.; Verma, A.K. Ultraviolet radiation-induced differential microRNA expression in the skin of hairless SKH1 mice, a widely used mouse model for dermatology research. Oncotarget 2016, 7, 84924–84937. References 2009, 160, 442–445. [CrossRef] [PubMed] 102. O’Shaughnessy, R.F.; Akgũl, B.; Storey, A.; Pﬁster, H.; Harwood, C.A.; Byrne, C. Cutaneous human papillomaviruses down-regulate AKT1, whereas AKT2 up-regulation and activation associates with tumors. Cancer Res. 2007, 67, 8207–8215. [CrossRef] [PubMed] 103. Hafner, C.; Landthaler, M.; Vogt, T. Activation of the PI3K/AKT signalling pathway in non-melanoma skin cancer is not mediated by oncogenic PIK3CA and AKT1 hotspot mutations. Exp. Dermatol. 2010, 19, e222–e227. [CrossRef] [PubMed] 104. Pickering, C.R.; Zhou, J.H.; Lee, J.J.; Drummond, J.A.; Peng, S.A.; Saade, R.E.; Tsai, K.Y.; Curry, J.L.; Tetzlaff, M.T.; Lai, S.Y.; et al. Mutational landscape of aggressive cutaneous squamous cell carcinoma. Clin. Cancer Res. 2014, 20, 6582–6592. [CrossRef] [PubMed] 105. Al-Rohil, R.N.; Tarasen, A.J.; Carlson, J.A.; Wang, K.; Johnson, A.; Yelensky, R.; Lipson, D.; Elvin, J.A.; Vergilio, J.A.; Ali, S.M.; et al. Evaluation of 122 advanced-stage cutaneous squamous cell carcinomas by Cancers 2017, 9, 86 16 of 18 16 of 18 comprehensive genomic proﬁling opens the door for new routes to targeted therapies. Cancer 2016, 122, 249–257. [CrossRef] [PubMed] comprehensive genomic proﬁling opens the door for new routes to targeted therapies. Cancer 2016, 122, 249–257. [CrossRef] [PubMed] 106. Durinck, S.; Ho, C.; Wang, N.J.; Liao, W.; Jakkula, L.R.; Collisson, E.A.; Pons, J.; Chan, S.W.; Lam, E.T.; Chu, C.; et al. Temporal dissection of tumorigenesis in primary cancers. Cancer Discov. 2011, 1, 137–143. [CrossRef] [PubMed] 107. Quinn, A.G.; Sikkink, S.; Rees, J.L. Basal cell carcinomas and squamous cell carcinomas of human skin show distinct patterns of chromosome loss. Cancer Res. 1994, 54, 4756–4759. [PubMed] 108. Kubo, Y.; Urano, Y.; Hida, Y.; Arase, S. Lack of somatic mutation in the PTEN gene in squamous cell carcinomas of human skin. J. Dermatol. Sci. 1999, 19, 199–201. [CrossRef] 109. Murao, K.; Kubo, Y.; Ohtani, N.; Hara, E.; Arase, S. Epigenetic abnormalities in cutaneous squamous cell carcinomas: Frequent inactivation of the RB1/p16 and p53 pathways. Br. J. Dermatol. 2006, 155, 999–1005. [CrossRef] [PubMed] 110. Hertzler-Schaefer, K.; Mathew, G.; Somani, A.K.; Tholpady, S.; Kadakia, M.P.; Chen, Y.; Spandau, D.F.; Zhang, X. Pten loss induces autocrine FGF signaling to promote skin tumorigenesis. Cell Rep. 2014, 6, 818–826. [CrossRef] [PubMed] 111. Darido, C.; Georgy, S.R.; Wilanowski, T.; Dworkin, S.; Auden, A.; Zhao, Q.; Rank, G.; Srivastava, S.; Finlay, M.J.; Papenfuss, A.T.; et al. Targeting of the tumor suppressor GRHL3 by a miR-21-dependent proto oncogenic network results in PTEN loss and tumorigenesis. Cancer Cell 2011, 20, 635–648. References [CrossRef] [PubMed] 112. Zhao, L.; Li, W.; Marshall, C.; Grifﬁn, T.; Hanson, M.; Hick, R.; Dentchev, T.; Williams, E.; Werth, A.; Miller, C.; et al. Srcasm inhibits Fyn-induced cutaneous carcinogenesis with modulation of Notch1 and p53. Cancer Res. 2009, 69, 9439–9447. [CrossRef] [PubMed] 113. Waterman, E.A.; Sakai, N.; Nguyen, N.T.; Horst, B.A.; Veitch, D.P.; Dey, C.N.; Ortiz-Urda, S.; Khavari, P.A.; Marinkovich, M.P. A laminin-collagen complex drives human epidermal carcinogenesis through phosphoinositol-3-kinase activation. Cancer Res. 2007, 67, 4264–4270. [CrossRef] [PubMed] 114. Segrelles, C.; Ruiz, S.; Perez, P.; Murga, C.; Santos, M.; Budunova, I.V.; Martínez, J.; Larcher, F.; Slaga, T.J.; Gutkind, J.S.; et al. Functional roles of Akt signaling in mouse skin tumorigenesis. Oncogene 2002, 21, 53–64. [CrossRef] [PubMed] 115. Wilker, E.; Lu, J.; Rho, O.; Carbajal, S.; Beltrán, L.; DiGiovanni, J. Role of PI3K/Akt signaling in insulin-like growth factor-1 (IGF-1) skin tumor promotion. Mol. Carcinog. 2005, 44, 137–145. [CrossRef] [PubMed] 116. DiGiovanni, J.; Bol, D.K.; Wilker, E.; Beltrán, L.; Carbajal, S.; Moats, S.; Ramirez, A.; Jorcano, J.; Kiguchi, K. Constitutive expression of insulin-like growth factor-1 in epidermal basal cells of transgenic mice leads to spontaneous tumor promotion. Cancer Res. 2000, 60, 1561–1570. [PubMed] 117. Segrelles, C.; Lu, J.; Hammann, B.; Santos, M.; Moral, M.; Cascallana, J.L.; Lara, M.F.; Rho, O.; Carbajal, S.; Traag, J.; et al. Deregulated activity of Akt in epithelial basal cells induces spontaneous tumors and heightened sensitivity to skin carcinogenesis. Cancer Res. 2007, 67, 10879–10888. [CrossRef] [PubMed] 118. Papadakis, E.S.; Cicho´n, M.A.; Vyas, J.J.; Patel, N.; Ghali, L.; Cerio, R.; Storey, A.; O’Toole, E.A. Axl promotes cutaneous squamous cell carcinoma survival through negative regulation of pro-apoptotic Bcl-2 family members. J. Investig. Dermatol. 2011, 131, 509–517. [CrossRef] [PubMed] 119. Elkabets, M.; Pazarentzos, E.; Juric, D.; Sheng, Q.; Pelossof, R.A.; Brook, S.; Benzaken, A.O.; Rodon, J.; Morse, N.; Yan, J.J.; et al. AXL mediates resistance to PI3Kα inhibition by activating the EGFR/PKC/mTOR axis in head and neck and esophageal squamous cell carcinomas. Cancer Cell 2015, 27, 533–546. [CrossRef] [PubMed] 120. Stephens, L.; Williams, R.; Hawkins, P. Phosphoinositide 3-kinases as drug targets in cancer. Curr. Opin. Pharmacol. 2005, 5, 357–365. [CrossRef] [PubMed] 121. Falasca, M. PI3K/Akt signalling pathway speciﬁc inhibitors: A novel strategy to sensitize cancer cells to anti-cancer drugs. Curr. Pharm. Des. 2010, 16, 1410–1416. [CrossRef] [PubMed] 122. Yap, T.A.; Bjerke, L.; Clarke, P.A.; Workman, P. Drugging PI3K in cancer: Reﬁning targets and therapeutic strategies. Curr. Opin. Pharmacol. References 2015, 23, 98–107. [CrossRef] [PubMed] 123. Stark, A.K.; Sriskantharajah, S.; Hessel, E.M.; Okkenhaug, K. PI3K inhibitors in inﬂammation, autoimmunity and cancer. Curr. Opin. Pharmacol. 2015, 23, 82–91. [CrossRef] [PubMed] 124. Thorpe, L.M.; Yuzugullu, H.; Zhao, J.J. PI3K in cancer: Divergent roles of isoforms, modes of activation and therapeutic targeting. Nat. Rev. Cancer 2015, 15, 7–24. [CrossRef] [PubMed] Cancers 2017, 9, 86 17 of 18 17 of 18 125. Zhao, W.; Qiu, Y.; Kong, D. Class I phosphatidylinositol 3-kinase inhibitors for cancer therapy. Acta Pharm. Sin. B 2017, 7, 27–37. [CrossRef] [PubMed] 126. Ndubaku, C.O.; Heffron, T.P.; Staben, S.T.; Baumgardner, M.; Blaquiere, N.; Bradley, E.; Bull, R.; Do, S.; Dotson, J.; Dudley, D.; et al. Discovery of 2-{3-[2-(1-isopropyl-3-methyl-1H-1,2-4-triazol-5-yl)-5,6- dihydrobenzo[f]imidazo[1,2-d][1,4]oxazepin-9-yl]-1H-pyrazol-1-yl}-2-methylpropanamide (GDC-0032): A β-sparing phosphoinositide 3-kinase inhibitor with high unbound exposure and robust in vivo antitumor activity. J. Med. Chem. 2013, 56, 4597–4610. [PubMed] 127. Ciraolo, E.; Iezzi, M.; Marone, R.; Marengo, S.; Curcio, C.; Costa, C.; Azzolino, O.; Gonella, C.; Rubinetto, C.; Wu, H.; et al. Phosphoinositide 3-kinase p110beta activity: Key role in metabolism and mammary gland cancer but not development. Sci. Signal. 2008, 1, ra3. [CrossRef] [PubMed] 128. Jia, S.; Liu, Z.; Zhang, S.; Liu, P.; Zhang, L.; Lee, S.H.; Zhang, J.; Signoretti, S.; Loda, M.; Roberts, T.M.; et al. Essential roles of PI(3)K-p110beta in cell growth, metabolism and tumorigenesis. Nature 2008, 454, 776–779. [PubMed] 129. Torbett, N.E.; Luna-Moran, A.; Knight, Z.A.; Houk, A.; Moasser, M.; Weiss, W.; Shokat, K.M.; Stokoe, D. A chemical screen in diverse breast cancer cell lines reveals genetic enhancers and suppressors of sensitivity to PI3K isoform-selective inhibition. Biochem. J. 2008, 415, 97–110. [CrossRef] [PubMed] 130. Wee, S.; Wiederschain, D.; Maira, S.M.; Loo, A.; Miller, C.; deBeaumont, R.; Stegmeier, F.; Yao, Y.M.; Lengauer, C. PTEN-deﬁcient cancers depend on PIK3CB. Proc. Natl. Acad. Sci. USA 2008, 105, 13057–13062. [CrossRef] [PubMed] 131. Zhu, Q.; Youn, H.; Tang, J.; Tawﬁk, O.; Dennis, K.; Terranova, P.F.; Du, J.; Raynal, P.; Thrasher, J.B.; Li, B. Phosphoinositide 3-OH kinase p85alpha and p110beta are essential for androgen receptor transactivation and tumor progression in prostate cancers. Oncogene 2008, 27, 4569–4579. [CrossRef] [PubMed] 132. Cai, Y.; Dodhia, S.; Su, G.H. Dysregulations in the PI3K pathway and targeted therapies for head and neck squamous cell carcinoma. Oncotarget 2017, 8, 22203–22217. [CrossRef] [PubMed] 133. Balagula, Y.; Kang, S.; Patel, M.J. References Synergism between mTOR pathway and ultraviolet radiation in the pathogenesis of squamous cell carcinoma and its implication for solid-organ transplant recipients. Photodermatol. Photoimmunol. Photomed. 2015, 31, 15–25. [CrossRef] [PubMed] 134. De Gruijl, F.R.; Koehl, G.E.; Voskamp, P.; Strik, A.; Rebel, H.G.; Gaumann, A.; de Fijter, J.W.; Tensen, C.P.; Bavinck, J.N.; Geissler, E.K. Early and late effects of the immunosuppressants rapamycin and mycophenolate mofetil on UV carcinogenesis. Int. J. Cancer 2010, 127, 796–804. [CrossRef] [PubMed] 135. Wulff, B.C.; Kusewitt, D.F.; VanBuskirk, A.M.; Thomas-Ahner, J.M.; Duncan, F.J.; Oberyszyn, T.M. Sirolimus reduces the incidence and progression of UVB-induced skin cancer in SKH mice even with co-administration of cyclosporine A. J. Investig. Dermatol. 2008, 128, 2467–2473. [CrossRef] [PubMed] 136. Rho, O.; Kiguchi, K.; Jiang, G.; DiGiovanni, J. Impact of mTORC1 inhibition on keratinocyte proliferation during skin tumor promotion in wild-type and BK5.AktWT mice. Mol. Carcinog. 2014, 53, 871–882. [CrossRef] [PubMed] 137. Amornphimoltham, P.; Leelahavanichkul, K.; Molinolo, A.; Patel, V.; Gutkind, J.S. Inhibition of Mammalian target of rapamycin by rapamycin causes the regression of carcinogen-induced skin tumor lesions. Clin. Cancer Res. 2008, 14, 8094–8101. [CrossRef] [PubMed] 138. Geissler, E.K. Skin cancer in solid organ transplant recipients: Are mTOR inhibitors a game changer? Transpl. Res. 2015, 4, 1. [CrossRef] [PubMed] 139. Chockalingam, R.; Downing, C.; Tyring, S.K. Cutaneous squamous cell carcinomas in organ transplant recipients. J. Clin. Med. 2015, 4, 1229–1239. [CrossRef] [PubMed] 140. Jung, J.W.; Overgaard, N.H.; Burke, M.T.; Isbel, N.; Frazer, I.H.; Simpson, F.; Wells, J.W. Does the nature of residual immune function explain the differential risk of non-melanoma skin cancer development in immunosuppressed organ transplant recipients? Int. J. Cancer 2016, 138, 281–292. [CrossRef] [PubMed] 141. Kauffman, H.M.; Cherikh, W.S.; Cheng, Y.; Hanto, D.W.; Kahan, B.D. Maintenance immunosuppression with target-of-rapamycin inhibitors is associated with a reduced incidence of de novo malignancies. Transplantation 2005, 80, 883–889. [CrossRef] [PubMed] 142. Euvrard, S.; Morelon, E.; Rostaing, L.; Gofﬁn, E.; Brocard, A.; Tromme, I.; Broeders, N.; del Marmol, V.; Chatelet, V.; Dompmartin, A.; et al. Sirolimus and secondary skin-cancer prevention in kidney transplantation. N. Engl. J. Med. 2012, 367, 329–339. [CrossRef] [PubMed] 18 of 18 18 of 18 Cancers 2017, 9, 86 143. Campbell, S.B.; Walker, R.; Tai, S.S.; Jian, Q.; Russ, G.R. Randomized controlled trial of sirolimus for renal transplant recipients at high risk of melanoma skin cancer. Am. J. Transpl. 2012, 12, 1146–1156. [CrossRef] [PubMed] 144. References Hoogendijk-van den Akker, J.M.; Harden, P.N.; Hoitsma, A.J.; Proby, C.M.; Wolterbeek, R.; Bouwes Bavinck, J.N.; de Fijter, J.W. Two-year randomized controlled prospective trial converting treatment of stable renal transplant recipients with cutaneous invasive squamous cell carcinomas to sirolimus. J. Clin. Oncol. 2013, 31, 1317–1323. [CrossRef] [PubMed] 145. Alberú, J.; Pascoe, M.D.; Campistol, J.M.; Schena, F.P.; Rial Mdel, C.; Polinsky, M.; Neylan, J.F.; Korth-Bradley, J.; Goldberg-Alberts, R.; Maller, E.S.; et al. Lower malignancy rates in renal allograft recipients converted to sirolimus-based, calcineurin inhibitor-free immunotherapy: 24 month results from the CONVERT trial. Transplantation 2011, 92, 303–310. [CrossRef] [PubMed] 146. Klempner, S.J.; Myers, A.P.; Cantley, L.C. What a tangled web we weave: Emerging resistance mechanisms to inhibition of the phosphoinositide 3-kinase pathway. Cancer Discov. 2013, 3, 1345–1354. [CrossRef] [PubMed] inhibition of the phosphoinositide 3-kinase pathway. Cancer Discov. 2013, 3, 1345–1354. [CrossRef] [PubMed] 147. Brown, K.K.; Toker, A. The phosphoinositide 3-kinase pathway and therapy resistance in cancer. F1000Prime Rep. 2015, 7, 13. [CrossRef] [PubMed] 147. Brown, K.K.; Toker, A. The phosphoinositide 3-kinase pathway and therapy resistance in cancer. F1000Prime Rep. 2015, 7, 13. [CrossRef] [PubMed] 148. O’Reilly, K.E.; Rojo, F.; She, Q.B.; Solit, D.; Mills, G.B.; Smith, D.; Lane, H.; Hofmann, F.; Hicklin, D.J.; Ludwig, D.L.; et al. mTOR inhibition induces upstream receptor tyrosine kinase signaling and activates Akt. Cancer Res. 2006, 66, 1500–1508. [CrossRef] [PubMed] 149. Carracedo, A.; Ma, L.; Teruya-Feldstein, J.; Rojo, F.; Salmena, L.; Alimonti, A.; Egia, A.; Sasaki, A.T.; Thomas, G.; Kozma, S.C.; et al. Inhibition of mTORC1 leads to MAPK pathway activation through a PI3K-dependent feedback loop in human cancer. J. Clin. Investig. 2008, 118, 3065–3074. [CrossRef] [PubMed] 150. Mavrommati, I.; Maffucci, T. mTOR inhibitors: Facing new challenges ahead. Curr. Med. Chem. 2011, 18, 2743–2762 [CrossRef] [PubMed] 149. Carracedo, A.; Ma, L.; Teruya-Feldstein, J.; Rojo, F.; Salmena, L.; Alimonti, A.; Egia, A.; Sasaki, A.T.; Thomas, G.; Kozma, S.C.; et al. Inhibition of mTORC1 leads to MAPK pathway activation through a PI3K-dependent feedback loop in human cancer. J. Clin. Investig. 2008, 118, 3065–3074. [CrossRef] [PubMed] Thomas, G.; Kozma, S.C.; et al. Inhibition of mTORC1 leads to MAPK pathway activation through a PI3K-dependent feedback loop in human cancer. J. Clin. Investig. 2008, 118, 3065–3074. [CrossRef] [PubMed] 150. Mavrommati, I.; Maffucci, T. mTOR inhibitors: Facing new challenges ahead. Curr. Med. Chem. 2011, 18, 2743–2762. [CrossRef] [PubMed] p p 150. Mavrommati, I.; Maffucci, T. mTOR inhibitors: Facing new challenges ahead. Curr. Med. Chem. References 2011, 18, 2743–2762. [CrossRef] [PubMed] © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
https://openalex.org/W2763971076	https://europepmc.org/articles/pmc5682656?pdf=render	English	null	Neuron and microglia/macrophage-derived FGF10 activate neuronal FGFR2/PI3K/Akt signaling and inhibit microglia/macrophages TLR4/NF-κB-dependent neuroinflammation to improve functional recovery after spinal cord injury	Cell death and disease	2,017	cc-by	9,583	Neuron and microglia/macrophage-derived FGF10 activate neuronal FGFR2/PI3K/Akt signaling and inhibit microglia/macrophages TLR4/NF-κB-dependent neuroinflammation to improve functional recovery after spinal cord injury Jian Chen1,4, Zhouguang Wang2,4, ZengMing Zheng1,2, Yu Chen1, Sinan Khor3, KeSi Shi1, ZiLi He1, Qingqing Wang1, Yingzheng Zhao2, Hongyu Zhang2, Xiaokun Li2, Jiawei Li1, Jiayu Yin2, Xiangyang Wang,1 and Jian Xiao,1,2 Therapeutics used to treat central nervous system (CNS) injury were designed to repair neurites and inhibit cell apoptosis. Previous studies have shown that neuron-derived FGF10 exerts potential neuroprotective effects after cerebral ischemia injury. However, little is known about the role of endogenous FGF10 in the recovery process after spinal cord injury (SCI). In this study, we found that FGF10 is mainly produced by neuron and microglia/macrophages, and its expression is increased after SCI. Exogenous treatment of FGF10 improved functional recovery after injury by reducing apoptosis, as well as repairing neurites via FGFR2/PI3K/ Akt pathway. On another hand, inhibiting the PI3K/Akt pathway with LY294002 partially reversed the therapeutic effects of FGF10. In addition, small interfering RNA knockdown of FGFR2 suppressed PI3K/Akt pathway activation by FGF10 and abolished its anti- apoptotic and neurite repair effects in vitro. Furthermore, FGF10 treatment inhibited the activation and proliferation of microglia/ macrophages through regulation of TLR4/NF-κB pathway, and attenuated the release of pro-inflammatory cytokines after SCI. Thus, the increased expression of FGF10 after acute SCI is an endogenous self-protective response, suggesting that FGF10 could be a potential treatment for CNS injury. p j y Cell Death and Disease (2017) 8, e3090; doi:10.1038/cddis.2017.490; published online 5 October 2017 has revolved around repairing injured dendrites and axons and promoting their outgrowth. 1Department of Orthopaedic Surgery, The Second Affiliated Hospital and Yuying Children's Hospital of Wenzhou Medical University, Wenzhou, China; 2Molecular Pharmacology Research Center, School of Pharmaceutical Sciences, Wenzhou Medical University, Wenzhou, China and 3Department of Molecular Pharmacology, Albert Einstein College of Medicine, Bronx, NY, USA Corresponding author: X Wang or J Xiao, Department of Orthopaedic Surgery, The Second Affiliated Hospital and Yuying Children's Hospital of Wenzhou Medical University, Wenzhou 325027, China. Tel: +86 577 88002814; Fax: +86 577 88002823; E-mail: xiangyangwang@wmu.edu.cn or xfxj2000@126.com 4These authors contributed equally to this work. Received 17.5.17; revised 07.8.17; accepted 28.8.17; Edited by B Joseph Citation: Cell Death and Disease (2017) 8, e3090; doi:10.1038/cddis.2017.490 Ofﬁcial journal of the Cell Death Differentiation Association Citation: Cell Death and Disease (2017) 8, e3090; doi:10.1038/cddis.2017.490 Ofﬁcial journal of the Cell Death Differentiation Association OPEN Neuron and microglia/macrophage-derived FGF10 activate neuronal FGFR2/PI3K/Akt signaling and inhibit microglia/macrophages TLR4/NF-κB-dependent neuroinflammation to improve functional recovery after spinal cord injury Cytoskeleton remodeling, such as microtubule assembly, occurs at the top of a growing neurite and is believed to be crucial for growth cone initiation and regrowth of injured neuritis.5 Microtubule dynamics give rise to a highly polarized morphology in neurons, as evidenced by a single axon and multiple dendrites.6 In recent years, it was reported that pharmacological treatment to stabilize micro- tubules promotes axon regeneration after SCI.7 In addition, it was reported that FGF13 acts as a microtubule-stabilizing protein to regulate neuronal migration and polarization in the cerebral cortex.8 Traumatic spinal cord injury (SCI) is a major cause of death and lifelong disability in the world, and over 250 000 people suffer SCI in the United States.1 The pathological process of SCI can be divided into two phases: (1) the primary injury characterized by direct local mechanical damage to the spinal cord at the time of injury, and (2) the secondary injury, which could possibly be counteracted using neuroprotective agents. Apart from local ischemia, other detrimental events such as local edema, focal hemorrhage, excitotoxicity and in particular, oxidative stress and post-ischemic neuroinflammation, con- tribute to prolonged secondary tissue injury after SCI.2,3 These detrimental secondary events result in neuronal cell death or structural damage of surviving neurons, leading to physical and functional deficits. Neurites of injured neurons in the adult CNS can seldom spontaneously regenerate in an inhibitory environment.9 Post- ischemic neuroinflammation is usually regarded as a deleter- ious factor to neurological function and leads to progressive deterioration of ventral horn motor neurons.10 The neuroin- flammatory changes are attributed to microglia, the resident immunocyte in the CNS. Microglia are scavenger, which remove dead cells and are related to both elimination and maintenance of synapses for neural signal transduction.11 After injury, microglia are rapidly activated, undergoing Neurons are highly polarized cells that are composed of dendrites, which are tapered, shorter extensions to receive information and an axon, which is a thin, long hair-like extension to transmit information. Several studies have demonstrated the presence of neuronal death and axonal interruption around a primary lesion, which is the main obstacle preventing recovery from secondary damage.4 As a result of this, treatment of central nervous system (CNS) injury A novel therapeutic intervention for SCI J Chen et al A novel therapeutic intervention for SCI J Ch t l Neuron and microglia/macrophage-derived FGF10 increases with the activation of FGFR2/PI3K/Akt signaling after acute traumatic SCI. Neuron and microglia/macrophage-derived FGF10 activate neuronal FGFR2/PI3K/Akt signaling and inhibit microglia/macrophages TLR4/NF-κB-dependent neuroinflammation to improve functional recovery after spinal cord injury I hi d i d l h i l When triggered by a stimulus such as direct mechanical trauma, activated microglia expresses high levels of Toll-like receptors (TLRs) in the CNS.16 Several studies have suggested that a wide array of TLRs, in particular TLR4, on microglia/macrophages can be further stimulated by secreted cytokines. This enhances a pro-inflammatory environment and exacerbates neuronal death and dysfunction.17 In a study performed using middle cerebral artery occlusion (MCAO) mice model, TLR4-deficient mice have less inflammatory response, contributing to minor infarct size after impact.18 These results link TLR4 signaling pathway and innate immunity with neuroinflammation triggered by ischemic injury. S f C S y gg y j y Some endogenous factors released by the CNS that are induced by injury may also be beneficial recovery from the injury.19 However, the factors and associated mechanisms have not been fully investigated. The fibroblast growth factors (FGFs) are a family of cell signaling molecules released by various tissues that share a broad spectrum of biochemical and biological properties. FGF10 is a typical paracrine FGF and was originally cloned from rat embryos.20 Numerous articles have reported that innate FGF signaling promotes wound repair and tissue regeneration, regulates multiple organs development and maintains tissue homeostasis.21,22 However, few articles have reported the role of FGF10 in CNS injury. A recent study reported for the first time that brain FGF10 is primarily produced from neurons and upregulated in a model for MCAO model.23 Exogenous FGF10 treatment ameliorated cerebral ischemic injury and reduced neuronal apoptosis as well.23 Nevertheless, whether there is a change in the expression of FGF10 and the signaling pathways activated by FGF10 after SCI have not been reported. To determine whether FGF10 activated PI3K/Akt signaling is mediated by FGFR2, FGFR2 small interfering RNA (siRNA) was used to knockdown FGFR2 in PC12 cell before FGF10 treatment (Supplementary Figures S2a-c). FGFR2 siRNA significantly lowered the ratio of p-Akt/t-Akt after FGF10 treatment, when compared with both the control group and the negative control siRNA group (Supplementary Figures S2a and d). Besides, FGFR2 knockdown further lowered the ratio of p-Akt/t-Akt without pretreatment of FGF10 (Supplementary Figures S2e and f). These results suggested that FGFR2 mediates FGF10- activation of the PI3K/Akt signaling pathway. Thus, we supposed that the FGFR2/PI3K/Akt signaling pathway is involved in the therapeutic effect of FGF10. Neuron and microglia/macrophage-derived FGF10 activate neuronal FGFR2/PI3K/Akt signaling and inhibit microglia/macrophages TLR4/NF-κB-dependent neuroinflammation to improve functional recovery after spinal cord injury (a and on data of FGF10 expression at several time points after acute SCI. (c-e) Double immunofluorescence of FGF10 and cellular markers for a/macrophages (Iba-1), in spinal cord tissue adjacent to lesion (scale bar: 50 μm). (e-g) Western blots and quantification data of FGFR2, p-Akt er surgery. (h) Double immunofluorescence of FGFR2 and NeuN in sections from tissue at 1 day after SCI (scale bar: 50 μm). Data repre differences between the SCI and sham groups are indicated as Po0.05, Po0.01, Po0.001, n = 5 J Chen et al 2 Figure 1 Neuron and microglia/macrophage-derived FGF10 increases with the activation of FGFR2/PI3K/Akt signaling after acute traumatic SCI. (a and b) Western blots and quantification data of FGF10 expression at several time points after acute SCI. (c-e) Double immunofluorescence of FGF10 and cellular markers for (c) neuron (NEUN) or (d) microglia/macrophages (Iba-1), in spinal cord tissue adjacent to lesion (scale bar: 50 μm). (e-g) Western blots and quantification data of FGFR2, p-Akt and Akt in each group at 1 day after surgery. (h) Double immunofluorescence of FGFR2 and NeuN in sections from tissue at 1 day after SCI (scale bar: 50 μm). Data represent the mean ± S.D. Significant differences between the SCI and sham groups are indicated as Po0.05, Po0.01, Po0.001, n = 5 Cell Death and Disease A novel therapeutic intervention for SCI J Chen et al 3 immunofluorescence staining in spinal cord tissue adjacent to the lesion showed that FGF10 is mainly present in neurons and microglia/macrophages rather than astrocytes (Figures 1c and d, Supplementary Figure S1a). And double immunofluorescence for FGF10 and CD68, an anther marker of microglia/ macrophages, showed the same result (Supplementary Figure S1b). Thus, we hypothesized that FGF10 may have a role in neurons and microglia/macro- phages. FGF10 mediates numerous biological responses by activating FGFR2/PI3K/Akt signaling in a paracrine manner. However, few articles have examined the expression of FGFR2 after acute SCI. Interestingly, we observed higher expression of FGFR2 and ratio of p-Akt/t-Akt after acute SCI by western blotting, and FGF10 treatment upregulated the expression of p-Akt, but slightly change the level of FGFR2 in SCI group (Figures 1e–g). Double immunofluorescence staining for FGFR2, as well as neurons (NEUN) showed that FGFR2 is significantly upregulated in neurons, similar to FGF10. The SCI group had increased FGFR2-positive puncta in neurons compared with sham group (Figure 1h). Neuron and microglia/macrophage-derived FGF10 activate neuronal FGFR2/PI3K/Akt signaling and inhibit microglia/macrophages TLR4/NF-κB-dependent neuroinflammation to improve functional recovery after spinal cord injury Furthermore, we found the increasing expression of other FGFs, such as FGF1, FGF2 and FGF7in SCI group, (Supplementary Figures S1c–g), which is consistent with other study for FGF1 and FGF2.24 However, our study first reported the increasing level of FGF7 after SCI, especially in acute stage, which may activate FGFR2 after SCI. morphological and molecular changes, which are related to neurotoxicity.12 As a result of the blood–brain barrier disruption,13 there are evidences that hematogenous macro- phages contribute to secondary tissue damage in acute CNS injury.14,15 However, the distinctions between microglia and macrophages in CNS have been elusive with the lack of discriminating marker. d sc at g a e When triggered by a stimulus such as direct mechanical trauma, activated microglia expresses high levels of Toll-like receptors (TLRs) in the CNS.16 Several studies have suggested that a wide array of TLRs, in particular TLR4, on microglia/macrophages can be further stimulated by secreted cytokines. This enhances a pro-inflammatory environment and exacerbates neuronal death and dysfunction.17 In a study performed using middle cerebral artery occlusion (MCAO) mice model, TLR4-deficient mice have less inflammatory response, contributing to minor infarct size after impact.18 These results link TLR4 signaling pathway and innate immunity with neuroinflammation triggered by ischemic injury. Some endogenous factors released by the CNS that are induced by injury may also be beneficial recovery from the injury.19 However, the factors and associated mechanisms have not been fully investigated. The fibroblast growth factors (FGFs) are a family of cell signaling molecules released by various tissues that share a broad spectrum of biochemical and biological properties. FGF10 is a typical paracrine FGF and was originally cloned from rat embryos.20 Numerous articles have reported that innate FGF signaling promotes wound repair and tissue regeneration, regulates multiple organs development and maintains tissue homeostasis.21,22 However, few articles have reported the role of FGF10 in CNS injury. A recent study reported for the first time that brain FGF10 is primarily produced from neurons and upregulated in a model for MCAO model.23 Exogenous FGF10 treatment ameliorated cerebral ischemic injury and reduced neuronal apoptosis as well.23 Nevertheless, whether there is a change in the expression of FGF10 and the signaling pathways activated by FGF10 after SCI have not been reported. Neuron and microglia/macrophage-derived FGF10 activate neuronal FGFR2/PI3K/Akt signaling and inhibit microglia/macrophages TLR4/NF-κB-dependent neuroinflammation to improve functional recovery after spinal cord injury In this study, we aimed to explore the potential neuropro- tective effects of FGF10 after acute SCI both in vivo and in vitro, as well as the mechanism by which it promotes neurite repair and prevents apoptosis. We further studied the mechanisms underlying the inflammatory response after SCI and the signaling pathways that mediate FGF10’s beneficial effects. Our results support that FGF10 may be a novel therapeutic intervention for SCI and pontentially could be useful for other traumatic CNS diseases. FGF10 decreases spinal cord tissue damage and motor neuron loss, and promotes locomotor recovery from SCI in vivo. As spinal FGF10 is increased in neurons after acute SCI, we explored if this could have a therapeutic effect on SCI by administering exogenous FGF10 in SCI model. The therapeutic effect of FGF10 is in part due to activation of the PI3K/Akt pathway in many biological processes, including the early ischemia/reperfusion injury.23,25,26 BBB scores and inclined plane test scores were using to assess the therapeutic effect of FGF10. BBB scores in the SCI and FGF10 groups were significantly below normal with no significant difference within the first week after surgery. However, BBB scores began to increase at 14 days after surgery in the FGF10 group (Figure 2a). Similarly, we noted higher inclined plane test scores in the FGF10-treated group at 14, 21 and 28 days after SCI (Figure 2b), suggesting that locomotor function was significantly improved compared with the SCI group. To further confirm the neuroprotective role of FGF10, we used a specific PI3K inhibitor, LY294002, in conjunction with FGF10 treatment. LY294002 significantly Results Neuron and microglia/macrophage-derived FGF10 increases activation of FGFR2/PI3K/Akt signaling after acute SCI. In order to detect the expression of FGF10 after SCI, the T7–T10 level around the lesion epicenter of spinal cord tissue were excised. We assessed the influence of acute SCI on FGF10 protein expression at several time points from spinal cord tissue. Western blot analyses showed that FGF10 was significantly increased after acute SCI and peaked at 1-day post-operation (Figures 1a and b). Double Cell Death and Disease A novel therapeutic intervention for SCI J Chen et al A novel therapeutic intervention for SCI J Chen et al igure 2 FGF10 decreases spinal cord tissue damage, motor neuron loss and promotes locomotor recovery after acute SCI in vivo. (a and b) The BBB scores and inc ane test scores of each group. (c and d) HE staining of each group at 28 days after surgery and quantification data of the percent of cavity necrotic tissue at each interval ( ar: 200 μm). (e) Nissl staining of each group to test the surviving neurons at 28 days after surgery. Data represent the mean ± S.D. Significant differences between the treat nd control groups are indicated as Po0.05, *Po0.01, n = 5 J Chen et al 4 Figure 2 FGF10 decreases spinal cord tissue damage, motor neuron loss and promotes locomotor recovery after acute SCI in vivo. (a and b) The BBB scores and inclined plane test scores of each group. (c and d) HE staining of each group at 28 days after surgery and quantification data of the percent of cavity necrotic tissue at each interval (scale bar: 200 μm). (e) Nissl staining of each group to test the surviving neurons at 28 days after surgery. Data represent the mean ± S.D. Significant differences between the treatment and control groups are indicated as Po0.05, *Po0.01, n = 5 Figure 2 FGF10 decreases spinal cord tissue damage, motor neuron loss and promotes locomotor recovery after acute SCI in vivo. (a and b) The BBB scores and inclined plane test scores of each group. (c and d) HE staining of each group at 28 days after surgery and quantification data of the percent of cavity necrotic tissue at each interval (scale bar: 200 μm). (e) Nissl staining of each group to test the surviving neurons at 28 days after surgery. Data represent the mean ± S.D. Results Significant differences between the treatment and control groups are indicated as Po0.05, *Po0.01, n = 5 tissue and decreasing motor neuron survival compared with FGF10 treatment alone (Figures 2c–e). Taken together, FGF10 could exert a neuroprotective effect on SCI in vivo. suppressed beneficial effect of FGF10 on functional recovery (Figures 2a and b). The HE and Nissl staining results revealed that the SCI group displayed greater destruction of central gray matter and peripheral white matter, which followed by remarkable motor neuron loss in the anterior horn. However, the FGF10-treated group had a decreased cavity of necrotic tissue around the injury site and decreased motor neuron loss in the anterior horn, showing that FGF10 protected against severe damage after SCI (Figures 2c and d). Moreover, LY294002 treatment significantly increased the damage caused by SCI, aggravating the cavity of necrotic FGF10 treatment decreases apoptosis through activation of the PI3K/Akt pathway. To test whether FGF10 treatment decreases apoptosis in SCI, TUNEL staining was performed, SCI significantly increased the number of apoptotic cells compared with the sham group. In comparison, FGF10 treatment greatly reduced apoptotic activity, but this was in part reversed by LY294002 (Figures 3a and b). Moreover, FGF10 treatment decreases apoptosis through activation of the PI3K/Akt pathway. To test whether FGF10 treatment decreases apoptosis in SCI, TUNEL staining was performed, SCI significantly increased the number of apoptotic cells compared with the sham group. In comparison, FGF10 treatment greatly reduced apoptotic activity, but this was in part reversed by LY294002 (Figures 3a and b). Moreover, Cell Death and Disease A novel therapeutic intervention for SCI J Chen et al FGF10 reduces apoptosis via activation of the PI3K/Akt pathway. After SCI in rats, (a and b) TUNEL assay was performed in each group at 7 days after SC ). (c and d) Western blots and quantification data of cleaved-caspase 3, Bax and Bcl-2 of each group at 3 days after surgery. (e) Double immunofluorescence eaved-caspase 3 (green) of each group at 3 days after surgery (scale bar: 50 μm). Data represent the mean ± S.D. Significant differences between the treatm are indicated as Po0.05, Po0.01, Po0.001, n = 5 J Chen et al 5 Figure 3 FGF10 reduces apoptosis via activation of the PI3K/Akt pathway. After SCI in rats, (a and b) TUNEL assay was performed in each group at 7 days after SCI (scale bar: 50 μm). Results And double immunofluorescence assay of pro- inflammatory cytokines (TNF-α, IL-6) and Iba-1+ cells showed that FGF10 treatment reduced the pro- inflammatory cytokine release in microglia/macrophages after SCI (Figures 5f and g). These results indicated that FGF10 treatment significantly inhibited microglia/macro- phages activation and migration, and tightly regulated the production of pro-inflammatory cytokines in microglia/macro- phages following SCI. FGF10 treatment suppresses the TLR4/NF-κB signaling pathway. To further confirm the underlying anti-inflammatory effect of FGF10, we explored whether the TLR4/NF-κB pathway was involved in rats after SCI. We observed increased expression of TLR4 in the SCI group compared with sham group, which was reversed by FGF10 treatment (Figures 6a and b). Immunostaining results showed that FGF10 treatment significantly lowered TLR4-expressing microglia/macrophages compared with SCI group (Figure 6c). We also examined the protein levels of p-IκBα, IκBα and NF-κB (p65). And found higher expression of p65 and p-IκBα in the SCI group compared with the sham group, but this was reduced by FGF10 treatment (Figures 6d–g). To further confirm whether the TLR4/NF-κB pathway mediates the anti-inflammatory mechanism of FGF10, we used LPS, a TLR4 ligand, to activate the TLR4/NF-κB pathway in BV-2 cells, which should increase neuroinflammation.30 Pretreat- ment with FGF10 significantly reduced TLR4 expression and attenuated NF-κB activation compared with LPS treatment alone (Supplementary Figures S5a–e). Similarly, immunos- taining results showed that pretreatment with FGF10 sig- nificantly decreased the amount of TLR4 in LPS-treated BV-2 cells (Supplementary Figure S5f). Nuclear translocation of NF-κB triggers transcription of many inflammatory genes. Immunostaining assays further showed that FGF10 treatment significantly attenuated nuclear translocation of NF-κB induced by LPS (Supplementary Figure S5g). FGF10 improves neurite repair and enhances axonal sprouting in acute SCI. Microtubule-associated protein 2 (MAP2), a specific structural protein in neuron, is mainly expressed in neuronal dendrites and is known to stabilize microtubules and regulate the length of dendrites.27,28 How- ever, the expression of MAP2 and acetylated tubulin (AcTub) after acute SCI is unclear. As shown in Figures 4a and b, the expression of AcTub and MAP2 protein decreased, reaching the lowest point at 1 or 2 days post-SCI (Figures 4a–c), showing that SCI reduced microtubule protein with limited repair capacity. Moreover, AcTub and MAP2 were upregulated in the FGF10 group on the first day after injury compared with the SCI group, but this effect was reversed by LY294002 treatment (Figures 4d–f). Results (c and d) Western blots and quantification data of cleaved-caspase 3, Bax and Bcl-2 of each group at 3 days after surgery. (e) Double immunofluorescence of NeuN (red) and cleaved-caspase 3 (green) of each group at 3 days after surgery (scale bar: 50 μm). Data represent the mean ± S.D. Significant differences between the treatment and SCI groups are indicated as Po0.05, Po0.01, Po0.001, n = 5 Cell Death and Disease Cell Death and Disease A novel therapeutic intervention for SCI J Chen et al A novel therapeutic intervention for SCI J Chen et al A novel therapeutic intervention for SCI J Ch t l 6 western blot analysis showed increased levels of cleaved- caspase 3 and Bax in the SCI group, which was significantly attenuated by FGF10 treatment. In contrast, FGF10 increased the level of Bcl-2 compare with SCI group. Moreover, LY294002 reversed the anti-apoptotic effect of FGF10 as shown by increased Bax and cleaved-caspase 3 and decreased Bcl-2, which is consistent with our double immunofluorescence staining results (Figures 3c–e). To further investigate the effect of FGF10 on cell viability, we knocked down FGFR2 using siRNA in PC12 cells before FGF10 treatment. For the in vitro study, H2O2 treatment was used to mimic neuronal injury after acute SCI. TUNEL assay results showed that FGFR2 knockdown increased the apoptotic activity compared with the FGF10-treated H2O2 group (Supplementary Figures S3a and b). On another hand, FGF10 markedly decreased the expression of cleaved- caspase 3 and Bax and increased the expression of Bcl-2 with H2O2 treatment. However, FGFR2 knockdown reversed the anti-apoptotic effect of FGF10 (Supplementary Figures S3c and d). Similarly, immunofluorescent staining revealed increased cleaved-caspase 3-positive puncta with FGFR2 knockdown compared with the FGF10-treated H2O2 group (Supplementary Figure S3e). In addition, FGF10 treatment increased the ratio of p-Akt/Akt induced by H2O2, which was suppressed by FGFR2 knockdown (Supplementary Figures S3f and g,). These results further demonstrated the anti- apoptotic effect of FGF10 after SCI. we examined expression of Iba-1 and pro-inflammatory cytokines IL-6 and TNF-α. The SCI group showed increased expression of Iba-1, IL-6 and TNF-α compared with sham group, which was significantly reversed by FGF10 treatment (Figures 5a-d). Importantly, the results of immunohistochem- ical staining of Iba-1 showed FGF10 reduced the Iba-1+ microglia/macrophages population at the injury area (Figure 5e). Results Immunofluorescent staining showed that FGF10 treatment promoted the outgrowth of AcTub labeled axons,29 which elongate into the distal regions of the SCI area compared with the untreated and LY294002 groups (Figure 4g), suggesting that FGF10 may have a role in stabilizing microtubule structure and repairing neurites after acute SCI. In neuronal cultures, FGFR2 knockdown reversed the increased microtubule stabilization seen with FGF10 treatment by reducing the expression of AcTub and MAP2 (Supplementary Figures S4a–c). Immunofluorescent staining showed that pretreatment with FGFR2 siRNA abolished the beneficial effect of FGF10 on neuronal repair (Supplementary Figures S4d and e). Taken together, FGF10 activated FGFR2/ PI3K/Akt signaling contributes to the repair of neurites. Discussion In recent years, various pharmacological treatments have focused on axonal and dendritic repair to enhance recovery from CNS injury.31,32 Some neurotrophins, including brain- derived neurotrophic factor and nerve growth factor (NGF), have been proven to effectively promote neurite outgrowth.19,33 However, most research neglects the body’s self-repair mechanisms after CNS injury. In this study, we found that endogenous FGF10 is significantly released after SCI from neurons and microglia/macrophages, especially in the acute phase. We further characterized the role of endogenous FGF10 after SCI, both in neuron and microglia/ macrophages. FGF10 treatment prevents microglia/macrophages acti- vation and reduces pro-inflammatory cytokine release. To determine whether FGF10 affected microglia/macro- phages activation and pro-inflammatory cytokine release, Cell Death and Disease A novel therapeutic intervention for SCI J Chen et al Figure 4 FGF10 promotes neurite repair in acute SCI. (a-c) Western blots and quantification data of AcTub and MAP2 expression at several time points after acute SCI. (d-f Western blots and quantification data of AcTub and MAP2 in each group at 1 day post-surgery. (g) Immunofluorescence of AcTub (green) and GFAP (red) of the injured spinal cord sections from tissue in each group at 28 days after surgery (scale bar: 50 μm). Data represent the mean ± S.D. Significant differences between the treatment and control group are indicated Po0.05, Po0.01, Po0.001, n = 5 Figure 4 FGF10 promotes neurite repair in acute SCI. (a-c) Western blots and quantification data of AcTub and MAP2 expression at several time points after acute SCI. (d-f) Western blots and quantification data of AcTub and MAP2 in each group at 1 day post-surgery. (g) Immunofluorescence of AcTub (green) and GFAP (red) of the injured spinal cord sections from tissue in each group at 28 days after surgery (scale bar: 50 μm). Data represent the mean ± S.D. Discussion Significant differences between the treatment and SCI groups are indicated as 1 d ft ( l b 200 ) (f d ) I fl f Figure 5 FGF10 prevents microglia/macrophages activation and reduces pro-inflammatory cytokine release in rats after acute SCI. (a-d) Western blots and quantification data of Iba-1, TNF-α and IL-6 in each group at 1 day after SCI. Data represent the mean ± S.D. Significant differences between the treatment and SCI groups are indicated as Po0.05, *Po0.01, n = 5. (e) Immunohistochemical staining of Iba-1 in each group at 1 day after surgery (scale bar: 200 μm). (f and g) Immunofluorescence of pro- inflammatory cytokines (TNF-α and IL-6, green) and Iba-1 (red) in each group at 1 day after surgery signaling may have a role in cellular death. Bcl-2, Bax and cleaved-caspase 3 were used as markers for apoptotic activation or inhibition. Among them, Bcl-2 has an anti- apoptotic effect, whereas the release of Bax and cleaved- caspase 3 are pro-apoptotic.45 Our results showed that FGF10 activated the PI3K/Akt pathway and significantly decreased the protein expression of Bax and cleaved- caspase 3, and upregulated the expression of Bcl-2 in rats of SCI. Interestingly, FGFR2 knockdown blocked activation of PI3K/Akt signaling pathway and abolished the anti-apoptotic effect of FGF10. These results showed that FGF10 activated the FGFR2/PI3K/Akt pathway as a neuroprotective mechan- ism after SCI to reduce neuronal apoptosis caused by oxidative stress. cytoskeletal proteins, such as microtubules. Microtubules consist of heterodimers of α-tubulin and β-tubulin, and are crucial structural components of neurites. Moreover, micro- tubules have pivotal roles in neuronal function, such ante- rograde and retrograde transport in the axon.48 As acetylated tubulin is abound in stable microtubules, activating histone deacetylase by calcium ions, accelerates microtubule depo- lymerization through tubulin deacetylation.49 Evidence sug- gests that MAP2 deletion reduces microtubule density and length in dendrites.50 In addition, MAP2 mediates a link between cellular signaling and cytoskeletal structure, acting as a molecular scaffold upon which cytoskeleton-modifying proteins dissociate and assemble during neuronal activity.51 In this study, we first demonstrated reduced expression of AcTub and MAP2 after acute SCI, directly contributing to neuronal dysfunction and death. Cell Death and Disease Discussion Significant differences between the treatment and control groups are indicated Po0.05, Po0.01, Po0.001, n = 5 After crushing SCI, the initial trauma is followed by prolonged secondary injury including many inflammatory, ischemic and neurotoxic events that structurally damage the neuronal integrity around the injury site.34 Endogenous ROS activate various intrinsic pathways, including the pro-apoptotic signaling pathways in neurons.35 In the 'intrinsic pathway', Bcl- 2 family proteins (such as cytochrome c, endonuclease G, caspase and AIF) combine with each other, leading to the release of pro-apoptotic proteins, as well as liberating caspase-activated DNase, triggering activation of the apoptosis.36,37 ROS can activate various upstream signaling mechanisms, including p53 and PI3K/Akt, which both regulate the intrinsic pathway.38,39 The PI3K/Akt pathway is critical for growth and survival in many biological processes, including early ischemia/reperfusion injury as shown in our previous work.40 Akt phosphorylates and inactivates Bad, a pro- apoptotic Bcl-2 family protein, reducing apoptosis after cerebral ischemia.41 Akt also suppresses the activation of pro-caspase-9, and caspase-9 phosphorylation, preventing apoptotic activation.42 The activation of PI3K/Akt signaling pathway by FGF10 could be attributed to FGFR2b, the receptor of FGF10.43 Several studies have reported that FGFR2 have a critical role in regulating oxidative stress and cellular apoptosis. In addition, FGFR2 is upregulated in myxoid liposarcoma, and inhibiting the expression of FGFR2 reduced cell proliferation and increased apoptosis.44 In this study, we observed higher FGFR2 expression in neurons and significant increase of FGF7 and FGF10 on the first day after acute SCI, which have been reported to activate the FGFR2. So we supposed that FGF10 activated FGFR2/PI3K/Akt Cell Death and Disease A novel therapeutic intervention for SCI J Chen et al Figure 5 FGF10 prevents microglia/macrophages activation and reduces pro-inflammatory cytokine release in rats after acute SCI. (a-d) Western blots and quantification data of Iba-1, TNF-α and IL-6 in each group at 1 day after SCI. Data represent the mean ± S.D. Significant differences between the treatment and SCI groups are indicated as Po0.05, *Po0.01, n = 5. (e) Immunohistochemical staining of Iba-1 in each group at 1 day after surgery (scale bar: 200 μm). (f and g) Immunofluorescence of pro- inflammatory cytokines (TNF-α and IL-6, green) and Iba-1 (red) in each group at 1 day after surgery 8 ory cytokine release in rats after acute SCI. (a-d) Western blots and quantification .D. Discussion These results were consistent with our in vivo results, suggesting that the TLR4/NF-κB pathway is involved in the underlying anti-inflammatory mechanism of FGF10. We also observed increased expression of FGF10 in microglia/macrophages after SCI, which led us to investigate if FGF10 also is important for microglia/macrophages’ func- tion. After SCI, the prolonged inflammatory response enhances resident microglia/macrophages’ activation and proliferation, which subsequently promotes production of pro-inflammatory factors, such as TNF-α and IL-6, creating an inhibitory environment for neurite regeneration.9 Activated microglia/macrophages produce a variety of pro-inflammatory mediators, as well as other toxic mediators, which trigger signaling cascades and neurotoxic responses in the second- ary phase of SCI. These events significantly contribute to both neuronal death and neurite injury.55,56 Many studies have been reported about the critical role of TLR4 in ischemic CNS. Activation of TLR4 signaling contributes to astrocyte-mediated inflammation, and may control pro-inflammatory astroglial conversion to the neurodegenerative phenotype.57 It have been reported to protects blood–brain barrier by inhibiting TLR4-mediated inflammatory pathway in ischemic brain.58,59 Figure 7 A schematic diagram depicting the potential molecular mechanisms underlying FGF10 protection via neurite repair, reducing apoptosis and decreasing inflammatory cytokines after acute SCI phosphorylated-Akt (Ser473) and cleaved-caspase 3 antibodies were purchased from Abcam (Cambridge, MA, USA). The reagents of cell culture were obtained from Gibco (Grand Island, NY, USA). All other reagents were purchased from Sigma-Aldrich (St. Louis, MO, USA) unless specified otherwise. phosphorylated-Akt (Ser473) and cleaved-caspase 3 antibodies were purchased from Abcam (Cambridge, MA, USA). The reagents of cell culture were obtained from Gibco (Grand Island, NY, USA). All other reagents were purchased from Sigma-Aldrich (St. Louis, MO, USA) unless specified otherwise. Surgical procedure. All the surgical interventions and postoperative animal care procedures were in strict accordance with the Animal Care and Use Committee of Wenzhou Medical College. All Sprague–Dawley rats were housed in the SPF Laboratory Animal Room. The rats were injected intraperitoneally with 10% chloral hydrate (3.6 ml/kg), and positioned on a cork platform as discussed previously.61 The operator incised the skin to expose the vertebral column in the dorsum, and then performed a laminectomy at the T9 vertebral section. And the spinal cord was clearly exposed and clamped by a vascular clip (30 g force; Oscar, Shanghai, China) for 1 min to simulate a moderate crushing injury model. For the sham group, a T9 laminectomy was performed and the exposed spinal cord for 1 min without compression injury. Discussion After surgery, we emptied bladder twice daily until the recovery of bladder function. FGF10 was dissolved in saline and administered intravenously (1mg/kg/day) until the rats were killed.62 After surgery, another group of rats was injected with 1mg/kg/day FGF10 and a specific PI3K inhibitor (LY294002, 0.3 mg/kg, i.v.) at the same time. The sham group was injected with saline. Cell culture treatment protocols. The PC12 cells and BV-2 cells were obtained from Cell Bank of Type Culture Collection of Chinese Academy of Sciences, Shanghai Institute of Cell Biology, Chinese Academy of Sciences. PC12 cells were cultured in RPMI-1640 medium with 10% (v/v) fetal bovine serum (FBS), 100 U/ml penicillin and 100 U/ml streptomycin. BV-2 cells were cultured in MEM with heat-inactivated 10% (v/v) FBS, 100 U/ml penicillin and 100 U/ml streptomycin. PC12 cells were treated with FGF10 (100 ng/ml) and H2O2 (100 μM) for 8 h. BV-2 cells were treated with FGF10 (100 ng/ml) and LPS (0.5 μg/ml) for 24 h. All experiments were performed at least three times. Locomotion recovery assessment. To assess the locomotion recovery in rats after SCI, the Basso, Beattie and Bresnahan (BBB) scores and the inclined plane test were used as mentioned previously.61 In short, the BBB scores range from 0 point (complete paralysis) to 21 points (normal locomotion) according to the muscle strength and joint movement of rats. Concurrently, rats were evaluated in two positions (right side or left side up) on a testing device. For each position, a rat could keep its position for 5 s without falling was recorded. In this study, BBB scores and the inclined plane test were performed by two blinded independent researchers at several time points after surgery. In conclusion, we first demonstrated that spinal cord- derived FGF10 significantly increased in neurons and micro- glia/macrophages after acute SCI. Exogenous FGF10 treat- ment facilitates better functional recovery through the FGFR2/ PI3K/Akt signaling pathway, Inhibiting the PI3K/Akt signaling pathway and FGFR2 knockdown abolished these therapeutic effects. Moreover, FGF10 treatment inhibited microglia/ macrophages activation and proliferation via regulation of the TLR4/NF-κB pathway, and attenuated the inflammatory response in animals with SCI (Figure 7). As endogenous FGF10 exerts neuroprotective effects following CNS injury, our results suggest that it may in turn be a potentially useful treatment for CNS injury. Hematoxylin–eosin (HE) and nissl staining. Discussion Moreover, activating PI3K/Akt signal- ing has been demonstrated to involve in NGF-induced neurite outgrowth in PC12 cells52 and suppressing the MEK/ERK/Akt pathway inhibits neurite outgrowth in N2a cells.53 FGF10 has been reported as a morphogen that is a critical for hypothalamic axon growth into the forming median eminence Moreover, neurons are cells with high energy requirements, and are sensitive to ROS stimulation, especially in their axons and dendrites.46 Previous studies including our work noted morphological alterations of neurites, described as bead formation, and reduced number and density of dendrites in ROS-treated granule cells.47 One mechanism of ROS- induced neurite degeneration occurs by disruption of Cell Death and Disease A novel therapeutic intervention for SCI J Chen et al 0 suppressed microglia/macrophages TLR4 expression and downstream NF-κB signaling in rats after SCI. (a and b) Western blots and qua at 1 day after SCI. (c) Double immunofluorescence staining for Iba-1 positive microglia/macrophages (green) and TLR4 (red) of sections g) Representative western blots and quantification data of p-IκBα, IκBα and p65 of each group at 1 day after SCI. Data represent the mean n the treatment and SCI groups are indicated as Po0.05, *Po0.01, n = 5 J Chen et al 9 Figure 6 FGF10 suppressed microglia/macrophages TLR4 expression and downstream NF-κB signaling in rats after SCI. (a and b) Western blots and quantification data of TLR4 in each group at 1 day after SCI. (c) Double immunofluorescence staining for Iba-1 positive microglia/macrophages (green) and TLR4 (red) of sections from the tissue at 1 day after SCI. (d-g) Representative western blots and quantification data of p-IκBα, IκBα and p65 of each group at 1 day after SCI. Data represent the mean ± S.D. Significant differences between the treatment and SCI groups are indicated as Po0.05, *Po0.01, n = 5 Figure 6 FGF10 suppressed microglia/macrophages TLR4 expression and downstream NF-κB signaling in rats after SCI. (a and b) Western blots and quantification data of TLR4 in each group at 1 day after SCI. (c) Double immunofluorescence staining for Iba-1 positive microglia/macrophages (green) and TLR4 (red) of sections from the tissue at 1 day after SCI. (d-g) Representative western blots and quantification data of p-IκBα, IκBα and p65 of each group at 1 day after SCI. Data represent the mean ± S.D. Discussion Significant differences between the treatment and SCI groups are indicated as Po0.05, **Po0.01, n = 5 Cell Death and Disease A novel therapeutic intervention for SCI J Chen et al 10 Figure 7 A schematic diagram depicting the potential molecular mechanisms underlying FGF10 protection via neurite repair, reducing apoptosis and decreasing inflammatory cytokines after acute SCI and neurohypophysis.29 FGF10 regulates neurogenesis and preserves neurogenic potential through its specific expression pattern in the adult mammalian brain.54 In this study, we found that FGF10 activated FGFR2/PI3K/Akt signaling pathway was critical for stabilizing microtubule structure and repairing neurites by regulation the expression of microtubule proteins and outgrowth of AcTub labeled neurites. and neurohypophysis.29 FGF10 regulates neurogenesis and preserves neurogenic potential through its specific expression pattern in the adult mammalian brain.54 In this study, we found that FGF10 activated FGFR2/PI3K/Akt signaling pathway was critical for stabilizing microtubule structure and repairing neurites by regulation the expression of microtubule proteins and outgrowth of AcTub labeled neurites. g We also observed increased expression of FGF10 in microglia/macrophages after SCI, which led us to investigate if FGF10 also is important for microglia/macrophages’ func- tion. After SCI, the prolonged inflammatory response enhances resident microglia/macrophages’ activation and proliferation, which subsequently promotes production of pro-inflammatory factors, such as TNF-α and IL-6, creating an inhibitory environment for neurite regeneration.9 Activated microglia/macrophages produce a variety of pro-inflammatory mediators, as well as other toxic mediators, which trigger signaling cascades and neurotoxic responses in the second- ary phase of SCI. These events significantly contribute to both neuronal death and neurite injury.55,56 Many studies have been reported about the critical role of TLR4 in ischemic CNS. Activation of TLR4 signaling contributes to astrocyte-mediated inflammation, and may control pro-inflammatory astroglial conversion to the neurodegenerative phenotype.57 It have been reported to protects blood–brain barrier by inhibiting TLR4-mediated inflammatory pathway in ischemic brain.58,59 Activating microglia/macrophages TLR4 signaling by exogen- ous or endogenous ligands, such as LPS, heme and fibrinogen, induces nuclear translation of NF-κB, which increases release of pro-inflammatory cytokines and leads to neuronal death.17,60 Strikingly, FGF10 treatment signifi- cantly decreased microglia/macrophages activation prolifera- tion, and production of pro-inflammatory cytokines in vivo. Using LPS, a TLR4 ligand, to activate downstream signaling, we found that FGF10 treatment was able to decrease TLR4 expression, leading to reduced p-IκB-α and IκB-α degradation and nuclear translocation of NF-κB transcription factors. A novel therapeutic intervention for SCI J Chen et al Of microtubules and memory: implications for microtubule dynamics in dendrites and spines. Mol Biol Cell 2017; 28: 1–8. 7. Hellal F, Hurtado A, Ruschel J, Flynn KC, Laskowski CJ, Umlauf M et al. Microtubule stabilization reduces scarring and causes axon regeneration after spinal cord injury. Science 2011; 331: 928–931. 8. Wu QF, Yang L, Li S, Wang Q, Yuan XB, Gao X et al. Fibroblast growth factor 13 is a microtubule-stabilizing protein regulating neuronal polarization and migration. Cell 2012; 149: 1549–1564. 9. Schwartz M, Cohen I, Lazarov-Spiegler O, Moalem G, Yoles E. The remedy may lie in ourselves: prospects for immune cell therapy in central nervous system protection and repair. J Mol Med 1999; 77: 713–717. 10. Witcher KG, Eiferman DS, Godbout JP. Priming the inflammatory pump of the CNS after traumatic brain injury. Trends Neurosci 2015; 38: 609–620. Immunohistochemical staining. Transverse and longitudinal sections (5-μm thick) were deparaffinized, rehydrated and then blocked by addition of 3% (v/v) H2O2 for 10 min followed by incubation in 5% BSA for 30 min. After incubation with primary antibodies (anti-FGF1, 2, 7, anti-Iba-1), the samples were incubated with the respective second antibodies and counterstained with hematoxylin. Images were obtained using a light microscope. 11. Tremblay ME, Stevens B, Sierra A, Wake H, Bessis A, Nimmerjahn A. The role of microglia in the healthy brain. J Neurosci 2011; 31: 16064–16069. 12. Schweitzer PJ, Fallon BA, Mann JJ, Kumar JS. PET tracers for the peripheral benzodiazepine receptor and uses thereof. Drug Discov Today 2010; 15: 933–942. 13. Popovich PG, Hickey WF. Bone marrow chimeric rats reveal the unique distribution of resident and recruited macrophages in the contused rat spinal cord. J Neuropathol Exp Neurol 2001; 60: 676–685. The TUNEL method. To test apoptotic DNA fragmentation, transverse sections were removed at 4–5 mm rostral and caudal of the lesion, and TUNEL staining was performed 7 days after SCI. The tissues (5 μm thick) were deparaffinized, and rehydrated. Cells were incubated with 4% PFA for 1 h. Then, tissues and cells were incubated with 0.1 % Triton X-100 for 30 min. Apoptotic cells of spinal cord tissue were stained with In Situ Cell Death Detection Kit (Roche Molecular Biochemicals, Basel, Switzerland) according to the manufacturer’s instructions, and DAPI. All apoptotic changes were tested under a Nikon ECLIPSE Ti microscope (Nikon). 14. Popovich PG, Guan Z, Wei P, Huitinga I, van Rooijen N, Stokes BT. 1. Ray SK, Samantaray S, Smith JA, Matzelle DD, Das A, Banik NL. Inhibition of cysteine proteases in acute and chronic spinal cord injury. Neurotherapeutics 2011; 8: 180–186. A novel therapeutic intervention for SCI J Chen et al 11 USA). Following blocking with 5% nonfat milk, the primary antibodies were incubated: anti-MAP2 (1:500), anti-acetyl-α-tubulin (1:1000), anti-FGFR2 (1:200), anti-cleaved-caspase 3 (1:500), anti-NeuN (1:1000), anti-GFAP (1:1000), anti-Iba-1 (1:500), anti-TLR4 (1:500), anti-GAPDH (1:1000), anti-NF-κB (1:400), anti-IκB (1:400) and anti-p-IκB (1:400), followed by their respective secondary antibodies. The bands were detected by the ChemiDicTM XRS + Imaging System (Bio-Rad), and the intensity of these bands were analyzed using Image Lab 3.0 software (Bio- Rad). Experiments were performed at least three times. USA). Following blocking with 5% nonfat milk, the primary antibodies were incubated: anti-MAP2 (1:500), anti-acetyl-α-tubulin (1:1000), anti-FGFR2 (1:200), anti-cleaved-caspase 3 (1:500), anti-NeuN (1:1000), anti-GFAP (1:1000), anti-Iba-1 (1:500), anti-TLR4 (1:500), anti-GAPDH (1:1000), anti-NF-κB (1:400), anti-IκB (1:400) and anti-p-IκB (1:400), followed by their respective secondary antibodies. The bands were detected by the ChemiDicTM XRS + Imaging System (Bio-Rad), and the intensity of these bands were analyzed using Image Lab 3.0 software (Bio- Rad). Experiments were performed at least three times. 2. Wang L, Yao Y, He R, Meng Y, Li N, Zhang D et al. Methane ameliorates spinal cord ischemia-reperfusion injury in rats: antioxidant, anti-inflammatory and anti-apoptotic activity mediated by Nrf2 activation. Free Radic Biol Med 2017; 103: 69–86. 3. Amar AP, Levy ML. Pathogenesis and pharmacological strategies for mitigating secondary damage in acute spinal cord injury. Neurosurgery 1999; 44: 1027–1039, discussion 1039-1040. 4. Slaets H, Nelissen S, Janssens K, Vidal PM, Lemmens E, Stinissen P et al. Oncostatin M reduces lesion size and promotes functional recovery and neurite outgrowth after spinal cord injury. Mol Neurobiol 2014; 50: 1142–1151. j y 5. Hur EM, Saijilafu, Zhou FQ. Growing the growth cone: remodeling the cytoskeleton to promote axon regeneration. Trends Neurosci 2012; 35: 164–174. Immunofluorescence staining. Transverse and longitudinal sections (5-μm thick) were deparaffinized and rehydrated. PC12 cells were fixed with 4% PFA for 1 h. And tissue and cells slices were blocked by 5% bovine serum albumin (BSA) for 30 min. Then, they were incubated with the following primary antibodies overnight: anti-acetyl-α-tubulin (1:1000), anti-FGFR2 (1:200), anti-cleaved-caspase 3 (1:500), anti-NeuN (1:1000), anti-GFAP (1:1000), anti-Iba-1 (1:500), anti-TLR4 (1:400) and anti-NF-κB (1:400). The next day, the following secondary antibodies were incubated for 1 h: Alexa-Fluor 488 donkey anti-mouse/rabbit, Alexa-Fluor 647 donkey anti-mouse/rabbit. Then, the slices labeled with DAPI for 7 min. All images were observed using a Nikon ECLIPSE Ti microscope (Nikon, Tokyo, Japan). 6. Dent EW. Conflict of Interest The authors declare no conflict of interest. nd activating PI3K/Akt survival signaling pathway in mice. Sci Rep 2016; 6: 198 24. Koshinaga M, Sanon HR, Whittemore SR. Altered acidic and basic fibroblast growth factor expression following spinal cord injury. Exp Neurol 1993; 120: 32–48. Acknowledgements. This study was partially supported by a research grant from the National Natural Science Funding of China (81722028, 81371988 and 81572237, 81772450), Zhejiang Provincial Program for the Cultivation of High-level Innovative Health Talents (to JX). Zhejiang Provincial Natural Science Foundation (R18H50001, LY17H090017). 25. Utley S, James D, Mavila N, Nguyen MV, Vendryes C, Salisbury SM et al. Fibroblast growth factor signaling regulates the expansion of A6-expressing hepatocytes in association with AKT-dependent beta-catenin activation. J Hepatol 2014; 60: 1002–1009. 26. Memarzadeh S, Xin L, Mulholland DJ, Mansukhani A, Wu H, Teitell MA et al. Enhanced paracrine FGF10 expression promotes formation of multifocal prostate adenocarcinoma and an increase in epithelial androgen receptor. Cancer Cell 2007; 12: 572–585. 27. Godena VK, Brookes-Hocking N, Moller A, Shaw G, Oswald M, Sancho RM et al. Increasing microtubule acetylation rescues axonal transport and locomotor deficits caused by LRRK2 Roc-COR domain mutations. Nat Commun 2014; 5: 5245. A novel therapeutic intervention for SCI J Chen et al Depletion of hematogenous macrophages promotes partial hindlimb recovery and neuroanatomical repair after experimental spinal cord injury. Exp Neurol 1999; 158: 351–365. 15. Gris D, Marsh DR, Oatway MA, Chen Y, Hamilton EF, Dekaban GA et al. Transient blockade of the CD11d/CD18 integrin reduces secondary damage after spinal cord injury, improving sensory, autonomic, and motor function. J Neurosci 2004; 24: 4043–4051. 16. Carpentier PA, Duncan DS, Miller SD. Glial Toll-like receptor signaling in central nervous system infection and autoimmunity. Brain Behav Immun 2008; 22: 140–147. system infection and autoimmunity. Brain Behav Immun 2008; 22: 140–147. 17. Li XQ, Wang J, Fang B, Tan WF, Ma H. Intrathecal antagonism of microglial TLR4 reduces inflammatory damage to blood-spinal cord barrier following ischemia/reperfusion injury in rats. Mol Brain 2014; 7: 28. 18. Caso JR, Pradillo JM, Hurtado O, Lorenzo P, Moro MA, Lizasoain I. Toll-like receptor 4 is involved in brain damage and inflammation after experimental stroke. Circulation 2007; 115: 1599–1608. Small interfering RNA transfection. PC12 cells were treated with FGFR2 siRNA (100 pmol, GeneChem, Shanghai, China) in serum-free medium containing Lipofectamine 2000 (Life Technologies, Carlsbad, CA, USA). After transfection for 6 h, medium was switched to medium containing 5% FBS for 24 h. After treatments, cells were harvested for further experiments. 19. Reibel S, Larmet Y, Carnahan J, Marescaux C, Depaulis A. Endogenous control of hippocampal epileptogenesis: a molecular cascade involving brain-derived neurotrophic factor and neuropeptide Y. Epilepsia 2000; 41(Suppl 6): S127–S133. 20. Yamasaki M, Miyake A, Tagashira S, Itoh N. Structure and expression of the rat mRNA encoding a novel member of the fibroblast growth factor family. J Biol Chem 1996; 271: 15918–15921. Statistical analysis. The results are expressed as the mean ± S.D. from at least three independent experiments. And statistical significance was analyzed using Graphpad Prism (La Jolla, CA, USA) (one-way analysis of variance (ANOVA) and Tukey’s test). Po0.05 was considered statistically significant. 21. Li X, Wang C, Xiao J, McKeehan WL, Wang F. Fibroblast growth factors, old kids on the new block. Semin Cell Dev Biol 2016; 53: 155–167. 22. Beenken A, Mohammadi M. The FGF family: biology, pathophysiology and therapy. Nat Rev Drug Discov 2009; 8: 235–253. 23. Li YH, Fu HL, Tian ML, Wang YQ, Chen W, Cai LL et al. Neuron-derived FGF10 ameliorates cerebral ischemia injury via inhibiting NF-kappaB-dependent neuroinflammation and activating PI3K/Akt survival signaling pathway in mice. Sci Rep 2016; 6: 19869. Discussion To measure the cavity area of spinal cord tissue in each group after surgery, rats in all groups were killed at 28 days and the spinal cord tissues were embedded in paraffin. Longitudinal sections (5 mm thick) were cut into 5-μm thickness for HE staining. Transverse sections were incubated in 1% cresyl violet acetate for Nissl staining to measure the surviving neurons. Western blot assay. Spinal cord tissue from T7 to T10 was collected at 1 day and 3 days after surgery. Briefly, spinal cord tissue and cells were lysed using RIPA with phosphatase inhibitors and protease inhibitors cocktail and then protein concentration was measured by bicinchoninic acid reagents (Thermo, Rockford, IL, USA), equivalent amounts of protein was separated with 8–12% SDS–PAGE gels, and transferred to polyvinylidene fluoride membranes (Bio-Rad, Hercules, CA, Materials and Methods Reagents. FGF10 was obtained from Grost (Grost Biotechnology, Zhejiang, China). Antibodies against FGFR2, Bax, Bcl-2, IL-6 and TNF-α were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). The MAP2, acetyl-α-tubulin (AcTub), TLR4, NF-κB, IκB, p-IκB antibodies and PI3K/Akt inhibitor, LY294002, were obtained from Cell Signaling Technologies (Danvers, MA, USA). The Akt, Cell Death and Disease A novel therapeutic intervention for SCI J Chen et al A novel therapeutic intervention for SCI J Chen et al A novel therapeutic intervention for SCI J Chen et al 12 30. Chen Z, Jalabi W, Shpargel KB, Farabaugh KT, Dutta R, Yin X et al. Lipopolysaccharide- induced microglial activation and neuroprotection against experimental brain injury is independent of hematogenous TLR4. J Neurosci 2012; 32: 11706–11715. 50. Harada A, Teng J, Takei Y, Oguchi K, Hirokawa N. MAP2 is required for dendrite elongation, PKA anchoring in dendrites, and proper PKA signal transduction. Mol Biol Cell 2002; 13: 325a–325a. p g 31. Miyake S, Muramatsu R, Hamaguchi M, Yamashita T. Prolyl hydroxylase regulates axonal rewiring and motor recovery after traumatic brain injury. Cell Death Dis 2015; 6: e1638. 51. Lim RW, Halpain S. Regulated association of microtubule-associated protein 2 (MAP2) with Src and Grb2: evidence for MAP2 as a scaffolding protein. J Biol Chem 2000; 275: 20578–20587. 32. Deng L, Ruan Y, Chen C, Frye CC, Xiong W, Jin X et al. Characterization of dendritic morphology and neurotransmitter phenotype of thoracic descending propriospinal neurons after complete spinal cord transection and GDNF treatment. Exp Neurol 2016; 277: 103–114. 52. Shibata A, Laurent CE, Smithgall TE. The c-Fes protein-tyrosine kinase accelerates NGF- induced differentiation of PC12 cells through a PI3K-dependent mechanism. Cell Signal 2003; 15: 279–288. 53. Wang ZY, Wang J, Li JJ, Wang XH, Yao YZ, Zhang XJ et al. MEK/ERKs signaling is essential for lithium-induced neurite outgrowth in N2a cells. Int J Dev Neurosci 2011; 29: 415–422. 33. Sarma T, Koutsouris A, Yu JZ, Krbanjevic A, Hope TJ, Rasenick MM. Activation of microtubule dynamics increases neuronal growth via the nerve growth factor (NGF)- and Galphas-mediated signaling pathways. J Biol Chem 2015; 290: 10045–10056. 54. Hajihosseini MK, De Langhe S, Lana-Elola E, Morrison H, Sparshott N, Kelly R et al. Localization and fate of Fgf10-expressing cells in the adult mouse brain implicate Fgf10 in control of neurogenesis. Mol Cell Neurosci 2008; 37: 857–868. 34. Moon YJ, Lee JY, Oh MS, Pak YK, Park KS, Oh TH et al. Inhibition of inflammation and oxidative stress by Angelica dahuricae radix extract decreases apoptotic cell death and improves functional recovery after spinal cord injury. J Neurosci Res 2012; 90: 243–256. 55. Guadagno J, Xu X, Karajgikar M, Brown A, Cregan SP. Microglia-derived TNFalpha induces apoptosis in neural precursor cells via transcriptional activation of the Bcl-2 family member Puma. Cell Death Dis 2013; 4: e538. p y p j y 35. A novel therapeutic intervention for SCI J Chen et al Onyango IG, Bennett JP Jr., Tuttle JB. Endogenous oxidative stress in sporadic Alzheimer's disease neuronal cybrids reduces viability by increasing apoptosis through pro-death signaling pathways and is mimicked by oxidant exposure of control cybrids. Neurobiol Dis 2005; 19: 312–322. 56. Teeling JL, Perry VH. Systemic infection and inflammation in acute CNS injury and chronic neurodegeneration: underlying mechanisms. Neuroscience 2009; 158: 1062–1073. 36. Lee BI, Lee DJ, Cho KJ, Kim GW. Early nuclear translocation of endonuclease G and subsequent DNA fragmentation after transient focal cerebral ischemia in mice. Neurosci Lett 2005; 386: 23–27. 57. Rosciszewski G, Cadena V, Murta V, Lukin J, Villarreal A, Roger T et al. Toll-like receptor 4 (TLR4) and triggering receptor expressed on myeloid cells-2 (TREM-2) activation balance astrocyte polarization into a proinflammatory phenotype. Mol Neurobiol 2017. (doi:10.1007/ s12035-017-0618-z). 37. Li P, Nijhawan D, Budihardjo I, Srinivasula SM, Ahmad M, Alnemri ES et al. Cytochrome c and dATP-dependent formation of Apaf-1/caspase-9 complex initiates an apoptotic protease cascade. Cell 1997; 91: 479–489. 58. Zhang T, Su J, Guo B, Wang K, Li X, Liang G. Apigenin protects blood-brain barrier and ameliorates early brain injury by inhibiting TLR4-mediated inflammatory pathway in subarachnoid hemorrhage rats. Int Immunopharmacol 2015; 28: 79–87. 38. Noshita N, Sugawara T, Lewen A, Hayashi T, Chan PH. Copper-zinc superoxide dismutase affects Akt activation after transient focal cerebral ischemia in mice. Stroke 2003; 34: 1513–1518. 59. Wang P, You SW, Yang YJ, Wei XY, Wang YZ, Wang X et al. Systemic injection of low-dose lipopolysaccharide fails to break down the blood-brain barrier or activate the TLR4-MyD88 pathway in neonatal rat brain. Int J Mol Sci 2014; 15: 10101–10115. 39. Endo H, Kamada H, Nito C, Nishi T, Chan PH. Mitochondrial translocation of p53 mediates release of cytochrome c and hippocampal CA1 neuronal death after transient global cerebral ischemia in rats. J Neurosci 2006; 26: 7974–7983. 60. Okada T, Suzuki H. Toll-like receptor 4 as a possible therapeutic target for delayed brain injuries after aneurysmal subarachnoid hemorrhage. Neural Regen Res 2017; 12: 193–196. 40. Fu H, Xu H, Chen H, Li Y, Li W, Zhu Q et al. Inhibition of glycogen synthase kinase 3 ameliorates liver ischemia/reperfusion injury via an energy-dependent mitochondrial mechanism. J Hepatol 2014; 61: 816–824. 61. Zhang HY, Wang ZG, Wu FZ, Kong XX, Yang J, Lin BB et al. A novel therapeutic intervention for SCI J Chen et al Regulation of autophagy and ubiquitinated protein accumulation by bFGF promotes functional recovery and neural protection in a rat model of spinal cord injury. Mol Neurobiol 2013; 48: 452–464. p 41. Kamada H, Nito C, Endo H, Chan PH. Bad as a converging signaling molecule between survival PI3-K/Akt and death JNK in neurons after transient focal cerebral ischemia in rats. J Cerebr Blood F Met 2007; 27: 521–533. 62. Han DS, Li F, Holt L, Connolly K, Hubert M, Miceli R et al. Keratinocyte growth factor-2 (FGF-10) promotes healing of experimental small intestinal ulceration in rats. Am J Physiol Gastrointest Liver Physiol 2000; 279: G1011–G1022. 42. Cardone MH, Roy N, Stennicke HR, Salvesen GS, Franke TF, Stanbridge E et al. Regulation of cell death protease caspase-9 by phosphorylation. Science 1998; 282: 1318–1321. 43. Vega-Hernandez M, Kovacs A, De Langhe S, Ornitz DM. FGF10/FGFR2b signaling is essential for cardiac fibroblast development and growth of the myocardium. Development 2011; 138: 3331–3340. Cell Death and Disease is an open-access journal published by Nature Publishing Group. This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ 44. Kunstlinger H, Fassunke J, Schildhaus HU, Brors B, Heydt C, Ihle MA et al. FGFR2 is overexpressed in myxoid liposarcoma and inhibition of FGFR signaling impairs tumor growth in vitro. Oncotarget 2015; 6: 20215–20230. 45. Adams JM, Cory S. The Bcl-2 protein family: arbiters of cell survival. Science 1998; 281: 1322–1326. 46. Fukui K, Ushiki K, Takatsu H, Koike T, Urano S. Tocotrienols prevent hydrogen peroxide- induced axon and dendrite degeneration in cerebellar granule cells. Free Radic Res 2012; 46: 184–193. 47. Fukui K, Takatsu H, Koike T, Urano S. Hydrogen peroxide induces neurite degeneration: prevention by tocotrienols. Free Radic Res 2011; 45: 681–691. 48. Miller RH, Lasek RJ. Cross-bridges mediate anterograde and retrograde vesicle transport along microtubules in squid axoplasm. J Cell Biol 1985; 101: 2181–2193. 48. Miller RH, Lasek RJ. Cross-bridges mediate anterograde and retrograde along microtubules in squid axoplasm. J Cell Biol 1985; 101: 2181–2193. 49. Cho Y, Cavalli V. Publisher’s Note 28. Soltani MH, Pichardo R, Song Z, Sangha N, Camacho F, Satyamoorthy K et al. Microtubule- associated protein 2, a marker of neuronal differentiation, induces mitotic defects, inhibits growth of melanoma cells, and predicts metastatic potential of cutaneous melanoma. Am J Pathol 2005; 166: 1841–1850. Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 29. Liu F, Pogoda HM, Pearson CA, Ohyama K, Lohr H, Hammerschmidt M et al. Direct and indirect roles of Fgf3 and Fgf10 in innervation and vascularisation of the vertebrate hypothalamic neurohypophysis. Development 2013; 140: 1111–1122. Cell Death and Disease A novel therapeutic intervention for SCI J Chen et al HDAC5 is a novel injury-regulated tubulin deacetylase controlling axon regeneration. EMBO J 2012; 31: 3063–3078. r The Author(s) 2017 pplementary Information accompanies this paper on Cell Death and Disease website (http://www.natu ation accompanies this paper on Cell Death and Disease website (http://www.nature.com/cddis) Cell Death and Disease Cell Death and Disease
https://openalex.org/W2072608896	https://pirineos.revistas.csic.es/index.php/pirineos/article/download/230/229/228	Spanish; Castilian	null	Impactos recientes de los cambios ambientales en los recursos hídricos superficiales de la cuenca del Duero	Pirineos	2,012	cc-by	15,646	Pirineos.Revista de Ecología de Montaña Vol. 167, 107-142 Jaca, Enero-Diciembre, 2012 ISSN: 0373-2568 eISSN: 1988-4281 doi: 10.3989/Pirineos.2012.167006 Pirineos.Revista de Ecología de Montaña Vol. 167, 107-142 Jaca, Enero-Diciembre, 2012 ISSN: 0373-2568 eISSN: 1988-4281 doi: 10.3989/Pirineos.2012.167006 Pirineos.Revista de Ecología de Montaña Vol. 167, 107-142 Jaca, Enero-Diciembre, 2012 ISSN: 0373-2568 eISSN: 1988-4281 doi: 10.3989/Pirineos.2012.167006 IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS SUPERFICIALES DE LA CUENCA DEL DUERO Impact of environmental change un surface water resources in the Douro basin ENRIQUE MORÁN TEJEDA Instituto Pirenaicos de Ecología, CSIC (Spanish Research Council), Campus de Aula Dei, P.O. Box 202, Zaragoza 50080, Spain. enrique.moran@unige.ch ABSTRACT.– In recent decades, the environmental changes due to human development have put pressure on water resources in the Mediterranean basin, a region where availability of water has been historically limited. In this work we ana- lyze the evolution and variability (1961-2005) of streamflows in one of the largest rivers basin of the Iberian Peninsula, the Duero River basin. Moreover, the factors responsible for such evolution are assessed. Results show a significant and general- ized decrease of water resources in the basin, together with changes in the timing of monthly distribution. Climate itself, with quasi-stationary precipitation and increasing temperatures in the long-term, is not enough to explain the decreasing streamflows. Thus, observed land-cover expansion in the headwaters is thought to be increasingly contributing to the hydrological depletion. On the other hand, impoundment of water through dams is increasing in the basin and consequently contributing the hydrological change. Results offer the basis for future projections of water availability in scenarios of water scarcity due to forthcoming climate change ENRIQUE MORÁN TEJEDA cos superficiales –el caudal en los ríos– en una de las cuencas hidrográficas de mayor entidad de la Península Ibérica, y los factores ambientales responsables de su evolu- ción. Los resultados del trabajo muestran un descenso notable y generalizado en los caudales en la región, acompañado de un cambio en los regímenes fluviales. La evo- lución del clima, con unas precipitaciones muy variables pero sin tendencias nota- bles a largo plazo, y unas temperaturas en aumento, explica en parte, pero no en su totalidad, el descenso hidrológico. En las cabeceras fluviales se ha detectado un incre- mento significativo de la cubierta vegetal durante el periodo de estudio, el cual pare- ce estar participando en gran medida en el descenso de caudales. Por otro lado, la regulación por medio de embalses está incrementando en la cuenca y con ello contri- buyendo al cambio hidrológico en la región. Los resultados obtenidos ofrecen la base conceptual para proyectar la disponibilidad futura de los recursos hídricos en los escenarios de mayor escasez como consecuencia del cambio climático venidero. Palabras clave: Cuenca del Duero, recursos hídricos, régimen fluvial, varia- bilidad climática, cubierta vegetal, gestión hidrológica. Keywords: Duero basin, water resources, fluvial regime, climate variability, land-cover, hydrological management. Keywords: Duero basin, water resources, fluvial regime, climate variability, land-cover, hydrological management. RESUMEN.– La disponibilidad de recursos hídricos ha sido históricamente un factor limitante de desarrollo en los países de la cuenca mediterránea. Durante las últimas décadas la presión sobre la disponibilidad se ha acentuado debido a los cam- bios ambientales observados como resultado del desarrollo industrial. En este traba- jo se analizan la evolución y variabilidad recientes (1961-2005) de los recursos hídri- Recibido: 14-10-2011. Aceptado: 17-12-2011. 107 108 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... Durante las dos últimas décadas han proliferado en la literatura científica los estudios enfocados a evaluar la evolución reciente de los recursos hídricos y los procesos relacionados con la misma (p.ej. Lettenmaier et al., 1994, Kahya & Kalayci, 2004, Birsan et al., 2005, García-Ruiz et al., 2011). El motivo princi- pal es la preocupación existente sobre la disponibilidad de agua en las déca- das futuras, en las que se pronostican cambios graduales en los valores de las principales variables climáticas (temperaturas y precipitaciones) en el contex- to del denominado “cambio climático”. De forma muy general un calenta- miento de la atmósfera a nivel planetario, derivado del aumento en la con- centración de gases invernadero, incrementará los ratios de evapotranspira- ción provocando a su vez un incremento en la variabilidad espacial y tempo- ral de las precipitaciones y causando notables cambios en el ciclo hidrológico (Milly et al., 2005, IPCC, 2007). y No obstante, en los últimos años también se ha puesto el énfasis en el papel hidrológico de un proceso que se observa de forma generalizada en las zonas de montaña de los países desarrollados: el incremento de la cubierta vegetal, como consecuencia del abandono de las actividades tradicionales de pastoreo y cultivo en las laderas de montaña. Los trabajos realizados en cuen- cas experimentales han demostrado que la cubierta vegetal y los bosques afectan al ciclo hidrológico y el balance de agua en la medida en que contro- lan los procesos de partición de la precipitación (Crockford & Richardson, 2000, Llorens & Domingo, 2007), antes de formar parte de la escorrentía superficial: la cubierta vegetal, o el dosel arbóreo retiene parte de las gotas de lluvia en el proceso denominado como“interceptación”, y consume agua para sus necesidades vitales, potenciando ambos procesos la evapotranspiración del agua a la atmósfera; asimismo favorece la infiltración de agua en el suelo a través de sus sistemas radiculares(Zhang et al., 2001, Cosandey et al., 2005, Morán, 2007).Por ello cualquier cambio de uso que suponga un incremento en densidad o superficie de la misma repercute directamente en un descenso en la escorrentía, mientras que una retirada de la cubierta vegetal suele aumentar los caudales de los ríos (p. ej., Bosch & Hewlett, 1982, Bent, 2001, Gallart & Llorens, 2003). Desde mediados del siglo XX se viene advirtiendo en la montaña españo- la un incremento paulatino de la cubierta vegetal. 1. Introducción La cuenca del río Duero es por su extensión y por el volumen de agua dre- nada, uno de los sistemas hidrográficos de mayor entidad de la Península Ibérica. Si bien su régimen hidrológico no tiene el carácter deficitario de las cuencas de la vertiente mediterránea, la presión a la que están sometidos los recursos hídricos, tanto por factores naturales como antrópicos, hace de su evaluación una tarea fundamental si se quiere pronosticar la disponibilidad futura los mismos. La localización de la cuenca en el noroeste peninsular y su disposición topográfica a modo de gran meseta rodeada de cadenas monta- ñosas le confieren unos rasgos climáticos diversos, destacando la mediterra- neidad y la continentalidad del interior y los ambientes húmedos y fríos del reborde montañoso y el sector noroccidental más cercano al océano Atlántico. Las montañas que bordean la cuenca son, gracias a su capacidad de retención y almacenamiento de las precipitaciones, la fuente principal de los recursos hídricos de la región, siendo ésta una característica común de las montañas en los ambientes mediterráneos (Viviroli & Weingartner 2004, García-Ruiz et al., 2011). Los recursos hídricos de la cuenca, su cantidad y variabilidad, depen- derán por lo tanto en gran medida de los procesos que tengan lugar en las zonas de montaña. Los gradientes altitudinales y energéticos confieren ade- más a las montañas una especial sensibilidad frente a los cambios ambienta- les (Beniston, 2005), por lo que el estudio de los mismos es fundamental para comprender la disponibilidad de los recursos hídricos aguas abajo, donde se produce mayoritariamente su consumo. 108 IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... Los dos factores que más peso han tenido en esta recuperación han sido el abandono de las actividades agrarias en las zonas de montaña, y las repoblaciones forestales. El abandono de la agricultura o la disminución de la presión del pastoreo sobre el suelo montano provoca una regeneración natural de los ecosistemas de montaña mediante la recolonización vegetal, que puede evolucionar hasta etapas más avanzadas, como el establecimiento de una cubierta forestal (Vicente-Serrano et al,. 2000). Este proceso ha sido observado en diferentes sistemas montaño- 109 ENRIQUE MORÁN TEJEDA sos españoles como los Pirineos (Poyatos et al., 2003) o el Sistema Ibérico (Lasanta-Martínez et al., 2005). Junto a ello, la sustitución de cultivos cerealis- tas por pastizales y las reforestaciones llevadas a cabo por las administracio- nes han sido los procesos de cambio más significativos que ha experimenta- do el suelo de la montaña española en las últimas décadas (García-Ruiz et al., 1996). ) Así como la variabilidad climática es un condicionante natural de la dis- ponibilidad de los recursos hídricos, y el aumento de la cubierta vegetal res- ponde a procesos mixtos de intervención antrópica y evolución natural de los ecosistemas, los caudales de los ríos de la mayor parte de sistemas hidrográ- ficos mundiales están sujetos a fuertes regulaciones estacionales e hiperanua- les a través de las presas. La construcción de embalses constituye uno de los impactos más destacados del hombre sobre el medio natural (Petts, 1984). Los embalses y su gestión causan alteraciones en el régimen de los ríos, la dismi- nución de caudales aguas abajo como resultado de la infiltración y evapora- ción de la lámina de agua, alteran los ritmos y tasas de erosión/sedimenta- ción de los ríos al actuar como trampas de sedimentos, y modifican incluso las propiedades físico-químicas del agua, por lo que inducen a graves altera- ciones en los ecosistemas riparios (Cosandey& Robinson, 2000, Verstraeten & Poesen, 2000, Maingi & Marsh, 2002, Bonacci & Roje-Bonacci, 2003, Nilsson et al., 2005). Pero quizás el impacto más estremecedor lo produzcan sobre las poblaciones humanas, con la desaparición de poblaciones enteras, desplaza- mientos masivos de personas, pérdida de campos de cultivo en las superficies inundadas, incluso pérdidas de vidas humanas producidas por accidentes (Berkun, 2010). No obstante, y a pesar de sus impactos negativos, numerosos embalses han sido construidos y se siguen construyendo en todo el mundo para atender a diferentes demandas. 110 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... Entre sus funciones más importantes destacan el aprovisionamiento de agua para la irrigación de cultivos, la pro- ducción hidroeléctrica, el control y laminación de avenidas o el suministro urbano e industrial (López-Moreno et al., 2002). En la cuenca del Duero exis- ten a día de hoy más de 90 embalses, con una capacidad cercana a los 7.500 hm3, y una demanda total de 3.870 hm3. De dicho volumen más del 93%, (3.600 hm3) se destina al regadío, aproximadamente el 6% (225 hm3) se desti- na a abastecimientos urbanos y domésticos, y el resto, unos 45 hm3 a usos industriales y otros (www.chduero.es). Además, los embalses con una locali- zación topográfica propicia para los saltos de agua son utilizados también para la producción de energía hidroeléctrica. El presente trabajo comprende una valoración global de la evolución de los recursos hídricos superficiales en la cuenca del Duero, de su variabilidad espacial y temporal durante la segunda mitad del siglo pasado y primeros años del presente, teniendo en cuenta los tres factores mencionados líneas IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... arriba. Se analizan por lo tanto los patrones de cambio en las variables climá- ticas (temperaturas y precipitaciones), usos del suelo y gestión de embalses y en qué medida han afectado a la evolución de los caudales de los ríos de la cuenca durante el periodo 1961-2005. Los resultados obtenidos servirán como punto de partida para contextualizar trabajos enfocados a inferir la disponibi- lidad futura de los recursos hídricos a través de la modelización hidrológica. 1 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 2. Zona de estudio: la cuenca del Duero La totalidad de la cuenca del Duero cubre una extensión de 97.290 km2, si bien este trabajo se ha enfocado en el sector español de la cuenca, que ocupa un 81% de dicha superficie y drena a Portugal unos aportes anuales de 13.800 hm3 a través del su río principal, el Duero (Figura 1). La mayor parte de la cuenca del Duero se corresponde con una depresión colmatada con sedimen- tos terciarios y cuaternarios, conformando una meseta de una altitud media relativamente elevada (700 m), y en la que se ha asentado la red fluvial desde la era terciaria. Desde el punto de vista hidrológico las unidades más impor- tantes son los bordes montañosos ya que constituyen el área fuente de los recursos hídricos. Los bordes de la cuenca corresponden a diferentes cadenas montañosas que alcanzan altitudes superiores a los 2.500 m, y tienen un ori- gen geológico y composición litológica heterogénea: i) Los Montes de León o Montañas Galaico-Leonesas representan el límite noroccidental de la cuenca, están formadas por materiales paleozoicos y presentan altitudes superiores a los 2.000 m. ii) El límite norte lo forma la Cordillera Cantábrica, con cumbres próximas a los 2.500 m. y una estructura geológica compleja en cuanto a su origen, con materiales diversos del Paleozoico y Cenozoico. iii) El Sistema Ibérico delimita la parte oriental de la cuenca y está formado por materiales mesozoicos con cimas que superan los 2.000 m. iv) Finalmente, por el sur la cuenca está delimitada por el Sistema Central con materiales paleozoicos –plutónicos y metamórficos–, fracturados y deformados por la orogenia alpi- na, y cumbres que alcanzan los 2.500 m. y q El régimen térmico de la cuenca es el característico de un clima de interior, con una apreciable oscilación térmica anual, entre las temperaturas inferiores a los 5º C en invierno, y los 20º C de media en verano. Por su parte, el régi- men pluviométrico es la característica que define la mediterraneidad del clima, con un período húmedo entre el otoño y la primavera y un período con escasa pluviosidad en los meses estivales. Espacialmente las precipitaciones se distribuyen de forma desigual entre el interior (se sobrepasan los 400 mm anuales) y el borde montañoso, donde se registran con más de 1.000 mm anuales. 3. Datos y métodos En esta sección presentamos un resumen de los métodos y técnicas utili- zados para alcanzar los objetivos generales del trabajo. Se obvia la exposición de los procedimientos estadísticos habituales sin perjuicio de que aparezcan mencionados a lo largo del texto y, para más detalles, se remite al lector inte- resado a las referencias bibliográficas citadas. 2. Zona de estudio: la cuenca del Duero A pesar del predominio de un clima mediterráneo continentalizado 111 ENRIQUE MORÁN TEJEDA en la mayor parte de la cuenca, en los rebordes montañosos sería más apro- piado hablar de climas tanto sub-mediterráneos como sub-atlánticos. Por último, la dualidad existente en la topografía de la cuenca también está presente en la distribución de los usos del suelo y de las formaciones vegetales. Con la excepción de pequeñas manchas forestales o bosques de ribera, el interior de la cuenca está enteramente ocupado por terrenos de cul- tivo dedicados al cereal, el viñedo y el regadío. El dominio forestal se extien- de por el sector suroccidental con formaciones de encina (Quercus ilexsp.ballo- ta) y melojos (Q. pirenaica), y en las cadenas montañosas, con un marcado con- traste bioclimático que explica la presencia de abedulares (Betula pendula), hayedos (Fagus sylvatica) y robledales (Q. robur) en las montañas septentrio- nales, y de melojares y pinares (P. pinaster y P. sylvestris) en las montañas orientales y meridionales. Figura 1. La cuenca del Duero y la localización de las estaciones hidrológicas y climáticas estu- diadas. Figure 1. The Duero basin and the location of streamflows and weather stations used. Figura 1. La cuenca del Duero y la localización de las estaciones hidrológicas y climáticas estu- diadas. Fi 1 Th D b i d h l i f fl d h i d Figure 1. The Duero basin and the location of streamflows and weather stations used. 112 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 3.2 Análisis de tendencias Además de la aplicación de pruebas estadísticas estándares como los aná- lisis de correlación y regresión lineal, el análisis central del trabajo se basa en la detección de tendencias en las series de datos, cuya metodología detalla- mos continuación: Se entiende como tendencia al cambio gradual a largo plazo en la media de una variable. Un procedimiento común para ilustrar si una serie de datos presenta una tendencia es el ajuste de una recta (regresión lineal) a la serie temporal, lo que permite comprobar la dirección (positiva o negativa) y mag- nitud (grado de inclinación) de dicha tendencia. Sin embargo, ésta se trata de una prueba paramétrica, esto es, que se basa en los valores de las series, y es por lo tanto vulnerable a la existencia de datos extremos, o a la no-normali- dad en la distribución de los datos. Para solventar este problema, desde hace décadas se han desarrollado las denominadas pruebas “no-paramétricas”, que se basan en el rango de los registros dentro de la serie en vez de en sus valores. Los test no-paramétricos más utilizados en la detección de tendencias han sido, y son, el test de rangos de Spearman(Lehmann, 1975, Sneyers, 1990) y el test de Mann-Kendall (Mann, 1945, Kendall, 1975). La validez y robustez de ambos en la detección de tendencias en series climáticas e hidrológicas ha sido ampliamente demostrada, siendo muy similar los resultados que de ellos se obtienen (Hirsch et al., 1982, Berryman et al., 1988, Zhang et al., 2000, Yue et al., 2002). El test de Mann-Kendall ha sido utilizado para la detección de ten- dencias en las series hidrológicas y climáticas de este trabajo,después de com- probar la similitud de resultados con el test de Spearman. La hipótesis nula (N0) representa el caso en que no existe un cambio gradual en la media de la serie de datos a lo largo del tiempo; y la hipótesis alternativa (N1) correspon- dería con el caso en el cual la media está aumentando o disminuyendo a lo largo del tiempo (Kundzewicz & Robson 2004). El test se ha aplicado median- te el paquete estadístico SPSS ®, y se basa en un algoritmo de correlación entre la variable X (años) y la variable Y (precipitación, temperaturas, y apor- taciones hídricas mensuales de cada año). Las observaciones se ordenan, separadamente, de forma ascendente y son reemplazadas por sus rangos. 3.1. Datos climáticos, hidrológicos y cartográficos Una base de datos climáticos, hidrológicos y de usos del suelo ha sido ela- borada para todo el territorio de la cuenca. Los datos climáticos, precipitacio- nes (mm) y temperaturas medias mensuales (º C), para el periodo 1961-2005 fueron obtenidos de la Agencia Española de Meteorología (AEMET). Los datos hidrológicos, aportaciones fluviales (hm3) mensuales y caudales diarios (m3/s), de la Confederación Hidrográfica del Duero, y los usos del suelo de los mapas forestales de España de 1966 a escala 1:400.000 en papel, y del 2003 a escala 1:50.000 y en formato digital, para las 9 provincias castellano-leonesas. Tanto los datos climáticos como hidrológicos pasaron un control de cali- dad basado en la detección de inhomogeneidades, en el relleno de datos ausentes, y en la eliminación de datos erróneos u outliers (ver detalles de dichos procedimientos en: Alexandersson, 1986, Lanzante, 1996, Peterson et al., 1998, García-Ruiz et al., 2001, Vicente-Serrano et al., 2009). Tras el proceso, que resultó en el descarte de un número considerable de series al no cumplir los requisitos de calidad (periodo de registro 1961-2005 y porcentaje de datos ausentes < 15%), un total de 214 series de precipitación, 57 series de tempe- ratura y 56 series de aportaciones fluviales, con una distribución más o menos homogénea en la cuenca (Figura 1), fueron utilizadas para los análisis. g g p Para analizar las variaciones en los usos del suelo se cartografiaron los mismos sobre los mapas forestales originales citados líneas arriba. Estos mapas contienen la cartografía de tipos de cubierta vegetal pero también de usos del suelo, como por ejemplo las formaciones forestales (pinos, esclerófi- las, y decíduas), matorrales, pastizales, campos de cultivo y cuerpos de agua. Sin embargo, el número y tipología de clases difieren enormemente entre las dos fuentes (1966 y 2003). Por ello se realizó un reclasificación y homogenei- zación de leyendas mediante la fusión de las clases originales en función de una similar cubierta del suelo. El resultado fueron cinco simples clases de uso del suelo: agua, urbano-suelo desnudo, cultivo, pastizal-matorral y bosque. Esto permitió el cálculo de la variación de superficie ocupada por cada clase entre las dos fechas señaladas. 113 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 114 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 3.2 Análisis de tendencias El test de Mann-Kendall se basa en el estadístico S, que se define como: p y p p g El test de Mann-Kendall se basa en el estadístico S, que se define como: 114 IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... dondeson los valores de la secuencia de datos, es la longitud de la serie de datos y dondeson los valores de la secuencia de datos, es la longitud de la serie de datos y El test de Mann-Kendall nos devuelve dos parámetros que nos permiten identificar la existencia de tendencias en las series así como su magnitud. El primero de ellos es el estadístico tau τ, y nos indica el signo (positivo o nega- tivo) de la tendencia, y su intensidad. Un tau = 0 indica la inexistencia de ten- dencia. Cuando los valores se alejan de cero, estaríamos ante la existencia de una tendencia, positiva si los valores son positivos y negativa si los valores son negativos. El segundo parámetro, y quizás el más importante, es la signi- ficación estadística (α), que mide, en términos de probabilidad, si el valor de la pendiente obtenida es diferente del rango de valores que tendría bajo la hipótesis nula. Ello nos indica, por lo tanto, la probabilidad de que podamos rechazar o no, con certeza, la hipótesis nula. 1 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 3.3. Análisis de componentes principales En un estudio de carácter regional como el presente, uno de los asuntos más interesantes es la identificación de patrones o modelos de comporta- miento común en distintos parámetros de las variables estudiadas, y com- probar si éstos presentan una distribución espacial determinada. Gracias al avance de las ciencias de la información y a la existencia de bases de datos cada vez más extensas, se ha hecho cada vez más frecuente la búsqueda de técnicas que permiten identificar patrones, comúnmente denominadas “herramientas de reducción de datos”. En el presente trabajo se ha utilizado en distintas ocasiones el conocido como “Análisis de Componentes Principales”, cuyo potencial en la detección de patrones climáticos e hidrológicos es ampliamente reconocido (p.ej. Widmann & Schär, 1997, Rodríguez-Puebla et al., 1998, Kalayci & Kahya, 2006). El análisis de componentes principales (PCA) es una técnica de análisis factorial, esto es, que se basa en la extracción de factores que definen la estruc- tura subyacente de un universo multivariante (matriz de datos). El objetivo general de esta técnica es reducir la dimensionalidad de la base datos forma- da por un número elevado de variables, y obtener grupos de variables, inde- pendientes entre sí y que retienen la mayor parte de la varianza contenida en las variables originales (Tabachnick & Fidell, 1996). Para que el análisis sea 115 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ENRIQUE MORÁN TEJEDA efectivo, las variables originales tienen que mostrar entre sí altos valores de correlación, lo cual significa que existe información redundante y que por ello puede ser reducida y explicada por una serie de variables nuevas. Las nuevas variables obtenidas (componentes principales) son combinaciones lineales de las variables originales y no muestran correlación entre sí (Jollife, 2002). El primer componente principal es aquella combinación lineal que explica el mayor porcentaje de varianza contenida en las variables originales, el segun- do componente será la combinación lineal no correlacionada con el compo- nente anterior y que absorbe el mayor porcentaje de varianza no explicada por el primero, y así sucesivamente. 116 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 4.1. Evolución y tendencias climáticas En la Figura 2 se muestran los mapas anual y mensuales de tendencias, resultado de interpolar (mediante un método kriging) los valores del tau de Mann-Kendall en cada serie de temperaturas anuales y mensuales. La figura refleja que en la totalidad de la cuenca las temperaturas medias anuales han experimentado una tendencia creciente durante el periodo de estudio. En todo el territorio el incremento térmico ha sido estadísticamente significativo (con un 95% de confianza) y asimismo parece existir una tendencia hacia un aumento más acentuado en la zona sureste de la cuenca. En cuanto a las tem- peraturas medias mensuales, en todos los meses se han registrado tendencias positivas en la mayor parte del territorio de la cuenca, aunque no con la misma intensidad ni significación estadística. Los meses en los que las ten- dencias han sido más acentuadas son marzo, junio, agosto, y en menor medi- da mayo y diciembre. Por su parte, enero, febrero, julio, septiembre y octubre y noviembre son los que ha registrado, aunque positivas, tendencias más sua- ves, y no significativas en la mayor parte del territorio. Estacionalmente se puede decir que han sido en los meses de primavera y verano donde el ascen- so térmico ha sido más pronunciado; y en el otoño donde el incremento ha sido más atenuado. 116 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... Figura 2. Tendencias de las temperaturas medias anuales y mensuales en la cuenca del Duero durante el periodo de estudio. Línea negra: α < 0,05 (estadísticamente significativo con un 95% de confianza). Figure 2. Trends in mean annual and monthly temperatures in the Duero basin, for the studied period. Black line: α < 0,05 (significant with a 95% of confidence). Figura 2. Tendencias de las temperaturas medias anuales y mensuales en la cuenca del Duero durante el periodo de estudio. Línea negra: α < 0,05 (estadísticamente significativo con un 95% de confianza). Figure 2. Trends in mean annual and monthly temperatures in the Duero basin, for the studied period. Black line: α < 0,05 (significant with a 95% of confidence). Figura 2. Tendencias de las temperaturas medias anuales y mensuales en la cuenca del Duero durante el periodo de estudio. Línea negra: α < 0,05 (estadísticamente significativo con un 95% de confianza). de confianza). Figure 2. ENRIQUE MORÁN TEJEDA El test de Mann-Kendall aplicado a una serie promedio para toda la región (Tabla 1) muestra una tendencia creciente acentuada y estadísticamente sig- nificativa con un 99% de confianza; además un ajuste lineal realizado entre las temperaturas y el tiempo revela un cambio entre el comienzo y el final del periodo superior a un 11%, pasando de una media de 12,2ºC de temperatura predicha en 1961 a 13,6ºC en el 2005. Esto supone una variación de + 1,4ºC a lo largo del periodo de estudio, o lo que es lo mismo, un aumento de casi 0,3ºC por década. El desglose mensual del ajuste lineal revela un aumento de en torno 2ºC en el mes de marzo y agosto, y de 2,3ºC en junio, es decir de casi 0,4ºC y 0,5ºC por década respectivamente. Septiembre y octubre son por su parte los meses en los que el incremento ha sido menor, con apenas 0,5ºC de diferencia entre el comienzo y el final del periodo. Ajuste lineal Tendencia n= 53 Tempera- Predicho Predicho tura R2 1961 2005 Cambio % tau MK α media (ºC) Enero 1,96 0,07 1,50 2,34 56,00 0,21* 0,04 Febrero 2,77 0,07 2,29 3,19 39,30 0,16 0,13 Marzo 4,35 0,33* 3,34 5,38 61,08 0,41* 0,00 Abril 9,57 0,09* 5,00 5,92 18,40 0,21* 0,04 Mayo 13,60 0,10* 7,40 8,52 15,14 0,23* 0,03 Junio 17,73 0,38* 9,60 11,90 23,96 0,46* 0,00 Julio 20,11 0,21* 12,20 13,50 10,66 0,29* 0,01 Agosto 12,69 0,34* 11,70 13,60 16,24 0,41* 0,00 Septiembre 10,55 0,03 10,15 10,80 6,40 0,11 0,29 Octubre 7,43 0,03 7,10 7,70 8,45 0,12 0,25 Noviembre 4,18 0,08* 3,70 4,60 24,32 0,17 0,10 Diciembre 2,41 0,17* 1,70 3,10 82,35 0,30* 0,00 AÑO 12,90 0,21* 12,16 13,55 11,43 0,29* 0,01 Tabla 1. Estadísticos de centralidad y cambio (ajuste lineal y test de Mann-Kendall) para las series regionales de temperaturas mensuales y anuales. * Indica estadísticamente significativo con un 95% de confianza. Table 1. Basic statistics for the regional series of temperature. Indicates significant with a 95% of confi- dence. Tabla 1. Estadísticos de centralidad y cambio (ajuste lineal y test de Mann-Kendall) para las series regionales de temperaturas mensuales y anuales. Indica estadísticamente significativo con un 95% de confianza. En la Figura 3 se muestran los mapas de tendencias en las precipitaciones anuales y mensuales. 4.1. Evolución y tendencias climáticas Trends in mean annual and monthly temperatures in the Duero basin, for the studied period. Black line: α < 0,05 (significant with a 95% of confidence). 117 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 117 118 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ENRIQUE MORÁN TEJEDA ENRIQUE MORÁN TEJEDA lo son en pequeñas partes del territorio. Los meses de octubre, diciembre, abril, mayo, julio y agosto presentan tendencias positivas, pero tan solo en agosto éstas son significativas en una parte importante del territorio. En tér- minos de ganancia o pérdida de agua de precipitación las tendencias más importantes son aquellas registradas para los meses de invierno, ya que estos meses acumulan la mayor parte de la precipitación. En cualquier caso, estas tendencias negativas (tampoco las positivas) no son estadísticamente signifi- cativas y generalizadas en todo el territorio de la cuenca, y ello se ve refleja- do asimismo en las tendencias para las precipitaciones anuales. A pesar de que se aprecia un patrón espacial en la distribución de las tendencias, con valores positivos en la mitad norte de la cuenca, y valores negativos en la mitad sur, tan sólo en una pequeña zona del sector suroccidental las tenden- cias son estadísticamente significativas, mientras que en la mayor parte del territorio los valores de tendencia no son significativos con un 95% de con- fianza. La Tabla 2 muestra además los estadísticos de dispersión y cambio (inclu- yendo las tendencias) para las precipitaciones medias en la cuenca; también se muestran las tendencias (tau de Mann-Kendall) y su significación estadís- tica. En la mayoría de los meses no se registran tendencias significativas, excepto en junio, con una tendencia regresiva, y en agosto, con una tendencia positiva. Las precipitaciones de febrero y octubre también han registrado una tendencia negativa y positiva, respectivamente, con unos valores que se encuentran en el límite de la significación estadística establecida. Mediante un ajuste lineal se ha estimado el porcentaje de cambio que se ha producido entre el comienzo y el final del periodo. Para los meses citados anteriormen- te, las precipitaciones han ascendido en más de un 130% y 104% en agosto y octubre, respectivamente, y se han reducido en un 56% y un 24% en febrero y junio. No obstante, la importancia de tales cambios, independientemente de su magnitud, radica en el volumen de precipitaciones registrado en cada mes. Por ello los más importantes a considerar son los meses de febrero y octubre (con tendencias opuestas), cuyas precipitaciones representan cerca de un 20% del total anual. ENRIQUE MORÁN TEJEDA En ellos se observa que no existe una homogeneidad en el signo de las tendencias en los distintos meses del año, además de que, en la mayoría de los casos, los valores de tendencia no son significativos, o sólo 118 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... 1 (Pirineos 2012 Vol 167 107-142 ISSN 0373-2568 eISSN: 1988-4281 doi: 10 3989/Pirineos 2012 167006) Figura 3. Tendencias de las precipitaciones anuales y mensuales en la cuenca del Duero duran el periodo de estudio. Línea negra (tendencias positivas) y blanca (tendencia negativa): α < 0, (estadísticamente significativo con un 95% de confianza). Figure 3. Trends in annual and monthly precipitation in the Duero basin for the studied period. Bla (positive trend) and white (negative trend) lines: α < 0,05 (significant with a 95% of confidence). Figura 3. Tendencias de las precipitaciones anuales y mensuales en la cuenca del Duero durante el periodo de estudio. Línea negra (tendencias positivas) y blanca (tendencia negativa): α < 0,05 (estadísticamente significativo con un 95% de confianza). Figure 3. Trends in annual and monthly precipitation in the Duero basin for the studied period. Black (positive trend) and white (negative trend) lines: α < 0,05 (significant with a 95% of confidence). Figura 3. Tendencias de las precipitaciones anuales y mensuales en la cuenca del Duero durante el periodo de estudio. Línea negra (tendencias positivas) y blanca (tendencia negativa): α < 0,05 (estadísticamente significativo con un 95% de confianza). Figure 3. Trends in annual and monthly precipitation in the Duero basin for the studied period. Black (positive trend) and white (negative trend) lines: α < 0,05 (significant with a 95% of confidence). Figura 3. Tendencias de las precipitaciones anuales y mensuales en la cuenca del Duero durante el periodo de estudio. Línea negra (tendencias positivas) y blanca (tendencia negativa): α < 0,05 (estadísticamente significativo con un 95% de confianza). Figure 3 Trends in annual and monthly precipitation in the Duero basin for the studied period Black 119 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 119 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ENRIQUE MORÁN TEJEDA Por el contrario, un aumento del 130% en las precipitaciones de agosto, no supondrá una gran contribución a la evolución de las precipi- taciones anuales, dado su escaso peso relativo (3%). Como resultado, las pre- cipitaciones anuales muestran una evolución, dentro de su variabilidad, más o menos estacionaria en el tiempo; permitiendo el test de Mann-Kendall con- firmar la inexistencia de una tendencia significativa para el periodo de estu- dio (tau = 0,04; α = 0,72). (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 120 IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... Tabla 2. Estadísticos de centralidad, dispersión, y cambio para las precipitaciones medias de la cuenca del Duero durante el periodo 1961-2005. * Estadísticamente significativo con un 95% de confianza. Table 2. Basic statistics for the regional series of precipitation. Indicates significant with a 95% of con- fidence. Precipi- Contri- Coeficiente n = 214 tación bución de varia- Ajuste lineal Tendencia media al total ción R2 Cambio % tau MK – (mm) anual % medio % Enero 67,73 10,68 79,33 0,01 -13,89 -0,09 0,38 Febrero 55,98 8,83 84,48 0,11 -56,41 -0,18 0,08 Marzo 46,95 7,40 80,85 0,02 -24,53 -0,10 0,35 Abril 57,85 9,12 65,47 0,00 8,93 -0,01 0,92 Mayo 62,86 9,91 59,48 0,02 18,97 0,08 0,45 Junio 39,63 6,25 79,17 0,08 -24,00 -0,23 0,02 Julio 21,67 3,42 112,43 0,00 4,55 0,01 0,89 Agosto 19,81 3,12 108,27 0,09 133,33 0,24* 0,02 Septiembre 40,55 6,39 83,86 0,00 -8,33 -0,01 0,95 Octubre 68,31 10,77 75,58 0,11 104,44 0,20 0,05 Noviembre 74,16 11,69 73,86 0,00 -7,79 -0,07 0,48 Diciembre 70,97 11,19 82,07 0,02 43,10 0,07 0,51 Año 634,29 100,00 23,33 0,00 0,16 0,04 0,72 4.2. Cambios en los usos del suelo 4.2. Cambios en los usos del suelo Total cuenca Franja norte Franja sur Tipo 1966 2003 ∆% 1966 2003 ∆% 1966 2003 ∆% Agua 313 422 25,89 55 94 41,33 19 44 56,45 Urbano-suelo desnudo 30 1.024 97,04 5 179 97,46 5 116 95,94 Pastizal-matorral 11.956 14.050 14,91 5.475 4.018 -36,25 2.251 3.460 34,94 Forestal 20.127 23.452 14,18 4.866 7.062 31,10 2.388 2.922 18,26 Cultivos 46.482 39.932 -16,40 5.962 5.041 -18,26 5.106 3.235 -57,83 f f q p y 4.2. Cambios en los usos del suelo La Figura 4 muestra el estado de los usos del suelo según los mapas foresta- les de 1966 y 2003 y la reclasificación efectuada (detalles en el apartado meto- dológico). A pesar de la diferencia de detalle (debido a las distintas escalas de los mapas originales), es fácil apreciar visualmente el crecimiento que se ha producido en la superficie cubierta por el bosque entre 1966 y 2003. De acuer- do con los análisis espaciales (Tabla 3), el crecimiento de la superficie forestal ha sido de un 14,2%, pasando a ocupar poco más de 20.000 km2 en 1966, a casi 23.500 en el 2003. La superficie correspondiente al pastizal-matorral también ha crecido en la misma proporción que el bosque, con un 15% de variación. La expansión de ambos se ha producido en detrimento del suelo ocupado por cultivos y uso agrícola, que, siendo el principal uso del suelo de la cuenca, ha sufrido un retroceso del 16% entre las dos fechas. Si consideramos el análisis comparado de las franjas norte y sur de la cuenca, donde se sitúan las cabe- ceras de la mayoría de los ríos que drenan al Duero, observamos que existen notables diferencias en la variación de los usos entre ambos territorios. Así, la superficie forestal ha aumentado en más de un 30%, mientras que en la fran- 121 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ENRIQUE MORÁN TEJEDA ja sur se ha producido un incremento más moderado de la superficie forestal, con un 18% de variación. Figura 4. Variación de los usos del suelo en la cuenca del Duero entre 1966 y 2003. Figure 4. Land-use changes in the Duero basin between 1966 and 2003. Figura 4. Variación de los usos del suelo en la cuenca del Duero entre 1966 y 2003. Figure 4. Land-use changes in the Duero basin between 1966 and 2003. Tabla 3. Variación de la superficie (km2) ocupada por los distintos usos del suelo. Table 3. Variation of the surface (square kilometers) occupied by various land-use classes. Tabla 3. Variación de la superficie (km2) ocupada por los distintos usos del suelo. Table 3. Variation of the surface (square kilometers) occupied by various land-use classes. 122 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 4.3 El cambio hidrológico En la Figura 5 se muestran las tendencias en las aportaciones anuales y mensuales durante el periodo de estudio. Nótese que en este caso se presen- tan los meses según el año hidrológico, con comienzo en octubre y fin en sep- tiembre. Un total de 38 estaciones (casi el 68%) muestra tendencias anuales negativas estadísticamente significativas, mientras que en 14 estaciones, los valores de tendencia negativa no son significativos. La última estación de afo- ros del Duero registra una tendencia regresiva significativa, así como las esta- 122 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS R IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... ciones del Esla, el Tormes, y el río Águeda que quedan aguas abajo, por lo que se puede afirmar que los recursos hídricos superficiales de la cuenca han experimentado un notable descenso durante el periodo de estudio. Mensualmente se observa que el descenso de los aportes fluviales no es homogéneo. Los meses centrales del invierno y la primavera, que son por su parte los más caudalosos del año, son en los que un mayor porcentaje de esta- ciones registran tendencias negativas, siendo especialmente numerosas en los meses de febrero y abril. Volviendo a las tendencias mensuales en la precipi- tación, recordamos que éstas tan sólo eran negativas en los meses de febrero y en menor medida en marzo, lo cual ayudaría a explicar las tendencias hidrológicas en estos meses; sin embargo las precipitaciones de abril mostra- ban tendencias ligeramente positivas. El descenso de las aportaciones fluvia- les en el mes de abril puede atribuirse a varias razones. En primer lugar, esto puede ser consecuencia de la inercia que presentan los procesos hidrológicos con respecto a la precipitación, y por ello las tendencias en los caudales de abril todavía reflejan las tendencias en la precipitación de invierno, incluyen- do una menor acumulación de nieve durante los meses de febrero y marzo. Por otro lado, podrían existir un factor o factores que estén contribuyendo, a pesar de la precipitación creciente, al descenso en los caudales de primavera, como podría ser el aumento en las temperaturas (y su papel sobre la fusión nival) o el propio incremento de la cobertura vegetal, como se demostrará más adelante. (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 124 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 4.3 El cambio hidrológico En los meses de verano y otoño, la variabilidad es mayor, aun- que el curso principal sigue registrando la inercia de la tendencia negativa registrada en los meses previos. En la Figura 6 se representa la evolución de las aportaciones promedio de todas las estaciones de aforo (arriba, izquierda) y la de los tramos más cau- dalosos con estación de aforo (los ríos Duero, Esla y Eresma), apreciándose en ambas que la tendencia regresiva no ha sido constante a lo largo de los años. Parece existir un patrón común de evolución con un descenso paulatino durante los primeros 15 años, un repentino incremento en 1975 que dura 3 años, seguido de otro periodo largo de descenso desde 1978 hasta mediados de los 90. En los últimos años, aunque la evolución es muy variable, parece haber una leve recuperación. El sombreado gris representa el rango inter- cuartil (es decir el rango que cubre entre el 25% y el 75% de los casos), con fil- trado de media móvil de 5 años. Éste nos permite observar un periodo de des- censo desde el inicio de la serie hasta mediados de los 90, interrumpido por el incremento de principios de los 70. Se aprecia además con más claridad el ascenso que se produce durante los últimos 15 años de la serie. A pesar de los ciclos existentes, la tendencia general durante el periodo de estudio es decre- ciente. Mediante un ajuste lineal a las series hidrológicas podemos obtener un valor aproximado del descenso que se ha producido durante el periodo de 123 ENRIQUE MORÁN TEJEDA Figura 5. Tendencias en las aportaciones fluviales anuales y mensuales durante el periodo de estudio. Figure 5. Trends in monthly and annual river flows during the studied period. Figura 5. Tendencias en las aportaciones fluviales anuales y mensuales durante el periodo de estudio. Figure 5. Trends in monthly and annual river flows during the studied period. 124 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 124 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS R IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... estudio, restando el valor predicho por la recta de ajuste al final de la serie del valor predicho al comienzo. 4.3 El cambio hidrológico En el río Duero, las pérdidas habrían sido supe- riores al 120%, pasando de unas aportaciones de 5.230 hm3 al comienzo del periodo, a las 2.303 hm3 de final del periodo. Algo similar ha ocurrido en los otros dos ríos más caudalosos, el Esla, con un descenso de un 48%, y el Eresma con pérdidas cercanas al 100%. Sumando las aportaciones del Esla y el Eresma, –que tributan al Duero aguas abajo del aforador estudiado– a las del propio Duero, contaríamos unos aportes de casi 15.000 hm3 a comienzos de la serie, y de casi 8.200 hm3 a finales del periodo, lo que supone unas pér- didas de casi la mitad de los recursos hídricos en tan sólo medio siglo. Figura 6. Evolución de las aportaciones fluviales medias de la cuenca (arriba, izquierda) y de los tramos más caudalosos. El sombreado gris representa el rango intercuartil de los casos de estudio, con una media móvil de 5 años. Figure 6. Evolution of river flows for the average series of the basin (upper-left panel) and in the three main courses of the basin. Grey shade represents the interquartil range of the studied cases with a moving average of 5 years. Figura 6. Evolución de las aportaciones fluviales medias de la cuenca (arriba, izquierda) y de los tramos más caudalosos. El sombreado gris representa el rango intercuartil de los casos de estudio, con una media móvil de 5 años. Figure 6. Evolution of river flows for the average series of the basin (upper-left panel) and in the three main courses of the basin. Grey shade represents the interquartil range of the studied cases with a moving average of 5 years. El descenso neto de los recursos hídricos no es el único cambio relevante en la hidrología de la cuenca. Mediante un análisis de componentes princi- pales se han caracterizado en la cuenca tres tipos de regímenes fluviales. Un régimen pluvial, localizado en los cursos medios y bajos de los ríos, también en alguna cabecera; un régimen nivo-pluvial, localizado en las zonas más altas de la cuenca; y un régimen alterado por la regulación fluvial. 126 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 4.3 El cambio hidrológico Si nos ate- nemos a los cursos con régimen natural, observamos en la Figura 7 los cam- bios que se han producido en los mismos entre la primera y segunda mitad del periodo de estudio en dos ejemplos de estaciones con cada tipo de régi- 125 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ENRIQUE MORÁN TEJEDA men. En los cursos con régimen pluvial se aprecia que el pico de caudal inver- nal se localiza en el mes de febrero durante la primera mitad del periodo de estudio. El cambio más notable que se observa en la segunda mitad del perio- do de estudio es el brusco descenso de este pico, con el consiguiente retroce- so de máximo fluvial al mes de enero. Por su parte en los casos de cursos con régimen nivo-pluvial, además del retroceso de los caudales en el mes de febrero, se produce también un descenso muy notable en el pico nival. Este máximo en la primera mitad del periodo se producía en el mes de abril o mayo, y en la segunda mitad, además de haber perdido magnitud, se ha ade- lantado también un mes. Además, en uno de los casos mostrados (arriba, derecha) vemos la transformación de un régimen natural en un régimen regu- lado, con la construcción de un embalse cuya gestión evidencia la retención de agua durante el invierno y la primavera, para ser desembalsada en el vera- no (de ahí los altos caudales de verano en la segunda mitad del periodo). Los resultados mostrados hasta ahora evidencian un descenso claro y de gran magnitud de los caudales de la cuenca. Las precipitaciones por su parte no muestran tendencias tan evidentes, por lo que se manifiesta un claro des- ajuste entre la evolución climática y la hidrológica. Figura 7. Ejemplos de cambio en los regímenes fluviales entre 1961-83 (línea negra) y 1984-05 (línea gris punteada). Figure 7. Examples of change of fluvial regimes between 1961-83 (black line) and 1984-05 (grey dotted line). Figura 7. Ejemplos de cambio en los regímenes fluviales entre 1961-83 (línea negra) y 1984-05 (línea gris punteada). Figura 7. Ejemplos de cambio en los regímenes fluviales entre 1961-83 (línea negra) y 1984-05 (línea gris punteada). Figure 7. Examples of change of fluvial regimes between 1961-83 (black line) and 1984-05 (grey dotted line). g p Figure 7. IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... Para inferir el papel que las variables climáticas han tenido sobre la evo- lución hidrológica se han desarrollado regresiones lineales múltiples con las precipitaciones y temperaturas como variables predictoras, para una selec- ción de cuencas de cabecera, en las que la intervención humana ha sido más bien escasa (Beguería et al., 2003, López-Moreno et al. 2011). Las series de pre- cipitaciones y temperaturas introducidas en las regresiones se han obtenido de la siguiente forma: con las 214 series de precipitación y 57 de temperatura se han construido grids climáticos con una resolución 100x100m mediante el método de interpolación explicado en Ninyerola et al. (2000). Para cada cuen- ca de cabecera se ha obtenido entonces una serie promedio de precipitación y otra de temperatura, que son las que se han utilizado para modelizar la evo- lución de las aportaciones. El método de regresión “paso a paso” permite introducir las variables en el modelo tan sólo si presentan una contribución estadísticamente significativa (α< 0,05) a la explicación de la variable depen- diente, y son descartadas si no cumplen con dicho criterio estadístico. En la Tabla 4 se muestran los resultados de dichas regresiones. En primer lugar hay que destacar que la principal variable que explica la evolución de las aporta- ciones es, en 20 de los 21 casos, la precipitación. Las temperaturas, al contra- rio de lo que cabría esperar dado su papel sobre la evapotranspiración, tan sólo intervienen de manera significativa en 7 de los modelos realizados. El dato más relevante de la tabla son los coeficientes de MK calculados para los residuales de los modelos, que expresan la tendencia teórica de la parte no explicada por el modelo. En la mayoría de los modelos los residuales presen- ta coeficientes negativos, aunque tan sólo en 7 casos, las tendencias son esta- dísticamente significativas. Una tendencia negativa en los residuales indica que las aportaciones están evolucionando con independencia del clima, o dicho de otra forma, que un factor que no ha sido incluido en los modelos está contribuyendo al descenso de las aportaciones. En cuencas de cabecera no reguladas, el único factor capaz de explicar tal descenso es el incremento de la cubierta vegetal, que como se ha demostrado anteriormente ha sido nota- ble en las zonas de montaña de la cuenca. 4.3 El cambio hidrológico Examples of change of fluvial regimes between 1961-83 (black line) and 1984-05 (grey dotted line). 126 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 12 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ENRIQUE MORÁN TEJEDA ENRIQUE MORÁN TEJEDA Tabla 4. Resultados de las regresiones lineales realizadas para estimar la evolución de las aporta- ciones en función de la precipitación y la temperatura, y coeficientes de Mann-Kendall para los residuales de los modelos. Se muestran los coeficientes de cada variable independiente, el por- centaje de varianza explicada (R2) y su nivel de significación estadística en el modelo (α). * indica significación estadística (α < 0,05). g Table 4. Results of linear regressions for predicting the evolution of streamflows as a function of precipita- tion and temperature, and Mann-Kendall coefficients for the values of residuals on time. * Indicates sig- nificant with a 95% of confidence. 128 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) Precipitación Temperatura Residuales Estación Constante Coef. R2 α Coef. R2 a R2 Tau α total MK 2000 4,50E-11 0,72 0,52* 0,00 - - 0,56 0,52 -0,10 0,32 2006 -3,90E-11 0,73 0,54* 0,00 - - 0,43 0,54 -0,22* 0,03 2009 -1,60E-11 - - 0,14 - - 0,38 0,01 -0,20 0,05 2019 1,40E-11 0,82 0,68* 0,00 0,21 0,04* 0,02 0,72 -0,17 0,11 2024 -1,90E-11 0,68 0,45* 0,00 0,23 0,05* 0,04 0,50 -0,14 0,19 2028 -4,40E-12 0,57 0,23* 0,00 0,62 0,37* 0,00 0,60 -0,15 0,15 2030 -4,90E-11 0,83 0,67* 0,00 0,34 0,12* 0,00 0,79 -0,08 0,42 2035 2,00E-11 0,81 0,66* 0,00 - - 0,38 0,66 -0,55* 0,00 2046 -5,60E-11 0,74 0,54* 0,00 - - 0,53 0,54 0,00 0,97 2047 5,20E-11 0,71 0,51* 0,00 - - 0,84 0,51 -0,21* 0,05 2050 5,70E-11 0,83 0,64* 0,00 0,3 0,09* 0,01 0,73 -0,19 0,08 2051 3,50E-11 0,74 0,54* 0,00 - - 0,81 0,54 0,00 0,98 2052 -1,20E-10 0,67 0,45* 0,00 - - 0,69 0,45 -0,09 0,37 2068 7,20E-12 0,68 0,47* 0,00 - - 0,11 0,47 -0,29* 0,00 2070 3,50E-12 0,28 0,36* 0,00 - - 0,44 0,36 -0,03 0,70 2078 -6,80E-11 0,49 0,28* 0,00 0,38 0,14* 0,03 0,42 -0,30* 0,00 2089 -4,10E-11 0,81 0,61* 0,00 - - 0,37 0,65 -0,02 0,82 2101 -3,70E-11 0,38 0,16* 0,00 - - 0,24 0,16 -0,34* 0,00 2104 -7,30E-11 0,78 0,61* 0,00 - - 0,76 0,61 -0,25* 0,02 2107 9,90E-11 0,84 0,70* 0,00 0,17 0,03* 0,05 0,73 -0,10 0,35 2109 1,40E-10 0,7 0,49* 0,00 - - 0,99 0,49 -0,02 0,87 f % f f nentes. Las series predichas y los residuales se representan asimismo en la Figura 8, con curvas grises y negra (en trazo grueso), respectivamente. IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... Mediante un análisis de componentes principales se han identificado dos grupos de estaciones (de entre las estaciones de cabecera seleccionadas) en función de la evolución de los aportes fluviales. Las series agregadas de apor- taciones de ambos grupos se muestran con una curva negra en la Figura 8, donde también se muestra la localización de las estaciones pertenecientes a cada grupo o componente principal. Salvo excepciones, la mayoría de esta- ciones pertenecientes al grupo 1 se encuentran en las cabeceras del norte de la cuenca, mientras que las del grupo 2 se encuentran en las montañas del sur. Al igual que para las series individuales, se ha modelizado la evolución de las aportaciones con las series climáticas agregadas de cada uno de los compo- 127 128 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ENRIQUE MORÁN TEJEDA En ambos casos se aprecia como los residuos presentan una evolución descen- dente, revelando que las aportaciones observadas descienden en el tiempo a un ritmo superior a las predichas en función del clima, lo cual indica la exis- tencia de un factor con una componente temporal que está influyendo en ese descenso de las aportaciones. El hecho de encontrar una tendencia regresiva más acentuada en los residuos del componente 1 (norte de la cuenca), donde la expansión de la cubierta vegetal ha sido de mayor magnitud, sugiere que Precipitación Temperatura Residuales Estación Constante Coef. R2 α Coef. R2 a R2 Tau α total MK 2000 4,50E-11 0,72 0,52* 0,00 - - 0,56 0,52 -0,10 0,32 2006 -3,90E-11 0,73 0,54* 0,00 - - 0,43 0,54 -0,22* 0,03 2009 -1,60E-11 - - 0,14 - - 0,38 0,01 -0,20 0,05 2019 1,40E-11 0,82 0,68* 0,00 0,21 0,04* 0,02 0,72 -0,17 0,11 2024 -1,90E-11 0,68 0,45* 0,00 0,23 0,05* 0,04 0,50 -0,14 0,19 2028 -4,40E-12 0,57 0,23* 0,00 0,62 0,37* 0,00 0,60 -0,15 0,15 2030 -4,90E-11 0,83 0,67* 0,00 0,34 0,12* 0,00 0,79 -0,08 0,42 2035 2,00E-11 0,81 0,66* 0,00 - - 0,38 0,66 -0,55* 0,00 2046 -5,60E-11 0,74 0,54* 0,00 - - 0,53 0,54 0,00 0,97 2047 5,20E-11 0,71 0,51* 0,00 - - 0,84 0,51 -0,21* 0,05 2050 5,70E-11 0,83 0,64* 0,00 0,3 0,09* 0,01 0,73 -0,19 0,08 2051 3,50E-11 0,74 0,54* 0,00 - - 0,81 0,54 0,00 0,98 2052 -1,20E-10 0,67 0,45* 0,00 - - 0,69 0,45 -0,09 0,37 2068 7,20E-12 0,68 0,47* 0,00 - - 0,11 0,47 -0,29* 0,00 2070 3,50E-12 0,28 0,36* 0,00 - - 0,44 0,36 -0,03 0,70 2078 -6,80E-11 0,49 0,28* 0,00 0,38 0,14* 0,03 0,42 -0,30* 0,00 2089 -4,10E-11 0,81 0,61* 0,00 - - 0,37 0,65 -0,02 0,82 2101 -3,70E-11 0,38 0,16* 0,00 - - 0,24 0,16 -0,34* 0,00 2104 -7,30E-11 0,78 0,61* 0,00 - - 0,76 0,61 -0,25* 0,02 2107 9,90E-11 0,84 0,70* 0,00 0,17 0,03* 0,05 0,73 -0,10 0,35 2109 1,40E-10 0,7 0,49* 0,00 - - 0,99 0,49 -0,02 0,87 nentes. Las series predichas y los residuales se representan asimismo en la Figura 8, con curvas grises y negra (en trazo grueso), respectivamente. ENRIQUE MORÁN TEJEDA En ambos casos se aprecia como los residuos presentan una evolución descen- dente, revelando que las aportaciones observadas descienden en el tiempo a un ritmo superior a las predichas en función del clima, lo cual indica la exis- tencia de un factor con una componente temporal que está influyendo en ese descenso de las aportaciones. El hecho de encontrar una tendencia regresiva más acentuada en los residuos del componente 1 (norte de la cuenca), donde la expansión de la cubierta vegetal ha sido de mayor magnitud, sugiere que nentes. Las series predichas y los residuales se representan asimismo en la Figura 8, con curvas grises y negra (en trazo grueso), respectivamente. En ambos casos se aprecia como los residuos presentan una evolución descen- dente, revelando que las aportaciones observadas descienden en el tiempo a un ritmo superior a las predichas en función del clima, lo cual indica la exis- tencia de un factor con una componente temporal que está influyendo en ese descenso de las aportaciones. El hecho de encontrar una tendencia regresiva más acentuada en los residuos del componente 1 (norte de la cuenca), donde la expansión de la cubierta vegetal ha sido de mayor magnitud, sugiere que 128 128 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... ACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS R el crecimiento forestal es el factor que ayuda a explicar el descenso de las aportaciones. Según dicha hipótesis, en aquellas cuencas donde el incremento de la superficie forestal ha sido mayor, los residuales de los modelos deben pre- sentar una tendencia regresiva más acusada. En el gráfico de dispersión de la Figura 9 se aprecia una relación lineal en esa dirección, si bien alguna cuenca se aleja de la recta de ajuste, dando lugar a un coeficiente de determinación de Pearson (R2), aunque significativo, no muy elevado. Figura 8. Aportaciones observadas (curva negra fina), predichas (curva gris) y residuos (curva negra, gruesa) para las series agregadas de cada uno de los componentes principales, suaviza- das mediante media móvil de 5 años. El ajuste lineal (recta), y los valores de Mann-Kendall corresponden a la evolución de los residuos. Figure 8. Observed river flows (black thin line), predicted river flows (grey line) and residuals (black thick line) for the aggregated series of each principal component. 129 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ENRIQUE MORÁN TEJEDA The linear trend and Mann-Kendall values correspond to the evolution of residuals. Figura 8. Aportaciones observadas (curva negra fina), predichas (curva gris) y residuos (curva negra, gruesa) para las series agregadas de cada uno de los componentes principales, suaviza- das mediante media móvil de 5 años. El ajuste lineal (recta), y los valores de Mann-Kendall corresponden a la evolución de los residuos. Figure 8. Observed river flows (black thin line), predicted river flows (grey line) and residuals (black thick line) for the aggregated series of each principal component. The linear trend and Mann-Kendall values correspond to the evolution of residuals. Figura 8. Aportaciones observadas (curva negra fina), predichas (curva gris) y residuos (curva negra, gruesa) para las series agregadas de cada uno de los componentes principales, suaviza- das mediante media móvil de 5 años. El ajuste lineal (recta), y los valores de Mann-Kendall corresponden a la evolución de los residuos. Figure 8. Observed river flows (black thin line), predicted river flows (grey line) and residuals (black thick line) for the aggregated series of each principal component. The linear trend and Mann-Kendall values correspond to the evolution of residuals. Figura 8. Aportaciones observadas (curva negra fina), predichas (curva gris) y residuos (curva negra, gruesa) para las series agregadas de cada uno de los componentes principales, suaviza- das mediante media móvil de 5 años. El ajuste lineal (recta), y los valores de Mann-Kendall corresponden a la evolución de los residuos. p Figure 8. Observed river flows (black thin line), predicted river flows (grey line) and residuals (black thick line) for the aggregated series of each principal component. The linear trend and Mann-Kendall values correspond to the evolution of residuals. 129 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ENRIQUE MORÁN TEJEDA Figura 9. Relación entre los residuales de los modelos y la variación en la superficie forestal. Figure 9. Relationship between residuals of models and variation of forest surface. Figura 9. Relación entre los residuales de los modelos y la variación en la superficie forestal. Figure 9. Relationship between residuals of models and variation of forest surface. 130 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 4.4. El papel regulador de los embalses Hasta ahora se han mostrado estadísticas de la evolución de los recursos hídricos en régimen natural, o escasamente alterado por el hombre. A conti- nuación se muestran unos ejemplos de cómo la gestión de algunos embalses modifica no sólo el régimen fluvial sino la disponibilidad de agua a largo plazo, en función de los patrones de gestión. En la Figura 10 se representa el régimen fluvial (promedio para el periodo de estudio) a la entrada y salida de 4 embalses de la cuenca, así como las reservas mensuales en los mismos. Se han seleccionado estos 4 ejemplos, ya que ilustran los modelos de gestión tipo que se dan en la cuenca: por un lado observamos un modelo en el que apenas se modifica el régimen fluvial (Cervera-Ruesga), tan sólo se retienen peque- ñas cantidades de agua en el invierno que sirven para aumentar las reservas en la primavera y el verano. El valor de correlación (R = 0,95) entre las entra- das y salidas mensuales indica que apenas se produce modificación del régi- men fluvial. En segundo lugar se observa un patrón en el que se retienen mayores cantidades de agua durante el invierno (La Requejada), con la con- secuente modificación del régimen fluvial (R = 0,24). Y por último encontra- mos dos ejemplos (Barrios de Luna y Cuerda del Pozo) en los que la retención 130 130 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS R IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... de caudales en el invierno es muy fuerte, y éstos son posteriormente libera- dos en el verano para abastecer diferentes demandas (riego, abastecimiento urbano, etc.). En estos casos se produce una inversión del régimen fluvial, de ahí los valores negativos de correlación entre entradas y salidas (R = -0,14 y -0,67 respectivamente). Figura 10. Entradas (línea negra), salidas (línea gris) y reservas (barras) mensuales en cuatro embalses de la cuenca del Duero, promediadas para el periodo de estudio. R: coeficiente de correlación de Pearson. Figure 10. Inflows (black line), outflows (grey line) and volume of water storage (bars) in 4 reservoirs of the basin. R: Pearson’s correlation coefficient. Figura 10. Entradas (línea negra), salidas (línea gris) y reservas (barras) mensuales en cuatro embalses de la cuenca del Duero, promediadas para el periodo de estudio. R: coeficiente de correlación de Pearson. Figure 10. (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 4.4. El papel regulador de los embalses Inflows (black line), outflows (grey line) and volume of water storage (bars) in 4 reservoirs of the basin. R: Pearson’s correlation coefficient. El coeficiente de correlación de Pearson informa del nivel de regulación o alteración del régimen fluvial por parte de los embalses. Mientras que la Figura 10 presenta una imagen estática de dicho nivel de regulación prome- diado durante el periodo de estudio, en la Figura 11 podemos ver la evolu- ción del mismo a lo largo de los años. Tanto para los embalses cuyo coefi- ciente de correlación medio indica una leve regulación, como para aquellos en los que la alteración del régimen es elevada, durante las últimas décadas se aprecia una ligera tendencia hacia un mayor nivel de regulación, esto es, hacia valores más bajos del coeficiente de correlación. Este aumento de la alte- ración fluvial ha ido acompañado de un incremento sostenido de los niveles de embalsado, sobre todo en los dos embalses con mayor nivel de regulación, 131 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 132 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ENRIQUE MORÁN TEJEDA ENRIQUE MORÁN TEJEDA Barrios de Luna y Cuerda del Pozo, como se aprecia en la Figura 11. En este sentido, cabe destacar la diferencia en la evolución de los niveles de embalsa- do entre ambos, a pesar de mostrar una tendencia del mismo signo. En el embalse de Cuerda del Pozo el incremento del nivel de embalsado se produ- ce con una notable variabilidad inter-anual, lo cual sugiere que se ha ido adaptando el nivel de embalsado a las demandas o bien a la disponibilidad de agua que gobierna la variabilidad climática. Por el contrario en el embalse de Barrios de Luna se observa cómo los niveles de embalsado se han mante- nido constantes durante periodos largos de tiempo, y en momentos puntua- les se han bajado o incrementado, dando lugar a un balance final de incre- mento de los niveles. Este patrón indica más una gestión que depende de los intereses particulares de los gestores del embalse, por encima de otras consi- deraciones como la disponibilidad o demanda de agua. En cualquier caso se observa que los ríos están experimentando cada vez niveles más altos de regulación, a costa de aumentar los embalses las reservas a lo largo de los años, de lo que se deduce igualmente que los embalses están contribuyendo al descenso de los caudales de los ríos de la cuenca. 132 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) Figura 11. Evolución del coeficiente de correlación entre entradas y salidas mensuales (curva negra) y del volumen de agua embalsada (barras grises) durante el periodo de estudio. Las líne- as rectas se corresponden con ajustes lineales para ilustrar la tendencia temporal. Nótese que el eje correspondiente a las reserves (derecha) tiene los valores invertidos. Figure 11. Evolution of the correlation coefficient between inflows and outflows (black line) and evolution of the water storage (grey bars) during the studied period. Straightlinesindicate linear trends. Figura 11. Evolución del coeficiente de correlación entre entradas y salidas mensuales (curva negra) y del volumen de agua embalsada (barras grises) durante el periodo de estudio. Las líne- as rectas se corresponden con ajustes lineales para ilustrar la tendencia temporal. Nótese que el eje correspondiente a las reserves (derecha) tiene los valores invertidos. Figure 11. Evolution of the correlation coefficient between inflows and outflows (black line) and evolution of the water storage (grey bars) during the studied period. Straightlinesindicate linear trends. 5. Discusión y conclusiones Los recursos hídricos de la cuenca del Duero han experimentado un retro- ceso notable durante la segundo mitad del siglo XX. En este trabajo, además de poner de manifiesto las características de este retroceso, se evalúan las que pensamos que son las principales causas del mismo. p q p p De forma general se puede atribuir el descenso global de los recursos hídricos a la reducción de las aportaciones fluviales en invierno y primavera, coincidiendo con los momentos de máximo caudal. En los ríos con máximo invernal el descenso es mayor en los meses de febrero, resultando en un des- plazamiento del pico a meses anteriores. Otra característica común es el retro- ceso significativo de los caudales de primavera, que implica también el ade- lantamiento en un mes del pico primaveral. Asistimos por lo tanto a dos pro- cesos complementarios, que a simple vista deberían implicar una preocupa- ción en los gestores de los recursos hídricos ya que se cuenta con menos agua para abastecer una demanda que a priori no ha descendido (Gil Olcina 1999). A ello hay que añadirle la incertidumbre derivada de las causas de dichos cambios, ya que, como se ha comprobado en este trabajo, y se discute a con- tinuación, pueden ser consecuencia de numerosos procesos, alguno de los cuales seguramente se escape de los contenidos de este estudio. g p En primer lugar el descenso y redistribución mensual de los caudales se debe explicar, como no podría ser de otra forma, desde el punto de vista de la evolución climática de la cuenca. El clima es el factor responsable de las entradas, y parte de las salidas, de agua en las cuencas hidrográficas. Si hay una característica que defina la evolución climática en la Península Ibérica, esta es la variabilidad (Bladé & Castro-Díez, 2010). Las precipitaciones, y en menor medida las temperaturas, en la región mediterránea, presentan una evolución variable en el tiempo. A esta incertidumbre climatológica se ha unido históricamente el conflicto entre disponibilidad y demanda, ya que normalmente el consumo de las mayores cantidades de agua se produce en las épocas en que la disponibilidad es menor (Iglesias, 2005). No obstante, superpuestas a esta variabilidad, en muchas ocasiones se encuentran tenden- cias a largo plazo en las variables climáticas, que pueden condicionar en mayor medida la disponibilidad de los recursos hídricos. ENRIQUE MORÁN TEJEDA 132 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 132 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 5. Discusión y conclusiones En el caso de las precipitaciones, si bien observamos que han descendido en el mes de febrero, no muestran por lo general descensos significativos. De hecho las series anua- les de precipitación se podría decir que no han experimentado tendencias cla- ras de ningún signo. Nuestras observaciones no contradicen en exceso los resultados obtenidos en la mayoría de los trabajos realizados sobre la evolu- ción de las precipitaciones a lo largo del siglo XX en el sur de Europa. Aunque los estudios suelen diferir en cuanto a periodos de tiempo analizados, escala 133 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ENRIQUE MORÁN TEJEDA espacial del análisis, número de observatorios, o incluso tests estadísticos uti- lizados, en general no se ha podido detectar un descenso significativo de las precipitaciones en la región (Giorgi, 2002, Norrant & Douguedroit, 2006, Gonzalez-Hidalgo et al., 2010). Por el contrario, la evolución de las tempera- turas en la región muestra un claro signo de incremento, siendo las tenden- cias más evidentes en los meses de diciembre, y entre la primavera y el vera- no. El aumento térmico demostrado para la cuenca del Duero coincide con el incremento general registrado en todo el territorio peninsular, cuyo origen se cifra por la mayoría de los autores a comienzos de la década de los 70 (De Castro et al., 2005, Brunet et al., 2007, Bladé & Castro-Díez, 2010). Asimismo, el análisis por regiones mundiales que ofrece el último informe del Panel Intergubernamental sobre Cambio Climático (IPCC 2007) muestra para el continente europeo un punto de inflexión a comienzos de la década de los 70, a partir del cual las temperaturas aumentan de forma sostenida hasta la actualidad, y pronostica una continuidad en el incremento térmico a lo largo del siglo XXI, como causa del aumento en la concentración atmosférica de gases invernadero de origen antrópico. g g p Volviendo a los cambios hidrológicos observados, el descenso en los cau- dales de invierno podría estar reflejando las tendencias negativas en la preci- pitación de febrero. Sin embargo observamos cómo ya en diciembre y enero se aprecian tendencias significativas en los caudales, las cuales no pueden ser explicadas por la evolución en la precipitación. Por otro lado observamos una caída y retroceso del pico nival en aquellos tramos fluviales en que el máxi- mo se produce en la primavera. 134 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 5. Discusión y conclusiones Esto se debe probablemente a una combina- ción de dos procesos. En primer lugar, las tendencias negativas en la precipi- tación de febrero sugieren un descenso también en la precipitación en forma de nieve, y por lo tanto menor acumulación de nieve en las partes altas de las montañas, lo que a su vez supone una menor cantidad de agua de escorren- tía una vez que se produce la fusión del manto nivoso en primavera. En segundo lugar, la ocurrencia más temprana del pico primaveral podría ser atribuida al incremento de las temperaturas de invierno y primavera, lo que provocaría tanto una menor acumulación de nieve, como una fusión más temprana. Este incremento térmico, durante el invierno y la primavera es considerado responsable de cambios en los regímenes fluviales en otros luga- res de la península. Por ejemplo, López-Moreno (2005) demostró en los Pirineos una evolución regresiva del manto nivoso de abril durante las últi- mas cinco décadas, la cual presentaba buena correlación con las tendencias en precipitación en los meses precedentes a la fusión, y con las tendencias de las temperaturas de abril. Más recientemente, Pons et al.,(2009) encontraron ten- dencias negativas en el número de “días de nieve” durante el invierno y la primavera en la mitad norte de España (incluyendo puntos de medición en la 134 ACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS R IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... cuenca del Duero) durante las últimas tres décadas. En este caso, los autores demostraron que estas tendencias se correlacionaban mejor con la evolución de las temperaturas que con las de las precipitaciones. Cambios en los regí- menes fluviales similares a los observados en este trabajo han sido detectados en otras zonas de montaña de la Península Ibérica y de la región Mediterránea. Por ejemplo López-Moreno et al.,(2008) y García-Ruiz et al.,(2001)demostraron una reducción significativa de los caudales primavera- les, con una ocurrencia más temprana del pico máximo en los ríos pirenaicos, lo cual atribuyeron a la reducción de la cantidad de nieve acumulada, por condicionantes climáticos. Kalayci and Kahya (2006) encontraron tendencias negativas en los caudales mensuales de distintos ríos de Turquía, y lo rela- cionaron con las crecientes temperaturas y evapotranspiración. (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 5. Discusión y conclusiones 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ENRIQUE MORÁN TEJEDA bocadura en Tortosa, de los cuales atribuyen un tercio al descenso en las pre- cipitaciones, otro tercio al consumo por los regadíos, y un tercio también al crecimiento en la cubierta forestal en las cabeceras fluviales. Un resultado relevante de nuestros análisis es la variable, pero en general escasa, participación de la “temperatura” en los modelos para explicar la evo- lución de las aportaciones hídricas. Desde un punto de vista teórico las cre- cientes temperaturas deberían estar provocando un aumento en la evapo- transpiración y afectando negativamente al balance hídrico de las cuencas; que esto no se vea reflejado en los modelos se debe probablemente a la esca- la de análisis utilizada. El hecho de trabajar con datos anuales puede estar enmascarando procesos significativos a escala estacional, pero no suficiente- mente relevantes a escala anual. Anteriormente se apuntó a la posible influen- cia de las crecientes temperaturas sobre los cambios en los regímenes fluvia- les, sin embargo su papel tendría más que ver con el adelanto del pico pri- maveral de origen nival que con el descenso neto de las aportaciones anuales. Lespinas et al. (2010), argumentan que el descenso en los caudales de ríos de montaña del sur de Francia está asociado con el papel de las crecientes tem- peraturas en el control de la acumulación y fusión de la nieve, sin embargo no consideró el posible impacto de los cambios en los usos del suelo ocurri- dos en su zona de estudio. La evapotranspiración es un proceso dependiente de la actividad vegetativa de las plantas, y el consumo de agua por parte de éstas se produce básicamente durante la primavera y el verano. Un aumento de las temperaturas significaría un incremento potencial de la necesidad hídrica de las plantas y por tanto mayor consumo sobre todo en los meses de verano, cuando el estrés hídrico es más notable. El escaso peso que represen- tan las aportaciones de verano con respecto al total anual (< 5%), explicaría que este proceso no se vea reflejado sobre la evolución interanual de las apor- taciones hídricas. En cualquier caso, el papel de los cambios en la cobertura vegetal y del aumento en las temperaturas deben estar íntimamente relacio- nados a través del proceso de evapotranspiración, y resultan extremadamen- te complejos de separar en base a estudios estadísticos. 5. Discusión y conclusiones p y p p A pesar de que las variables climáticas pueden explicar en parte el des- censo observado en las aportaciones fluviales, encontramos evidencias de una evolución dispar, sobre todo entre caudales y precipitaciones que hace pensar que un factor no climático está contribuyendo al descenso hidrológi- co. Dada esta disparidad entre la evolución de las precipitaciones y los cau- dales, nuestra principal hipótesis para explicar el descenso hidrológico radi- ca en el incremento observado de la cubierta vegetal y el papel hidrológico de la misma demostrado en cuencas experimentales. En el presente trabajo se pone de manifiesto que el crecimiento de la cubierta vegetal ha podido tener un papel relevante en el descenso de los aportes fluviales. Tres observaciones verifican dicha hipótesis: (i) existe en todos los casos de estudio una tenden- cia regresiva en los aportes fluviales más marcada que en las precipitaciones; (ii) Los residuos de los modelos presentan coeficientes de Mann-Kendall negativos, lo cual indica una separación progresiva entre la evolución de las aportaciones y la evolución climática, y (iii) la disparidad entre la evolución de las aportaciones y de las precipitaciones, y las tendencias regresivas en los aportes son más evidentes en las cabeceras fluviales localizadas en el norte de la cuenca, donde la expansión del bosque y el matorral de montaña ha afec- tado a mayores áreas. Tomando esta relación como hipótesis principal distin- tos investigadores, a partir de métodos estadísticos, han estimado el papel de los cambios en los usos del suelo sobre la evolución interanual de los cauda- les en grandes cuencas hidrológicas o conjuntos regionales. Por ejemplo, Begueríaet al., (2003), Gallart& Llorens (2004), y López-Moreno et al.,(2011) comprueban para un conjunto de cabeceras fluviales tributarias del Ebro una separación sistemática de la evolución de las aportaciones respecto a las pre- cipitaciones, así como una tendencia significativa en los residuos, que acha- can al crecimiento de la cubierta vegetal en las cuencas de drenaje, y lo cuan- tifican en un 30% y un 17% respectivamente. También Gallart & Llorens (2002) perciben un importante descenso en los aportes del Ebro en su desem- 135 (Pirineos, 2012, Vol. 136 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 5. Discusión y conclusiones Por último, se ha mostrado la capacidad de algunos embalses para modi- ficar las condiciones hidrológicas, y se atisba una tendencia hacia mayores niveles de regulación y embalsado, con el consiguiente descenso de los cau- dales, aguas abajo de los mismos. Distintos estudios realizados hasta la fecha han comprobado igualmente la capacidad de los embalses para modificar los regímenes fluviales aguas abajo de la presa. En nuestro ámbito más cercano, el trabajo de López-Moreno (2006) demuestra que los embalses pirenaicos presentan, por lo general, modificaciones del régimen fluvial más suaves que los embalses de la cuenca del Duero, y raramente se observan inversiones del régimen fluvial como las demostradas en este trabajo. Por otro lado, Batalla et ACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS R IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... al. (2004), estudiaron los cambios hidrológicos producidos por los embalses de la cuenca del Ebro y observaron patrones de modificación de los regíme- nes fluviales similares a los demostrados en este trabajo. Así, mientras unos embalses apenas modificaron la distribución estacional de los caudales aguas abajo, otros, principalmente destinados al riego durante el verano, producían una inversión total del régimen. En función de nuestras observaciones pode- mos argumentar que la política hidrológica de la cuenca está apostando por una gestión de los recursos en su oferta a través del incremento de las reser- vas para hacer frente a una disponibilidad de agua cada vez menor. Una polí- tica hidrológica basada en el aprovisionamiento de recursos será a corto plazo insostenible, en caso de que continúe el descenso de caudales en la cuenca observado en este trabajo, acorde con las previsiones de evolución climática para las próximas décadas (Iglesias, 2005). p p g La disponibilidad futura de recursos hídricos presenta altos niveles de incertidumbre. Los modelos climáticos predicen un descenso de hasta un 20% en los recursos hídricos de la región mediterránea durante el curso del siglo presente (IPCC, 2007). Para la cuenca del Duero hemos demostrado que el clima no puede explicar únicamente el descenso en los caudales. Por ello el principal reto que se plantea, derivado de los resultados obtenidos, es la modelización de una de las variables más importantes para explicar las pér- didas de agua, la evapotranspiración. 5. Discusión y conclusiones Ésta conjuga tanto condiciones climáti- cas como condiciones de cubierta vegetal y es por tanto imprescindible su conocimiento a la hora de plantear proyecciones futuras de disponibilidad de agua en escenarios de cambio climático. Los resultados de este trabajo nos permiten concluir que el descenso hidrológico demostrado no debería ser estudiado solamente desde el ámbito científico. El reconocimiento del mismo por parte de los responsables o ges- tores públicos del recurso es esencial a corto plazo. Cualquier intervención relativa a la gestión de los recursos hídricos de la cuenca debe ser considera- da desde una perspectiva global que considere las sinergias entre las monta- ñas y el llano, lugares respectivos de origen y consumo de los mismos. Solamente una gestión integrada del territorio podría minimizar los impactos del descenso hidrológico sobre los ecosistemas, la población y las actividades económicas de la cuenca del Duero. 137 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) Agradecimientos El presente estudio constituye una síntesis de los resultados de la tesis doctoral defendida por el autor en la Facultad de Geografía e Historia de la Universidad de Salamanca, en marzo de 2011. El trabajo ha sido posible gra- 137 ENRIQUE MORÁN TEJEDA cias una beca de investigación concedida por la Junta de Castilla y León (Becas de formación del personal investigador, orden EDU/918/2006 del 30 de mayo) disfrutada entre los años 2006 y 2010. El autor expresa su más sin- cero agradecimiento a Antonio Ceballos y Nacho López por la supervisión; y a Sergio Vicente, Javier Zabalza, José Manuel Llorente, Jorge Lorenzo y Santiago Beguería por su colaboración en las distintas etapas de elaboración del trabajo. 138 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) Referencias Alexandersson, H. 1986. A Homogeneity test applied to precipitation data.Journal of Climatology, 6:661-675. Beguería, S., López-Moreno, J. I., Lorente, A., Seeger, M. & García-Ruiz, J. M. 2003. Assessing the effect of climate oscillations and land-use changes on streamflow in the Central Spanish Pyrenees. Ambio, 32:283-286. Beniston, M. 2005. Mountain climates and climatic change: An overview of processes focusing on the European Alps. Pure and Applied Geophysics, 162:1587-1606. Doi:10.1007/s00024-005-2684-9. Bent, G. C. 2001. Effects of forest-management activities on runoff compo- nents and ground-water recharge to Quabbin Reservoir, central Massachusetts. Forest Ecology and Management, 143:115-129. Berkun, M. 2010. Hydroelectric potential and environmental effects of multi- dam hydropower projects in Turkey. Energy for Sustainable Development, 14:320-329. Berryman, D., Bobee, B., Cluis, D. & Haemmerli, J. 1988. Nonparametric tests for trend detection in water-quality time series. Water Resources Bulletin, 24:545-556. Birsan, M. V., Molnar, P., Burlando, P. & Pfaundler, M. 2005. Streamflow trends in Switzerland. Journal of Hydrology, 314:312-329. Bladé, I. & Castro-Díez, Y. 2010. Tendencias atmosféricas en la Península Ibérica durante el periodo instrumental en el contexto de la variabilidad natural. En F. Pérez & R. Boscolo, (Eds). Clima en España: pasado, presente y futuro. 25-42 pp. Bonacci, O. & Roje-Bonacci, T. 2003. The influence of hydroelectrical develop- ment on the flow regime of the karstic river Cetina. Hydrological Processes, 17:1-15. Doi:10.1002/hyp.1190. y Bosch, J. M. & Hewlett, J. D. 1982. A review of catchment experiments to determine the effect of vegetation changes on water yield and evapo-tran- spiration. Journal of Hydrology, 55:3-23. 138 IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... R. & Smith, R. A. 1982. Techniques of trend analysis for monthly water quality data. Water Resources Research, 18:107-121. Iglesias, A., Estrela, T., Gallart, F. 2005. Impactos sobre los recursos hídricos. In: Moreno Rodríguez, J.M. (Coord.).Evaluación preliminar de los impactos en España por efecto del cambio climático. Ministerio de Medio Ambiente y Universidad de Castilla-La Mancha. 303-354. IPCC. 2007. Climate Change 2007: Synthesis Report. [Core Writing Team, Pachauri RK and Reisinger A (eds.)]. IPCC: Geneva, Switzerland, 104 pp. C t ib ti f W ki G I II d III t th F th A t Pachauri RK and Reisinger A (eds.)]. IPCC: Geneva, Switzerland, 104 pp. Contribution of Working Groups I, II and III to the Fourth Assessment g ( )] , , pp Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. J llif I T 2002 P i i l C t A l i 2 diti S i N Y k Report of the Intergovernmental Panel on Climate Change. Jollife, I. T. 2002. Principal Component Analysis. 2 edition. Springer, New York. J , p p y p g , Kahya, E. & Kalayci, S. 2004. Trend analysis of streamflow in Turkey. Journal of Hydrology, 289:128-144. f y gy Kalayci, S. & Kahya, E. 2006. Assessment of streamflow variability modes in Turkey: 1964-1994. Journal of Hydrology, 324:163-177. Kendall, M. G. 1975. Kendall Rank Correlation Methods. Griffin, London. Kundzewicz, Z. W. & Robson, A. J. 2004. Change detection in hydrological records-a review of the methodology / Revue méthodologique de la détection de changements dans les chroniques hydrologiques. Hydrological Sciences Journal, 49: 7-19. Lanzante, J. R. 1996. Resistant, robust and non-parametric techniques for the analysis of climate data: Theory and examples, including applications to historical radiosonde station data. International Journal of Climatology, 16:1197-1226. Lasanta-Martínez, T., Vicente-Serrano, S. M. & Cuadrat-Prats, J. M. 2005. Mountain Mediterranean landscape evolution caused by the abandon- ment of traditional primary activities: a study of the Spanish Central Pyrenees. Applied Geography, 25:47-65. Doi:10.1016/j.apgeog.2004.11.001. y pp g p y j pg g Lehmann, E. L. 1975. Nonparametrics Statistical Methods Based on Ranks. Holden-Day, Inc., California. Lespinas, F., Ludwig, W. & Heussner, S. 2010. Impact of recent climate change on the hydrology of coastal Mediterranean rivers in Southern France. Climatic Change, 99:425-456. g Lettenmaier, D. P., Wood, E. F. & Wallis, J. R. 1994. IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... Brunet, M., Jones, P. D., Sigro, J., Saladie, O., Aguilar, E., Moberg, A., Della- Marta, P. M., Lister, D., Walther, A. & Lopez, D. 2007. Temporal and spa- tial temperature variability and change over Spain during 1850-2005. Journal of Geophysical Research-Atmospheres, 112. Doi:10.1029/2006jd008249. Cosandey, C., Andreassian, V., Martin, C., Didon-Lescot, J. F., Lavabre, J., Folton, N., Mathys, N. & Richard, D. 2005. The hydrological impact of the mediterranean forest: a review of French research. Journal of Hydrology, 301:235-249. Doi:10.1016/j.jhydrol.2004.06.040. j j y Cosandey, C. & Robinson, M. 2000. Hydrologie continentale. Armand Colin, Paris, 360 pp. pp Crockford, R. H. & Richardson, D. P. 2000. Partitioning of rainfall into throughfall, stemflow and interception: effect of forest type, ground cover and climate. Hydrological Processes,14:2903-2920. De Castro, M., Martín-Vide, J. & Alonso, S. 2005. El clima de España: pasado, presente y escenarios de clima para el siglo XXI. pp 1-64 En J. M. C. Moreno Rodríguez, (Ed). Evaluación preliminar de los impactos en España por efecto del cambio climático. Ministerio de Medio Ambiente y Universidad de Castilla-La Mancha. Gallart, F. & Llorens, P. 2003. Catchment management under environmental change: Impact of land cover change on water resources. Water International, 28:334-340. García-Ruiz, J. M., Beguería Portugués, S., López Moreno, J. I., Lorente Grima, A. & Seeger, M. 2001. Los recursos hídricos superficiales del Pirineo aragonés y su evolución reciente. Geoforma Ediciones, Logroño. García-Ruiz, J. M., Lasanta, T., Ruiz-Flano, P., Ortigosa, L., White, S., González, C. & Martí, C. 1996. Land-use changes and sustainable devel- opment in mountain areas: a case study in the Spanish Pyrenees. Landscape Ecology, 11:267-277. García-Ruiz, J. M., López-Moreno, J. I., Vicente-Serrano, S. M., Lasanta- Martínez, T. & Beguería, S. 2011. Mediterranean water resources in a glob- al change scenario. Earth-Science Reviews, 105:121-139. g Gil Olcina, A. 1999. Los usos del agua en España: una perspectiva histórica. En A. Gil Olcina & A. Morales Gil, (Eds). Los Usos del Agua en España. Caja de Ahorros del Mediterráneo e Instituto Universitario de Geografía (Universidad de Alicante), Alicante, pp 13-48. Giorgi, F. 2002. Variability and trends of sub-continental scale surface climate in the twentieth century. Part I: observations. Climate Dynamics, 18:675-691. Doi:10.1007/s00382-001-0204-x. Gonzalez-Hidalgo, J. C., Brunetti, M. & de Luis, M. 2010. Precipitation trends in Spanish hydrological divisions, 1946-2005. Climate Research, 43:215-228. Doi:10.3354/cr00937. 139 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) ENRIQUE MORÁN TEJEDA Hirsch, R. M., Slack, J. 140 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... Hydro-climatological trends in the continental United States, 1948-88. Journal of Climate, 7:586- 607. Llorens, P. & Domingo, F. 2007. Rainfall partitioning by vegetation under Mediterranean conditions. A review of studies in Europe. Journal of Hydrology, 335:37-54. Doi:10.1016/j.jhydrol.2006.10.032. López-Moreno, J. I. 2005. Recent variations of snowpack depth in the central Spanish Pyrenees. Arctic, Antarctic and Alpine Research, 37:253-260. 140 141 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) IMPACTOS RECIENTES DE LOS CAMBIOS AMBIENTALES EN LOS RECURSOS HÍDRICOS... López-Moreno, J. I. 2006. Cambio ambiental y gestión de los embalses en el Pirineo Central español. Consejo de Portección de la Naturaleza de Aragón, Zaragoza, 208 pp. g pp López-Moreno, J. I., Begueria, S. & Garcia-Ruiz, J. M. 2002. Influence of the Yesa reservoir on floods of the Aragon River, central Spanish Pyrenees. Hydrology and Earth System Sciences, 6:753-762. López-Moreno, J. I., Beniston, M. & García-Ruiz, J. M. 2008. Environmental change and water management in the Pyrenees: Facts and future perspec- tives for Mediterranean mountains. Global and Planetary Change, 61:300- 312. López-Moreno, J. I., Vicente-Serrano, S. M., Morán-Tejeda, E., Zabalza, J., Lorenzo-Lacruz, J. & García-Ruiz, J. M. 2011. Impact of climate evolution and land use changes on water yield in the Ebro basin. Hydrology and Earth System Sciences,15:311-322. Maingi, J. K. & Marsh, S. E. 2002. Quantifying hydrologic impacts following dam construction along the Tana River, Kenya. Journal of Arid Environments, 50:53-79. Mann, H. B. 1945. Non parametric test against trend. Econometrica, 13:245-259. h Milly, P. C. D., Dunne, K. A. & Vecchia, A. V. 2005. Global pattern of trends in streamflow and water availability in a changing climate. Nature, 438:347- 350. Morán, C. 2007. Trascolación en un bosque de roble melojo (Quercus pyrenaica willd) en la sierra de Tamames (Salamanca). Trabajo de Grado Inédito. Universidad de Salamanca. Nilsson, C., Reidy, C. A., Dynesius, M. & Revenga, C. 2005. Fragmentation and flow regulation of the world’s large river systems. Science, 308:405- 408. Doi:10.1126/science.1107887. Ninyerola, M., Pons, X. & Roure, J. M. 2000. A methodological approach of cli- matological modelling of air temperature and precipitation through GIS techniques. International Journal of Climatology, 20:1823-1841. Norrant, C. & Douguedroit, A. 2006. Monthly and daily precipitation trends in the Mediterranean (1950-2000). Theoretical and Applied Climatology, 83:89-106. Doi:10.1007/s00704-005-0163-y. y Peterson, T. C., Easterling, D. R., Karl, T. R., Groisman, P., Nicholls, N., Plummer, N., Torok, S., Auer, I., Boehm, R., Gullett, D., Vincent, L., Heino, R., Tuomenvirta, H., Mestre, O., Szentimrey, T., Salinger, J., Forland, E. J., Hanssen-Bauer, I., Alexandersson, H., Jones, P. & Parker, D. 1998. Homogeneity adjustments of in situ atmospheric climate data: A review. International Journal of Climatology, 18:1493-1517. J f gy Petts, G. E. 1984. Impounded rivers. John Willey & sons Eds, Chichester, 326 pp. ENRIQUE MORÁN TEJEDA ENRIQUE MORÁN TEJEDA Pons, M. R., San-Martín, D., Herrera, S. & Gutiérrez, J. M. 2009. Snow trends in Northern Spain: analysis and simulation with statistical downscaling methods. International Journal of Climatology, 30:1795-1806. f gy Poyatos, R., Latron, J. & Llorens, P. 2003. Land use and land cover change after agricultural abandonment - The case of a Mediterranean Mountain Area (Catalan Pre-Pyrenees). Mountain Research and Development, 23:362- 368. Rodríguez-Puebla, C., Encinas, A. H., Nieto, S. & Garmendia, J. 1998. Spatial and temporal patterns of annual precipitation variability over the Iberian Peninsula. International Journal of Climatology, 18:299-316. J f gy Sneyers, R. 1990. On the statistical analysis of series of observations. Technical Note - World Meteorological Organization, Geneva. Tabachnick, B. & Fidell, L. 1996. Using multivariate statistics. 3 edition. Harper Collins, New York. Verstraeten, G. & Poesen, J. 2000. Estimating trap efficiency of small reser- voirs and ponds: methods and implications for the assessment of sedi- ment yield. Progress in Physical Geography, 24:219-251. Vicente-Serrano, S., Beguería, S. & Lasanta-Martínez, T. 2000. El proceso de revegetación en un área de montaña como consecuencia de los cambios de gestión: Aplicación de un SIG al Valle Borau (Pirineo Central Español).En:Tecnologías Geográficas para el Desarrollo Sostenible, pp 448-469. Departamento de Geografía, Universidad de Alcalá. Vicente-Serrano, S. M., Beguería, S., López-Moreno, J. I., García-Vera, M. A. & Stepanek, P. 2009. A complete daily precipitation database for northeast Spain: reconstruction, quality control, and homogeneity. International Journal of Climatology. Doi:10.1002/joc.1850. Viviroli, D. & Weingartner, R. 2004. The hydrological significance of moun- tains: from regional to global scale. Hydrology and Earth System Sciences, 8:1016-1029. Widmann, M. & Schär, C. 1997. A principal component and long-term trend analysis of daily precipitation in Switzerland. International Journal of Climatology, 17:1333-1356. gy Yue, S., Pilon, P. & Cavadias, G. 2002. Power of the Mann-Kendall and Spearman’s rho tests for detecting monotonic trends in hydrological series. Journal of Hydrology, 259:254-271. f y gy Zhang, L., Dawes, W. R. & Walker, G. R. 2001. Response of mean annual evap- otranspiration to vegetation changes at catchment scale. Water Resources Research, 37:701-708. Zhang, X. B., Vincent, L. A., Hogg, W. D. & Niitsoo, A. 2000. Temperature and precipitation trends in Canada during the 20th century. Atmosphere-Ocean, 38:395-429. 142 (Pirineos, 2012, Vol. 167, 107-142, ISSN 0373-2568, eISSN: 1988-4281, doi: 10.3989/Pirineos.2012.167006) 142
https://openalex.org/W2768966896	https://europepmc.org/articles/pmc5824371?pdf=render	English	null	Dihydroartemisinin up-regulates VE-cadherin expression in human renal glomerular endothelial cells	Journal of Cellular and Molecular Medicine	2,017	cc-by	2,899	Abstract The antimalarial agent dihydroartemisinin (DHA) has been shown to be anti-inﬂammatory. In this study, we found that DHA increased the expression of the junctional protein vascular endothelial (VE)-cadherin in human renal glomerular endothelial cells. In addition, DHA inhibited TGF-b RI-Smad2/3 signalling and its downstream effectors SNAIL and SLUG, which repress VE-cadherin gene transcription. Correspondingly, DHA decreased the binding of SNAIL and SLUG to the VE-cadherin promoter. Together, our results suggest an effect of DHA in regulating glomerular permeability by elevation of VE-cadherin expression. Keywords: dihydroartemisinin glomerular endothelial cells VE-cadherin TGF-b signalling Introduction In the kidney, vascular permeability is regulated by the glomerular ﬁl- tration barrier (GFB), a highly specialized blood ﬁltration interface maintaining the balance of ion and metabolite concentrations [1]. The impairment of the GFB is the important feature of various renal inﬂammatory diseases [2]. The GFB is composed of glomerular endothelium, the glomerular basement membrane (GBM) and the podocyte layer [3]. The glomerular endothelium is a semipermeable membrane formed by glomerular endothelial cells (GECs), which are a unique microvascular cell type with round shape and fenestrations [2]. GECs are exposed to circulating elements of the blood and are sensitive to various inﬂammatory factors [2]. With dysfunction of the GFB, glomerular capillaries become highly permeable to water, solutes and plasma proteins, resulting in oedema and albuminuria [4]. GECs are connected by adherens, tight and gap junctions, which maintain cell to cell adhesion and control vascular permeability [5]. VE-cadherin is expressed exclusively in endothelial cells and is a major component of vascular adherens junctions [6]. drug due to its ability to inhibit the sarcoplasmic and endoplasmic reticulum calcium ATPase of Plasmodium falciparum [8]. DHA is a water-soluble derivative of artemisinin that produces few adverse side effects [8]. Artemisinin and its derivatives displayed strong anti- inﬂammatory effects [9]. However, the underlying mechanisms have not been fully understood. In this study, we evaluated the effects of DHA on the expression of VE-cadherin in human renal glomerular endothelial cells (HRGECs). We found that DHA signiﬁcantly elevated the expression of VE-cadherin and inhibited transforming growth factor receptor I (TGF-b RI)-Smad2/ 3 signalling in HRGECs. In addition, DHA down-regulated expression of SNAIL and SLUG, the transcriptional repressors of the VE-cadherin gene. ChIP assay demonstrated that DHA signiﬁcantly decreased the binding of SNAIL and SLUG to the VE-cadherin promoter. Journal of Cellular and Molecular Medicine published by John Wiley & Sons Ltd and Foundation for Cellular and Molecular Medicine.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. Materials and methods Artemisinin is a sesquiterpene lactone endoperoxide extracted from the Artemisia annua plant [7]. It is widely used as an antimalarial Liqun Li a, Xiaocui Chen a, Fengyun Dong a, Qiang Liu a, Caiqing Zhang b, Dongmei Xu c, Thaddeus D. Allen d, Ju Liu a, * a Laboratory of Microvascular Medicine, Medical Research Center, Shandong Provincial Qianfoshan Hospital, Shandong University, Jinan, Shandong, China b Department of Respiratory and Critical Care Medicine, Shandong Provincial Qianfoshan Hospital, Shandong, University, Jinan, Shandong, China c Department of Nephrology, Shandong Provincial Qianfoshan Hospital, Shandong University, Jinan, Shandong, China d Tradewind BioScience, Daly City, CA, USA CA, USA) and cultured in Dulbecco s modiﬁed Eagle s medium (DMEM) E-mail: ju.liu@sdu.edu.cn ª 2017 The Authors. Journal of Cellular and Molecular Medicine published by John Wiley & Sons Ltd and Foundation for Cellular and Molecular Medicine.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. doi: 10.1111/jcmm.13448 J. Cell. Mol. Med. Vol 22, No 3, 2018 pp. 2028-2032 J. Cell. Mol. Med. Vol 22, No 3, 2018 pp. 2028-2032 Short Communication Dihydroartemisinin up-regulates VE-cadherin expression in human renal glomerular endothelial cells Liqun Li a, Xiaocui Chen a, Fengyun Dong a, Qiang Liu a, Caiqing Zhang b, Dongme Thaddeus D. Allen d, Ju Liu a, * ª 2017 The Authors. Chromatin immunoprecipitation (ChIP) assay Chromatin fragments of HRGECs were prepared as previously described [11]. Immunoprecipitation was performed with a ChIP assay kit (Upstate Biotechnology Inc. Lake Placid, NY, USA) with the antibodies against SNAIL, SLUG or control IgG (Abcam) according to the manufacturer’s instructions. The DNA fragments were detected by semi-quantitative PCR. The primer sequences were as follows: sense, 50-GGGTGGACAAG CACCTTAAA-30; antisense, 50-ACCCCACTTGAACCCCTACT-30. The detailed materials and methods was described in Data S1. Western blotting Western blotting was performed as previously described [10]. The pri- mary antibodies were rabbit anti-VE-cadherin, mouse anti-SNAIL, rabbit anti-SLUG, rabbit anti-Smad2 and rabbit anti-phospho-Smad2 (pSer255) (Abcam, Cambridge, MA, USA), rabbit anti-Smad3, rabbit anti-phospho- Smad3 (pSer423/425) and rabbit anti-GAPDH (Cell Signaling Technology, Beverly, MA, USA) and rabbit anti-TGF-b RI (Santa Cruz Biotechnology, Santa Cruz, CA, USA). The secondary antibodies were HRP-conjugated goat anti-rabbit IgG and HRP-conjugated goat anti- mouse IgG (Proteintech, Chicago, IL, USA). Cell culture and treatments HRGECs were obtained from Sciencell Research Laboratories (Carlsbad, CA, USA) and cultured in Dulbecco’s modiﬁed Eagle’s medium (DMEM) HRGECs were obtained from Sciencell Research Laboratories (Carlsbad, CA, USA) and cultured in Dulbecco’s modiﬁed Eagle’s medium (DMEM) Journal of Cellular and Molecular Medicine published by John Wiley & Sons Ltd and Foundation for Cellular and Molecular Medicine.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. J. Cell. Mol. Med. Vol 22, No 3, 2018 calculated using b-actin or GAPDH as an endogenous internal control. The primer sequences were listed in Table S1. (Corning Inc., Corning, NY, USA), supplemented with 10% foetal bovine serum (Lonza, Basel, Switzerland), 100 IU/ml penicillin and 100 lg/ml streptomycin. DHA was purchased from Sigma-Aldrich (St. Louis, MO, USA) and applied to HRGEC cultures with a ﬁnal concentration 25 lM for 24 hrs before measurements. ª 2017 The Authors. Quantitative real-time PCR VE-cadherin is a transmembrane adhesion molecule bridging adjacent endothelial cells [6]. Inside cells, VE-cadherin is complexed with b-catenin and p120, which, in turn, bind to a-catenin, an actin-bind- ing protein [12]. The VE-cadherin–catenin complex is essential for the maintenance of vascular integrity. The effect of DHA on the expres- sion of VE-cadherin in HRGECs has been examined. We found a Total cellular RNA was extracted from HRGECs with the E.Z.N.A. total RNA Kit II (OMEGA Bio-tek, Inc., Norcross, GA, USA) following the manufacturer’s protocol. Synthesis of cDNA was performed with the RevertAid First strand cDNA Synthesis kit (Thermo Fisher, Grand Island, NY, USA). QRT-PCR was performed with a ViiA7 Real-Time PCR Sys- tem (Applied Biosystems, Waltham, MA, USA). Relative expression was Fig. 1 DHA up-regulates the expression of VE-cadherin and inhibits TGF-b signalling in HRGECs. (A) Relative VE-cadherin mRNA expression in HRGECs treated with vehicle or DHA (n = 4; P < 0.01). (B) Immunoblots of VE-cadherin protein from HRGECs treated with vehicle or DHA. GAPDH was used as loading control. (C) Representative images of VE-cadherin immunostaining on HRGECs treated with vehicle or DHA. Magniﬁcation: 200X. (D) Immu- noblots of TGF-b RI protein from HRGECs treated with vehicle or DHA. (E) Immunoblots of phospho-Smad2 and total Smad2 from HRGECs treated with vehicle or DHA (F) Immunoblots of phospho-Smad3 and total Smad3 from HRGECs treated with vehicle or DHA. GAPDH was used as loading control. Fig. 1 DHA up-regulates the expression of VE-cadherin and inhibits TGF-b signalling in HRGECs. (A) Relative VE-cadherin mRNA expression in HRGECs treated with vehicle or DHA (n = 4; P < 0.01). (B) Immunoblots of VE-cadherin protein from HRGECs treated with vehicle or DHA. GAPDH was used as loading control. (C) Representative images of VE-cadherin immunostaining on HRGECs treated with vehicle or DHA. Magniﬁcation: 200X. (D) Immu- noblots of TGF-b RI protein from HRGECs treated with vehicle or DHA. (E) Immunoblots of phospho-Smad2 and total Smad2 from HRGECs treated with vehicle or DHA (F) Immunoblots of phospho-Smad3 and total Smad3 from HRGECs treated with vehicle or DHA. GAPDH was used as loading control. 2029 Cellular and Molecular Medicine published by John Wiley & Sons Ltd and Foundation for Cellular and Molecular Medicine. bits the expression of SNAIL/SLUG in HRGECs and suppresses SNAIL/SLUG binding afﬁnity to the human VE-cadherin prom reated with 25 lM DHA for 24 hrs before measurements. ª 2017 The Authors. ª 2017 The Authors Journal of Cellular and Molecular Medicine published by John Wiley & Sons Ltd and Foundation for Cellular and Molecular Medicine Quantitative real-time PCR (A, C) Relative mRNA expression of SNAIL (A) and SLUG (C) measur GECs treated with vehicle or DHA (n = 4; P < 0.01). (B, D) Immunoblots of SNAIL (B) and SLUG (D) protein from HRGECs GAPDH was used as loading control. (E, G) Representative images of PCR products from the DNA fragments pulled down by S G) antibodies. Primers were designed to detect the -240 E-box on the human VE-cadherin promoter. (F, H) Binding ratios relat of chromatin used for ChIP with the SNAIL (F) and SLUG (H) antibodies (n = 4; P < 0.01). Fig. 2 DHA inhibits the expression of SNAIL/SLUG in HRGECs and suppresses SNAIL/SLUG binding afﬁnity to the human VE-cadherin promoter. HRGECs were treated with 25 lM DHA for 24 hrs before measurements. (A, C) Relative mRNA expression of SNAIL (A) and SLUG (C) measured by qRT-PCR in HRGECs treated with vehicle or DHA (n = 4; P < 0.01). (B, D) Immunoblots of SNAIL (B) and SLUG (D) protein from HRGECs trea- ted with DHA. GAPDH was used as loading control. (E, G) Representative images of PCR products from the DNA fragments pulled down by SNAIL (E) and SLUG (G) antibodies. Primers were designed to detect the -240 E-box on the human VE-cadherin promoter. (F, H) Binding ratios relative to the total input of chromatin used for ChIP with the SNAIL (F) and SLUG (H) antibodies (n = 4; P < 0.01). 2030 J. Cell. Mol. Med. Vol 22, No 3, 2018 J. Cell. Mol. Med. Vol 22, No 3, 2018 signiﬁcant up-regulation of VE-cadherin mRNA following 25 lM DHA treatments for 24 hrs (P < 0.01, Fig. 1A). Consistently, VE-cadherin protein was also increased after 24 hrs of DHA treatment (Fig. 1B). Immunoﬂuorescent staining of HRGECs monolayers demonstrated that the intensity of VE-cadherin plasma membrane staining was sig- niﬁcantly increased in HRGECs after 24 hrs of DHA exposure (Fig. 1C). These results indicated that DHA increased VE-cadherin expression in HRGECs. Up-regulation of VE-cadherin in HRGECs reduces vascular permeability [10]. Therefore, it is likely that DHA- induced up-regulation of VE-cadherin directly antagonizes glomerular hyperpermeability in renal inﬂammation diseases. to the speciﬁc nucleotide sequence CANNTG, called the E-box motif, through highly conserved C2H2-type zinc-ﬁnger domains [15]. In endothelial cells, SNAIL and SLUG bind to the proximal E-boxes (at - 240) of human VE-cadherin promoter and suppress its promoter activity [15]. Quantitative real-time PCR To examine the binding afﬁnity of SNAIL and SLUG, ChIP assays were performed with HRGECs treated with DHA. Native chromatin was immunoprecipitated with antibodies raised against SNAIL, SLUG or control IgG, and the immunoprecipitated fragments were subjected to PCR using speciﬁc primers ﬂanking the -240 SNAIL/SLUG binding site on the promoter of the human VE-cadherin gene. The binding of both SNAIL (Fig. 2E and F) and SLUG (Fig. 2G and H) to the VE-cadherin promoter was remarkably reduced by DHA treatment (P < 0.01). This suggests that DHA increases the expres- sion of VE-cadherin via a cascade of events that include inhibitory effects on SNAIL and SLUG. Acknowledgements This study was supported by grants from the Traditional Chinese Medicine Research Projects of Shandong Province (no.2015-285), the National Natural Science Foundation of China (no. 81370269) and the Shandong Taishan Schol- arship (Ju Liu). Supporting information Additional Supporting Information may be found online in the supporting information tab for this article: Table S1. Quantitative RT-PCR primer sequences Table S1. Quantitative RT-PCR primer sequences Data S1. Materials and Methods Data S1. Materials and Methods DHA inhibits TGF-b RI-Smad2/3 signalling in HRGECs In conclusion, we found that DHA signiﬁcantly increases the expression of VE-cadherin in HRGECs. DHA inhibits TGF-b RI-Smad2/ 3 signalling and the downstream transcriptional activation of SNAIL and SLUG. In addition, DHA treatment decreased the binding of SNAIL and SLUG protein to the VE-cadherin promoter. Together, DHA-induced increase in VE-cadherin expression arises through down-regulation of basal levels of TGF-b RI-Smad2/3 signalling, lower expression of SNAIL and SLUG and ultimately a relief of tran- scription repression of the VE-cadherin promoter. Upon activation with TGF-b, the TGF-b RI induces a downstream sig- nalling cascade that includes the phosphorylation of Smad2/3 [13]. In endothelial cells, phosphorylated Smad2/3 translocates into the nucleus and bind to the Smad-binding element (SBE), activating the expression of target genes [13]. We examined the effects of DHA treatment on the TGF-b pathway in HRGECs. Western blot analysis showed that the expression of TGF-b RI protein was decreased in HRGECs after incubation with DHA (Fig. 1D). Levels of phosphory- lated Smad2 and phosphorylated Smad3 were also decreased follow- ing DHA treatment, whereas total levels of Smad2 and Smad3 remained unchanged (Fig. 1E and F). Previous studies showed that TGF-b signalling impairs the barrier function of microvascular endothelial monolayers through down-regulating the expression VE-cadherin [11, 14]. Thus, suppression of TGF-b RI-Smad2/3 sig- nalling might contribute to DHA-induced up-regulation of VE-cadherin. DHA suppresses the expression of SNAIL and SLUG in HRGECs The authors declare no conﬂict of interest. The SNAIL family of zinc-ﬁnger transcription factors is known regula- tors of VE-cadherin [15]. SNAIL and SLUG are downstream effectors of TGF-b RI-Smad2/3 signalling [16]. Smad proteins bind directly to the SBE of the SNAIL and SLUG promoters to elevate their transcrip- tion [16]. Therefore, we examined the effect of DHA on SNAIL and SLUG expression in HRGECs. QRT-PCR analysis demonstrated that DHA treatment markedly decreased the mRNA expression of SNAIL (P < 0.01, Fig. 2A) and SLUG (P < 0.01, Fig. 2C). Western blot anal- ysis conﬁrmed that both SNAIL and SLUG protein were decreased in HRGECs after DHA treatment (Fig. 2B and D). SNAIL and SLUG bind 3. Deen WM, Bridges CR, Brenner BM. Biophysical basis of glomerular permselectivity. J Membr Biol. 1983; 71: 1–10. ª 2017 The Authors. Journal of Cellular and Molecular Medicine published by John Wiley & Sons Ltd and Foundation for Cellular and Molecular Medicine. 2031 ª 2017 The Authors. References 2. Obeidat M, Obeidat M, Ballermann BJ. Glomerular endothelium: a porous sieve and formidable barrier. Exp Cell Res. 2012; 318: 964–72. 3. Deen WM, Bridges CR, Brenner BM. Biophysical basis of glomerular permselectivity. J Membr Biol. 1983; 71: 1–10. 3. Deen WM, Bridges CR, Brenner BM. Biophysical basis of glomerular permselectivity. J Membr Biol. 1983; 71: 1–10. 2. Obeidat M, Obeidat M, Ballermann BJ. Glomerular endothelium: a porous sieve and formidable barrier. Exp Cell Res. 2012; 318: 964–72. 1. Hausmann R, Grepl M, Knecht V, et al. The glomerular ﬁltration barrier function: new concepts. Curr Opin Nephrol Hypertens. 2012; 21: 441–9. 2031 he Authors. Cellular and Molecular Medicine published by John Wiley & Sons Ltd and Foundation for Cellular and Molecular Medicine. 4. Mehta D, Malik AB. Signaling mechanisms regulating endothelial permeability. Physiol Rev. 2006; 86: 279–367. 9. Wang JX, Tang W, Zhou R, et al. The new water-soluble artemisinin derivative SM905 ameliorates collagen-induced arthritis by sup- pression of inﬂammatory and Th17 responses. Br J Pharmacol. 2008; 153: 1303–10. 13. Goumans MJ, Lebrin F, Valdimarsdottir G. Controlling the angiogenic switch: a balance between two distinct TGF-b receptor signal- ing pathways. Trends Cardiovasc Med. 2003; 13: 301–7. 5. Molema G, Aird WC. Vascular heterogeneity in the kidney. Semin Nephrol. 2012; 32: 145–55. 10. 14. Maroni D, Davis JS. TGFB1 disrupts the angiogenic potential of microvascular endothelial cells of the corpus luteum. J Cell Sci. 2011; 124: 2501–10. 10. Du L, Dong F, Guo L, et al. Interleukin-1beta increases permeability and upregulates the expression of vascular endothelial-cadherin in human renal glomerular endothelial cells. Mol Med Rep. 2015; 11: 3708–14. 6. Alexander JS, Alexander BC, Eppihimer LA, et al. Inﬂammatory mediators induce sequestration of VE-cadherin in cultured human endothelial cells. Inﬂammation. 2000; 24: 99–113. 15. Kataoka H, Murayama T, Yokode M, et al. A novel snail-related transcription factor Smuc regulates basic helix-loop-helix tran- scription factor activities via speciﬁc E-box motifs. Nucleic Acids Res. 2000; 28: 626– 33. 11. 11. Li Z, Jimenez SA. Protein kinase Cdelta and c-Abl kinase are required for transforming growth factor beta induction of endothelial- mesenchymal transition in vitro. Arthritis Rheum. 2011; 63: 2473–83. 7. Tran TH, Dolecek C, Pham PM, et al. Dihy- droartemisinin-piperaquine against mul- tidrug-resistant Plasmodium falciparum malaria in Vietnam: randomised clinical trial. Lancet. 2004; 363: 18–22. 16. Smith AP, Verrecchia A, Faga G, et al. ª 2017 The Authors. Journal of Cellular and Molecular Medicine published by John Wiley & Sons Ltd and Foundation for Cellular and Molecular Medicine. ª 2017 The Authors Journal of Cellular and Molecular Medicine published by John Wiley & Sons Ltd and Foundation for Cellular and Molecular Medicine ª 2017 The Authors. References A positive role for Myc in TGFbeta-induced Snail transcription and epithelial-to- mesenchymal transition. Oncogene. 2009; 28: 422–30. 12. Vestweber D. VE-cadherin: the major endothelial adhesion molecule controlling cellular junctions and blood vessel forma- tion. Arterioscler Thromb Vasc Biol. 2008; 28: 223–32. 8. Eckstein-Ludwig U, Webb RJ, Van Goethem ID, et al. Artemisinins target the SERCA of Plasmodium falciparum. Nature. 2003; 424: 957–61. 2032
https://openalex.org/W4224211512	https://repository.ubn.ru.nl//bitstream/handle/2066/249483/249483.pdf	English	null	Case Series on the Long-Term Effect of Three Different Types of Maxillary Implant-Supported Overdentures on Clinical Outcomes and Complications	Journal of clinical medicine	2,022	cc-by	9,715	Case Series on the Long-Term Effect of Three Different Types of Maxillary Implant-Supported Overdentures on Clinical Outcomes and Complications Naeini, E.N.; Bruyn, H. de; Bronkhorst, E.M.; D'Haese, J. 2022, Article / Letter to editor (Journal of Clinical Medicine, 11, 8, (2022), article 2251) Doi link to publisher: https://doi.org/10.3390/jcm11082251 rticle Case Series on the Long-Term Effect of Three Different Types of Maxillary Implant-Supported Overdentures on Clinical Outcomes and Complications itis Natali Naeini 1,* , Hugo De Bruyn 1,2 , Ewald M. Bronkhorst 1 and Jan D’haese 1 1 Department of Dentistry, Radboud University Medical Centre, 6525 GA Nijmegen, The Netherlands; hugo.debruyn@radboudumc.nl (H.D.B.); ewald.bronkhorst@radboudumc.nl (E.M.B.); jan.dhaese@radboudumc.nl (J.D.) 1 Department of Dentistry, Radboud University Medical Centre, 6525 GA Nijmegen, The Netherlands; hugo.debruyn@radboudumc.nl (H.D.B.); ewald.bronkhorst@radboudumc.nl (E.M.B.); jan.dhaese@radboudumc.nl (J.D.) 1 Department of Dentistry, Radboud University Medical Centre, 6525 GA Nijmegen, The Netherlands; hugo.debruyn@radboudumc.nl (H.D.B.); ewald.bronkhorst@radboudumc.nl (E.M.B.); jan.dhaese@radboudumc.nl (J.D.) j 2 Department of Periodontology and Oral Implantology, Faculty of Medicine an University of Ghent, 9000 Gent, Belgium j 2 Department of Periodontology and Oral Implantology, Faculty of Medicine and Health Sciences, University of Ghent, 9000 Gent, Belgium * Correspondence: natali n@hotmail co uk or natali naeini@radboudumc nl y g * Correspondence: natali_n@hotmail.co.uk or natali.naeini@radboudumc.nl y g * Correspondence: natali_n@hotmail.co.uk or natali.naeini@radboudumc.nl Abstract: (1) Long-term data on maxillary implant overdentures (IODs) are scarce. This case series evaluated three types of IODs supported by six, four or three implants (Anyridge®, Mega’Gen Implant Co., Ltd., Daegu, South-Korea), after 3–5 years in function. (2) A total of 31 patients, with 132 implants, were non-randomly allocated based on available bone or ﬁnancial limitations. IOD-6 received a telescopic overdenture; IOD-4 a bar; and IOD-3, non-connected implants with locator abutments. Implant survival, bone level changes, probing pocket depth (PPD), plaque index, bleeding on probing (BOP), and technical, biological and aesthetic complications were registered. Impact of suprastructures on bone loss and PPD was analyzed using mixed-effect linear regression models. Differences between groups were analyzed using the ANOVA test for BOP, and Kruskal Wallis test for complications. (3) In total, 23 patients participated in the follow-up (9 female, 14 male), with average age of 62.2 years; 7, 11 and 5 patients in IOD-6, IOD-4 and IOD-3, respectively. Implant survival after 4.4 years on average, was 98% in total; 100%, 97.8% and 93.3% for IOD-6, IOD-4 and IOD-3, respectively. Mean bone loss corresponded to 0.68 mm (SD 1.06, range −4.57–1.51), 0.39 mm (SD 1.06, range −3.6–2.43), and 1.42 mm (SD 1.68, range −5.11–0.74) for IOD-6, IOD-4 and IOD-3, respectively. A statistically signiﬁcant difference was seen in bone level when comparing IOD-6 to IOD-3 (p = 0.044), and IOD-4 to IOD-3 (p = 0.018). Citation: Naeini, E.N.; De Bruyn, H.; Bronkhorst, E.M.; D’haese, J. Case Series on the Long-Term Effect of Three Different Types of Maxillary Implant-Supported Overdentures on Clinical Outcomes and Complications. J. Clin. Med. 2022, 11, 2251. https://doi.org/10.3390/ jcm11082251 Academic Editors: Stefan Vandeweghe and Mieszko Wieckiewicz Received: 14 March 2022 Accepted: 16 April 2022 Published: 18 April 2022 Citation: Naeini, E.N.; De Bruyn, H.; Bronkhorst, E.M.; D’haese, J. Case Series on the Long-Term Effect of Three Different Types of Maxillary Implant-Supported Overdentures on Clinical Outcomes and Complications. J. Clin. Med. 2022, 11, 2251. https://doi.org/10.3390/ jcm11082251 Keywords: dental implants; maxillary overdenture; prosthodontics; clinical outcome; peri-implantitis Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- iations. Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- iations. rticle Case Series on the Long-Term Effect of Three Different Types of Maxillary Implant-Supported Overdentures on Clinical Outcomes and Complications Mean PPD was 3.8 mm (SD: 0.69; range 2.5–5.3), 3.5 mm (SD 0.59; range 2.33–5), and 3.2 mm (SD 0.56; range 2–4) for IOD-6, IOD-4 and IOD-3, respectively, and differed signiﬁcantly between IOD-6 and IOD-3 (p = 0.029). Incidence of peri-implantitis was 1%. No differences were seen for complications between groups. (4) Maxillary IOD supported by four to six implants is the most reliable treatment regarding implant survival and peri-implant health. More research is needed in the clinical outcomes, in particular the peri-implant health, and complications of maxillary IODs, especially with a reduced number of implants. Note: To cite this publication please use the final published version (if applicable). Journal of Clinical Medicine Journal of Clinical Medicine Journal of Clinical Medicine 1. Introduction This, too, is dependent on several factors, including patient genetic disorders, patient smoking and biomechanical factors such as cement or impression material remnants in the peri-implant sulcus [10]; bacterial contamination of the implant components in the presence of a microgap at the interface between the ﬁxture and abutment [11]; and technical issues such as loose screws, mobile components and fractured materials [10]. Cement remnants in the soft tissues is an example of problems associated with clinical handling. Cement remnants can lead to marginal bone resorption and infection due to a foreign body reaction that may also be linked to the leakage of titanium, which is inevitable around oral implants [10]. However, a recent 10-year retrospective study compared the long-term survival and complication rate of screw-retained vs. cement-retained single implant crowns, and found mean bone loss to be signiﬁcantly greater for the screw-retained group [12] g y g g p [ ] Peri-implantitis is deﬁned as a plaque-associated pathological condition occurring in tissues around dental implants, characterized by inﬂammation in the peri-implant mucosa and subsequent progressive loss of supporting bone [13]. According to the 2017 Consen- sus report of the World Workshop on the Classiﬁcation of Periodontal and Peri-Implant Diseases and Conditions [13], the diagnosis of peri-implantitis requires the presence of bleeding and/or suppuration, increased probing depths compared to previous examina- tions and crestal bone loss beyond normal initial bone remodeling [13]. In the absence of previous examination data, peri-implantitis should then be deﬁned by the presence of bleeding and/or suppuration, in combination with probing depths ≥6 mm, and bone levels ≥3 mm apical to the most coronal portion of the intraosseous part of the implant [13]. The prevalence of peri-implantitis varies signiﬁcantly among studies due to inconsistent deﬁnitions, various reporting methods and different study characteristics. A recent critical review reported that the prevalence of peri-implantitis, after a mean follow-up time of 5 years, was between 0% and 39.7%, based on 15 different case deﬁnitions [14]. y In terms of long-term complications of IODs, veneer fractures are reported to be the most frequent technical complication during a follow-up period of 5 to 15 years, followed by abutment screw loosening/fractures and framework fractures [6]. Biological complications are not always reported and if they are mentioned in studies, they are described generally as ‘soft-tissue inﬂammation’ [15]. Most of the time, a high level of patient-centered outcomes have been observed [16]. 1. Introduction For over a century, the standard of care for edentulous patients has been a conventional complete removable denture. This allowed edentulous patients to eat, speak and function again, albeit often with issues of reduced comfort and impaired well-being due to lack of stability or retention of the denture. As the alveolar ridge reduces over time, this results in social, psychological and functional disabilities [1,2]. The use of dental implants for oral rehabilitation has become a highly predictable treatment for fully and partially edentulous patients. According to the literature, good short- and long-term results have been reported for various treatment indications with an implant survival that ranges from 82–98% [1–4]. However, it has often been the case that implants in the maxilla have a lower success rate Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). https://www.mdpi.com/journal/jcm J. Clin. Med. 2022, 11, 2251. https://doi.org/10.3390/jcm11082251 J. Clin. Med. 2022, 11, 2251 2 of 13 than in the mandible [5]. Nowadays, the use of overdentures retained on two implants (IOD-2) has, especially in the mandible, become a reliable and often standardized treatment option for improving retention [2]. Given the historically lower implant survival in the maxilla, a full-arch ﬁxed prosthesis (FAFDP on four to six implants, is the standard of care with a 15-year implant survival of 90.9% [6]). From a cost–beneﬁt point of view, this is a costly treatment, not affordable to many patients. Hence, cheaper solutions for the maxilla have been introduced. The all-on-four concept, using a ﬁxed 10-unit bridge on four implants, yielded survival of 98% after 5 years [7]. Implant treatment success is predominantly based on the stability of the bone sur- rounding the implant. According to Ästrand et al. [8], major changes in peri-implant bone level can take place between implant placement and prosthetic loading. This peri-implant bone remodeling occurs in order to re-establish the biological width, especially in patients or at sites with thin, soft tissues [9]. This early remodeling is dependent on multiple factors: type of implant used, surgery and prosthetic aspects. Bone loss in the long term is most often a consequence of peri-implant diseases, also inducing pocket formation and suppura- tion. 2.1. Patient Selection and Treatment Allocation Originally, 31 referred patients presenting with a complaint of lack of retention in their removable maxillary denture, were included in this clinical case series, all of whom had a sufﬁcient amount of bone to install at least 3 implants as conﬁrmed on CBCT. The patients were in good health. If necessary, patients were scheduled for periodontal treatment of the residual mandibular teeth prior to implant placement. Smokers up to 10 cigarettes/day were also included. Patients suffering from severe osteoporosis and/or non-controlled diabetes mellitus or taking anticoagulants that could not easily be substituted were also excluded. A healing period of approximately 2 months after tooth extraction was allowed before implant surgery. Hence, initial post-extraction bone resorption occurred before surgery. This study used a practice-based approach and group allocation was not random- ized. Patients were allocated into one of the 3 treatment groups based on the availability of bone on the CBCT ﬁrstly, and secondly, based on their ﬁnancial possibility. So, in the case of sufﬁcient volume of bone on the CBCT, the patient was advised to receive 6 implants and restored with a type of telescopic overdenture (IOD-6). Where there was a lack of bone volume, or ﬁnancial restrictions from the patient, 4 implants were advised to support a bar-retained overdenture (IOD-4). As the last option, 3 non-splinted implants were placed, each with a locator abutment to retain the overdenture (IOD-3). All patients signed a written consent before treatment and approval was obtained from the ethical committee of the Ghent University hospital (ID B670201420643). 1. Introduction Apart from ﬁnancial restrictions, other limitations such as insufﬁcient bone volume or an unfavorable jaw relation can sometimes prevent the placement of a sufﬁcient number of implants required for a ﬁxed prosthesis [17,18]. Therefore, an IOD, retained on a reduced number of implants, could be a viable alternative for edentulous patients with compromised oral function. Maxillary IODs have previously been reported to have a relatively high rate of implant loss compared to other implant treatment modalities [17,19]. Implant number, length and inclination may be compromised by anatomic limitations and bone morphology [20]. Moreover, controversy persists regarding factors critical for implant and prosthetic success [21]. The number of J. Clin. Med. 2022, 11, 2251 3 of 13 3 of 13 implants in a maxillary IOD affects the survival rate, due to the forces on the overdenture being distributed by the bone surrounding the implants. Thus, the more implants, the more evenly the forces are distributed [22]. There are no speciﬁc guidelines concerning the number of implants necessary to retain a maxillary overdenture [23]. Bergendal et al. [17] compared two IODs in both the maxilla and the mandible and described a cumulative implant survival rate of 75.4% and 100% respectively, after 7 years in function; it is obvious that this IOD-2 yields too many failures and complications to be advisable. On the other hand, for an IOD without palatal coverage, a minimum of four implants is favorable [21,24–26]. These recommendations are further supported by recent studies by Doornewaard et al. [27], who reported 96% implant survival up to 4 years, for maxillary IODs retained on a bar on four implants. p Currently, long-term data on maxillary IODs with various designs of retention are scarce. The aim of the present study was to evaluate the effect of three different types of maxillary overdentures on the clinical outcome, including survival, peri-implant bone loss, peri-implant health and complications of implants in the edentulous maxilla, after 3–5 years in function. 2.2. Surgical and Prosthetic Procedure The implants used in this case series (Anyridge®, Mega’Gen Implant Co. Ltd., Daegu, Korea) had a width varying from 3.5 to 5.5 mm and a length varying from 7 to 13 mm and a 5◦morse taper internal connection. It has a moderately rough surface with an aver- age Sa value of 1.3 µm and a 0.8 mm thread pitch. All the implants were placed using a conventional one-stage surgical approach by the same periodontist (JD). Following local anaesthesia, a crestal incision was made to elevate the full-thickness mucoperiostal ﬂap and expose the bone. Using a duplicate of the pre-existing denture as a surgical guide plate (with palatal support), the osteotomies were prepared at 1500 rpm with abundant irrigation. All implants were planned to be placed equicrestally or slightly subcrestal, using a maximum insertion torque of 35 Ncm. Healing abutments were connected using hand torque (±15 Ncm), prior to ﬂap closure. Peri-apical intraoral radiographs were obtained by the surgeon immediately after implant placement (baseline), with commercially available ﬁlm holders using the parallel long-cone technique in order to visualize the implant threads and alveolar bone level. A conventional relined complete denture served as a provisional J. Clin. Med. 2022, 11, 2251 4 of 13 solution. A prescription was given for Ibuprofen 600 mg (3 times a day, for as long as necessary), Amoxicillin 1 g (2 times a day for 5 days) and 0.12% chlorhexidine mouth rinse (2 times a day, for 2 weeks). One week after surgery, patients returned for suture removal, and afterwards they returned for regular post-operative maintenance, including relining of the denture if necessary. Three months after surgery, the stability of the implants was clinically evaluated. A ﬁrst impression was made using an irreversible hydrocolloid material (Cavex, Haarlem, The Netherlands). Based on this impression, an individual open tray was made and an impression on abutment ((IOD-6, IOD-4) or implant (IOD-3) level was obtained using a polyether impression material (Impregum, 3 M, ESPE, St. Paul, MN, USA). After a screw- retained jaw registration, try-in and additional ﬁt of the metallic structure (for IOD-6 and IOD-4), placement of the ﬁnal prosthesis was carried out. During placement of the ﬁnal restoration the occlusion was carefully checked and hygiene instructions were given to obtain good oral hygiene maintenance. Figure 1 shows the intra-oral implant position and suprastructures for each group, 3 months after placement. Figure 1. 2.2. Surgical and Prosthetic Procedure Restorative procedure for group IOD-6 (A), IOD-4 (B) and IOD-3 (C), showing the intra- oral implant position and suprastructures 3 months after placement, ﬁnal prosthesis and periapical radiographs obtained after the placement of ﬁnal prosthesis. Figure 1. Restorative procedure for group IOD-6 (A), IOD-4 (B) and IOD-3 (C), showing the intra- oral implant position and suprastructures 3 months after placement, ﬁnal prosthesis and periapical radiographs obtained after the placement of ﬁnal prosthesis. 2.3. Clinical Assessment Patients were assessed at the implant center after 6 months and 1 year, after which they returned to their referring dentist for regular maintenance. As part of this research investigation, the originally treated patients were invited to participate in a clinical research investigation after at least 3 years in function. An external periodontal specialist (ENN) from the University of Ghent, not involved in the original surgical interventions, evaluated all the listed clinical and radiographic parameters. The peri-implant health was evaluated using the following parameters: probing pockets depth (PPD), plaque score using the Silness—Loë plaque index, and bleeding on probing (BoP) using the modiﬁed bleeding index (mBI) by Mombelli et al. [28]. In this study, the periodontal parameters were later dichotomized into a variable of 1 or 0, so that when an implant presented with any bleeding, it was subsequently given a score of 1, and 0 when no bleeding was observed. This score was then divided by the number of implants to give the relative number of implants with bleeding per patient. An implant was classiﬁed as a failure when it was removed because of mobility, loss of integration, ongoing bone loss, signs of infection and/or patient discomfort [29]. Early failures (prior to loading) have been excluded from this study. Baseline peri-apical radiographic exposures were obtained immediately after implant insertion. The bone level was measured per treatment group, at baseline and at the time J. Clin. Med. 2022, 11, 2251 5 of 13 5 of 13 of scientiﬁc recall (3–5 years after implant surgery). The software used was AxioVision Rel. 4.8 (Carl Zeiss MicroImaging GmbH, Germany), using an accuracy of 0.01 mm. The known distance between the implant thread (0.8 mm) was used for the calibration of the radiographs. The implant–abutment connection was determined as the baseline reference point (0 mm) from which point the closest bone implant contact was measured (Figure 2). Both mesial and distal bone level was measured from the implant. Figure 2. Radiograph obtained at the day of implant placement (A), and after 3–5 years of function (B). The yellow arrow indicates the reference point located at implant–abutment interface; the red arrow indicates the bone-to-implant contact level. Figure 2. Radiograph obtained at the day of implant placement (A), and after 3–5 years of function (B). The yellow arrow indicates the reference point located at implant–abutment interface; the red arrow indicates the bone-to-implant contact level. 2.3. Clinical Assessment Complications were divided into technical (implant fracture, screw/abutment frac- ture, screw/abutment loosening, loss of retention), aesthetic (fracture of veneering mate- rial/framework of prosthesis) and biological (mucositis, peri-implantitis). 2.4. Statistical Analysis Implant number was used as the unit for the statistical analysis on bone loss, PPD and BoP. The impact of different suprastructures on bone loss and PPD was analyzed using mixed-effect linear regression models that accounted for the clustering of implants per patient by incorporating a random intercept for variable patient. In both analyses, the baseline measurement was included in the model as independent variable, together with the variable suprastructure, using IOD-6 as a reference group. These two analyses were per- formed with the library lme4 in R version 3.6.2. The comparison of the scoring of bleeding upon probing between the suprastructures was carried out by means of an ANOVA test, while the comparisons of the complications per group was done with the Kruskal Wallis test. The ANOVA and the Kruskal Wallis tests were carried out using IBM®SPSS® 25.0 (SPSS Inc., Chicago, IL, USA). 3.2. Bone Level and Probing Pocket Depth The effect of the suprastructure on bone loss was calculated by using the bone level at baseline and the suprastructure as independent variables. IOD-6 was used as the reference for the mixed model analysis, as shown in Table 2. No statistically signiﬁcant difference was seen when comparing IOD-6 and IOD-4 (p = 0.828). However, a statistically signiﬁcant difference was seen when comparing IOD-6 and IOD-3 (p = 0.044). Initially, the mixed model analysis only used IOD-6 as the reference; however, to give more clarity to the reader, IOD-4 was also compared with IOD-3 and a statistically signiﬁcant difference was seen here as well (p = 0.018). Hence, IOD-3 shows more bone loss as compared to the other two treatment groups. Figure 3 shows a boxplot of the bone level at baseline and after 3–5 years, and the overall bone loss (mm) over time for all three treatment groups. With regards to other clinical parameters, there was a statistically signiﬁcant difference in probing pocket depths between IOD-6 and IOD-3 (p = 0.029). Table 2. Table demonstrating the effect of suprastructure on bone loss and probing pocket depth (PPD). IOD-6 used as the reference for the mixed model analysis. When comparing IOD-6 to IOD-3, a statistically signiﬁcant difference was seen for the effect of the suprastructure on the bone level (p = 0.044), as well as on PPD (p = 0.029). Table 2. Table demonstrating the effect of suprastructure on bone loss and probing pocket depth (PPD). IOD-6 used as the reference for the mixed model analysis. When comparing IOD-6 to IOD-3, a statistically signiﬁcant difference was seen for the effect of the suprastructure on the bone level (p = 0.044), as well as on PPD (p = 0.029). Table 2. Table demonstrating the effect of suprastructure on bone loss and probing pocket depth (PPD). IOD-6 used as the reference for the mixed model analysis. When comparing IOD-6 to IOD-3, a statistically signiﬁcant difference was seen for the effect of the suprastructure on the bone level (p = 0.044), as well as on PPD (p = 0.029). Variable Effect 95% CI p-Value Bone Loss (mm) Intercept −0.030 [−0.642 . . . 0.583] 0.925 Bone level at baseline 0.074 [−0.142 . . . 0.290] 0.503 IOD-6 vs. IOD-4 0.085 [−0.685 . . . 0.855] 0.828 IOD-6 vs. IOD-3 −0.985 [−1.940 . . . −0.028] 0.044 IOD-4 vs. IOD-3 * −1.070 [−1.959 . . 3.1. Clinical Outcome Bone Level and Probing Pocket Depth The effect of the suprastructure on bone loss was calculated by using the bone level at baseline and the suprastructure as independent variables. IOD-6 was used as the reference for the mixed model analysis, as shown in Table 2. No statistically signiﬁcant difference was seen when comparing IOD-6 and IOD-4 (p = 0.828). However, a statistically signiﬁcant difference was seen when comparing IOD-6 and IOD-3 (p = 0.044). Initially, the mixed model analysis only used IOD-6 as the reference; however, to give more clarity to the reader, IOD-4 was also compared with IOD-3 and a statistically signiﬁcant difference was seen here as well (p = 0.018). Hence, IOD-3 shows more bone loss as compared to the other two treatment groups. Figure 3 shows a boxplot of the bone level at baseline and after 3–5 years, and the overall bone loss (mm) over time for all three treatment groups. With regards to other clinical parameters, there was a statistically signiﬁcant difference in probing pocket depths between IOD-6 and IOD-3 (p = 0.029). Table 2. Table demonstrating the effect of suprastructure on bone loss and probing pocket depth (PPD). IOD-6 used as the reference for the mixed model analysis. When comparing IOD-6 to IOD-3, a statistically signiﬁcant difference was seen for the effect of the suprastructure on the bone level (p = 0.044), as well as on PPD (p = 0.029). Variable Effect 95% CI p-Value Bone Loss (mm) Intercept −0.030 [−0.642 . . . 0.583] 0.925 Bone level at baseline 0.074 [−0.142 . . . 0.290] 0.503 IOD-6 vs. IOD-4 0.085 [−0.685 . . . 0.855] 0.828 IOD-6 vs. IOD-3 −0.985 [−1.940 . . . −0.028] 0.044 IOD-4 vs. IOD-3 * −1.070 [−1.959 . . . −0.180] 0.018 PPD (mm) Intercept 3.754 [3.427 . . . 4.080] <0.001 IOD-6 vs. IOD-4 0.188 [−0.612 . . . 0.237] 0.387 IOD-6 vs. IOD-3 −0.599 [−1.136 . . . −0.062] 0.029 IOD-4 vs. IOD-3 * −0.411 [−0.917 . . . −0.094] 0.111 * Additional comparison has been added for clarity to the reader to see the difference between these groups as well. Table 1. Table showing descriptive results of the population and clinical outcomes at baseline and at time of follow-up. * Additional comparison has been added for clarity to the reader to see the difference between these groups as well. 3.1. Clinical Outcome Total Patient Group IOD-6 IOD-4 IOD-3 Number of patients Baseline 31 9 12 10 Follow-up 23 7 (−1 §) 11 (10 + 1 §) 5 Number of implants Baseline 132 54 48 30 Follow-up 102 42 46 (44 + 2 §) 15 Bone level at baseline (mm) Mean (SD; range) 0.55 (0.74; −1.06–2.48) 0.70 (0.77; −1.06–2.13) 0.45 (0.77; −0.87–2.48) 0.39 (0.39; −0.50–1.01) Bone level at follow-up (mm) Mean (SD; range) −0.10 (1.01; −4.49–2.14) 0.02 (0.91; −2.8–1.81) 0.06 (0.70; −1.33–2.14) −1.03 (1.61; −4.49–1.15) Bone loss (mm) Mean (SD; range) 0.65 (1.20; −2.44–5.11) 0.68 (1.06; −1.51–4.57) 0.39 (1.06; −2.43–3.6) 1.43 (1.68; −0.74–5.11) PPD (mm) Mean (SD; range) 3.6 (0.66; 2.0–5.3) 3.8 (0.69; 2.5–5.3) 3.5 (0.59; 2.33–5) 3.2 (0.56; 2–4) BOP (relative %) Mean (SD; range) 61.6% (31.1%; 0–100%) 78.6% (24.9%; 33.0–100%) 54.5% (33.2%; 0–100%) 53.3% (29.8%; 33.0–100%) (§) represents 1 patient originally selected for IOD-6 switched to an overdenture on a bar on 6, instead of 4 implants. Table 1. Table showing descriptive results of the population and clinical outcomes at baseline and at time of follow-up. Table 1. Table showing descriptive results of the population and clinical outcomes at baseline and at time of follow-up. 3.2. Bone Level and Probing Pocket Depth 3.2. Bone Level and Probing Pocket Depth 3.1. Clinical Outcome A total of 132 maxillary implants were originally inserted in 31 patients (17 male, 14 female); 3 implants failed prior to loading and were replaced, and 1 failed after loading; 23 patients (102 implants; 62.2 years (SD 6.9, range 48–77) participated in the follow-up study (9 female, 14 male). Of the eight patients who dropped out, ﬁve were untraceable, two refused to participate and one was unable to attend due to illness. One patient originally selected for IOD-6, requested a bar-retained overdenture be made, and was later allocated to prosthetic group IOD-4. The mean follow-up time was 4.4 years (range 3.4–5.1), and comparable between groups. The age of the subjects was comparable between groups as well. With the limitation of the dropouts (8/31) for which no information is available, the current overall survival rate was 98% for the whole patient group, and 100%, 97.8% and 93.3% for IOD-6, IOD-4 and IOD-3, respectively. Table 1 summarizes the descriptive results for clinical outcomes at baseline and at time of follow-up. The fraction of bleeding implants for IOD-6, IOD-4 and IOD-3 were 78.6%, 54.5% and 53.3%, respectively (p = 0.229). J. Clin. Med. 2022, 11, 2251 6 of 13 Table 1. Table showing descriptive results of the population and clinical outcomes at baseline and at time of follow-up. Total Patient Group IOD-6 IOD-4 IOD-3 Number of patients Baseline 31 9 12 10 Follow-up 23 7 (−1 §) 11 (10 + 1 §) 5 Number of implants Baseline 132 54 48 30 Follow-up 102 42 46 (44 + 2 §) 15 Bone level at baseline (mm) Mean (SD; range) 0.55 (0.74; −1.06–2.48) 0.70 (0.77; −1.06–2.13) 0.45 (0.77; −0.87–2.48) 0.39 (0.39; −0.50–1.01) Bone level at follow-up (mm) Mean (SD; range) −0.10 (1.01; −4.49–2.14) 0.02 (0.91; −2.8–1.81) 0.06 (0.70; −1.33–2.14) −1.03 (1.61; −4.49–1.15) Bone loss (mm) Mean (SD; range) 0.65 (1.20; −2.44–5.11) 0.68 (1.06; −1.51–4.57) 0.39 (1.06; −2.43–3.6) 1.43 (1.68; −0.74–5.11) PPD (mm) Mean (SD; range) 3.6 (0.66; 2.0–5.3) 3.8 (0.69; 2.5–5.3) 3.5 (0.59; 2.33–5) 3.2 (0.56; 2–4) BOP (relative %) Mean (SD; range) 61.6% (31.1%; 0–100%) 78.6% (24.9%; 33.0–100%) 54.5% (33.2%; 0–100%) 53.3% (29.8%; 33.0–100%) (§) represents 1 patient originally selected for IOD-6 switched to an overdenture on a bar on 6, instead of 4 implants. 3.2. 3.2. Bone Level and Probing Pocket Depth . −0.180] 0.018 PPD (mm) Intercept 3.754 [3.427 . . . 4.080] <0.001 IOD-6 vs. IOD-4 0.188 [−0.612 . . . 0.237] 0.387 IOD-6 vs. IOD-3 −0.599 [−1.136 . . . −0.062] 0.029 IOD-4 vs. IOD-3 * −0.411 [−0.917 . . . −0.094] 0.111 * Additi l i h b dd d f l it t th d t th diff b t th ll 7 of 13 J. Clin. Med. 2022, 11, 2251 Figure 3. Boxplot of the median bone level in mm at baseline (start) and after 3–5 years (end) for all 3 treatment groups, and the overall bone loss (change). Figure 3. Boxplot of the median bone level in mm at baseline (start) and after 3–5 years (end) for all 3 treatment groups, and the overall bone loss (change). 3.3. Peri-Implant Health The incidence of peri-implantitis in the study population is 1% according to the 2017 The consensus report [13], as one of the implants showed bone levels of 3 mm apical to the most coronal portion of the intraosseous part of the implant, as well as a probing depth of >5 mm combined with bleeding on probing, as shown in Table 3. Table 3. Table showing number of implants with corresponding bone loss (mm) and probing pocket depth (mm). The number in brackets is the number of implants that showed bleeding on probing, followed by the percentage of implants which showed bleeding out of the total number of implants (n = 102). Only one of the implants was classiﬁed as having peri-implantitis, denoted in bold. Pocket Depth Bone Loss ≤3 mm >3 mm and ≤4 mm >4 mm and ≤5 mm >5 mm Total <0 mm 5 (2,1.9%) 23 (12, 11.8%) 4 (3, 2.9%) 1 (1, 0.9%) 33 ≤1 mm 8 (4, 3.9%) 23 (17, 16.7%) 5 (4, 3.9%) 0 36 >1 ≤2 mm 7 (4, 3.9%) 9 (7, 6.9%) 3 (3, 2.9%) 0 19 >2 ≤3 mm 2 (0) 9 (5, 4.9%) 2 (2, 1.9%) 1 (1, 0.9%) 14 Total 22 (10, 10.2%) 64 (41, 41.8%) 14 (12, 12.2%) 2 (2, 1.9%) 102 3 4 Complications 4. Discussion Currently there is no consensus regarding the number of implants required for an implant-supported maxillary overdenture. The present study assessed the clinical outcomes and complications with three types of IODs retained on three, four or six implants, after 3 to 5 years of function. 3.4. Complications Table 4 shows each complication encountered per patient, per treatment group, from the time of implant placement to the time of follow-up. For six patients in IOD-6, a technical complication occurred. In these cases, the patients incurred a loss of retention rubbers (resolved by changing the retention rubbers), and a crown fracture (patients 5 and 6). One patient experienced a biological complication, peri-implantitis, and was treated with antibiotics (patient 1). For IOD-4, 1 patient experienced a biological complication whereby J. Clin. Med. 2022, 11, 2251 8 of 13 8 of 13 the implant had not osseointegrated (patient 12). This implant was removed and a new implant replaced it. Four patients experienced aesthetic complications in wear of the IOD. No intervention was carried out for these patients. Two biological complications were reported in group IOD-3; one patient had peri-implantitis and was treated by carrying out ﬂap surgery (patient 19); another patient experienced late implant failure and the implant was removed (patient 23). Three patients experienced technical complications, including prosthetic fracture (patients 20, 21, 23), and had new prostheses made. Loss of retention rubbers were also reported and replaced (patients 21 and 23). The Kruskal Wallis test was used for analysis between groups with an uneven distribution; there was no statistically signiﬁcant difference. Table 4. Table showing distribution of all complications encountered per patient, per treatment group, at the time of follow-up, and corresponding mean bone loss (a negative number indicates bone gain) and probing pocket depth (PPD). and probing pocket depth (PPD). Complications Suprastructure Patient ID Total No. Technical Biological Aesthetic Mean Bone Loss (mm) Mean PPD (mm) IOD-6 1 1 1 1.84 4.42 2 2 2 0.60 3.69 3 2 2 −0.06 3.86 4 4 4 0.72 3.03 5 1 1 0.77 3.53 6 2 1 0.25 3.14 7 1 1 0.61 4.61 IOD-4 8 0.98 2.92 9 2 2 −0.01 3.58 10 1 1 0.16 4.08 11 1.15 3.38 12 2 1 * 1 −1.05 3.88 13 0.14 4.29 14 1 1 0.64 3.75 15 −0.20 3.42 16 1 1 0.18 3.38 17 2.10 3.21 18 −0.12 3.42 IOD-3 19 1 1 3.49 3.44 20 1 1 1.83 3.33 21 4 4 −0.67 3.11 22 1.91 2.72 23 3 2 1 ** 0.20 3.17 * Implant lost before functional loading and replaced. ** Implant lost during follow-up. 4.1. Survival Three implants failed before functional loading and were replaced. Uncontrolled premature contacts of the healing abutments with the provisionally relined prosthesis could be a possible reason for this failure. Although there is no evidence for the suggestion that bruxism may cause an overload on dental implants, this could be a plausible explanation for the fact that the other implant failed after functional loading of an overdenture with a locator system, in a patient with bruxism. However, it remains difﬁcult to draw strong conclusions regarding the cause of implant failure. In the present study, two implants failed J. Clin. Med. 2022, 11, 2251 9 of 13 after loading, resulting in an overall survival rate of 98% after a mean follow-up period of 4.4 years. This is in accordance with a systematic review by Slot et al. [22], where survival rates of >95% were reported. Implant survival for IOD-6, IOD-4 and IOD-3 was 100%, 97.8% and 93.3%, respectively, suggesting an intense relation between number of implants and risk of failure. A possible explanation could be the design of the anchorage system. Meijer et al. [30] found that in the case of a bar structure between the implants, the force distribution is spread to the bone surrounding the bar on two implants, whereas when solitary ball attachments are loaded, the forces are distributed to the surrounding bone of a single implant. Raghoebar et al. [31] described in a systematic review an implant survival rate of 98.1% after 1 year in a case of six or more implants and a splinted anchorage. In a case of less than four implants and a splinted anchorage or a non-splinted anchorage, an implant survival rate of 97.0% or 88.9% was observed, respectively. They concluded that an implant-supported (less than four implants) maxillary overdenture provided with a splinted anchorage is accompanied with a high implant survival rate, while there is an increased risk of implant loss when less than four implants with a non-splinted anchorage are used. This is in accordance with the present study. Higher survival rates of 100% and 99.2% were reported in another RCT comparing four and six implants, respectively, supporting splinted maxillary overdentures after an observation period of 5 years [32]. It can be argued that in this RCT, all included patients initially had enough bone to receive six implants; however, for the purpose of the study, some received only four [32]. 4.1. Survival The current study is not only not an RCT, but uses a practice-based approach, which describes realistic situations whereby patients are treated with fewer implants due to not having enough bone volume to being with. Hence the high failure rate with reduced number of implants may also reﬂect the more critical bone condition of the treated patients in the locator-retained group (IOD-3). Conversely, lower survival rates of 82.3% were found in a prospective study evaluating free-handed ﬂaplessly placed one-piece maxillary mini dental implants with ball attachments, supporting a metal-reinforced, horse-shoe removable denture, after just 2 years in function [33]. In these cases, the patients had reduced bone volume to begin with and could not be treated with conventional implant diameters [33]. A drawback of the present study is the small sample size, due to the study site being a referral center whereby general dentists would refer the patient back to the surgeon only when issues occurred. 4.2. Bone Loss Between baseline (i.e., implant insertion) and an average of 4.4 years in function, a mean bone loss of 0.65 mm was observed for the whole patient group, and 0.68 mm, 0.39 mm and 1.42 mm for IOD-6, IOD-4 and IOD-3, respectively. Bone loss around implants can occur for several reasons, including as a consequence of initial bone remodeling [8] and of peri-implant disease. In this study, we controlled certain co-factors, which could affect the bone level; implant type and implant surgery was comparable for the three treatment modalities and all IODs were screw-retained. The latter may explain the steady state in bone level over time and only 1% incidence of peri-implantitis. The reported bone loss between implant insertion and long-term follow-up in this study includes the initial bone remodeling and additional bone loss after functional loading. This may therefore cause an overestimation in comparison with many clinical studies that do not take initial bone remodeling into account because they consider the time of loading, being many months after bone healing took place, as baseline for bone loss calculation. One notable ﬁnding in this study is the statistically signiﬁcant effect of the suprastructure on the bone levels over time, namely when comparing non-splinted, locator attachment–supported overdentures (IOD-3) to the splinted alternatives (IOD-4 and IOD-6). A mean bone loss of 1.42 mm was reported for IOD-3. This signiﬁcant difference could be related to the movement of the denture caused by having only three implants with locator attachments. Pressure of the denture on the mucosa, in combination with uneven force distribution, could explain this, as also suggested by Kamei et al. [34]. The latter used ﬁnite element analysis and found that maxillary denture retained on four implants results in less displacement than in models J. Clin. Med. 2022, 11, 2251 10 of 13 10 of 13 with fewer implants. Stress is also reduced when implants are inserted in the premolar area [34]. Similar conclusions were drawn from a study carried out by Dimililer et al. [35], aiming to ﬁnd the optimal implant location, number and diameter to support a maxillary implant-supported overdenture. The implant diameter had no signiﬁcant effect on stresses; however, as the number of implants increased, decreased stress values were observed in the peri-implant bone and implants [35]. It is well established that using four implants to support a more rigid suprastructure, may overcome this possible loading effect [31,34,35]. 4.3. Peri-Implant Health In the current study, peri-implant bone loss was not inﬂuenced by plaque and bleeding. This corroborates with studies that have also found no relation between the probing depth or bleeding and the mean bone loss, mean follow-up time, and reported prevalence of peri- implantitis [14,37]. Verhoeven et al. [38] reported a rather poor speciﬁcity and sensitivity for the plaque and bleeding index and considered these periodontal parameters as unreliable for clinical evaluation in implant dentistry. They suggest that radiographs remain the most important source of information to assess peri-implant bone level changes around dental implants. Generally speaking, the mean pocket depth for this study was low as compared to other studies [14]. A statistically signiﬁcant difference was seen between IOD-6 and IOD-3 (p = 0.029). IOD-6 showed a higher relative number of implants showing mucositis as compared to IOD-4 and IOD-3. The total score for the total patient group was 62%. Despite IOD-6 showing a higher fraction of bleeding implants in this study, the Anova test showed no statistically signiﬁcant difference when comparing the effect of suprastructure on bleeding on probing (p = 0.229). A plausible reason for this could be that IOD-6 has more implants than other groups, and so a higher relative bleeding score is to be expected, as well as increased difﬁculty maintaining oral hygiene the more implants there are. Currently, literature reporting on the epidemiology as opposed to patient-related outcomes and prosthetic complications of peri-implant diseases in this speciﬁc group of patients is scarce. Onclin et al. [39] reported patient-level incidence of peri-implant mucositis of 37.7% after 5 years and 64.6% after 10 years. When peri-implantitis is deﬁned according to the aforementioned criteria [13], then one implant was reported as having peri-implantitis in this study. In a recent review [19], the prevalence of peri-implantitis on implant level ranged from 0–40%. The deﬁnition of peri-implantitis varies considerably between studies, mostly due to dissimilar thresholds for bone loss. This obviously makes comparisons between studies difﬁcult. For example, Meyle et al. [40] reported a high prevalence of 24% after 10 years. The threshold for bone loss for the diagnosis for peri- implantitis was ‘any bone loss’, which explains the high reported prevalence. If one were to apply the guidelines of the 8th European Workshop on Periodontology on their material, the prevalence would be 0%. 4.2. Bone Loss Finally, another reason could be the lack of splinting in the IOD-3 system, although it is interesting to see that in a recent systematic review [36] comparing splinted vs. unsplinted designs for a maxillary overdenture supported by four implants, no statistical difference was detected in the survival rate of implants between the splinted implant group and the unsplinted implant group. It can be disputed that this systematic review included many RCT’s whereby implants were placed in patients who had sufﬁcient bone volume. 4.5. Limitations One drawback of the present study is the small sample size. This is due to the study being private practice-based research, with no ﬁnancial compensation for patients taking part in the study. The same can be said for the number of drop-outs, as the research is based on referred patients, which makes long-term follow-up more difﬁcult. For the same reason, one can only analyze the baseline radiographs available; therefore, not being able to guarantee that the follow-up radiographs would be obtained at the same angulation as baseline, which could have limited the study. However, only radiographs whereby the threads of the implants were clearly visible were analyzed, and calibration was always carried out using the thread pitch of the implant. Furthermore, peri-implant bone loss is often described as a circumferential disease, suggesting that the direction of the beam on the implant does not alter the result [43]. The lack of control in the condition of the antagonistic jaw was also a limitation, due to the limited number of cases. On the other hand, the denture design was not different in terms of occlusion and articulation pattern, and the prosthodontist strived to attain a balanced occlusion and articulation, thereby equally distributing the load on the dental implants. 4.4. Complications There was no statistically signiﬁcant difference in the effect of the suprastructures on complications experienced. Each IOD group reported technical, biological and aesthetic complications, which is comparable to what is reported in the literature; maxillary IODs are described to have a relatively high number of complications in general, especially during the ﬁrst year in function [21], with both technical and biological problems being observed. The majority of complications are related to the weakness of the anchorage components connecting bar and overdenture [41] or loosening and fracturing of the retention system [21]. Conversely, Slot et al. [32] reported that prosthetic complications were scarce, and only J. Clin. Med. 2022, 11, 2251 11 of 13 11 of 13 restricted to the repair of denture base and teeth when an IOD with built-in cobalt chromium reinforcement structure and gold retentive clips was used. This would explain the minimal prosthetic complications [42]. In this study, ﬁnancial limitations of the patients meant that such reinforcements could not be provided, and a simpler prothesis was created. 5. Conclusions A maxillary overdenture supported by four to six implants is a reliable treatment regarding survival of the implants and peri-implant health. Within the aforementioned limitations of the study, there was no clinical difference in IODs supported by four or six implants, which conﬁrms the existing evidence. More bone loss was observed when only three non-splinted implants support a maxillary overdenture, suggesting possible overloading. However, longer follow-up time is needed to assess whether this reaches a stable situation or instead leads to future complications such as ongoing bone loss, resulting in peri-implantitis or even implant loss. More research is needed in terms of complications, patient-centered outcomes and value-based care with various types of maxillary implant overdentures. Author Contributions: E.N.N.: conceptualization and research design, data collection, data analysis and interpretation, writing and drafting of article and statistical analysis; H.D.B.: conceptualization and research design, critical revision and ﬁnal approval of article; E.M.B.: statistical analysis; J.D.: conceptualization and research design, patient selection and performance of implant surgeries, critical revision and ﬁnal approval of article. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee) of Ghent University hospital (ID B670201420643). Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: The data presented in this study are available on request from the corresponding author. The data are not publicly available due to ethical restrictions. Conﬂicts of Interest: The authors declare no conﬂict of interest. Conﬂicts of Interest: The authors declare no conﬂict of interest. 12 of 13 12 of 13 J. Clin. Med. 2022, 11, 2251 References [CrossRef] p y p , ( pp ), [ ] 16. Mertens, C.; Steveling, H.G. Implant-supported ﬁxed prostheses in the edentulous maxilla: 8-year prospective results. Clin. Oral Implant. Res. 2011, 22, 464–472. [CrossRef] p 17. Bergendal, T.; Engquist, B. Implant-supported overdentures: A longitudinal prospective study. Int. J. Oral Maxillofac. Implants 1998, 13, 253–262. 18. Visser, A.; Raghoebar, G.M.; Meijer, H.J.; Vissink, A. Implant-retained maxillary overdentures on milled bar suprastructures: A 10-year follow-up of surgical and prosthetic care and aftercare. Int. J. Prosthodont. 2009, 22, 181–192. 19. Jemt, T.; Chai, J.; Harnett, J.; Heath, M.R.; Hutton, J.E.; Johns, R.B.; McKenna, S.; McNamara, D.C.; van Steenberghe, D.; Taylor, R.; et al. A 5-year prospective multicenter follow-up report on overdentures supported by osseointegrated implants. Int. J. Oral Maxillofac. Implants 1996, 11, 291–298. 20. Sadowsky, S.J. Mandibular implant-retained overdentures: A literature review. J. Prosthet. Dent. 2001, 86, 468–473. [CrossRef] 21. Sadowsky, S.J. Treatment considerations for maxillary implant overdentures: A systematic review. J. Prosthet. Dent. 2007, 97, 340–348. [CrossRef] 21. Sadowsky, S.J. Treatment considerations for maxillary implant overdentures: A systematic review. J. Prosthet. Dent. 2007, 97, 340–348. [CrossRef] 22. Slot, W.; Raghoebar, G.M.; Vissink, A.; Huddleston Slater, J.J.; Meijer, H.J. A systematic review of implant-supported maxillary overdentures after a mean observation period of at least 1 year. J. Clin. Periodontol. 2010, 37, 98–110. [CrossRef] [PubMed] 23. Roccuzzo, M.; Bonino, F.; Gaudioso, L.; Zwahlen, M.; Meijer, H.J. What is the optimal number of implants for removable reconstructions? A systematic review on implant supported overdentures Clin Oral Implants Res 2012 23 (Suppl 6) 229 237 22. Slot, W.; Raghoebar, G.M.; Vissink, A.; Huddleston Slater, J.J.; Meijer, H.J. A systematic review of implant-supported maxillary overdentures after a mean observation period of at least 1 year. J. Clin. Periodontol. 2010, 37, 98–110. [CrossRef] [PubMed] 22. Slot, W.; Raghoebar, G.M.; Vissink, A.; Huddleston Slater, J.J.; Meijer, H.J. A systematic review of implant supported maxillary overdentures after a mean observation period of at least 1 year. J. Clin. Periodontol. 2010, 37, 98–110. [CrossRef] [PubMed] 23. Roccuzzo, M.; Bonino, F.; Gaudioso, L.; Zwahlen, M.; Meijer, H.J. What is the optimal number of implants for removable reconstructions? A systematic review on implant-supported overdentures. Clin. Oral Implants Res. 2012, 23 (Suppl 6), 229–237. [CrossRef] [PubMed] 23. Roccuzzo, M.; Bonino, F.; Gaudioso, L.; Zwahlen, M.; Meijer, H.J. What is the optimal number of implants for removable reconstructions? A systematic review on implant-supported overdentures. Clin. References “Peri-Implantitis”: A Complication of a Forei “Disease”. Facts and Fiction. Clin. Implant Dent. Relat. Res. 2016, 18, 840–849. [CrossRef] 11. Sinjari, B.; D’Addazio, G.; De Tullio, I.; Traini, T.; Caputi, S. Peri-Implant Bone Resorption during Healing Abutment Placement: The Effect of a 0.20% Chlorhexidine Gel vs. Placebo-A Randomized Double Blind Controlled Human Study. BioMed Res. Int. 2018, 2018, 5326340. [CrossRef] [PubMed] 12. Sinjari, B.; D’Addazio, G.; Traini, T.; Varvara, G.; Scarano, A.; Murmura, G.; Caputi, S. A 10-year retrospe study on screw-retained versus cemented dental implant abutments. J. Biol. Regul. Homeost. Agents 201 13. Berglundh, T.; Armitage, G.; Araujo, M.G.; Avila-Ortiz, G.; Blanco, J.; Camargo, P.M.; Chen, S.; Cochran, D.; Derks, J.; Figuero, E.; et al. Peri-implant diseases and conditions: Consensus report of workgroup 4 of the 2017 World Workshop on the Classiﬁcation of Periodontal and Peri-Implant Diseases and Conditions. J. Periodontol. 2018, 89, S313–S318. [CrossRef] [PubMed] et al. Peri-implant diseases and conditions: Consensus report of workgroup 4 of the 2017 World Workshop on the Classiﬁcation of Periodontal and Peri-Implant Diseases and Conditions. J. Periodontol. 2018, 89, S313–S318. [CrossRef] [PubMed] p p g p p of Periodontal and Peri-Implant Diseases and Conditions. J. Periodontol. 2018, 89, S313–S318. [CrossRef] [PubMed] 14. Doornewaard, R.; Jacquet, W.; Cosyn, J.; De Bruyn, H. How do peri-implant biologic parameters c survival and peri-implantitis? A critical review. Clin. Oral. Implants Res. 2018, 29 (Suppl. 18), 100–123. ewaa d, ; Jacquet, W ; Cosy , J ; e uy , ow do pe p a t b o og c pa a ete s co espo d val and peri-implantitis? A critical review. Clin. Oral. Implants Res. 2018, 29 (Suppl. 18), 100–123. [CrossRef] survival and peri implantitis? A critical review. Clin. Oral. Implants Res. 2018, 29 (Suppl. 18), 100 123. [CrossRef] 15. Heydecke, G.; Zwahlen, M.; Nicol, A.; Nisand, D.; Payer, M.; Renouard, F.; Grohmann, P.; Muhlemann, S.; Joda, T. What is the optimal number of implants for fixed reconstructions: A systematic review. Clin. Oral Implants Res. 2012, 23 (Suppl 6), 217–228. [CrossRef] Zwahlen, M.; Nicol, A.; Nisand, D.; Payer, M.; Renouard, F.; Grohmann, P.; Muhlemann, S.; Joda, T. What is th y y J p number of implants for fixed reconstructions: A systematic review. Clin. Oral Implants Res. 2012, 23 (Suppl 6), 217–228. [CrossRef] 16. Mertens, C.; Steveling, H.G. Implant-supported ﬁxed prostheses in the edentulous maxilla: 8-year prospective results. Clin. Oral Implant. Res. 2011, 22, 464–472. References 1. Holm-Pedersen, P.; Lang, N.P.; Muller, F. What are the longevities of teeth and oral implants? Clin. Oral Implant. Res. 2007, 18 (Suppl. 3), 15–19. [CrossRef] [PubMed] ( pp ) 2. Feine, J.S.; Carlsson, G.E.; Awad, M.A.; Chehade, A.; Duncan, W.J.; Gizani, S.; Head, T.; Lund, J.P.; MacEntee, M.; Mericske-Stern, R.; et al. The McGill consensus statement on overdentures. Mandibular two-implant overdentures as ﬁrst choice standard of care for edentulous patients. Montreal, Quebec, May 24–25, 2002. Int. J. Oral Maxillofac. Implant. 2002, 17, 601–602. p y f p 3. Pjetursson, B.E.; Tan, K.; Lang, N.P.; Brägger, U.; Egger, M.; Zwahlen, M. A systematic review of the survival and complication rates of ﬁxed partial dentures (FPDs) after an observation period of at least 5 years. Clin. Oral Implant. Res. 2004, 15, 667–676. [CrossRef] [PubMed] [ ] [ ] 4. Simonis, P.; Dufour, T.; Tenenbaum, H. Long-term implant survival and success: A 10–16-year follow-up of non-submerged dental implants. Clin. Oral Implant. Res. 2010, 21, 772–777. [CrossRef] 5. Esposito, M.; Hirsch, J.M.; Lekholm, U.; Thomsen, P. Biological factors contributing to fai (II). Etiopathogenesis. Eur. J. Oral Sci. 1998, 106, 721–764. [CrossRef] 5. Esposito, M.; Hirsch, J.M.; Lekholm, U.; Thomsen, P. Biological factors contributing to failures of osseointegrated oral implants. (II). Etiopathogenesis. Eur. J. Oral Sci. 1998, 106, 721–764. [CrossRef] 6. Jemt, T.; Johansson, J. Implant treatment in the edentulous maxilla provided with ﬁxed prostheses. Clin. Implant Dent. Relat. Res. 2006, 8 6. Jemt, T.; Johansson, J. Implant treatment in the edentulous maxillae: A 15-year follow-up study on 76 consecutive patients provided with ﬁxed prostheses. Clin. Implant Dent. Relat. Res. 2006, 8, 61–69. [CrossRef] 7. Maló, P.; de Araújo Nobre, M.; Lopes, A.; Francischone, C.; Rigolizzo, M. “All-on-4” immediate-function concept for completely edentulous maxillae: A clinical report on the medium (3 years) and long-term (5 years) outcomes. Clin. Implant Dent. Relat. Res. 2012, 14 (Suppl 1), e139–e150. [CrossRef] pp 8. Astrand, P.; Engquist, B.; Dahlgren, S.; Grondahl, K.; Engquist, E.; Feldmann, H. Astra Tech and Branemark system implants: A 5-year prospective study of marginal bone reactions. Clin. Oral Implant. Res. 2004, 15, 413–420. [CrossRef] 9. Linkevicius, T.; Apse, P.; Grybauskas, S.; Puisys, A. The inﬂuence of soft tissue thickness on crestal bone changes around implants: A 1-year prospective controlled clinical trial. Int. J. Oral Maxillofac. Implant. 2009, 24, 712–719. y p p f p 10. Albrektsson, T.; Canullo, L.; Cochran, D.; De Bruyn, H. References Oral Implants Res. 2012, 23 (Suppl 6), 229–237. [CrossRef] [PubMed] 24. Klemetti, E. Is there a certain number of implants needed to retain an overdenture? J. Oral Rehabil. 2008, 35 (Suppl 1), 80–84. [CrossRef] 25. Naert, I.; Gizani, S.; van Steenberghe, D. Rigidly splinted implants in the resorbed maxilla to retain a hinging overdenture: 24. Klemetti, E. Is there a certain number of implants needed to retain an overdenture? J. Oral Rehabil. 2008, 35 (Suppl 1), 80–84. [CrossRef] 24. Klemetti, E. Is there a certain number of implants needed to retain an overdenture? J. Oral Rehabil. 2008, 35 (Suppl 1), 80–84. [CrossRef] 25. Naert, I.; Gizani, S.; van Steenberghe, D. Rigidly splinted implants in the resorbed maxilla to retain a hinging overdenture: A series of clinical reports for up to 4 years. J. Prosthet. Dent. 1998, 79, 156–164. [CrossRef] 25. Naert, I.; Gizani, S.; van Steenberghe, D. Rigidly splinted implants in the resorbed maxilla to re A series of clinical reports for up to 4 years. J. Prosthet. Dent. 1998, 79, 156–164. [CrossRef] p p y J 26. Kiener, P.; Oetterli, M.; Mericske, E.; Mericske-Stern, R. Effectiveness of maxillary overdentures supported by implants: Mainte- nance and prosthetic complications. Int. J. Prosthodont. 2001, 14, 133–140. J. Clin. Med. 2022, 11, 2251 13 of 13 13 of 13 27. Doornewaard, R.; Sakani, S.; Matthys, C.; Glibert, M.; Bronkhorst, E.; Vandeweghe, S.; Vervaeke, S.; De Bruyn, H. Four-implant- supported overdenture treatment in the maxilla. Part I: A randomized controlled split mouth trial assessing the effect of microthreads and abutment connection type on 4 years peri-implant health. Clin. Implant Dent. Relat. Res. 2021, 23, 671–679. [CrossRef] yp y p p p 28. Mombelli, A.; van Oosten, M.A.; Schurch, E., Jr.; Land, N.P. The microbiota associated with successful o titanium implants. Oral Microbiol. Immunol. 1987, 2, 145–151. [CrossRef] g g titanium implants. Oral Microbiol. Immunol. 1987, 2, 145–151. [CrossRef] 29. Zarb, G.A.; Albrektsson, T. Consensus report: Towards optimized treatment outcomes for dental implants. J. Prosthet. Dent. 1998, 80, 641. [CrossRef] p 29. Zarb, G.A.; Albrektsson, T. Consensus report: Towards optimized treatment outcomes for dental implants. J. Prosthet. Dent. 1998, 80, 641. [CrossRef] [ ] 30. Meijer, H.J.; Kuiper, J.H.; Starmans, F.J.; Bosman, F. Stress distribution around dental implants: Inﬂuence of superstructure, length f i l d h i h f dibl J P h D 68 96 102 [C R f] 30. References Meijer, H.J.; Kuiper, J.H.; Starmans, F.J.; Bosman, F. Stress distribution around dental implants: Inﬂuence of superstructure, length of implants, and height of mandible. J. Prosthet. Dent. 1992, 68, 96–102. [CrossRef] 30. Meijer, H.J.; Kuiper, J.H.; Starmans, F.J.; Bosman, F. Stress distribution around dental implants: In of implants, and height of mandible. J. Prosthet. Dent. 1992, 68, 96–102. [CrossRef] p g 31. Raghoebar, G.M.; Meijer, H.J.; Slot, W.; Slater, J.J.; Vissink, A. A systematic review of implant-supported overdentures in the d t l ill d t th dibl H i l t ? E J O l I l t 2014 7 (S l 2) S191 S201 p g 31. Raghoebar, G.M.; Meijer, H.J.; Slot, W.; Slater, J.J.; Vissink, A. A systematic review of implant-supported overdentures in the edentulous maxilla, compared to the mandible: How many implants? Eur. J. Oral Implant. 2014, 7 (Suppl 2), S191–S201. edentulous maxilla, compared to the mandible: How many implants? Eur. J. Oral Implant. 2014, 7 (Suppl 2), S191 S201. 32. Slot, W.; Raghoebar, G.M.; Cune, M.S.; Vissink, A.; Meijer, H.J. Maxillary overdentures supported by four or six implants in the anterior region: 5-year results from a randomized controlled trial. J. Clin. Periodontol. 2016, 43, 1180–1187. [CrossRef] [PubMed] 33. Van Doorne, L.; Gholami, P.; D’Haese, J.; Hommez, G.; Meijer, G.; De Bruyn, H. Three-Dimensional Radiographic Outcome of p y p p pp 32. Slot, W.; Raghoebar, G.M.; Cune, M.S.; Vissink, A.; Meijer, H.J. Maxillary overdentures supported by four or six implants in the anterior region: 5-year results from a randomized controlled trial. J. Clin. Periodontol. 2016, 43, 1180–1187. [CrossRef] [PubMed] g y 33. Van Doorne, L.; Gholami, P.; D’Haese, J.; Hommez, G.; Meijer, G.; De Bruyn, H. Three-Dimensional Radiographic Outcome of Free-Handed Flaplessly Placed Mini Dental Implants in Edentulous Maxillae after 2-Years Function. J. Clin. Med. 2020, 9, 2120. [CrossRef] 33. Van Doorne, L.; Gholami, P.; D Haese, J.; Hommez, G.; Meijer, G.; De Bruyn, H. Three-Dimensional Radiographic Outcome of Free-Handed Flaplessly Placed Mini Dental Implants in Edentulous Maxillae after 2-Years Function. J. Clin. Med. 2020, 9, 2120. [CrossRef] 34 Kamei K ; Gonda T ; Takahashi T ; Maeda Y Mechanical Study of Optimal Implant Position for Maxillary Implant-Supported Free-Handed Flaplessly Placed Mini Dental Implants in Edentulous Maxillae after 2-Years Function. J. Clin. Med. 2020, 9, 2120. [CrossRef] 34. Kamei, K.; Gonda, T.; Takahashi, T.; Maeda, Y. References Mechanical Study of Optimal Implant Position for Maxillary Implant-Supported 35. Dimililer, G.; Kücükkurt, S.; Cetiner, S. Biomechanical effects of implant number and diameter on stress distributions in maxillary implant-supported overdentures. J. Prosthet. Dent. 2018, 119, 244–249.e6. [CrossRef] p pp 36. Di Francesco, F.; De Marco, G.; Sommella, A.; Lanza, A. Splinting vs Not Splinting Four Implants Supporting a Maxillary Overdenture: A Systematic Review. Int. J. Prosthodont. 2019, 32, 509–518. [CrossRef] 37. Dierens, M.; Vandeweghe, S.; Kisch, J.; Nilner, K.; De Bruyn, H. Long-term follow-up of turned single implants placed in periodontally healthy patients after 16–22 years: Radiographic and peri-implant outcome. Clin. Oral Implants Res. 2012, 23, 197–204. [CrossRef] 38. Verhoeven, J.W.; Cune, M.S.; de Putter, C. Reliability of some clinical parameters of evaluation in implant dentistry. J. Oral Rehabil. 2000, 27, 211–216. [CrossRef] 39. Onclin, P.; Slot, W.; Vissink, A.; Raghoebar, G.M.; Meijer, H.J.A. Incidence of peri-implant mucositis and peri-implantitis in patients with a maxillary overdenture: A sub-analysis of two prospective studies with a 10-year follow-up period. Clin. Implant Dent. Relat. Res. 2022, 1–8. [CrossRef] 40. Meyle, J.; Gersok, G.; Boedeker, R.H.; Gonzales, J.R. Long-term analysis of osseointegrated implants in non-smoker patients with a previous history of periodontitis. J. Clin. Periodontol. 2014, 41, 504–512. [CrossRef] 41. Mangano, C.; Mangano, F.; Shibli, J.A.; Ricci, M.; Sammons, R.L.; Figliuzzi, M. Morse taper connection implants supporting “planned” maxillary and mandibular bar-retained overdentures: A 5-year prospective multicenter study. Clin. Oral Implant. Res. 2011, 22, 1117–1124. [CrossRef] [PubMed] 42. Slot, W.; Raghoebar, G.M.; van Dijk, G.; Meijer, H.J. Attachment of clips in a bar-retained maxillary implant overdenture: A clinical report. J. Prosthet. Dent. 2012, 107, 353–357. [CrossRef] p 43. Schwarz, F.; Derks, J.; Monje, A.; Wang, H.L. Peri-implantitis. J. Periodontol. 2018, 89 (Suppl. 1), S267–S290. [CrossRef] [PubMed]
https://openalex.org/W2036649899	https://europepmc.org/articles/pmc2263061?pdf=render	English	null	South Beach Diet associated ketoacidosis: a case report	Journal of medical case reports	2,008	cc-by	2,050	Abstract Introduction: It has been previously unclear whether a "mild" degree of low carbohydrate or "starvation" ketonemia and acidosis induced by a low carbohydrate diet is clinically relevant to a patient. Case presentation: A 30-year-old Caucasian male on a low carbohydrate diet presented with nausea, vomiting and abdominal pain. The patient's bicarbonate level was 12 and he had hyperglycemia and ketonemia. He was felt to be in diabetic ketoacidosis and was started on intravenous insulin and isotonic saline infusions and responded well. Following cessation of insulin therapy, the patient remained normoglycemic for the remainder of his hospital stay. He later admitted to having been on the South Beach Diet, which is a low carbohydrate diet, for the three weeks prior to his presentation and during which time he had lost 16 pounds. On admission his BMI was 27.1. On presentation, the patient was felt to be in diabetic ketoacidosis but, interestingly, he was subsequently euglycemic without therapy. Following discharge, the patient discontinued the diet plan and he has remained asymptomatic and euglycemic over the following two years. Conclusion: The hyperglycemic ketoacidosis in this patient may have been caused by increased concentrations of free fatty acids in the absence of carbohydrate-induced inhibition of beta- oxidation of fatty acids and in the presence of an abnormally high ratio of glucagons to insulin. Given the present day popularity of low-carbohydrate diet plans, healthcare providers should be aware of the apparent association between such diets and symptomatic ketoacidosis. In a patient with ketoacidosis suspected to be secondary to a low carbohydrate diet, all other causes of high anion gap acidosis should be ruled out before attributing the acidosis to the low carbohydrate diet. Published: 11 February 2008 Journal of Medical Case Reports 2008, 2:45 doi:10.1186/1752-1947-2-45 This article is available from: http://www.jmedicalcasereports.com/content/2/1/45 © 2008 Chalasani and Fischer; licensee BioMed Central Ltd. ; This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. South Beach Diet associated ketoacidosis: a case report Swapna Chalasani* and Jacqueline Fischer Address: Department of Internal Medicine, University of Illinois College of Medicine at Peoria, OSF Saint Francis Medical Centre, 530 NE Glen Oak Avenue, Peoria, IL 61637, USA Email: Swapna Chalasani* - swapna@uic.edu; Jacqueline Fischer - fischer@uicompim.org * Corresponding author Email: Swapna Chalasani* - swapna@uic.edu; Jacqueline Fischer - fischer@uicompim.org * Corresponding author Received: 13 June 2007 Accepted: 11 February 2008 Received: 13 June 2007 Accepted: 11 February 2008 Published: 11 February 2008 BioMed Central Journal of Medical Case Reports Open Access Page 1 of 3 (page number not for citation purposes) Case presentation A 30-year-old Caucasian male without significant past medical history presented with a two day history of nau- sea, vomiting and diffuse abdominal pain. The patient denied use of any medications (prescription or nonpre- scription) or any illicit substances. He did admit to occa- sional ethanol ingestion stating that he consumed four alcoholic beverages (approximately 0.6 ounces ethanol each) the night prior to the onset of symptoms. The patient had a family history of diabetes mellitus type 2 on both the paternal and maternal side. On presentation, the patient appeared in mild distress sec- ondary to his stated abdominal pain. BMI on admission was 27.1 (weight 91 kilograms), vital signs were within normal limits, and the patient appeared euvolemic. Com- plete physical examination was normal including a nor- mal abdominal examination. Initial laboratory studies revealed a high anion gap metabolic acidosis (arterial ph 7.34, arterial PCO2 23 mmHg, serum bicarbonate 12 mmol/L, serum anion gap 21) and hyperglycemia (serum glucose 267 mg/dL). The patient was found to have both ketonemia and ketonuria. Additional data, including a complete blood count, serum sodium, serum chloride, serum potassium, liver chemistries, lipid fractionation, serum lipase, serum amylase, plain chest radiography, and computed tomography of the abdomen and pelvis, were within normal limits. Serum osmolality, urine toxi- cology and lactic acid levels were not performed. Low-carbohydrate, fat-rich meals stimulate glucagon secretion, lower insulin secretion, and increase insulin resistance [2,3]. Dietary and endogenous fat are cat- abolized to form ketone bodies as an energy source [4]. Plasma fatty acid concentrations can be two-fold higher during low-versus normo-carbohydrate diets in the postabsorptive period [5]. When the body has no free car- bohydrates available, fat must be broken down into acetyl-CoA to generate energy. Acetyl-CoA is not being recycled through the citric acid cycle because the citric acid cycle intermediates (mainly oxaloacetate) have been depleted to feed the gluconeogenesis pathway, and the resulting accumulation of acetyl-CoA activates ketogene- sis and this might have led to the ketoacidosis in our patient. Discussion Here we present a case of hyperglycemic ketoacidosis associated with a low carbohydrate diet. The South Beach Diet is a popular diet plan which primarily relies on the restriction of dietary carbohydrates to achieve weight loss [1]. Our patient strictly adhered to 10 to 15 grams of car- bohydrate per day for 3 weeks prior to presentation and lost 16 pounds. He was following the most stringent form of this diet, namely that being the form in which total car- bohydrate consumption is limited to less than 20 grams daily. On presentation, our patient was felt to be in dia- betic ketoacidosis but, interestingly, the patient was sub- sequently euglycemic without therapy and, even after two years of follow up, remained asymptomatic and euglyc- emic. Introduction starches) are limited or replaced with foods containing a higher percentage of proteins, fats and/or fiber. By con- trast, if the diets are very low in starches and sugars (low- carbohydrate diets) the blood sugar level can fall so low that there is insufficient glucose to fuel the cells in the body. This state causes the pancreas to produce glucagon. Glucagon causes the conversion of stored glycogen to glu- cose and, once the glycogen stores are exhausted, causes Low carbohydrate diets are nutritional programs that advocate restricted carbohydrate consumption based on research that ties consumption of certain carbohydrates with increased blood insulin levels, and overexposure to insulin with metabolic syndrome (the most recognized symptom of which is obesity). Under these dietary pro- grams, foods high in digestible carbohydrates (sugars and Page 1 of 3 (page number not for citation purposes) Page 1 of 3 (page number not for citation purposes) http://www.jmedicalcasereports.com/content/2/1/45 Journal of Medical Case Reports 2008, 2:45 http://www.jmedicalcasereports.com/content/2/1/45 asymptomatic and euglycemic over the following two years while maintaining a BMI of 27. asymptomatic and euglycemic over the following two years while maintaining a BMI of 27. the liver to synthesize ketones (ketosis) and glucose (glu- coneogenesis) from fats and proteins. It has been previ- ously unclear whether this "mild" degree of low carbohydrate or "starvation" ketonemia and acidosis induced by a low carbohydrate diet is clinically relevant to a patient. Conclusion Despite the widespread use of weight reducing low carbo- hydrate diets for many years now, few reports to date have highlighted their association with clinically relevant ketoacidosis [6,7]. This either means that it is a rare com- plication, or that it has, so far, not been recognized as a possible complication of a very strict low carbohydrate diet. The hyperglycemic ketoacidosis could easily, in the past, have simply been passed off as a complication of type 2 diabetes mellitus or metabolic syndrome (the low carbohydrate diet being viewed as an irrelevancy). It could also be that some people are applying the diet in an ever increasingly more fanatical way. A final possibility is that the syndrome is brought about by some, as yet unknown, trigger in persons on a very low carbohydrate diet. The patient was felt to be in diabetic ketoacidosis and was started on intravenous insulin and isotonic saline infu- sions to which he responded well with rapid resolution of the acidosis and abdominal pain within ten hours. Fol- lowing cessation of the insulin therapy, the patient remained normoglycemic for the remainder of his hospi- tal stay (24 hours). Hemoglobin A1C was 5.1% (4.4%– 6.4%) and C peptide was 4.1 ng/mL (0.8–3.1 ng/mL). The patient later admitted to having been on the South Beach Diet at the time of presentation, having adhered to a particularly strict (less than 20 grams carbohydrate daily) form of this low carbohydrate diet plan. The patient stated that he had eliminated virtually all forms of carbo- hydrate from his diet for the three weeks prior to his pres- entation and had lost 16 pounds (7.3 kg) over the same time period. Following discharge, the patient discontin- ued the low carbohydrate diet plan and he has remained Given the present day popularity of low-carbohydrate diet plans, healthcare providers should be aware of the appar- ent association between such diets and symptomatic ketoacidosis. In a patient with ketoacidosis suspected sec- Page 2 of 3 (page number not for citation purposes) http://www.jmedicalcasereports.com/content/2/1/45 http://www.jmedicalcasereports.com/content/2/1/45 Journal of Medical Case Reports 2008, 2:45 ondary to a low carbohydrate diet, all other causes of high anion gap acidosis should be ruled out before attributing the acidosis to the low carbohydrate diet. Competing interests p g The author(s) declare that they have no competing inter- ests. Consent Written informed consent was obtained from the patient for the publication of this study. Authors' contributions All authors have read and approved the final manuscript. SC: Involved in the conception of the report and literature review along with manuscript preparation, editing and submission. JF: Involved in the literature review, manu- script editing and manuscript review. Conclusion Although these laboratory tests were not performed in our patient, serum osmolal gap, lactic acid levels and salicylate levels, in addition to the tests which were performed in our patient, may be useful in ruling out other causes of acidosis. References 1. [http://www.southbeachdiet.com]. (accession date April, 2007) 2 C G SC C f [ p ] ( p ) 2. Exton JH, Corbin JG, Harper SC: Control of gluconeogenesis in liver. V. Effects of fasting, diabetes, and glucagons on lactate and endogenous metabolism in the perfused rat liver. J Biol Chem 1972, 247:4996-5003. 3. Gutniak M, Grill V, Effendic S: Effect of composition of mixed meals-low versus high carbohydrate content-on insulin, glu- cagons, and somatostatin release in healthy humans and in patients with NIDDM. Diabetes care 1986, 9:244-9. 3. Gutniak M, Grill V, Effendic S: Effect of composition of mixed meals-low versus high carbohydrate content-on insulin, glu- cagons, and somatostatin release in healthy humans and in patients with NIDDM. Diabetes care 1986, 9:244-9. p 4. Jesica Pagano, David Katz L: Low-Down on Low-Carbohydrate Diets. The Nurse Practitioner 2003. p 4. Jesica Pagano, David Katz L: Low-Down on Low-Carbohydrate Diets. The Nurse Practitioner 2003. 5. Bisschop PH, De Sain-Van Der Velden MG, Stellard F: Dietary car- bohydrate deprivation increases 24-hour nitrogen excretion without affecting postabsoptive hepatic or whole body pro- tein metabolism in healthy men. J Clin Endocrinol Metab 2003, 88:3801-5. 6. Shah Pankaj, Isley William L: Ketoacidosis during a Low-Carbo- hydrate Diet. NEJM 2006, 354(1):97-98. y J ( ) 7. Chen TY, Smith W, Rosenstock JL, Lessnau KD: A life-threatening complication of Atkins diet. Lancet 2006, 367:958. y J ( ) 7. Chen TY, Smith W, Rosenstock JL, Lessnau KD: A life-threatening complication of Atkins diet. Lancet 2006, 367:958. References Publish with BioMed Central and every scientist can read your work free of charge "BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime." Sir Paul Nurse, Cancer Research UK Your research papers will be: available free of charge to the entire biomedical community peer reviewed and published immediately upon acceptance cited in PubMed and archived on PubMed Central yours — you keep the copyright Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp BioMedcentral Page 3 of 3 (page number not for citation purposes) Publish with BioMed Central and every scientist can read your work free of charge "BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime." Sir Paul Nurse, Cancer Research UK Your research papers will be: available free of charge to the entire biomedical community peer reviewed and published immediately upon acceptance cited in PubMed and archived on PubMed Central yours — you keep the copyright Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp BioMedcentral Publish with BioMed Central and every scientist can read your work free of charge
https://openalex.org/W2533936737	https://kclpure.kcl.ac.uk/ws/files/61360090/acs_2Ejpclett_2E6b02181.pdf	English	null	Controlling Structural Transitions in AuAg Nanoparticles through Precise Compositional Design	The journal of physical chemistry letters	2,016	cc-by	6,633	Citing this paper Pl h C t g t s pape Please note that where the full-text provided on King's Research Portal is the Author Accepted Manuscript or Post-Print version this may differ from the final Published version. If citing, it is advised that you check and use the publisher's definitive version for pagination, volume/issue, and date of publication details. And where the final published version is provided on the Research Portal, if citing you are again advised to check the publisher's website for any subsequent corrections. Citation for published version (APA): Gould, A. L., Rossi, K., Catlow, C. R. A., Baletto, F., & Logsdail, A. J. (2016). Controlling Structural Transitions in AuAg Nanoparticles through Precise Compositional Design. Journal of physical chemistry letters, 7(21), 4414- 4419. https://doi.org/10.1021/acs.jpclett.6b02181 Controlling Structural Transitions in AuAg Nanoparticles through Precise Compositional Design Anna L. Gould,†,‡ Kevin Rossi,§ C. Richard A. Catlow,†,‡,∥Francesca Baletto,,§ and Andrew J. Logsdail,†,∥ N anoparticles (NPs) consist of between a few and many thousands of atoms or molecules, interacting to form discrete particles.1 NPs have properties that are distinct from bulk and atomic systems, which, when considered alongside their nanoscale size, make them very suitable for applications in scientiﬁc ﬁelds such as medicine, optics, and catalysis.2−7 Recent research has shown that the nuclearity, composition and chemical ordering of bimetallic NPs can be well controlled during synthesis,8−14 but it remains an experimental challenge to control the shape of these systems.5,9,15 From an applied catalysis perspective, it would be beneﬁcial if one could preferentially form greater quantities of reactive sites, such as vertices and edges,5,8,16−21 over unreactive sites. However, synthesis, characterization, and application of the NPs is generally performed under conditions that do not aid structure stabilization: A multitude of geometric arrangements coexist on most energy landscapes,8,22,23 possibly with similar energies and separated by transition barriers that can be overcome at room temperature.15,24,25 The elevated operating temperatures for catalytic applications further exacerbates the problem of structural instability, and therefore understanding how to control the morphologies of NPs is of paramount importance to further their commercial applicability. The size of NPs makes them an attractive problem to address with computational techniques, and previous work using thermodynamic sampling has identiﬁed key transition pathways and barriers between structural motifs at a range of diﬀerent nuclearities,16,26−30 as well as demonstrating how the energy barriers change with chemical ordering in bimetallic NPs.31−34 One well-characterized pathway is the martensitic trans- formation between icosahedral (Ih) and cuboctahedral (CO) motifs via the so-called sextuple diamond−square−diamond (DSD) mechanism:26,35 Triangular facets are stretched and then rotated to form a diamond one and then transformed to a square facet26 (DS), with the reverse square−diamond process (SD) leading to reformation of the Ih from the CO. Recently, forward (SD) and backward (DS) transition pathways have been characterized at ﬁnite temperatures through molecular dynamics modeling for 147-atom Ag and AuAg NPs;32,33 however, no such transition occurs for Au NPs,32,34,36 showing that structural stability or transition barriers change consid- erably as a function of chemical composition. From a structural design perspective, controlling these transitions via chemical Received: September 23, 2016 Accepted: October 19, 2016 Published: October 19, 2016 © 2016 American Chemical Society 4414 DOI: 10.1021/acs.jpclett.6b02181 J. Phys. Chem. Lett. pubs.acs.org/JPCL DOI: 10.1021/acs.jpclett.6b02181 J. Phys. Chem. Lett. 2016, 7, 4414−4419 © 2016 American Chemical Society General rights General rights Copyright and moral rights for the publications made accessible in the Research Portal are retained by the authors and owners and it is a condition of accessing publications that users recognize and abide by the legal requirements associ ral rights for the publications made accessible in the Research Portal are retained by the authors and/or other copyright condition of accessing publications that users recognize and abide by the legal requirements associated with these right •Users may download and print one copy of any publication from the Research Portal for the purpose of private study or research. •You may not further distribute the material or use it for any profit-making activity or commercial gain •You may freely distribute the URL identifying the publication in the Research Portal •Users may download and print one copy of any publication from the Research Portal for the purpose of private study •You may not further distribute the material or use it for any profit-making activity or commercial gain y y p g y •You may freely distribute the URL identifying the publication in the Research Portal Take down policy If you believe that this document breaches copyright please contact librarypure@kcl.ac.uk providing details, and we will remove access to the work immediately and investigate your claim. p y If you believe that this document breaches copyright please contact librarypure@kcl.ac.uk providing details, and we w the work immediately and investigate your claim. Download date: 24. Oct. 2024 Letter pubs.acs.org/JPCL This is an open access article published under a Creative Commons Attribution (CC-BY) License, which permits unrestricted use, distribution and reproduction in any medium, provided the author and source are cited. This is an open access article published under a Creative Commons Attribution (CC-BY) License, which permits unrestricted use, distribution and reproduction in any medium, provided the author and source are cited. This is an open access article published under a Creative Commons Attribution (CC-BY) License, which permits unrestricted use, distribution and reproduction in any medium, provided the author and source are cited. This is an open access article published under a Creative Commons Attribution (CC-BY) License, which permits unrestricted use, distribution and reproduction in any medium, provided the author and source are cited. Controlling Structural Transitions in AuAg Nanoparticles through Precise Compositional Design Anna L. Gould,†,‡ Kevin Rossi,§ C. Richard A. Catlow,†,‡,∥Francesca Baletto,,§ and Andrew J. Logsdail,†,∥ Anna L. Gould,†,‡ Kevin Rossi,§ C. Richard A. Catlow,†,‡,∥Francesca Baletto,,§ and Andrew J. Logsdail,†,∥ †University College London, Kathleen Lonsdale Materials Chemistry, Department of Chemistry, 20 Gordon Street, London WC1H 0AJ, United Kingdom †University College London, Kathleen Lonsdale Materials Chemistry, Department of Chemistry, 20 Gor 0AJ, United Kingdom ‡The U.K. Catalysis Hub, Research Complex at Harwell, Rutherford Appleton Laboratory, Oxfordshire OX11 0FA, United Kingdom §Physics Department, King’s College London, London WC2R 2LS, United Kingdom ∥CardiﬀCatalysis Institute, School of Chemistry, CardiﬀUniversity, CardiﬀCF10 3AT, United Kingdom * S Supporting Information ABSTRACT: We present a study of the transitional pathways between high- symmetry structural motifs for AgAu nanoparticles, with a speciﬁc focus on controlling the energetic barriers through chemical design. We show that the barriers can be altered by careful control of the elemental composition and chemical arrangement, with core@shell and vertex-decorated arrangements being speciﬁcally inﬂuential on the barrier heights. We also highlight the complexity of the potential and free energy landscapes for systems where there are low-symmetry geometric motifs that are energetically competitive to the high-symmetry arrangements. In particular, we highlight that some core@shell arrangements preferentially transition through multistep restructuring of low- symmetry truncated octahedra and rosette-icosahedra, instead of via the more straightforward square-diamond transformations, due to lower energy barriers and competitive energetic minima. Our results have promising implications for the continuing eﬀorts in bespoke nanoparticle design for catalytic and plasmonic applications. ABSTRACT: We present a study of the transitional pathways between high- symmetry structural motifs for AgAu nanoparticles, with a speciﬁc focus on controlling the energetic barriers through chemical design. We show that the barriers can be altered by careful control of the elemental composition and chemical arrangement, with core@shell and vertex-decorated arrangements being speciﬁcally inﬂuential on the barrier heights. We also highlight the complexity of the potential and free energy landscapes for systems where there are low-symmetry geometric motifs that are energetically competitive to the high-symmetry arrangements. In particular, we highlight that some core@shell arrangements preferentially transition through multistep restructuring of low- symmetry truncated octahedra and rosette-icosahedra, instead of via the more straightforward square-diamond transformations, due to lower energy barriers and competitive energetic minima. Our results have promising implications for the continuing eﬀorts in bespoke nanoparticle design for catalytic and plasmonic applications. Controlling Structural Transitions in AuAg Nanoparticles through Precise Compositional Design Anna L. Gould,†,‡ Kevin Rossi,§ C. Richard A. Catlow,†,‡,∥Francesca Baletto,,§ and Andrew J. Logsdail,†,∥ 2016, 7, 4414−4419 N N anoparticles (NPs) consist of between a few and many thousands of atoms or molecules, interacting to form discrete particles.1 NPs have properties that are distinct from bulk and atomic systems, which, when considered alongside their nanoscale size, make them very suitable for applications in scientiﬁc ﬁelds such as medicine, optics, and catalysis.2−7 Recent research has shown that the nuclearity, composition and chemical ordering of bimetallic NPs can be well controlled during synthesis,8−14 but it remains an experimental challenge to control the shape of these systems.5,9,15 From an applied catalysis perspective, it would be beneﬁcial if one could preferentially form greater quantities of reactive sites, such as vertices and edges,5,8,16−21 over unreactive sites. However, synthesis, characterization, and application of the NPs is generally performed under conditions that do not aid structure stabilization: A multitude of geometric arrangements coexist on most energy landscapes,8,22,23 possibly with similar energies and separated by transition barriers that can be overcome at room temperature.15,24,25 The elevated operating temperatures for catalytic applications further exacerbates the problem of structural instability, and therefore understanding how to control the morphologies of NPs is of paramount importance to further their commercial applicability. The size of NPs makes them an attractive problem to address with computational techniques, and previous work using thermodynamic sampling has identiﬁed key transition pathways and barriers between structural motifs at a range of diﬀerent nuclearities,16,26−30 as well as demonstrating how the energy barriers change with chemical ordering in bimetallic NPs.31−34 One well-characterized pathway is the martensitic trans- formation between icosahedral (Ih) and cuboctahedral (CO) motifs via the so-called sextuple diamond−square−diamond (DSD) mechanism:26,35 Triangular facets are stretched and then rotated to form a diamond one and then transformed to a square facet26 (DS), with the reverse square−diamond process (SD) leading to reformation of the Ih from the CO. Recently, forward (SD) and backward (DS) transition pathways have been characterized at ﬁnite temperatures through molecular dynamics modeling for 147-atom Ag and AuAg NPs;32,33 however, no such transition occurs for Au NPs,32,34,36 showing that structural stability or transition barriers change consid- erably as a function of chemical composition. The Journal of Physical Chemistry Letters The reduction of ΔE for Au13@Ag134 is matched for ΔE(Ih →CO), where the same compositions have thermodynamic barriers of 3.25 and 2.78 eV, respectively, implying that the transition state is reduced in energy. Substitution of a single Au dopant in to the Ag147 NP generally results in minor variation of the forward and barrier, by ±0.02 eV. Exceptional changes in ΔE(Ih →CO) occur when the Au is positioned: (i) subsurface below an edge atom (3.33 eV) and (ii) in the middle of the (111) surface facet (3.30 eV). For Ag dopants in the Au147 NPs, slight variation was again observed: the barriers are generally reduced by at most 0.05 eV from the Au147 values. An exception occurs when Ag is positioned at a surface vertex, with ΔE increased by 0.02 eV in both directions; however, the transition mechanism remains identical. Ag doping in the middle of the (100) facet also reduces ΔE(CO →Ih) and ΔE(Ih →CO) by 0.06 and 0.09 eV, respectively, to 1.08 and 2.24 eV. gy In contrast, the DNEB pathways are more complicated for Ag13@Au134 and Ag55@Au92, proceeding via the r-Ih intermediates already observed for Ag1@Au146; ΔE(CO → Ih) is 1.52 and 3.07 eV for these compositions, respectively, while ΔE(Ih →CO) is 2.34 and 3.98 eV. It is noted that the rate of transformation is determined by the size of the Ag core: at low Ag concentration, deformation of the CO is the time- consuming process, with the crossover point between CO- and Ih-like structures higher in energy than for the starting structures (Figure 1, top), whereas at higher Ag concentration, the transition between CO- and Ih-like motifs is via motifs that are lower in energy than the starting structures, and thus r-Ih deformation becomes rate-limiting (Figure 1, bottom). A detailed structural analysis is presented in the Supporting Information, with a distorted TO formed in each case (Figure 3a) before transforming to an r-Ih (Figure 3b) via rotation and stretching similar to the SD mechanism. Dissection of the p y A distinct variation occurs when Ag is positioned at the center of Au NP, akin to a core@shell arrangement. This Ag1@ Au146 arrangement is the energetically least favorable for an Ag dopant,40 and we discover that the thermodynamic pathway between CO and Ih motifs contains 44 local minima (Figure 1, top). The Journal of Physical Chemistry Letters A key is provided in each image, with energies (ΔE) given relative to that of the CO motif, that is, akin to ΔE(CO →Ih). A vertical dashed line in each plot indicates where the transition from CO-like to Ih-like motifs occurs, and () indicates the lowest energy rosette-Ih minima encountered for Ag55@Au92 and is illustrated in detail in Figure 3. TO →r-Ih →Ih transitions. Through the transformation, the Ag atom is maintained in the central position of the NP, and a ring of Au atoms forms at the surface to expose a subsurface atom; the lowest energy structure encountered has just one of these rosette indentations. Calculations of the SD and DS pathways for Ag1@Au146 give ΔE(CO →Ih) and ΔE(Ih → CO) as 1.19 and 2.22 eV, which are lower than the r-Ih route and similar to the other values for Ag-doped Au NPs. Analysis of the r-Ih transition pathway shows that the initial energy barriers are marginally lower than for the SD mechanism (Figure 2, top) and that the Ih and r-Ih motifs are close in energy. TO →r-Ih →Ih transitions. Through the transformation, the Ag atom is maintained in the central position of the NP, and a ring of Au atoms forms at the surface to expose a subsurface atom; the lowest energy structure encountered has just one of these rosette indentations. Calculations of the SD and DS pathways for Ag1@Au146 give ΔE(CO →Ih) and ΔE(Ih → CO) as 1.19 and 2.22 eV, which are lower than the r-Ih route and similar to the other values for Ag-doped Au NPs. Analysis of the r-Ih transition pathway shows that the initial energy barriers are marginally lower than for the SD mechanism (Figure 2, top) and that the Ih and r-Ih motifs are close in energy. Initially, the stabilities of the monometallic Ag147 and Au147 NPs were calculated: DNEB pathways were identiﬁed with one barrier for the forward (CO →Ih) and backward (Ih →CO) transitions. The pathways follow the SD and DS routes, respectively, with the respective barrier energies, ΔE(CO →Ih) and ΔE(Ih →CO), being 0.50 (1.14) and 3.37 (2.33) eV for Ag147 (Au147). We expanded our calculations to include larger cores of two and three concentric shells of atoms, corresponding to the central 13 and 55 atoms, respectively. DNEB calculations for Au13@Ag134 and Au55@Ag92 give ΔE(CO →Ih) as 0.40 and 0.55 eV, with all transformations via the SD mechanism. The Journal of Physical Chemistry Letters The Journal of Physical Chemistry Letters methods would be ideal, but there has been limited investigation of how to use NP composition to achieve this goal. Thus we present work in this Letter that shows how the transition barriers between diﬀerent geometries for bimetallic AuAg NPs can be altered by careful chemical arrangement and stoichiometric control, which could allow NP design through improved stability of certain structures. methods would be ideal, but there has been limited investigation of how to use NP composition to achieve this goal. Thus we present work in this Letter that shows how the transition barriers between diﬀerent geometries for bimetallic AuAg NPs can be altered by careful chemical arrangement and stoichiometric control, which could allow NP design through improved stability of certain structures. p y We have focused on the transition between the Ih and CO motifs, which are high-symmetry NP morphologies with identical nuclearities but signiﬁcant structural diﬀerences: The Ih has low internal volume and the surface is composed of 20 (111) facets, while the CO is an FCC fragment with greater volume and a mix of (100) and (111) surfaces.37 The relative stability of these motifs varies with elemental composition and also depends on the number of atoms (N) in the NPs.8,37,38 We have focused on the “magic” number of N = 147, which ensures geometric closed-shells in the structures of interest.37 We have used a range of complementary modeling techniques, coupled to empirical potentials,32,39,40 to investigate transition pathways between structures, and subsequent energy barriers, in both the thermodynamic and kinetic regime. A pathway-exploring doubly nudged elastic band (DNEB) algorithm,41 as implemented in the OPTIM packages,42,43 was used for identiﬁcation of transition pathways at 0 K, while for exploration of the free-energy landscape we used the complementing discrete path-sampling (DPS) routines from OPTIM, namely, PATHSAMPLE,42−46 as well as our own metadynamics software.30 In particular, metadynamics includes anharmonic eﬀects that are disregarded by DPS and also oﬀers a quasi-agnostic exploration of the energy landscape due to the open-ended search of the conformational space.47,48 When not identiﬁed during the transition pathway calculations, the DS/ SD pathways between local minima were tested manually using the optimization routines in GULP.49 Figure 1. Transition pathways for CO ↔Ih. Top: Ag1@Au146 and Ag13@Au134; Bottom: Ag55@Au92 and Ag55@Au80Ag12. DOI: 10.1021/acs.jpclett.6b02181 J. Phys. Chem. Lett. 2016, 7, 4414−4419 Controlling Structural Transitions in AuAg Nanoparticles through Precise Compositional Design Anna L. Gould,†,‡ Kevin Rossi,§ C. Richard A. Catlow,†,‡,∥Francesca Baletto,,§ and Andrew J. Logsdail,†,∥ From a structural design perspective, controlling these transitions via chemical The size of NPs makes them an attractive problem to address with computational techniques, and previous work using thermodynamic sampling has identiﬁed key transition pathways and barriers between structural motifs at a range of diﬀerent nuclearities,16,26−30 as well as demonstrating how the energy barriers change with chemical ordering in bimetallic NPs.31−34 Received: September 23, 2016 Accepted: October 19, 2016 Published: October 19, 2016 Received: September 23, 2016 Accepted: October 19, 2016 Published: October 19, 2016 © 2016 American Chemical Society 4414 Figure 1. Transition pathways for CO ↔Ih. Top: Ag1@Au146 and Ag13@Au134; Bottom: Ag55@Au92 and Ag55@Au80Ag12. A key is provided in each image, with energies (ΔE) given relative to that of the CO motif, that is, akin to ΔE(CO →Ih). A vertical dashed line in each plot indicates where the transition from CO-like to Ih-like motifs occurs, and () indicates the lowest energy rosette-Ih minima encountered for Ag55@Au92 and is illustrated in detail in Figure 3. Letter Letter The Journal of Physical Chemistry Letters The key is as for Figure 1, with energy barriers (ΔEi) given as a function of each individual transition between minima i and i+1 in Figure 1. A vertical dashed line in each plot indicates where the transition from CO-like to Ih-like motifs occurs, and the horizontal gray line on each panel represents the barrier height for the more direct SD transition. Figure 3. Illustrations of key structures from the transformation of Ag55@Au92. Panels a and b are the CO-like and Ih-like minima either side of the transition marked with vertical dashed blue lines in Figures 1 (bottom) and 2 (second bottom). Au and Ag atoms are shown in gold and silver, respectively. Panels c and d are the r-Ih local minima identiﬁed as the lowest energy arrangement in the transition pathway for Ag55@Au92, as marked on Figure 1. Au atoms that form parts of the ﬁve-, six- and seven-member rosette rings highlighted in red. In panel c all atoms are included, whereas in panel d 16 Au atoms have been removed from the front of the NP to show the underlying Ih symmetry of the Ag core. When calculations were repeated for Ag-decorated vertices on Au-rich NPs, rather diﬀerent behavior was observed: Au135Ag12 has forward (SD) and backward (DS) barriers of 1.27 and 2.45 eV, that is, an increase in 0.13 and 0.12 eV, respectively, compared with Au147. For the core@shell arrangements, Ag1@Au134Ag12, Ag13@Au122Ag12, and Ag55@ Au80Ag12, only the latter now transforms via the r-Ih pathway. For Ag1@Au134Ag12 and Ag13@Au122Ag12, ΔE(CO →Ih) is via the SD pathway and equal to 1.17 and 1.54 eV, respectively, but for Ag55@Au80Ag12, ΔE(CO →Ih) is 2.37 eV, which is a large decrease of 0.70 eV compared with the initial core@shell arrangements (Figure 1, bottom). We note that the transition pathway has signiﬁcantly fewer steps and that the TO →r-Ih transition is in the center of the pathway, as the complex rotation of Au atoms on the r-Ih surface is prevented. A calculation for Ag55@Au80Ag12 transforming via the SD pathway gives a barrier of only 0.64 eV for ΔE(CO →Ih), and so again we conclude that the r-Ih dominates the DNEB calculations due to its low initial barriers and low-energy intermediate minima (Figure 2, bottom). For the backward transition (Ih →CO), ΔE is 2.20, 2.30, and 3.31 eV for Ag1@ Au134Ag12, Ag13@Au122Ag12, and Ag55@Au80Ag12, respectively. The Journal of Physical Chemistry Letters In this case, ΔE(CO →Ih) increases to 2.03 eV and ΔE(Ih →CO) to 3.01 eV compared with Au147. The transition goes through a variety of truncated octahedra (TO) and rosette-Ih (r-Ih) motifs,16,29 proceeding via sequential CO → 4415 DOI: 10.1021/acs.jpclett.6b02181 J. Phys. Chem. Lett. 2016, 7, 4414−4419 The Journal of Physical Chemistry Letters Letter Figure 2. Individual barriers for transition along the CO →Ih pathways in Figure 1. Top: Ag1@Au146; Second: Ag13@Au134; Third: Ag55@Au92; Bottom: Ag55@Au80Ag12. The key is as for Figure 1, with energy barriers (ΔEi) given as a function of each individual transition between minima i and i+1 in Figure 1. A vertical dashed line in each plot indicates where the transition from CO-like to Ih-like motifs occurs, and the horizontal gray line on each panel represents the barrier height for the more direct SD transition. The Journal of Physical Chemistry Letters and high-symmetry Ih are competitive in energy, but for Ag55@ Au92 the r-Ih is substantially lower in energy (Figure 1, bottom). Interestingly, calculations of the SD and DS pathway for Ag13@Au134 and Ag55@Au92 give ΔE(CO →Ih) as 1.28 and 0.90 eV and ΔE(Ih →CO) as 2.08 and 1.66 eV, respectively. These direct barriers are lower than via the r-Ih motifs, but the lower initial barriers to form the low-symmetry r-Ih motifs, especially in the case of the Ag13 core, mean that the transformation is taken along (and trapped in) the r-Ih pathway (Figure 2). Next, we studied the eﬀect of decorating the NP vertices with a secondary species, further increasing the mixing of the Au and Ag atoms. Such controlled decoration has been achieved for AuPd nanoparticles50 and is a logical progression when trying to design chemically bimetallic NPs. Initially, the 12 vertices of pure Ag147 were decorated to create Ag135Au12, with ΔE(CO → Ih) and ΔE(Ih →CO) calculated as 0.35 and 3.50 eV, thus showing a decrease and an increase, respectively, compared with Ag147. When the vertices of Au1@Ag146, Au13@Ag134, and Au55@Ag92 were decorated to give Au1@Ag134Au12, Au13@ Ag122Au12, and Au55@Ag80Au12, similar trends were observed: ΔE(CO →Ih) universally decreases by ∼0.1 eV to give 0.38, 0.27, and 0.38 eV, respectively; for ΔE(Ih →CO), the transition barriers are 3.38, 3.40, and 2.80 eV, which is a slight increase compared with Ag147. Figure 2. Individual barriers for transition along the CO →Ih pathways in Figure 1. Top: Ag1@Au146; Second: Ag13@Au134; Third: Ag55@Au92; Bottom: Ag55@Au80Ag12. DOI: 10.1021/acs.jpclett.6b02181 J. Phys. Chem. Lett. 2016, 7, 4414−4419 The Journal of Physical Chemistry Letters The Journal of Physical Chemistry Letters Letter Figure 4. Metadynamics landscape reconstruction in the stacking fault number (SFN) and maximum pair distribution distance (MPDD) collective variable space30 for simulations starting in the Ih basin, with free energy (ΔF) relative to the Ih motif reported at 100 K. Top: Au55@Ag92 free-energy landscape connecting Ih (A) and CO (B) basins via a DSD mechanism, as shown by the putative saddle point (C). Bottom: Rosette formation hinders the appearance of the CO basin in the chosen collective variable space for Ag55@Au92. Potential energy (ΔE) of the most relevant minima is reported to show their relative stability against the Ih motif. Morphologies are displayed using atomic models with gray and yellow spheres representing silver and gold atoms, respectively. Figure 4. Metadynamics landscape reconstruction in the stacking fault number (SFN) and maximum pair distribution distance (MPDD) collective variable space30 for simulations starting in the Ih basin, with free energy (ΔF) relative to the Ih motif reported at 100 K. Top: Au55@Ag92 free-energy landscape connecting Ih (A) and CO (B) basins via a DSD mechanism, as shown by the putative saddle point (C). Bottom: Rosette formation hinders the appearance of the CO basin in the chosen collective variable space for Ag55@Au92. Potential energy (ΔE) of the most relevant minima is reported to show their relative stability against the Ih motif. Morphologies are displayed using atomic models with gray and yellow spheres representing silver and gold atoms, respectively. the anharmonic contributions can lower energetic barriers, which explains part of the diﬀerence between our DPS and metadynamics results. Importantly, an increase in T does not change the mechanism of structural transition for either calculation method, with chemical composition and ordering having a much stronger inﬂuence on barriers. similar to those observed in the DNEB calculations (Figure 4, bottom). In general, the metadynamics simulations give more rapidly decreasing barriers, with ΔF(CO →Ih) reduced from 0.4 eV at 50 K to 0.1 eV at T = 150 K for all Ag-rich systems, including those with Au-decorated vertices. More detailed calculations were also pursued for single Au dopants in Ag147, that is, Au1Ag146, with the position of the Au atoms altering the barriers by at most 0.1 eV, and all barriers inversely correlated with T. The Journal of Physical Chemistry Letters In conclusion, we have shown that the transition barrier between high-symmetry geometric motifs for bimetallic NPs is not merely a linear interpolation between systems but instead is strongly dependent on chemical arrangements. In particular, we have highlighted the complexity of the potential and free- energy landscapes for systems where there are low-symmetry geometric motifs energetically competitive with the “high- symmetry” arrangements often pursued by theoretical and experimental scientists alike. In addition, we have also shown that the transition pathway can be controlled via careful construction of the NP with respect to stoichiometry and chemical arrangements, speciﬁcally by the use of vertex-doping to restrict surface-based structural transformations. Our results oﬀer potential for future work in bespoke nanocatalyst and nanoplasmonic design, where structural stability due to speciﬁc geometric features, such as surface facets and vertex decoration, could be facilitated by careful consideration of the composition of the NPs in question. For the backward transition, DPS shows that ΔF(Ih →CO) decreases at varying rates depending on the composition: For Ag147, ΔF(Ih →CO) decreases by 0.02 eV per 100 K, but for Au147 the rate is doubled to 0.04 eV per 100 K. For Ag-rich core@shell motifs, that is, with an Ag shell, the rate of decrease remains at 0.02 eV per 100 K, while for Au-rich motifs the clear anomaly is Ag55@Au92, for which ΔF(Ih →CO) initially decreases from 3.98 (0 K) to 3.94 eV (300 K) and then increases back up to 3.98 eV (500 K) due to the energetic variation in TO structures close to the CO motif. Again, analysis with metadynamics gives slightly lower barriers than DPS, with Ag-rich systems displaying ΔF(Ih →CO) = 2.85 ± 0.15 eV at 50 K, and these barrier heights are maintained at 100 and 150 K (Figure 4, bottom). For single atom dopants in Au1Ag146, the barriers are calculated to be 2.8 to 3.0 eV at 50 K and steady in this energy range up to 150 K. The Journal of Physical Chemistry Letters To put our results in the context of experimental investigations, we proceed to free-energy calculations, which we have achieved via both DPS and metadynamics method- ologies for pure and core@shell motifs. Calculations using DPS for temperatures (T) of 100 to 500 K show an almost universal decrease in the free energy of activation for the forward transition, ΔF(CO →Ih): a drop of 0.03 to 0.04 eV occurs when going from 0 to 100 K, whereafter the rate of decrease is reduced to ∼0.01 eV per 100 K. The only exception is the complicated case of Ag55@Au92, where ΔF(CO →Ih) increases by 0.04 eV per 100 K due to the transition states increasing in energy; it is also noted that the r-Ih has high entropy with respect to the CO and Ih, making it a more favorable minima with increasing T.51 The open-ended metadynamics simu- lations identify various sets of Ih- and CO-like geometries, Figure 3. Illustrations of key structures from the transformation of Ag55@Au92. Panels a and b are the CO-like and Ih-like minima either side of the transition marked with vertical dashed blue lines in Figures 1 (bottom) and 2 (second bottom). Au and Ag atoms are shown in gold and silver, respectively. Panels c and d are the r-Ih local minima identiﬁed as the lowest energy arrangement in the transition pathway for Ag55@Au92, as marked on Figure 1. Au atoms that form parts of the ﬁve-, six- and seven-member rosette rings highlighted in red. In panel c all atoms are included, whereas in panel d 16 Au atoms have been removed from the front of the NP to show the underlying Ih symmetry of the Ag core. lowest energy NP reveals that the Ag core is an Ih motif with ﬁve-, six-, and seven-membered rosette rings formed on the NP surface (Figure 3c,d). For Ag13@Au134 the low-symmetry r-Ih 4416 DOI: 10.1021/acs.jpclett.6b02181 J. Phys. Chem. Lett. 2016, 7, 4414−4419 ■ACKNOWLEDGMENTS (17) Haruta, M.; Tsubota, S.; Kobayashi, T.; Kageyama, H.; Genet, M. J.; Delmon, B. Low-Temperature Oxidation of CO over Gold (17) Haruta, M.; Tsubota, S.; Kobayashi, T.; Kageyama, H.; Genet, M. J.; Delmon, B. Low-Temperature Oxidation of CO over Gold Supported on TiO2, α-Fe2O3, and Co3O4. J. Catal. 1993, 144, 175− 192. A.J.L. acknowledges the Ramsay Memorial Trust and University College London for the provision of a Ramsay Fellowship. A.J.L. and C.R.A.C. acknowledge ﬁnancial support from the EPSRC, U.K. (Grant Reference: EP/IO30662/1), as does K.R. (Grant Reference ER/M506357/1) and F.B. (Grant References: EP/J010812/1 and EP/G003146/1), the latter of which is through Critical Mass Grant No. 408. K.R. and F.B. also acknowledge ﬁnancial support from the Royal Society (RG 120207). We acknowledge D. Wales for his support using the OPTIM software package, N. Dimitratos and A. A. Sokol for several stimulating and useful conversations, and Jörg Sassmanshausen for continued IT support. A.G., C.R.A.C., and A.J.L. acknowledge the use of the UCL and ARCHER high-performance computing facilities, and associated support services, with the latter supported via their membership of the UK HPC Materials Chemistry Consortium (EP/L000202). Supported on TiO2, α-Fe2O3, and Co3O4. J. Catal. 1993, 144, 175− 192. (18) Calle-Vallejo, F.; Loffreda, D.; Koper, M. T.; Sautet, P. Introducing structural sensitivity into adsorption-energy scaling relations by means of coordination numbers. Nat. Chem. 2015, 7, 403−410. (19) Calle-Vallejo, F.; Tymoczko, J.; Colic, V.; Vu, Q. H.; Pohl, M. D.; Morgenstern, K.; Loffreda, D.; Sautet, P.; Schuhmann, W.; Bandarenka, A. S. Finding optimal surface sites on heterogeneous catalysts by counting nearest neighbors. Science 2015, 350, 185−189. (20) Viñes, F.; Gomes, J. R.; Illas, F. Understanding the reactivity of metallic nanoparticles: beyond the extended surface model for catalysis. Chem. Soc. Rev. 2014, 43, 4922−4939. (19) Calle-Vallejo, F.; Tymoczko, J.; Colic, V.; Vu, Q. H.; Pohl, M. D.; Morgenstern, K.; Loffreda, D.; Sautet, P.; Schuhmann, W.; Bandarenka, A. S. Finding optimal surface sites on heterogeneous catalysts by counting nearest neighbors. Science 2015, 350, 185−189. ( ) (20) Viñes, F.; Gomes, J. R.; Illas, F. Understanding the reactivity of metallic nanoparticles: beyond the extended surface model for catalysis. Chem. Soc. Rev. 2014, 43, 4922−4939. (21) Rogers, S. M.; Catlow, C. R. A.; Chan-Thaw, C. E.; Gianolio, D.; Gibson, E. K.; Gould, A. L.; Jian, N.; Logsdail, A. J.; Palmer, R. E.; Prati, L.; et al. Tailoring Gold Nanoparticle Characteristics and the Impact on Aqueous-Phase Oxidation of Glycerol. ■REFERENCES (22) Dessens-Félix, M.; Pacheco-Contreras, R.; Barcaro, G.; Sementa, L.; Fortunelli, A.; Posada-Amarillas, A. Structural Motifs of Bimetallic Pt101−x Au x Nanoclusters. J. Phys. Chem. C 2013, 117, 20967− 20974. (1) Johnston, R. L. Atomic and Molecular Clusters; Taylor and Francis: London, 2002. (1) Johnston, R. L. Atomic and Molecular Clusters; Taylor and Francis: London, 2002. (2) Heiz, U.; Landman, U. Nanocatalysis; Springer-Verlag: Berlin, 2007. (2) Heiz, U.; Landman, U. Nanocatalysis; Springer-Verlag: Berlin, 2007. (23) Bochicchio, D.; Negro, F.; Ferrando, R. Competition between structural motifs in gold-platinum nanoalloys. Comput. Theor. Chem. 2013, 1021, 177−182. (3) Ray, P. C. Size and Shape Dependent Second Order Nonlinear Optical Properties of Nanomaterials and Their Application in Biological and Chemical Sensing. Chem. Rev. 2010, 110, 5332−5365. (24) Wang, Z. W.; Palmer, R. E. Direct Atomic Imaging and Dynamical Fluctuations of the Tetrahedral Au20 Cluster. Nanoscale 2012, 4, 4947−4949. g g (4) Liu, J. P.; Fullerton, E.; Gutﬂeisch, O.; Sellmyer, D. J. Nanoscale Magnetic Materials and Applications; Springer: New York, 2009. ( ) (5) Corain, B.; Shemid, G.; Toshima, N. Metal Nanoclusters in Catalysis and Materials Science: The Issue of Size Control; Elsevier: Amsterdam, The Netherlands, 2008. (25) Wang, Z. W.; Palmer, R. E. Determination of the ground-state atomic structures of size-selected Au nanoclusters by electron-beam- induced transformation. Phys. Rev. Lett. 2012, 108, 245502. (6) Asara, G. G.; Paz-Borbón, L. O.; Baletto, F. “Get in Touch and Keep in Contact”: Interface Effect on the Oxygen Reduction Reaction (ORR) Activity for Supported PtNi Nanoparticles. ACS Catal. 2016, 6, 4388−4393. y (26) Lipscomb, W. N. Framework Rearrangement in Boranes and Carboranes. Science 1966, 153, 373−378. (27) Baletto, F.; Mottet, C.; Ferrando, R. Reentrant morphology transition in the growth of free silver nanoclusters. Phys. Rev. Lett. 2000, 84, 5544. (7) Baletto, F.; Ferrando, R. Doped golden fullerene cages. Phys. Chem. Chem. Phys. 2015, 17, 28256−28261. y (8) Ferrando, R.; Jellinek, J.; Johnston, R. L. Nanoalloys: From Theory to Applications of Alloy Clusters and Nanoparticles. Chem. Rev. 2008, 108, 845−910. (28) Baletto, F.; Mottet, C.; Ferrando, R. Microscopic mechanisms of the growth of metastable silver icosahedra. Phys. Rev. B: Condens. Matter Mater. Phys. 2001, 63, 155408. (9) Li, Z. Y.; Young, N. P.; Di Vece, M. D.; Palomba, S.; Palmer, R. E.; Bleloch, A. L.; Curley, B. C.; Johnston, R. L.; Jiang, J.; Yuan, J. Three-dimensional Atomic-scale Structure of Size-selected Gold Nanoclusters. Nature 2008, 451, 46−48. ■ACKNOWLEDGMENTS ACS Catal. 2015, 5, 4377−4384. The authors declare no competing ﬁnancial interest. The authors declare no competing ﬁnancial interest. ■ASSOCIATED CONTENT The overall diﬀerences in barriers calculated using DPS and metadynamics are deemed insigniﬁcant with the small under- estimate from metadynamics probably due to the eﬀect of reduced dimensionality. The identical pathways and similar barrier heights identiﬁed by both methods, which include the r- Ih-assisted transformations for Ag@Au, show that anharmonic eﬀects do not play a signiﬁcant role in the SD path. In addition, * S Supporting Information The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpclett.6b02181. * S Supporting Information Detailed structural analysis for the non-DSD transitions between CO and Ih of Ag1@Au146, Ag13@Au134, Ag55@ Au92, and Ag55@Au80Ag12. (PDF) Detailed structural analysis for the non-DSD transitions between CO and Ih of Ag1@Au146, Ag13@Au134, Ag55@ Au92, and Ag55@Au80Ag12. (PDF) DOI: 10.1021/acs.jpclett.6b02181 J. Phys. Chem. Lett. 2016, 7, 4414−4419 4417 Letter ■AUTHOR INFORMATION Corresponding Authors control of the activity in dealloyed core-shell fuel cell catalysts. Nat. Chem. 2010, 2, 454−460. , , (15) Wells, D. M.; Rossi, G.; Ferrando, R.; Palmer, R. E. Metastability of the atomic structures of size-selected gold nanoparticles. Nanoscale 2015, 7, 6498−6503. The Journal of Physical Chemistry Letters The Journal of Physical Chemistry Letters Notes (16) Baletto, F.; Ferrando, R. Structural properties of nanoclusters: Energetic, thermodynamic, and kinetic effects. Rev. Mod. Phys. 2005, 77, 371−423. DOI: 10.1021/acs.jpclett.6b02181 J. Phys. Chem. Lett. 2016, 7, 4414−4419 ■REFERENCES y (29) Aprà, E.; Baletto, F.; Ferrando, R.; Fortunelli, A. Amorphization Mechanism of Icosahedral Metal Nanoclusters. Phys. Rev. Lett. 2004, 93, 065502. , (30) Pavan, L.; Rossi, K.; Baletto, F. Metallic nanoparticles meet metadynamics. J. Chem. Phys. 2015, 143, 184304. (10) José-Yacamán, M.; Ascencio, J. A.; Liu, H. B.; Gardea-Torresdey, J. Structure shape and stability of nanometric sized particles. J. Vac. Sci. Technol., B: Microelectron. Process. Phenom. 2001, 19, 1091−1103. ( ) (31) Mottet, C.; Rossi, G.; Baletto, F.; Ferrando, R. Single impurity effect on the melting of nanoclusters. Phys. Rev. Lett. 2005, 95, 035501. (11) Palmer, R. E.; Pratontep, S.; Boyen, H. G. Nanostructured Surfaces from Size-selected Clusters. Nat. Mater. 2003, 2, 443−448. (32) Gould, A. L.; Logsdail, A. J.; Catlow, C. R. A. Influence of Composition and Chemical Arrangement on the Kinetic Stability of 147-Atom Au-Ag Bimetallic Nanoclusters. J. Phys. Chem. C 2015, 119, 23685−23697. (12) Yin, Z.; Zhang, Y.; Chen, K.; Li, J.; Li, W.; Tang, P.; Zhao, H.; Zhu, Q.; Bao, X.; Ma, D. Monodispersed bimetallic PdAg nano- particles with twinned structures: Formation and enhancement for the methanol oxidation. Sci. Rep. 2014, 4, 4288. (33) Chen, F.; Johnston, R. L. Martensitic transformations in Ag-Au bimetallic core-shell nanoalloys. Appl. Phys. Lett. 2008, 92, 023112. p (13) Lu, J.; Low, K.-B.; Lei, Y.; Libera, J. A.; Nicholls, A.; Stair, P. C.; Elam, J. W. Toward atomically-precise synthesis of supported bimetallic nanoparticles using atomic layer deposition. Nat. Commun. 2014, 5, 3264. (34) Chen, F.; Curley, B. C.; Rossi, G.; Johnston, R. L. Structure, melting, and thermal stability of 55 atom Ag-Au nanoalloys. J. Phys. Chem. C 2007, 111, 9157−9165. (35) Mackay, A. L. A dense non-crystallographic packing of equal spheres. Acta Crystallogr. 1962, 15, 916−918. (14) Strasser, P.; Koh, S.; Anniyev, T.; Greeley, J.; More, K.; Yu, C.; Liu, Z.; Kaya, S.; Nordlund, D.; Ogasawara, H.; et al. Lattice-strain 4418 DOI: 10.1021/acs.jpclett.6b02181 J. Phys. Chem. Lett. 2016, 7, 4414−4419 Letter The Journal of Physical Chemistry Letters The Journal of Physical Chemistry Letters (36) Chen, F.; Li, Z. Y.; Johnston, R. L. Surface reconstruction precursor to melting in Au[sub 309] clusters. AIP Adv. 2011, 1, 032105. (37) Baletto, F.; Ferrando, R.; Fortunelli, A.; Montalenti, F.; Mottet, C. Crossover among structural motifs in transition and noble-metal clusters. J. Chem. Phys. 2002, 116, 3856−3863. (38) Logsdail, A. J.; Johnston, R. L. Interdependence of Structure and Chemical Order in High Symmetry (PdAu)N Nanoclusters. RSC Adv. 2012, 2, 5863−5869. (39) Cleri, F.; Rosato, V. Tight-binding potentials for transition metals and alloys. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 48, 22−33. (40) Gould, A.; Heard, C.; Logsdail, A.; Catlow, C. Segregation effects on the properties of (AuAg) 147. Phys. Chem. Chem. Phys. 2014, 16, 21049−21061. (41) Trygubenko, S. A.; Wales, D. J. A doubly nudged elastic band method for finding transition states. J. Chem. Phys. 2004, 120, 2082− 2094. (42) Wales, D. Energy Landscapes: Applications to Clusters, Biomolecules and Glasses; Cambridge University Press, 2003. (43) Wales, D. J. Energy landscapes: calculating pathways and rates. Int. Rev. Phys. Chem. 2006, 25, 237−282. y (44) Wales, D. J. Discrete path sampling. Mol. Phys. 2002, 100, 3285−3305. (45) Wales, D. J. Some further applications of discrete path sampling to cluster isomerization. Mol. Phys. 2004, 102, 891−908. (46) Trygubenko, S. A.; Wales, D. J. Kinetic analysis of discrete path sampling stationary point databases. Mol. Phys. 2006, 104, 1497−1507. (47) Laio, A.; Parrinello, M. Escaping free-energy minima. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 12562−12566. (48) Laio, A.; Gervasio, F. L. Metadynamics: a method to simulate rare events and reconstruct the free energy in biophysics, chemistry and material science. Rep. Prog. Phys. 2008, 71, 126601. g (49) Gale, J. D.; Rohl, A. L. The General Utility Lattice Program (GULP). Mol. Simul. 2003, 29, 291−341. (50) Zhang, H.; Watanabe, T.; Okumura, M.; Haruta, M.; Toshima, N. Catalytically Highly Active Top Gold Atom on Palladium Nanocluster. Nat. Mater. 2011, 11, 49−52. (51) Doye, J. P. K.; Calvo, F. Entropic Effects on the Size Dependence of Cluster Structure. Phys. Rev. Lett. 2001, 86, 3570− 3573. 4419
https://openalex.org/W1506020195	https://sciendo.com/pdf/10.1515/pjct-2015-0017	English	null	Monitoring and remediation technologies of organochlorine pesticides in drainage water	Polish Journal of Chemical Technology	2,015	cc-by	6,170	INTRODUCTION monitored. The efﬁ ciency of advanced oxidation pro- cesses with different nano materials and bioremediation with effective microorganisms (EMs) were evaluated to achieve the total degradation of lindane. The histological changes in liver and kidney of rats treated with remediate water relative control were investigated to conﬁ rm the complete detoxiﬁ cation of lindane-contaminated water after remediation. One class of organic pollutants which has rightly gained greater attention in environmental studies is the orga- nochlorine pesticides (OCPs). They are highly persistent and toxic in nature, and one of them, dieldrin, has been suspected to be carcinogenic1. Due to their persistence in the environment and biological accumulation through the food chain, OCPs can cause environmental damage and affect human health2–3. Moreover, due to the limits of water resources in Egypt and the sharp increase of human population, the re-use of drainage water for some purpo- ses (agriculture irrigation and some industrial activities) considered a source of a major concern. However, the re-use of wastewater in agriculture purposes have a great risk on human health in Egypt. Therefore, monitoring of organic pollutants in drainage water and searching for effective remediation technologies to remove these pollutants are in demand to improve the water quality. Chemicals Organochlroine mixture standard (aldrin, dieldrin, en- dosulfan, endrin, heptachlor, heptachlor epoxide, lindane, dichlorodiphenyltrichloroethane (DDT), Dichlorodiphe- nyldichloroethylene (DDE), and dichlorodiphenyldichlo- roethane (DDD) was obtained from Chem. Service, Inc 660,USA. Lindane with purity of 99.5% was obtained from Central laboratory for Pesticides, Agriculture Re- search Centre, Giza, Egypt. Zinc oxide (99.99%) and ferric oxide (99.9%) nanoparticles were obtained from Egypt Nanotech Company Limited, El-Wahat road, 6th October, Cairo, Egypt. The zinc and ferric oxides particles size are 50 and 40 nm, respectively with a surface area of 60 and 80 m2/g, respectively. Hexane and methanol analytical grade solvents were obtained from Sigma – Aldrich Company from Chemicals , U\|SA. Hydrogen peroxide and ferric chloride El-Gomhoria Company for Chemical and Glasses, Cairo, Egypt. Advanced oxidation processes (AOPs), which are constituted by the combination of several oxidants, are characterized by the generation of very reactive and oxi- dizing free radicals in aqueous solution such as hydroxyl radicals, which posses a great destruction power4–7. Bioremediation considered one of the most envi- ronmentally-sound and cost-effective methods for the decontamination and detoxiﬁ cation of pesticides diffe- rent environmental compartments8. The technology of Effective Microorganisms (EMs) was developed during the 1970’s at the University of Ryukyus, Okinawa, Japan9. Studies have suggested that EMs may have a number of applications, including agriculture, livestock, gardening and landscaping, composting, bioremediation, cleaning septic tanks, algal control and household uses10. Monitoring of the organochlorine compounds in dra- inage water Monitoring and remediation technologies of organochlorine pesticides in drainage water Ahmed Ismail1, Aly Derbalah1, Sabry Shaheen2 1Kafr El-Sheikh University, Pesticides Chemistry and Toxicology Department, Faculty of Agriculture, 33516, Kafr El-Sheikh, Egypt 2Kafr El-Sheikh University, Soil Department, Faculty of Agriculture, 33516, Kafr El-Sheikh, Egypt Corresponding author: e-mail: aliderbalah@yahoo.com Ahmed Ismail1, Aly Derbalah1, Sabry Shaheen2 1Kafr El-Sheikh University, Pesticides Chemistry and Toxicology Department, Faculty of Agricu El-Sheikh, Egypt 2Kafr El-Sheikh University, Soil Department, Faculty of Agriculture, 33516, Kafr El-Sheikh, Egypt Corresponding author: e-mail: aliderbalah@yahoo.com This study was carried out to monitor the presence of organochlorine in drainage water in Kafr-El-Sheikh Gover- norate, Egypt. Furthermore, to evaluate the efﬁ ciencies of different remediation techniques (advanced oxidation processes [AOPs] and bioremediation) for removing the most frequently detected compound (lindane) in drainage water. The results showed the presence of several organochlorine pesticides in all sampling sites. Lindane was detected with high frequency relative to other detected organochlorine in drainage water. Nano photo-Fenton like reagent was the most effective treatment for lindane removal in drainage water. Bioremediation of lindane by ef- fective microorganisms (EMs) removed 100% of the lindane initial concentration. There is no remaining toxicity in lindane contaminated-water after remediation on treated rats relative to control with respect to histopathologi- cal changes in liver and kidney. Advanced oxidation processes especially with nanomaterials and bioremediation using effective microorganisms can be regarded as safe and effective remediation technologies of lindane in water. Keywords: lindane, remediation, toxicity, degradation, water. 015 015 Pol. J. Chem. Tech., Vol. 17, No. 1, 2 Polish Journal of Chemical Technology, 17, 1, 115 — 122, 10.1515/pjct-2015-0017 Photochemical remediation cover Kafr El-Sheikh Governorate drainage water areas. These sampling sites were selected according to their proximity to residential areas and agricultural activities. Amber glass bottles were used for sampling and were cleaned by detergent, putted in acid bath, sterilized in the oven and solvent-washed by acetone and hexane before using. Three-liter water samples were collected twice (in the spring and summer) from sampling sites. Glass bottles rinsed twice with the sample water prior to ﬁ lling and closing. Sampling of water was carried out from the body of running water; the mouth of the bottle was pointed upstream and hands downstream to avoid contamination. The samples were collected for ﬁ ve days during each season. Water samples were acidiﬁ ed with know amount of 1 molar hydrochloric acid to inhibit the biological activity of the possible excite microorganisms. All samples clearly labeled by site number and sampling date. Three replicates were collected from each sampling site and all samples were transferred to the laboratory in ice container for further treatments. The scope of the experiments included the fol- lowing treatments: Nano photo-Fenton-like reagent [Fe2O3(nano)/H2O2/UV], nano photo zinc oxide combined with hydrogen peroxide [ZnO (nano)/H2O2/UV], photo Fenton like reagent (Fe3+/H2O2/UV), and photo zinc oxide combined with hydrogen peroxide (ZnO/H2O2/ UV). For the photo-Fenton-like reagent, a UV mercury lamp model VL-4.LC with a wavelength range of 254 to 365 nm was used for the irradiation of lindane in the aqueous solution. Ferric chloride and ferric oxide nanoparticles were used as sources of the iron catalyst. The solution was prepared by adding a desired amount of lindane (5 mg/L) to ﬁ ltered El-Hokess drainage (i.e., the highly contaminated site with organochlorine compo- unds) and carefully agitating the solution. Then, freshly prepared ferric chloride or ferric oxide nanoparticles at a concentration of 50 mg/L as Fe3+ were added followed by the addition of H2O2 at a concentration of 0.05%. After that, the solution was completed with water up to 1000 mL. The initial pH of the solution was adjusted to 2.8 by using hydrochloric acid 1 M for all experiments5. The solution was transferred from the standard ﬂ ask to a quartz glass cell (1000 mL) and exposed to irradiation of the UV lamp under a constant temperature of 25°C with steering. Recovery evaluation The efﬁ cacy of the analytical steps was evaluated by fortifying distilled water samples with the mixture stan- dard of organochlorine pesticides at concentration level of 1 mg L–1 and then the analytical steps (extraction, cleaning up and determination) that mentioned were performed and replicated three times. Good recovery range (90.8–98.6%) of the tested pesticides was obtained (data not published). Gas Chromatography – Mass Spectrometry Analysis Helium (purity 99.99%) was used as a carrier gas at a constant ﬂ ow of 1 mL min–1. Initial oven temperature was set at 100°C for 2 min, followed by a linear ramp to 180°C at a rate of 5oC min–1 (hold for 2 min). Sub- sequently, the temperature was raised to 200°C at a rate of 1.5°C min–1 followed by a ramp to 250 at a rate of 20°C min–1. A ﬁ nal ramp to 280°C was performed at a rate of 30°C and a hold time of 7 min. A split–spitless injector set at 250oC was always used and injections of 1 μL were performed in the splitless mode. Transfer line temperature was set at 285°C and the source temperature at 220°C. The mass spectrometer was operated in the electron impact mode (EI). Electron multiplier voltage was set at 1700 V and the dwell time at 25 ms12. The three replicates of each sample were injected to calculate the mean concentration. Sampling sites However after remediation of pesticide residues in water, toxicity assessment is needed to directly assess the potential hazard of both original pollutants and its metabolites7. Kafr El-Sheikh (Kotshinar Drainage), Fowa (Fowa Drainage No.11), Metobess (El-Hokss Drainage), Beila (Karakat Drainage), Balteem (Hafeer Shihabeldeen Dra- inage), Nashart Drainage and El-Hamoul (El-Hamoul Drainage) were selected to be the sampling sites to In this study the presence of organochlorine pesticides in drainage water in Kafr El-Sheikh governorate was Pol. J. Chem. Tech., Vol. 17, No. 1, 2015 116 Photochemical remediation The solutions (100 μL) from the irradiated samples were removed at regular intervals (i.e., 10, 20, 40, 80, 160 and 320 min) for high-performance liquid chromatography (HPLC) analysis. HPLC analysis The irradiated samples were analyzed directly by HPLC (1100 series; Agilent Technologies, Palo Alto, California). The HPLC column used (i.d. of 4.6 mm; length of 250 mm) was ﬁ lled with Wakosil-II 5 C18-100 (Wako Pure Chemical Industries, Ltd., Osaka, Japan). A mixture of methanol and distilled water (30:70) was used as mo- bile phase under the isocratic elution mode. The ﬂ ow rate was maintained at 1 mL/min and the UV detector wavelength was adjusted to be 202 nm13. Extraction procedure Water samples (500 ml) were extracted twice with 100 ml n-hexane each time. The extracts were combined and ﬁ ltered through nylon 66 ﬁ lter (47 mm x 0.45 μm, Supelco, USA). The ﬁ ltrate was concentrated by rotary evaporation at 50oC to a volume of about 1 mL. For clean up the concentrate was then transferred directly to an activated ﬂ orisil column, and the OCP fractions were eluted with a mixture of diethyl ether/n-hexane (5:95) at a ﬂ ow rate of 5 mL/min. Finally, the eluate was concentrated again by rotary evaporation and the ﬁ nal volume of the concentrate was made up to 1.0 mL volume by hexane for the GC-MS analysis11. Each sample was extracted and cleaned up three times. g p y ( ) y For the ZnO catalyst, 5 mg/L of lindane, with the appropriate amount of ZnO or ZnO nanoparticles (300 mg/L), was shacked carefully before illumination followed by the addition of H2O2 at a concentration of 0.05%. Then, the pH was adjusted to 7, which was the optimum pH for the ZnO catalyst (Derbalah 2009). The suspension was kept in the dark for 30 min before illumination to achieve maximum adsorption of the pe- sticide onto the semiconductor surface7. The solutions from the irradiated samples were removed at regular intervals for HPLC analysis as mentioned before else- where. Each experiment was replicated three times for accurate data. Blank experiments were carried out with the tested insecticide alone under the optimum pH and dark conditions were run in parallel at all intervals to assess biotic loss of lindane. The data was negligible due to the high persistence of lindane and the short time. Monitoring of organochlorine pesticides in drainage water The analytical parameters of organochlrine pesticides and maximum residue limits of these pesticides are shown in Table (1). The results of wastewater analysis from different sampling sites in Kafr El-sheikh governorate showed the presence of several organochlorine pesticide residues (aldrin, dieldrin, endosulfan, endrin, heptachlor, heptachlor epoxide, lindane, p, p-DDT, p, p-DDE and DDD) in the two sampling times (Tables 2–3). The concentrations of organochlorine pesticides ranged from 0.01 to 0.980 μg L–1 in drainage water at all sampling sites. With the concerning the sampling sites, the results showed that El-Hokss drainage was the highest conta- minated site with organochlorine pesticides while Fowa drainage no. 11 was the lowest contaminated one. With the respect to the detection frequency, lindane was the Bioremediation technique The effective microorganisms formulation (EMs1) used for bioremediation of lindane was obtained from the Egyptian Ministry of Agriculture, Giza, Egypt. This for- mulation contains 60 species of beneﬁ cial microorganisms Pol. J. Chem. Tech., Vol. 17, No. 1, 2015 117 grown in special media and produced in Egypt under supervision of the Japanese EMRO Scientiﬁ c Organiza- tion (Okinawa, Japan). Enrichment and propagation were carried out in sterilized 250-mL Erlenmeyer ﬂ asks using 190 mL mineral salt medium (MSM)[14] and 10 mL of effective microorganisms (5 mL from the formulation) supplemented with lindane at a concentration of 5 mg/L. The cultures were incubated at 30°C, pH 7 and 150 rpm as optimum conditions for the growth of the tested effective microorganisms15. Samples were collected at 0, 3, 8, 11, 15, 19, and 23 days for monitoring the degra- dation of the tested insecticide. Control ﬂ asks of equal volumes of mineral salt liquid (MSL) medium and the tested insecticide without the effective microorganisms were run in parallel at all intervals to assess biotic loss. The collected water samples of the tested insecticide were ﬁ ltered using syringe ﬁ lter (0.2 mm)15 followed by HPLC analysis as mentioned before. Each experiment was replicated three times for accurate data. access to drinking water and food. The animals were randomly divided into ﬁ ve groups, each comprising of three animals and water samples (possibly contain lindane or its toxic metabolite) after remediation by different treatments were given to rats as oral administration. Water samples were adjusted to neutral pH, ﬁ ltered and was free of hydrogen peroxide before orally adminis- trated to rats. Control group rats was fed with normal diet and given oral dose containing no lindane. After 21 days, the rats were scariﬁ ed under anesthesia and the kidney and liver organs were removed and prepared for histopathological examination according to the method described by Bancroft and Stevens16. The histopathology test was carried out at Department of Histopathology, Faculty of Veterinary Medicine, Cairo University Egypt. RESULTS AND DISCUSSION Monitoring of organochlorine pesticides in drainage water Toxicity test It is important to note that most of these organochlorines were virtually phased out many years ago and their presence in water residues were from past applications. Firstly, this is attributable to the persistent nature of these compounds. Secondly, water from the Nile originates from the African plateau and crosses eight countries before reaching Egyptian territory (e.g., Sudan, Ethiopia, Uganda, Tanzania, Kenya, Zaire, Rwanda and Burundi). While ﬂ owing through these countries, the Nile River is loaded with various types of pesticides and many other contaminants. Thus, it arrives in Egypt after already being contaminated with different pollutants, including the persistent chlorinated pesticides19. Thirdly, combustion of domestic wastes is a potential source of PTS in the Egyptian environment with a decreasing abundance in the order PAHs>PCBs> DDTs> HCBs>chlordane>HCHs> endosulfan20. Fo- urthly, Nile River ﬂ owing through Kafr El-Zayat City which contained one of the largest pesticides factory in Egypt that ﬂ ows his drainage contaminated water to Nile water , therefore, the Nile River is loaded with various types of pesticides before reaching Kafr-El-Sheikh Governorate20. In addition, organochlorines still have limited use in Egypt as a rodenticide and termiticide20. Figure 1. Degradation of lindane at initial concentration of 5 mg/L in wastewater Fe2O3(nano)/H2O2/UV, Fe+3/ H2O2/UV, ZnO(nano)/H2O2/UV and ZnO/H2O2/UV systems Figure 1. Degradation of lindane at initial concentration of 5 mg/L in wastewater Fe2O3(nano)/H2O2/UV, Fe+3/ H2O2/UV, ZnO(nano)/H2O2/UV and ZnO/H2O2/UV systems The results showed that, the degradation rate of lin- dane was enhanced by irradiation under Fe2O3 (nano)/ H2O2/UV and ZnO (nano)/H2O2/UV systems relative to the degradation under other photochemical remediation systems. This enhancement in lindane degradation rate under Fe2O3(nano)/H2O2/UV and ZnO (nano)/H2O2/UV systems compared to other photochemical irradiation systems may be due to the fact that the stabilized nano- particles offer much greater surface area and reactivity which lead to higher generation rate of hydroxyl radicals relative to the bulk materials21–22. gyp Finally, the misuse of these pesticides by concerned individuals in addition to the lack of or weak national control is behind the presence of these pesticides in water18. The occurrence of such pesticide residues in wastewater represents an environmental and health ha- zard due to the re-use in agriculture purposes. Frequent monitoring program had urgently needed in order to assess health risks associated with such contaminates especially with chronic exposure or a life-long intake of contaminated drinking water. Toxicity test To conﬁ rm the complete detoxiﬁ cation of lindane in treated water, toxicity test was conducted on rats. Lin- dane contaminated-waters after treatment with Fe3+/ H2O2/UV, Fe2O3(nano)/H2O2/UV, ZnO/H2O2/UV, ZnO (nano)/H2O2/UV and EMs were orally administrated to the tested rats. This test was carried out to measure the effect of the possible remaining lindane (parent or metabolites) in the water samples after remediation on rats with respect to histological changes in liver and kidney relative to control. Adult rats (Sprague dauley) with 100–120 g of weight, obtained from Faculty of Veterinary Medicine, Kafr- -El-Sheikh University were used. Rats were housed in polypropylene cages under standard conditions with free Table 1. Analytical method parameters of OCPs by the proposed method Table 1. Analytical method parameters of OCPs by the proposed method Table 2. Mean concentration of detected organochlorine pesticides (μg L–1) at all sampling sites in spring season ation of detected organochlorine pesticides (μg L–1) at all sampling sites in spring season Pol. J. Chem. Tech., Vol. 17, No. 1, 2015 118 Table 3. Mean concentration of detected pesticide residues (μg L–1) at all sampling sites in summer season Mean concentration of detected pesticide residues (μg L–1) at all sampling sites in summer season The results in Figure 1 showed that, the irradiation under Fe2O3(nano)/H2O2/UV system gave the highest degradation rate of lindane followed by ZnO(nano)/H2O2/ UV, Fe3+/H2O2/UV and ZnO/H2O2/UV systems, respec- tively. A complete degradation of lindane (100%) was achieved under Fe2O3(nano)/H2O2/UV system followed ZnO(nano) /H2O2/UV (98%), Fe3+/H2O2/UV (96.8%) and ZnO/H2O2/UV (95.2%) systems within 320 min of irradiation time, respectively (Fig. 1). highly detected compounds while DDD was the lowest detected one in all sampling sites. The detection frequ- ency and concentration level of the detected pesticides were higher in spring relative to summer season. highly detected compounds while DDD was the lowest detected one in all sampling sites. The detection frequ- ency and concentration level of the detected pesticides were higher in spring relative to summer season. The results of pesticides monitoring showed the pre- sence of several organochlorine compounds in drainage water and this are in agreement with those reported by Abd-Allah and Hesham17 and Ashry et al.18. Spite of some pesticides still present in wastewater after treatment, their concentration level was lower than the maximum residue limits (MRLs) according to Egyptian water qu- ality Standards (Tables 2–3). Toxicity test After 100 min of irradiation time, the degradation rate of the remaining lindane was quite slower than the ﬁ rst 100 min under all photochemical remediation systems. This is might be due to the low remaining concentration of lindane (lower than 20% of its initial concentration) after 100 min of irradiation time which lead to high delivery rate of Fe3+/H2O2 and ZnO/H2O2 systems corresponds to higher concentrations of these reagents, and this subsequently increase their ability to Degradation of lindane by advanced oxidation processes Biodegradation of lindane at initial concentration of 5 mg/L in wastewater by effective microorganisms (EMs) With the concerning the bioremediation of lindane, effective microorganisms showed high degradation ability against lindane in drainage water. This is may be due to that the effective microorganisms is not one microorga- nism but a mixture of microorganisms40 It is also described as a multi-culture of coexisting anaerobic and aerobic beneﬁ cial microorganisms41. Therefore, its degradation ability to lindane may be faster and effective than using one microorganism. The main species involved in EMs include: lactic acid bacteria (Lactobacillus plantarum, Lactobacillus casei and Streptoccus lactis), photosynthetic bacteria (Rhodopseudomonas palustrus and Rhodobacter spaeroides) yeasts (Saccharomyces cerevisiae and Candida utilis), actinomycetes (Streptomyces albus, and Streptomy- ces griseus) and fermenting fungi (Aspergillus oryzae and Mucor hiemalis)42. The degradation rate of lindane under Fe2O3(nano)/ H2O2/UV system was higher than that under Fe3+/ H2O2/UV system and this is may be due to the effect of nano ferric oxide particle size which agree with28–29 who developed a new catalyst using nanosize particles with a high surface area that can accelerate the photo Fenton-like reaction by increasing the hydroxyl radicals generation rate. The ferric and zinc oxide nanocatalysts are very reac- tive because the active sites are located on the surface. As such, they have a low diffusional resistance, and are easily accessible to the substrate molecules. Nanocatalysis is but one of the many practical applications of nano- technology which is concerned with the synthesis and functions of materials at the nanoscale range (lower than 100 nm)30–32. An important feature of nanomaterials is that their surface properties can be very different from those shown by their macroscopic or bulk counterparts33. As the term suggests, ‘nanocatalysis’ uses nanoparticles and nanosize porous supports with controlled shapes and sizes34. The application of nanoparticles as catalysts of the Fenton-like and photo-Fenton reactions has been described by several investigators28, 29, 35–37. In comparison with their microsize counterparts, nanoparticles show higher catalytic activities because of their large speciﬁ c surface where catalytically active sites are exposed38. The advantage of using nanoparticles as catalysts for Fenton-like reagent would more than offset the disadvan- tage (associated with the use of iron(III) catalysts) of requiring ultraviolet radiation to accelerate the reaction. Form all previous discussion, ferric oxide and zinc oxide nanoparticles are potentially useful for remediation of lindane polluted sites39. Degradation of lindane by advanced oxidation processes The ﬁ rst parameter considered in this study was the losses in lindane concentration with the irradiation time. Pol. J. Chem. Tech., Vol. 17, No. 1, 2015 119 compete with lindane as hydroxyl radical scavengers (eqs. 1, 2)5, 7, 23–25. Also, chloride and carbonate ions naturally present in water react as hydroxyl radical scavengers26 as shown in equations 3 and 4. Biodegradation of Lindane using effective microorga- nisms (EMs) The degradation ability of the effective microorganisms to lindane was illustrated in Figure 2. The effective mi- croorganisms showed high potential in the degradation of the tested insecticide. Nearly 99% of lindane initial concentration (5 mg/L) was degraded within three weeks of incubation with the effective microorganisms. q Fe2+ + .OH → Fe3+ + + OH (1) .OH + H2O2 → HO2 . + H2O (2) Cl– + .OH → Cl. + + OH– (3) CO3 –2 + .OH → CO3 –. + OH– (4) (4) Figure 2. Biodegradation of lindane at initial concentration of 5 mg/L in wastewater by effective microorganisms (EMs) The degradation rate of lindane under Fe2O3(nano)/ H2O2/UV and Fe3+/H2O2/UV systems was higher than that under ZnO(nano)/H2O2/UV and ZnO/H2O2/UV systems. The degradation rate of lindane under Fe2O3(nano)/ H2O2/UV and Fe3+/H2O2/UV systems was higher than that under ZnO(nano)/H2O2/UV and ZnO/H2O2/UV systems. This is may be due to the high generation rate of hy- droxyl radicals under photo Fenton like reagent (nano or normal) relative to photo zinc oxide combined with hydrogen peroxide (nano or normal)7. This is may be due to the high generation rate of hy- droxyl radicals under photo Fenton like reagent (nano or normal) relative to photo zinc oxide combined with hydrogen peroxide (nano or normal)7. The degradation rate of lindane under ZnO(nano)/ H2O2/UV system was higher than that under ZnO/H2O2/ UV system and this is may be due to the effect of par- ticle size of nano zinc oxide. The effect of particle size on the photodegradation efﬁ ciency can be ascribed to two reasons. 1) When the size of ZnO crystals decreases, the amount of the dispersion particles per volume in the solution will increase, resulting in the enhancement of the photon absorbance. 2) The surface area of ZnO photocatalyst will increase as the size of ZnO crystals decreases, which will promote the adsorption of more insecticide molecules on the surface27. Figure 2. Degradation of lindane by advanced oxidation processes ) As a conclusion, effective microorganisms could be used in various kinds of aerobic and anaerobic systems for treating agricultural wastes which represent the ﬁ rst point of discharge of many chemicals into environment. The effective and stable degradation capacity of this EMs technology in utilizing and degrading this compound reﬂ ected their efﬁ cacy in biotechnological application for the bioremediation of such contaminated water. These results indicated that EMs are more stable in retaining their ability to completely degrade lindane because these effective microorganisms live in symbio- tic relationships and their inﬂ uence on the lindane are sum of all activities of these microorganisms. Where the metabolites formed by one type of microorganism may be utilized by other group of microorganisms. This study so far suggested that microorganisms endowed with this property of degradation of toxic pollutants are a boon to mankind. Future studies on the genes responsible for enhanced biodegradation will enable us to elucidate the exact degradation pathway involved in its microbial biodegradation. Pol. J. Chem. Tech., Vol. 17, No. 1, 2015 120 Toxicity assessment the different remediation processes, some rats treated with Fe2O3(nano)/H2O2/UV (Fig. 3B), Fe3+/H2O2/UV (Fig. 3C), ZnO(nano)/H2O2/UV (D), ZnO/H2O2/UV (Fig. 3E) and effective microorganisms (Fig. 3F) without lindane and the kidney tissues were normal like control (data not published). The histopathological changes in the kidney The normal structure of kidney tissue is shown in Figure 3A. For the rats treated with lindane after re- mediation with Fe2O3(nano)/H2O2/UV (Fig. 3B), Fe3+/ H2O2/UV (Fig. 3C), ZnO(nano)/ H2O2/UV (Fig. 3D), ZnO/H2O2/UV (Fig. 3E) and effective microorganisms (Fig. 3F), the tissues were normal like control (Fig. 3B) but for preiveascular oedema (Fig. 3C), small vaculations of epithelial lining renal tubules (Figs. 3D, E) as will as glomeular tults and epithelial lining renal tubules (Fig. 3F). To conﬁ rm the safety of materials used in LITRATURE CITED 1. Tomkins, B.A., Merriweather, R., Jenkins, R.A. & Bayne, C.K. (1992). J. Assoc. Off. Anal. Chem. Int. 75, 1091–1099. ( ) ff 2. Colborn, T., Dumanoski, D. &Myers, J.P. (1996). Our Stolen Future. Dutton, NY, SA. 3. Li, J., Zhang, G., Guo, L.L., Xu, W.H., Li, X., Dlee, C.S.L., Ding, A.J. & Wanf, T. (2007). Organochlorine pesticides in the atmosphere o Guangzhou and Hong Kong: regional sources and long-range atmospheric transport. Atmospheric Environ. 41, 3889–3903. DOI: 10.1016/j.atmosenv.2006.12.052. 4. Benitez, F.J., Acero, J.L. & Real, F.J. (2002). Degradation of carbofuran by using ozone, UV radiation and advanced oxidation processes. J. Hazard. Mat. 89, 51–65. In terms of the removal of used nonmaterials from water after treatment, these could be removed by the addition of natural colloids that make aggregation and sedimentation of these materials and remove them from water43. In addition, membranes with suitable porous structures and homogeneous pore-size distribution can separate nanoparticles (NPs) that are less than 10 nm in size. Hence, ultraﬁ ltration and nanoﬁ ltration membranes are ideal for separating NPs and large molecules such as proteins because their pore sizes range from 1 to 100 nm44. Regarding removal of effective microorganisms, this bioformulation is very safe and has multiple uses such as serving as growth promoters for humans and poultry as well as it used to improve soil fertility etc. Spite of its safety and beneﬁ ts, it can be remove easy by passing treated water through nanoﬁ lter membrane. Since poly- mer ultra-ﬁ ltration membranes have been used for the separation of various foods, biological, pharmaceutical systems as well as for water puriﬁ cation44. 5. Derbalah, A.S., Nakatani, N. & Sakugawa, H. (2004). Photocatalytic removal of fenitrothion in pure and natural waters by photo-Fenton reaction. Chemosphere 57, 635–644. DOI: 10.1016/j. 6. Evgenidou, E., Konstantinou, I., Fytianos, K. & Poulios, I. (2007). Oxidation of tow organophosphorus insecticides by the photo-assisted Fenton reaction. Wat. Res. 41, 2015–2027. DOI: 10.1016/j.watres.2007.01.027. 7. Derbalah, A.S. (2009). Chemical remediation of carbofuran insecticide in aquatic system by advanced oxidation processes. J. Agric. Res. Kafr Elsheikh Univ. 35 (1), 308–327. 8. Derbalah, A.S.H. & Belal, E.B. (2008). Biodegradation kinetics of cymoxanil in Aquatic system. Chem. Ecol. 3, 169–180. DOI: 10.1080/02757540802032173. 9. Sangakkara, U.R. (2002). The Technology of Effective Microorganisms-Case Studies of Application’ Royal Agricultural College, Cirencester, UK Research. 10. EM Technology (1998). Effective Microorganisms for a Sustainable Agriculture and Environment. LITRATURE CITED From Link http:// emtech.org/prod01.htmm This study relative to other previous studies used ef- fective microorganisms’ formulation for the ﬁ rst time in lindane biodegradation or bioremediation. This is considered a ﬁ rst step for using this safe and effective formulation in the ﬁ eld of wastewater treatment. More- over, the histology technique to conﬁ rm the total detoxi- ﬁ cation of lindane in remediated water sound interest. Moreover, the toxicity of nanomaterials and effective microorganisms itself was evaluated with respect to histological change in kidney and liver relative to control (data not published) and this also reﬂ ect the need to evaluate the side effects or the safety of nanomaterials not only its efﬁ cacy. 11. Xu, X., Yang, H., Li, O.,Yang, B., Frank, X.W. & Lee, S.C. (2007). Residues of organochlorine pesticides in near shore waters of Lai Zhou Bay and Jiao Zhou Bay, Shandong Peninsula, China. Chemosphere 68, 126–139. DOI: 10.1016/j. chemosphere.2006.12.021. 12. Papadakis, E.N., Vryzas, Z. & Papadopoulou-Mourkidou, E. (2006). Rapid method for the determination of 16 orga- nochlorine pesticides in sesame seeds by microwave-assisted extraction and analysis of extracts by gas chromatography–mass spectrometry. J. Chromat. A, 1127, 6–11. 13. Ezemonye, L.I., Ikpesu, T.O. & Tongo, I. (2008). Distri- bution of lindane in water, sediment, and ﬁ sh from the Warri river of the Niger delta, Nigeria. Arh. High Rad. Toksikol. 59, 261–270. DOI: 10.2478/10004-1254-59-2008-1906. ACKNOWLEDGEMENT The authors thanks the ﬁ nancial supports providing from University of Kafrelsheikh, sector of postgraduate and research affairs, University research fund (Project No. KFUR03) in 2010. ( p ) To evaluate the efﬁ cacy of different tested remediation techniques in removing lindane from wastewater, toxicity assessment was carried out with respect to histology test. The histology test for all remediation techniques of linda- ne in wastewater showed no signiﬁ cant changes in kidney or liver of treated rats relative to control treatment. This is implies the complete detoxiﬁ cation of lindane and its possible toxic products in treated wastewater with diffe- rent remediation techniques. Also, this is implying the safety of all tested chemical and biological remediation techniques on human health especially when we extend the remediation time. The histopathological changes in the liver The normal structure of liver tissue is shown in Figure 4A. For the rats treated with lindane after re- mediation with Fe2O3(nano)/H2O2/UV (Fig. 4B), Fe3+/ H2O2/UV (Fig. 4C), ZnO (nano)/H2O2/UV (Fig. 4D), ZnO/H2O2/UV (Fig. 4E) and Effective microorganisms (Fig. 4F), the tissues were like control but for hydropic Figure 3. Sections in kidney of rats treated with lindane after remediation with Fe2O3(nano)/H2O2/UV (B), Fe+3/H2O2/UV (C), ZnO(nano)/H2O2/UV (D) and ZnO/H2O2/UV (E) and Effective microorganisms (F) relative to control (A) Figure 3. Sections in kidney of rats treated with lindane after remediation with Fe2O3(nano)/H2O2/UV (B), Fe+3/H2O2/UV (C), ZnO(nano)/H2O2/UV (D) and ZnO/H2O2/UV (E) and Effective microorganisms (F) relative to control (A) Figure 4. Sections in liver of rats treated with lindane after remediation with Fe2O3(nano)/H2O2/UV (B), Fe+3/H2O2/UV (C), ZnO (nano)/ H2O2/UV (D) and ZnO/H2O2/UV (E) and Effective microorganisms (F) relative to control (A) Figure 4. Sections in liver of rats treated with lindane after remediation with Fe2O3(nano)/H2O2/UV (B), Fe+3/H2O2/UV (C), ZnO (nano)/ H2O2/UV (D) and ZnO/H2O2/UV (E) and Effective microorganisms (F) relative to control (A) Pol. J. Chem. Tech., Vol. 17, No. 1, 2015 121 degeneration of hepatocytes (Figs. 4B, C, F), congestion of central vein and hydropic degeneration in hepatocytes (Fig. 4D) and congestion of hepatic sensoids as well as vacuolization of hepatocytes (Fig. 4E). To conﬁ rm the safety of materials used in the different remediation processes, some rats treated with Fe2O3(nano)/H2O2/UV (Fig. 4B), Fe3+/H2O2/UV (Fig. 4C), ZnO(nano)/H2O2/ UV (Fig. 4D), ZnO/H2O2/UV (Fig. 4E) and effective microorganisms (F) without lindane and liver tissues were normal like control (data not published). degradation and detoxiﬁ cation of lindane in contaminated drainage water. Effective microorganisms’ formulation is promising as effective and safe bioremediation technique for lindane removal in drainage water. CONCULSIONS 14. Abdel Megeed, A.A. & Elnakeeb, A. (2008). Bioremedia- tion of dimethoate by effective microorganisms in contaminated water. Terrestrial Aquat. Environ. Toxicol. 2, 1–4. These results clearly indicate the presence of numerous organochlorine compounds in drainage water sampling sites in Kafr El-Sheikh Governorate. The photo-Fenton like reagent and photo zinc oxide combined with hy- drogen peroxide showed much promise in the complete 15. Derbalah, A.S., Massoud, A.H. & Belal, E.B. (2008). Biodegrability of famoxadone by various microbial isolates in aquatic system. Land Contamination & Reclama 16 (1), 13–23. DOI: 10.2462/09670513.876. 122 Pol. J. Chem. Tech., Vol. 17, No. 1, 2015 35. Feng, J., Hu, X. & Yue, P.L. (2004 a). Novel bentonite clay-based Fe-nanocomposite as a heterogeneous catalyst for photo-Fenton discoloration and mineralization of Orange II. Environ. Sci. Technol. 38, 269–275. DOI: 10.1021/es034515c. 16. Bancroft, J.D. & Stevens, A. (1996). Theory and Practi- ceof Histological Techniques. (4th ed.). Churchill Livingstone. Edinburg, London, Melbourne and New York. 17. Abd-Allah, S.W. & Hesham, M.G. (2003). Monitoring of pesticide residues in different sources of drinking water in some rural areas. Alex. J. Agric. Res. 48 (3), 187–199. 36. Feng, J., Hu, X. & Yue, P.L. (2004 b). Discoloration and mineralization of Orange II using different heterogeneous catalysts containing Fe: a comparative study. Environ. Sci. Technol. 38, 5773–5778. DOI: 10.1021/es049811j. 18. Ashry, M.A., Bayoumi, O.C., El-Fakharany, I.I., Der- balah, A.S. & Ismail, A.A. (2006). Monitoring and removal of pesticidesresidues in drinking water collected from Kafr El-Sheikhgovernorate, Egypt. J. Agric. Res. Tanta Univ. 32 (3), 691–704. 37. Zelmanov, G., Semiat, R. (2008). Iron(3) oxide-based nanoparticles as catalysts in advanced organic aqueous oxida- tion. Wat. Res. 42, 492–498. DOI: 10.1016/j.watres.2007.07.045. 38. Nurmi, J., Tratnyek, P.G., Sarathy, V., Baer, D.R., Amo- nette, J.E., Pecher, K., Wang, C., Linehan, J.C., Matson, D.W., Penn, R.L. & Driessen, M.D. (2005). Characterization and properties of metallic iron nanoparticle: spectroscopy, elec- trochemistry,and kinetics. Environ. Sci. Technol. 39, 1221–1230. DOI: 10.1021/es049190u. 19. El-Sebae, A.H., Abou-Zeid, M.M. & Saleh, M.A. (1993). Status and environmental impact of toxaphene in the Third World a case study of African agriculture. Chemosphere 27, 2063–2072. 20. Barakat, A.O. (2003). Persistent organic pollutants in smoke particles emitted during open burning of municipal solid wastes. Bull. Environ. Contamin. Toxicol. 70, 174–181. DOI: 10.1007/s00128-002-0172-9. 39. Liu, W.T. (2006). Nanoparticles and their biological and environmental applications. J. Biosci. Bioeng. 102, 1–7. DOI: 10.1263/jbb.102.1 21. He, F., Zhao, D., Liu, J. & Roberts, C.B. CONCULSIONS (2007). Sta- bilization of Fe–Pd nanoparticles with sodium carboxymethyl cellulose for enhanced transport and dechlorination of trichlo- roethylene in soil and groundwater. Indian Engineer. Chem. Res. 46, 29–34. DOI: 10.1021/ie0610896. 40. Higa, T. (1995). What is EM Technology. College of Agriculture, University of Ryukyus, Okinawa, Japan. 41. EM Trading (2000). Effective Microorganisms (EM) from Sustainable Community Development. From EM Technology Product Link http://www.emtrading.com.html 22. He, F. & Zhao, D. (2005). Preparation and characteriza- -for degradation of chlorinated hydrocarbons in water. Environ. Sci. Technol. 39, 3314–3320. DOI:10.1021/es048743y. 42. Diver, S. (2001). Nature Farming and Effective Microor- ganisms’,Rhizosphere II: Publications. from Steve Diver Link http://ncatark.uark.edu/~steved/Nature-Farm-EM.html y 23. Catastini, C., Sarakha, M., Mailhot, G. & Bolte, M. (2002). Iron (III) Aquacomplesxes as effective photocatalysis for the degradation of pesticides in homogenous aqueous solutions. Sci. Tot. Environ. 298, 219–228. 43. Quik, J.T.K., Velzeboer, I., Wouterse, M., Koelmans, A.A. & Van de Meent, D. (2014). Heteroaggregation and Sedimen- tation Rates for Nanomaterials in Natural Waters. Wat. Res. 48, 269–279. DOI: 10.1016/j.watres.2013.09.036. 24. El-Morsi, T.M., Emara, M.M., Abd El Bary, H.M.H., Abd-El-Aziz, A. & Friesen K.J. (2002). Homogeneous de- gradation of 1,2<10 tetrachlorodecane in aqueous solutions using hydrogen peroxide, iron and UV light. Chemosphere 47, 343–348. DOI: 10.1016/S0045-6535(01)00305-8. j 44. El-Safty, S.A., Hoa, N.D. & Shenashen M.A. (2012). Topical Developments of Nanoporous Membrane Filters for Ultraﬁ ne Noble Metal Nanoparticles. Eur. J. Inorg. Chem. 5439–5450. DOI: 10.1002/ejic.201200629. ( ) 25. Wang, Q. & Lemely, A.T. (2002). Oxidation of diazi- non by anodic Fenton treatment. Wat. Res. 36, 3237–3244. DOI: 10.1016/S0043-1354(02)00041-6. 26. Pare, B.P., Singh, S. & Jonnalagadda, B. (2008). Visible light induced heterogeneous advanced oxidation process to degrade pararosanilin dye in aqueous suspension of ZnO. Indian J. Chem. 4, 830–835. 27. Wang, H., Xie, C., Zhang, W., Cai, Z., Cai, S., Yang, Z. & Gui, Y. (2007). Comparison of dye degradation efﬁ ciency using ZnO powders with various size scales. J. Hazard. Mat.141, 645–652. DOI: org/10.1016/j.jhazmat.2006.07.021. Á 28. Valdés-Solís, T.P., Valle-Vigón, P., Álvarez, S., Marbán, G. & Fuertes, A.B. (2007 a). Encapsulation of nanosized catalysts in the hollow core of a mesoporous carbon capsule. J. Catal. 251, 239–243. DOI: 10.1016/j.jcat.2007.07.006. Á 29. Valdés-Solís, T.P., Valle-Vigón, P., Álvarez, S., Marbán, G. & Fuertes, A.B. (2007 b). Manganese ferrite nanoparticles synthesized through a nanocasting route as a highly active Fenton catalyst. Catal. Commun. 8, 2037–2042. DOI: 10.1016/j. catcom.2007.03.030. 32. Lines, M.G. (2008). Nanomaterials for practical functional uses. J. CONCULSIONS Alloys Compd. 449, 242–245. DOI: 10.1016/j. 30. Mamalis, A.G. (2007). Recent advances in nanotechno- logy. J. Mat. Process. Technol. 181, 52–58. DOI:org/10.1016/j. jmatprotec.2006.03.052. j p 31. Miyazaki, K. & Islam, N. (2007). Nanotechnology sys- tems of innovation — an analysis of industry and academia research activities. Technovation 27, 661–675. DOI: 10.1016/j. technovation.2007.05.009. 33. Yuan, G. & Theng, B.K.G. (2008). Nanopaticles in the soil environment. Elements 4, 395–399. DOI:10.2113/gs 34. Bell, A.T. (2003). The impact of nanoscience on he- terogeneous catalysis. Sci. 299, 1688–1691. DOI: 10.1126/ science.1083671.
https://openalex.org/W2886777409	https://hal.archives-ouvertes.fr/hal-03184964/file/Clinical%20Infectious%20Diseases.pdf	English	null	Community-based Malaria Screening and Treatment for Pregnant Women Receiving Standard Intermittent Preventive Treatment With Sulfadoxine-Pyrimethamine: A Multicenter (The Gambia, Burkina Faso, and Benin) Cluster-randomized Controlled Trial	Clinical infectious diseases/Clinical infectious diseases (Online. University of Chicago. Press)	2,018	cc-by	9,286	Community-based Malaria Screening and Treatment for Pregnant Women Receiving Standard Intermittent Preventive Treatment With Sulfadoxine-Pyrimethamine: A Multicenter (The Gambia, Burkina Faso, and Benin) Cluster-randomized Controlled Trial Susana Scott, Umberto D’alessandro, Lindsay Kendall, John Bradley, Kalifa Bojang, Simon Correa, Fanta Njie, Halidou Tinto, Maminata Traore-Coulibaly, Hamtandi Magloire Natama, et al. To cite this version: Susana Scott, Umberto D’alessandro, Lindsay Kendall, John Bradley, Kalifa Bojang, et al.. Community-based Malaria Screening and Treatment for Pregnant Women Receiving Standard In- termittent Preventive Treatment With Sulfadoxine-Pyrimethamine: A Multicenter (The Gambia, Burkina Faso, and Benin) Cluster-randomized Controlled Trial. Clinical Infectious Diseases, 2019, 68 (4), pp.586-596. 10.1093/cid/ciy522. hal-03184964 COSMIC Consortiuma Background. We investigated whether adding community scheduled malaria screening and treatment (CSST) with arte- mether-lumefantrine by community health workers (CHWs) to standard intermittent preventive treatment in pregnancy with sulf- adoxine-pyrimethamine (IPTp-SP) would improve maternal and infant health. Methods. In this 2-arm cluster-randomized, controlled trial, villages in Burkina Faso, The Gambia, and Benin were randomized to receive CSST plus IPTp-SP or IPTp-SP alone. CHWs in the intervention arm performed monthly CSST during pregnancy. At each contact, filter paper and blood slides were collected, and at delivery, a placental biopsy was collected. Primary and secondary end- points were the prevalence of placental malaria, maternal anemia, maternal peripheral infection, low birth weight, antenatal clinic (ANC) attendance, and IPTp-SP coverage.h Results. Malaria infection was detected at least once for 3.8% women in The Gambia, 16.9% in Benin, and 31.6% in Burkina Faso. There was no difference between study arms in terms of placenta malaria after adjusting for birth season, parity, and IPTp-SP doses (adjusted odds ratio, 1.06 [95% confidence interval, .78–1.44]; P = .72). No difference between the study arms was found for peripheral maternal infection, anemia, and adverse pregnancy outcomes. ANC attendance was significantly higher in the interven- tion arm in Burkina Faso but not in The Gambia and Benin. Increasing number of IPTp-SP doses was associated with a significantly lower risk of placenta malaria, anemia at delivery, and low birth weight. Conclusions. Adding CSST to existing IPTp-SP strategies did not reduce malaria in pregnancy. Increasing the number of IPTp-SP doses given during pregnancy is a priority. g g g y y Clinical Trials Registration. NCT01941264; ISRCTN37259296. g ; Keywords. malaria; pregnancy; sulfadoxine-pyrimethamine; artemether-lumefantrine; community-based malaria screening. pregnancy; sulfadoxine-pyrimethamine; artemether-lumefantrine; community-based malaria screening. in turn depends on antenatal clinic (ANC) attendance. In many sub-Saharan African countries, both ANC and IPTp-SP cover- age remains low [5, 6]. In addition, sulfadoxine-pyrimethamine (SP) resistance is increasing and may have an impact on cur- rent IPTp-SP policy [7]. In West Africa, where SP resistance is low [8], an alternative strategy of intermittent screening and treatment in pregnancy (ISTp) was noninferior to IPTp-SP in preventing low birth weight, anemia, and placental malaria. In southeast Africa, it was associated with a higher malaria risk, possibly because of the low sensitivity of currently available rapid diagnostic tests (RDTs) in detecting low-density infec- tions [9, 10]. WHO does not recommend ISTp alone. HAL Id: hal-03184964 https://hal.science/hal-03184964v1 Submitted on 30 Mar 2021 L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. Clinical Infectious Diseases M A J O R A R T I C L E Community-based Malaria Screening and Treatment for Pregnant Women Receiving Standard Intermittent Preventive Treatment With Sulfadoxine-Pyrimethamine: A Multicenter (The Gambia, Burkina Faso, and Benin) Cluster-randomized Controlled Trial © The Author(s) 2018. Published by Oxford University Press for the Infectious Diseases Society of America. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. DOI: 10.1093/cid/ciy522 Received 5 March 2018; editorial decision 2 June 2018; accepted 27 June 2018; published online June 29, 2018. aMembers of the COSMIC (community-based scheduled screening and treatment of malaria in pregnancy for improved maternal and infant health) Consortium are listed in the Notes. Correspondence: S. Scott, Department of Infectious Disease Epidemiology, London School of Hygiene and Tropical Medicine, London, UK, WC1E 7HT. (susana.scott@lshtm.ac.uk). Clinical Infectious Diseases® 2019;68(4):586–96 Received 5 March 2018; editorial decision 2 June 2018; accepted 27 June 2018; published online June 29, 2018. Received 5 March 2018; editorial decision 2 June 2018; accepted 27 June 2018; published online June 29, 2018. COSMIC Consortiuma However, there is the need to both improve ANC coverage and protect pregnant women against malaria between ANC visits. ISTp at the village level could be beneficial if given in addition to IPTp-SP at ANC. Community health workers (CHWs) have been trained in many sub-Saharan African countries to per- form community case management of malaria, and could be trained to encourage pregnant women to attend the ANC and to systematically screen and treat them between ANC visits [11]. Malaria causes significant adverse pregnancy outcomes, such as maternal anemia, preterm delivery, low birth weight [1], and even maternal and infant death [2, 3]. The World Health Organization (WHO) recommends several interventions to control malaria during pregnancy, namely effective case man- agement, long-lasting insecticidal nets, and intermittent pre- ventive treatment with sulfadoxine-pyrimethamine (IPTp-SP) from the second trimester onward [4]. The protective efficacy of IPTp-SP against malaria infection depends on the number of IPTp-SP doses administered, which At Time of Delivery for All Communities A blood sample for hemoglobin measurement and on slide and filter paper for later parasitological diagnosis was collected just before delivery, and a placenta biopsy was collected at delivery. Current health status and birth outcomes were collected. All newborns were physically examined and weighed on digital scales immediately after delivery. Gestational age was estimated using the Ballard score [13]. Randomization and Blinding Thirty villages (clusters) (village population: 1000–2000) with CHWs in each country were randomly selected from all eligible clusters. Distance from the center of each village to the near- est health facility was calculated and used to group the clusters into 3 distance categories. Randomization was performed using computer-based randomization (Stata software, StataCorp, College Station, Texas) and was stratified by the 3 distance categories. Clinical Infectious Diseases® 2019;68(4):586–96 © The Author(s) 2018. Published by Oxford University Press for the Infectious Diseases Society of America. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. DOI: 10.1093/cid/ciy522 586 • CID 2019:68 (15 February) • COSMIC Consortium booked in. A health assessment was carried out at each ANC visit. Suspected malaria cases had an RDT (SD Bioline; spec- ificity 99.5%) and women testing positive were treated with artemether-lumefantrine (AL). A cluster-randomized controlled trial was designed to establish whether adding community scheduled malaria screening and treatment (CSST) by CHWs to standard IPTp-SP would further reduce placental malaria compared to IPTp-SP alone [12]. Community Health Worker Home Visits in the Intervention Arm CHWs in the intervention arm were trained in malaria case management and malaria in pregnancy, including the benefit of early ANC attendance and IPTp-SP. CHWs were asked to continuously identify all pregnant women and encourage them to attend the ANC as early as possible. Thereafter, at monthly intervals up to the last week of gestation, CHWs performed an RDT at home visits and collected a blood slide, regardless of malaria symptoms. They gave AL to all positive women. Severely ill women were referred to the health center for further care. The CHWs in the control villages did not take part in any of the study training. Laboratory Methods Giemsa-stained thick blood films were read by 2 experienced microscopists, with discrepancies resolved by a third one [12]. Maternal hemoglobin was measured using Hb301 Hemocue (Radiometer Group, Sweden). Blood spots on filter paper were analyzed by Plasmodium falciparum diagnostic polymerase chain reaction (PCR) [14, 15]. Study Sites and Participants A description of the study methods has been published else- where [12]. The study was implemented in 3 West African countries: Burkina Faso (Nanoro health district), The Gambia (Upper River region), and Benin (Glo-Djigbe, Zinvie, and Ze districts). In Burkina Faso and The Gambia, malaria is highly seasonal (July–December), whereas in Benin it is perennial with peaks during the rainy seasons (April–July and October– November). In all villages, community consent was obtained after sensitization meetings. All resident pregnant women were invited to participate after individual signed informed consent. Malaria Treatment in Pregnancy • CID 2019:68 (15 February) • 587 Statistical Analysis It was assumed that the intervention would decrease placental malaria from 15% to 10.5%, with a coefficient of variation of 0.15. Within each country, 15 clusters, each with 60 pregnant women per arm, would be able to detect a significant difference with 80% power and at the 5% significance level. Thus, 90 clus- ters with a total of 5400 women were required. All data were double entered using OpenClinica databases. Ethical Approval The study was done in accordance with the principles set forth in the Declaration of Helsinki and the International Conference on Harmonisation Tripartite Guidelines for Good Clinical Practice. Independent trial monitors visited each site throughout the study to ensure compliance with Good Clinical Practice standards. The trial was approved by the Gambia Government/Medical Research Council Joint Ethics Committee (reference number SCC1336), the Comité d’Ethique Institutionnel du Centre Muraz in Burkina Faso (reference number A20-2013/CE-CM), and the Comité National d’Ethique pour la Recherche en Santé in Benin (reference number 0126/MS /DC/SGM/DFR/CNERS/SA). A DSMB to review the trial procedures and results was set up. The trial is registered at Current Controlled Trials: ISRCTN37259296 (5 July 2013), and ClinicalTrials.gov: NCT01941264 (10 September 2013). Outcomes The primary outcome was placental malaria (any category). However, in Benin, rumors on placental biopsies had a negative impact on recruitment and, after discussions with the data and safety monitoring board (DSMB) and the local ethics commit- tee, it was decided to collect only peripheral blood. Secondary endpoints were maternal anemia (hemoglobin <11 g/dL) at delivery, maternal P. falciparum peripheral infection at delivery (PCR) and during pregnancy (microscopy), low birth weight (<2500 g), IPTp-SP coverage, the number of ANC visits, and the number of IPTp-SP doses. Serious adverse events (SAEs) were defined as any untoward medical occurrence that resulted in death, hospitalization, persistent or significant disability/ incapacity, or congenital anomaly/birth defect or that was life-threatening. All SAEs were reported to the DSMB. At the Antenatal Clinic Recruitment was done at first ANC (Figure 1). All women had a physical examination, a blood slide, and a blood sample on filter paper. Information on health and socioeconomic factors was also collected. Pregnant women in the second or third tri- mester were given their first IPTp-SP and their second dose Details of biopsy methods can be found elsewhere [12]. Placental biopsy slides were read by trained microscopists and Fi 1 S h d l f t d t d f th T i l t l [12] Abb i ti ANC t t l li i IPT i t itt t ti t t t i Figure 1. Schedule of events, updated from the Trial protocol paper [12]. Abbreviations: ANC, antenatal clinic; IPTp, intermittent preventive treatment in pregnancy. pdated from the Trial protocol paper [12]. Abbreviations: ANC, antenatal clinic; IPTp, intermittent preventive treatment in pregnancy. number of SP and AL doses was carried out. All other binary endpoints were examined with random effects logistic regres- sion. Count data were analyzed with mixed effects Poisson regression and continuous data were analyzed with mixed effects linear regression. classified as infected (acute: parasites and malaria pigment absent; chronic: parasites and malaria pigment; or past: only malaria pigment) or not infected (no parasites or pigment) [16]. Outcomes RESULTS Recruitment, Baseline Characteristics, and Follow-up Recruitment, Baseline Characteristics, and Follow-up Between November 2013 and November 2015, 4731 preg- nant women were recruited (Figure 2 and Supplementary Figure 1A–C), with Benin having recruited only half of the expected sample size. Loss to follow-up was low in The Gambia (113/1960 [5.8%]) and in Burkina Faso (62/1800 [3.4%]) but high in Benin (290/971 [29.9%]). Overall, 4266 (90.2%) women completed the follow-up and delivered in study (Figure 2). The primary endpoint (prevalence of placental malaria) was examined with logistic regression. Random effects for trial clus- ter were used to account for intracluster correlation. An analy- sis of the primary endpoint adjusted for season, gravidity, and Figure 2. Flowchart of the cohort of pregnant women enrolled in the study. Primary outcome is placental malaria for the Burkina Faso and The Gambia study sites and maternal peripheral infection at time of delivery for the Benin study site. Figure 2. Flowchart of the cohort of pregnant women enrolled in the study. Primary outcome is placental malaria for the Burkina Faso and The Gambia study sites and maternal peripheral infection at time of delivery for the Benin study site. 588 • CID 2019:68 (15 February) • COSMIC Consortium 588 • CID 2019:68 (15 February) • COSMIC Consortium In each country, baseline characteristics between interven- tion and control groups were similar. Use of insecticide-treated bed nets was >70%, and malaria prevalence was 5.3% in The Gambia (101/1888), 27.9% in Burkina Faso (501/1797), and 30.0% in Benin (288/959). About 20% were primigravidae (Table 1). malaria infection at least once during pregnancy (2 in the inter- vention arm and 1 in the control arm), and 2 received 1 course of AL (1 in each arm), with no drug-related serious adverse event recorded. There was no evidence that taking AL during pregnancy was associated with any adverse pregnancy outcome (at least 1 course of AL: OR, 0.82 [95% CI, .35–1.93]; P = .656). There were 9 maternal deaths (6 in the control arm and 3 in the intervention arm) and 45 perinatal deaths that occurred at time of delivery (21 in control vs 24 in intervention arm). Causes attributed to the SAEs, including maternal and perinatal death, are reported in Supplementary Table 8. The Effect of IPTp-SP on Pregnancy Outcomes Increasing number of IPTp-SP doses was associated with a significantly lower risk of placenta malaria (Table 4), mainly because of Burkina Faso where 49.2% (886/1800) of women had received ≥3 IPTp-SP doses (Supplementary Table 3A), whereas in The Gambia this figure was only 3.7% (73/1960) (Supplementary Table 3B). Placenta malaria occurred signifi- cantly less during the rainy than the dry season (aOR, 0.59 [95% CI, .49–.71]; P < .001), and was more frequent with increasing number of AL treatments (Table 4). The Effect of the Intervention on Placenta Malaria Data from placental biopsies were available for 88% (3171/3585) of deliveries in Burkina Faso and The Gambia. There was no dif- ference between intervention and control arms in terms of pla- centa malaria or for different categories of infection (Table 3), and this did not change after adjusting for season of birth, par- ity, and number of IPTp-SP administered (adjusted OR [aOR], 1.06 [95% CI, .78–1.44]; P = .72) (Table 4). Similar results were observed at country level (Supplementary Tables 2 and 3). Increasing number of IPTp-SP doses tended to decrease the risk of anemia at delivery (P = .06), whereas increasing number of AL treatments had the opposite effect (P = .02) (Table 6 and Supplementary Table 4). Similarly, the risk of low birth weight decreased significantly with the increasing number of IPTp-SP doses (P < .001; Table 7 and Supplementary Table 5). DISCUSSION Adding CSST by CHWs to the standard IPTp-SP at ANC did not reduce the risk of placental malaria or peripheral malaria infection at delivery. The intervention also aimed at increasing ANC attendance, particularly in early pregnancy, and at iden- tifying and treating infections between scheduled ANC visits. Scheduled ANC attendance did improve in Burkina Faso, sug- gesting that the intervention had the expected effect. During the trial, the 2013 WHO recommendations of at least 4 ANC visits and of administering IPTp-SP at each of them [17] had been implemented in Burkina Faso, with some women having as many as 6–7 scheduled ANC visits. In The Gambia and Benin, the national policy was still 2 scheduled visits; thus, we did not expect an increase in the number of ANC visits but rather an increase in coverage of 2 ANC visits. ANC attendance remained Malaria Treatment in Pregnancy • CID 2019:68 (15 February) • 589 RESULTS Most women in the intervention arm had at least 1 CSST by CHWs, with coverage higher in Burkina Faso (860/900 [96%]) and The Gambia (978/1008 [97%]) compared with Benin (474/542 [87.5%]). Overall, CHWs performed 7236 CSSTs, with an average of 3–4 home visits per woman (Supplementary Table 1). The mean number of slides per woman was signifi- cantly higher in the intervention than in the control group (Burkina Faso: means ratio [MR], 1.62 [95% confidence interval {CI}, 1.50–1.74]; The Gambia: MR, 2.08 [95% CI, 1.98–2.19]; Benin: MR, 2.04 [95% CI, 1.87–2.23]) (P < .001). Malaria infec- tion was detected by RDT at least once for 75 (3.8%) women in The Gambia, 164 (16.9%) in Benin, and 568 (31.6%) in Burkina Faso, and most received AL (Table 2 and Supplementary Table 1). The number of pregnant women diagnosed with a malaria infection was significantly higher in the intervention arm than in the control arm (P < .01 for all study sites). A high proportion of infected women received AL, with no difference between study arms in Burkina Faso and The Gambia. In Benin, the proportion of treated women was higher in the intervention arm (172/191 [90.0%]) than in the control arm (14/33 [42.4%]) (odds ratio [OR], 1.71 [95% CI, .99–2.94]; P = .055). ANC Attendance and IPTp-SP Uptake During Pregnancy Antenatal clinic attendance (scheduled visits) was significantly higher in the intervention arm in Burkina Faso but not in The Gambia and Benin (Table 9 and Supplementary Table 7). However, IPTp-SP coverage (ie, mean number of doses and per- centage of women who received at least 2 or 4 doses) was not significantly different between intervention and control arms. The Effect of the Intervention on Other Secondary Outcomes at Delivery The Effect of the Intervention on Other Secondary Outcomes at Delivery There was no difference between arms in peripheral maternal infection (OR, 0.97 [95% CI, .78–1.21]; P = .7), even after adjust- ing for several factors (aOR, 0.92 [95% CI, .74–1.15]; P = .4) (Table 5; country-specific data in Supplementary Table 3). Anemia (OR, 1.09 [95% CI, .92–1.28]; P = .3; Table 6 and Supplementary Table 4), mean hemoglobin (mean difference, 0.01 [95% CI, –.29 to .31]; P = .9), low birth weight (OR, 1.06 [95% CI, .82–1.38]; P = .6; Table 7 and Supplementary Table 5), and adverse pregnancy outcomes (Table 8 and Supplementary Table 6) did not differ between study arms. Nevertheless, in Burkina Faso, the odds of miscarriage tended to be higher in the intervention arm (OR, 4.54 [95% CI, .98–21.05]; P = .054; Supplementary Table 6A). Among these women, 3 had a eline Characteristics of Study Clusters at Start of the Trial, by Country Table 1. Baseline Characteristics of Study Clusters at Start of the Trial, by Country extremely low in Benin, where only slightly more than half of the women had 2 visits, and this may reflect the general poor was already high, with 80% of the women attending at least 2 scheduled visits, and this may explain why the intervention did y , y y Characteristic Burkina Faso Gambia Benin Control % Intervention % Control % Intervention % Control % Intervention % No. of clusters 15 … 15 … 15 … 15 … 15 … 15 … Median No. of women per cluster (IQR) 60 (60–60) 60 (60–60) 69 (61–98) 67 (53–95) 41 (19–55) 48 (27–62) Median age, y (IQR) 25 (20–30) 26 (21–30) 25 (20–29) 25 (20–29.5) 25 (21–29.5) 25 (20–30) Ethnic group Mandinka … … … … 238 (25.0) 332 (32.9) … … … … Fula … … … … 199 (20.9) 98 (9.7) … … … … Serahuleh … … … … 510 (53.6) 577 (57.2) … … … … Mossi 854 (94.9) 794 (88.2) … … … … … … … … Aizo/Ouemenou … … … … … … … … 389 (90.7) 473 (87.27) Other 46 (5.1) 106 (11.8) 5 (0.5) 1 (0.1) 40 (9.3) 69 (12.73) Median gestational age, wk, at time of recruitment (IQR) 22 (20–24) 22 (20–24) 20 (17–22) 20 (17–22) 20 (18–24) 20 (18–24) No. The Effect of the Intervention on Other Secondary Outcomes at Delivery of previous pregnancies 0 194 (21.6) 185 (20.6) 162 (17.1) 226 (22.4) 90 (21.0) 117 (21.6) 1 153 (17.0) 136 (15.1) 175 (18.4) 172 (17.1) 84 (19.6) 104 (19.2) 2 169 (18.8) 145 (16.1) 144 (15.2) 136 (13.5) 67 (15.6) 92 (17.0) 3 136 (15.1) 141 (15.7) 142 (15.0) 143 (14.2) 70 (16.3) 83 (15.3) ≥4 248 (27.6) 293 (32.6) 327 (34.4) 330 (32.8) 118 (27.5) 146 (26.9) Marital status Married 759 (86.0) 832 (94.7) 930 (97.7) 987 (98.0) 206 (48.0) 245 (45.2) Not married 124 (14.0) 47 (5.4) 22 (2.3) 20 (2.0) 223 (52.0) 297 (54.8) Religion Christianity 521 (58.0) 466 (52.0) … … … … 386 (90.0) 499 (92.1) Islam 215 (23.9) 306 (34.2) 935 (98.2) 986 (97.9) … … … … Traditional African religion 134 (14.9) 111 (12.4) … … … … … … … … Other 28 (4.9) 13 (1.5) 17 (1.8) 21 (2.1) 43 (10.0) 43 (7.9) Woman’s occupation Housewife 818 (91.4) 802 (90.3) 109 (11.5) 122 (12.2) 114 (26.6) 156 (29.1) Farmer/herder/gardener 47 (5.3) 67 (7.6) 719 (75.8) 760 (75.7) 42 (9.8) 46 (8.6) Sell at market/shopkeeper … … … … 62 (6.5) 65 (6.5) 144 (33.6) 151 (28.1) Childcare/domestic helper … … … … 45 (4.7) 50 (5.0) … … … … Seamstress … … … … … … … … 65 (15.2) 95 (17.7) Hairdresser … … … … … … … … 44 (10.3) 57 (10.6) Other 30 (3.4) 19 (2.1) 14 (1.5) 7 (0.7) 19 (4.4) 32 (6.0) Husband’s occupation Farmer/herdsman/fisherman/ gardener 707 (93.3) 789 (94.8) 578 (63.2) 591 (61.2) 103 (50.2) 103 (42.2) Bricklayer/carpenter/welder … … … … 36 (3.9) 46 (4.8) 20 (9.8) 28 (11.5) Sell at market/shopkeeper … … … … 84 (9.2) 66 (6.8) … … … … Imam/marabout/VHW … … … … 23 (2.5) 20 (2.1) … … … … Other 51 (6.7) 43 (5.2) 73 (8.0) 118 (12.2) 61 (29.8) 80 (32.8) Traveler … … … … 105 (11.5) 107 (11.1) … … … … Driver/motorbike driver … … … … … … … … 21 (10.2) 33 (13.5) Teacher … … … … 15 (1.6) 17 (1.8) … … … … Slept under treated net last night Yes 657 (73.2) 706 (78.7) 755 (80.2) 726 (72.9) 312 (72.7) 406 (74.9) No 158 (17.6) 125 (13.9) 114 (12.1) 126 (12.7) 84 (19.6) 103 (19.0) Do not have one 76 (8.5) 61 (6.8) 73 (7.6) 144 (14.5) 33 (7.7) 33 (6.1) Don’t know if have one 6 (0.7) 5 (0.6) … … … … … … … … Malaria positive by microscopy at recruitment No 640 (71.3) 656 (73.0) 870 (95.2) 917 (94.2) 300 (71.3) 371 (69.0) Yes 258 (28.7) 243 (27.0) 44 (4.8) 57 (5.9) 121 (28.7) 167 (31.0) Data are presented as No. The Effect of the Intervention on Other Secondary Outcomes at Delivery Total Number of Malaria Cases Diagnosed at Home Visits (Intervention Arm Only) and Antenatal Clinic Visits (Both Arms) by Rapid Diagnostic Test and Total Artemether-Lumefantrine Treatments Given During Study for All Countries Downloaded from https://academic.oup.com/cid/article/68/4/586/5046692 by guest on 30 March 2021 cORs obtained from logistic regression models. In all countries, the higher number of malaria infections diagnosed in the intervention group and the high proportion of malaria-positive women treated with AL indicate that CHWs implemented the intervention according to the instructions received. In Burkina Faso, RDT sensitivity and specificity per- formed by CHWs and compared to microscopy was 81.5% (95% CI, 67.9%–90.2%) and 92.1% (95% CI, 89.9%–93.9%), respec- tively [14], further confirming CHWs can both correctly use RDTs and adhere to test results and treatment guidelines [18]. Previous studies have shown that CHWs are able to diagnose and treat malaria in children [19–22] and are able to distribute IPTp to pregnant women [23].hf 3 days after CSST, and 96% of women reported having com- pleted the full course; in contrast, a qualitative study carried out in the same study area reported low adherence to antimalarial treatment by pregnant women despite good knowledge about malaria in pregnancy [24]. Increasing doses of IPTp-SP significantly decreased the risk of placental malaria. This effect was mainly seen in Burkina Faso, where women received up to 6 IPTp-SP doses. In The Gambia and Benin, the large majority of women had taken only 2 IPTp-SP doses and in both countries malaria prevalence was lower than in Burkina Faso. In these 2 countries, CSST should have had some effect as the time between IPTp-SP doses was longer than in Burkina Faso, so diagnosing and treating infec- tions during this period should have been beneficial. Malaria screening by the CHWs was done with an RDT whose detection threshold is at most 200 parasites/μL [25]. However, the large majority of infections during pregnancy are asymptomatic, with low parasite densities, often not detected by microscopy or Though the intervention did not have a direct effect on the prevalence of past placenta malaria, probably because of infec- tions before the first ANC visit, one would have expected a reduced risk of active or chronic placenta infections. This was not the case, and this may have been due to low adherence to AL treatment. Malaria Treatment in Pregnancy • CID 2019:68 (15 February) • 591 The Effect of the Intervention on Other Secondary Outcomes at Delivery (%) unless otherwise indicated. Abbreviations: IQR, interquartile range; VHW, village health workers. Downloaded from https://academic.oup.com/cid/article/68/4/586/5046692 by guest on 30 March 2021 was already high, with 80% of the women attending at least 2 scheduled visits, and this may explain why the intervention did not have the expected effect. was already high, with 80% of the women attending at least 2 scheduled visits, and this may explain why the intervention did not have the expected effect. extremely low in Benin, where only slightly more than half of the women had 2 visits, and this may reflect the general poor attendance in southern Benin. In The Gambia, ANC attendance 590 • CID 2019:68 (15 February) • COSMIC Consortium Table 2. Total Number of Malaria Cases Diagnosed at Home Visits (Intervention Arm Only) and Antenatal Clinic Visits (Both Arms) by Rapid Diagnostic Test and Total Artemether-Lumefantrine Treatments Given During Study for All Countries Cases Overall Control Intervention OR/RR 95% CI P Value No. % No. % No. % Burkina Faso Total No. of positive malaria casesa 994 … 388 … 606 … 1.56b 1.13–2.15 .006 Total No. of AL treatment given (% out of cases) 761 76.6 302 77.8 459 75.7 1.08b .94–1.25 .274 Total No. of women tested positive at least once 568 31.6 237 26.3 331 36.8 1.67c 1.22–2.30 .002 The Gambia Total No. of positive malaria casesa 87 … 15 … 72 … 4.55b 2.48–8.34 <.001 Total No. of AL treatment given (% out of cases) 71 81.6 12 80.0 59 81.9 1.23b .66–2.29 .515 Total No. of women tested positive at least once 75 3.8 15 1.6 60 6.0 3.94c 2.05–7.59 <.001 Benin Total No. of positive malaria casesa 224 … 33 … 191 … 4.67b 2.93–7.44 <.001 Total No. of AL treatment given (% out of cases) 186 83.0 14 42.4 172 90.1 1.71b .99–2.94 .055 Total No. of women tested positive at least once 164 16.9 20 4.7 144 26.6 7.53c 4.49–12.64 <.001 Abbreviations: AL, artemether-lumefantrine; CI, confidence interval; OR, odds ratio; RR, rate ratio. aWoman can be positive more than once during pregnancy. bRRs obtained from Poisson regression models. cORs obtained from logistic regression models. Table 2. Total Number of Malaria Cases Diagnosed at Home Visits (Intervention Arm Only) and Antenatal Clinic Visits (Both Arms) by Rapid Diagnostic Test and Total Artemether-Lumefantrine Treatments Given During Study for All Countries Table 2. The Effect of the Intervention on Other Secondary Outcomes at Delivery However, CHWs who visited treated women Malaria Treatment in Pregnancy • CID 2019:68 (15 February) • 591 Table 3. Placental Malaria in Burkina Faso and The Gambia Histology Control Intervention Odds Ratio 95% CI P Value No. % No. % No. with biopsy data 1544 1622 … … … Placental histology Any infection (acute, chronic, or past) 494 32 533 33 1.09 .80–1.48 .588 No infection 1050 68 1089 67 … … … Active infection (acute or chronic) 64 4 65 4 0.99 .66–1.49 .974 No active infection 1480 96 1557 96 … … … Past or chronic infection 472 31 509 31 1.09 .81–1.46 .571 No past or chronic infection 1072 69 1113 69 … … … Abbreviation: CI, confidence interval. Table 3. Placental Malaria in Burkina Faso and The Gambia Malaria Treatment in Pregnancy • CID 2019:68 (15 February) • 591 Table 4. Overall Adjusted Analysis for Placental Malaria in The Gambia and Burkina Faso Characteristic No. Positive, No. (%) Unadjusted OR 95% CI P Value Adjusted ORa 95% CI P Value Arm Intervention 1621 533 (33) … … … … … … Control 1543 494 (32) 1.09 .80–1.48 .588 1.06 .78–1.44 .722 Seasonality Delivery in rainy reason 1684 440 (26) 0.71 .59–.84 <.001 0.59 .49–.71 <.001 Delivery in dry season 1480 587 (40) 1 … … … … … Gravida First or second pregnancy 1124 451 (40) … … … … … … >2 pregnancies 2040 576 (28) 0.47 .39–.56 <.001 0.47 .39–.57 <.001 No. of SP doses 0 9 1 (11) 0.90 .10–.77 … 0.54 .06–5.14 … 1 496 140 (28) 1 … … 1 … … 2 1892 571 (30) 0.87 .66–1.14 … 0.85 .65–1.12 … 3 460 203 (44) 0.41 .29–.57 … 0.37 .26–.53 … 4 251 95 (38) 0.23 .16–.34 … 0.23 .15–.34 … 5 50 15 (30) 0.15 .08–.30 … 0.13 .07–.27 … 6 6 2 (33) 0.15 .03–.90 <.001 0.10 .01–.64 <.001 No. of AL treatments given 0 2651 695 (26) 1 … … 1 … … 1 364 223 (61) 2.10 1.62–2.72 … 2.01 1.53–2.64 … 2 75 75 (70) 2.60 1.66–4.08 … 1.86 1.15–3.02 … ≥3 34 34 (81) 4.52 2.01–10.15 <.001 2.98 1.28–6.95 <.001 Data are shown for any infection (acute, chronic, or past). Abbreviations: AL, artemether-lumefantrine; CI, confidence interval; OR, odds ratio; SP, sulfadoxine-pyrimethamine. aAdjusted for all other variables in the table Table 4. Abbreviations: AL, artemether-lumefantrine; CI, confidence interval; OR, odds ratio; SP, sulfadoxine-pyrimethamine. aAdjusted for all other variables in the table. Abbreviations: AL, artemether lumefantrine; CI, confidence interval; OR, odds ratio; SP, sulfadoxine pyrimethamine. aAdjusted for all other variables in the table. 592 • CID 2019:68 (15 February) • COSMIC Consortium Abbreviations: AL, artemether-lumefantrine; CI, confidence interval; OR, odds ratio; SP, sulfadoxine-pyrimethamine. aAdj d f ll h i bl i h bl Abbreviations: AL, artemether-lumefantrine; CI, confidence interval; OR, odds ratio; SP, sulfadoxine-pyrimethamine. aAdjusted for all other variables in the table. The Effect of the Intervention on Other Secondary Outcomes at Delivery Overall Adjusted Analysis for Placental Malaria in The Gambia and Burkina Faso Downloaded from https://academic.oup.com/cid/article/68/4/586/5046692 by guest on 30 March 2021 of SP doses 1 754 407 (54) 1 … … 1 … … 2 2454 1187 (48) 0.80 .67–.95 … 0.81 .68–.96 … 3 549 147 (27) 0.68 .52–.89 … 0.72 .54–.95 … 4 312 74 (24) 0.70 .50–.98 … 0.73 .52–1.02 … 5 69 20 (29) 0.94 .53–1.66 … 0.99 .99–1.75 … 6 8 2 (25) 0.76 .15–3.86 … 0.78 .15–4.06 … .022 .065 Abbreviations: AL, artemether-lumefantrine; CI, confidence interval; OR, odds ratio; SP, sulfadoxine-pyrimethamine. aAnemia defined as hemoglobin level <11 g/dL. Table 6. Anemia at Delivery, by Study Arm (All Countries) Characteristic No. Anemica, No. (%) OR 95% CI P Value Adjusted 95% CI P Value Arm Intervention 1989 861 (43) 1 … … 1 … … Control 2167 979 (45) 1.08 .92–1.28 .351 1.09 .92–1.28 .321 Seasonality Delivery in rainy season 2194 1089 (50) 1 … … 1 … … Delivery in dry season 1962 751 (38) 0.72 .63–.82 <.001 0.73 .64–.83 <.001 Gravida First or second pregnancy 1480 634 (43) 1 … … 1 … … >2 pregnancies 2674 1205 (45) 1.13 .99–1.30 .078 1.14 1.00–1.32 .052 No. of AL treatments given 0 3458 1623 (47) 1 … … 1 … … 1 505 147 (29) 0.85 .68–1.06 … 0.83 .66–1.04 … 2 146 56 (38) 1.54 1.07–2.21 … 1.58 1.09–2.28 … ≥3 (max 6) 47 14 (30) 1.23 .64–2.35 … 1.26 .65–2.41 … .029 .017 No. of SP doses 1 754 407 (54) 1 … … 1 … … 2 2454 1187 (48) 0.80 .67–.95 … 0.81 .68–.96 … 3 549 147 (27) 0.68 .52–.89 … 0.72 .54–.95 … 4 312 74 (24) 0.70 .50–.98 … 0.73 .52–1.02 … 5 69 20 (29) 0.94 .53–1.66 … 0.99 .99–1.75 … 6 8 2 (25) 0.76 .15–3.86 … 0.78 .15–4.06 … .022 .065 Abbreviations: AL, artemether-lumefantrine; CI, confidence interval; OR, odds ratio; SP, sulfadoxine-pyrimethamine. aAnemia defined as hemoglobin level <11 g/dL. Abbreviations: AL, artemether-lumefantrine; CI, confidence interval; OR, odds ratio; SP, sulfadoxine-pyrimethamine. aAnemia defined as hemoglobin level <11 g/dL. Table 7. Low Birth Weight, by Study Arm (All Countries) Characteristic No. LBW, No. Malaria Treatment in Pregnancy • CID 2019:68 (15 February) • 593 Abbreviations: AL, artemether-lumefantrine; CI, confidence interval; LBW, low birth weight; OR, odds ratio; SP, sulfadoxine-pyrimethamine. Downloaded from https://academic.oup.com/cid/article/68/4/586/5046692 by guest on 30 March 2021 Downloaded from https://academic.oup.com/cid/article/68/4/586/5046692 by guest on 30 March 2021 Table 5. Overall Adjusted Analysis for Maternal Peripheral Infection at Delivery (as Measured by Polymerase Chain Reaction) Characteristic No. Positive, No. (%) Unadjusted OR 95% CI P Value Adjusted ORa 95% CI P Value Arm Intervention 1951 208 (10.66) 1 … … 1 … … Control 1826 200 (10.95) 0.97 .78–1.21 .798 0.92 .74–1.15 .456 Seasonality Delivery in rainy reason 1796 125 (6.96) 2.94 2.34–3.69 <.001 2.84 2.25–3.58 <.001 Delivery in dry season 1981 283 (14.29) 1 … … 1 … … Gravida First or second pregnancy 1329 165 (12.42) 1 … … 1 … … >2 pregnancies 2448 243 (9.93) 0.76 .61–.84 .012 0.83 .66–1.04 .101 No. of AL treatments given 0 3107 273 (8.79) 1 … … 1 … … 1 478 88 (18.41) 1.43 1.09–1.88 … 1.33 1.00–1.77 … 2 145 33 (22.76) 1.76 1.16–2.66 … 1.76 1.13–2.73 … ≥3 (max 6) 47 14 (29.79) 2.35 1.23–4.48 <.001 2.26 1.15–4.47 .006 No. of SP doses 0 4 0 (0) … … … … … … 1 677 83 (12.26) 1 … … 1 … … 2 2192 219 (9.99) 0.72 .54–.96 … 0.74 .55–.98 … 3 531 50 (9.42) 0.30 .20–.44 … 0.37 .24–.56 … 4 298 46 (15.44) 0.47 .31–.71 … 0.55 .36–.85 … 5 67 9 (13.43) 0.39 .18–.84 … 0.44 .20–.94 … 6 8 1 (12.50) 0.38 .05–3.15 <.001 0.28 .03–2.42 <.001 Abbreviations: AL, artemether-lumefantrine; CI, confidence interval; OR, odds ratio; SP, sulfadoxine-pyrimethamine. aAdjusted for all other variables in the table. ed Analysis for Maternal Peripheral Infection at Delivery (as Measured by Polymerase Chain Reaction) Table 5. Overall Adjusted Analysis for Maternal Peripheral Infection at Delivery (as Measured by Polym 592 • CID 2019:68 (15 February) • COSMIC Consortium Table 6. Anemia at Delivery, by Study Arm (All Countries) Table 7. Low Birth Weight, by Study Arm (All Countries) Characteristic No. LBW, No. Downloaded from https://academic.oup.com/cid/article/68/4/586/5046692 by guest on 30 March 2021 (%) OR 95% CI P Value Adjusted 95% CI P Value Arm Intervention 2127 219 (10) 1 … … 1 … … Control 1950 201 (10) 1.06 .82–1.38 .643 1.06 .81–1.38 .695 Seasonality Delivery in rainy season 2159 213 (10) 1 … … 1 … … Delivery in dry season 1918 207 (11) 1.09 .89–1.34 .417 1.13 .92–1.40 .247 Gravida First or second pregnancy 1456 212 (15) 1 … … 1 … … >2 pregnancies 2619 208 (8) 0.50 .40–.61 <.001 0.50 .40–.61 <.001 No. of AL treatments given 0 3393 334 (10) 1 … … 1 … … 1 493 62 (13) 1.36 1.00–1.85 … 1.24 .91–1.71 … 2 144 16 (11) 1.18 .68–2.04 … 0.88 .50–1.55 … ≥3 (max 6) 47 8 (17) 1.87 .84–4.12 … 1.34 .62–3.12 … .128 .429 No. of SP doses 1 731 111 (15) 1 … … 1 … … 2 2415 230 (10) 0.55 .28–6.96 … 0.54 .42–.70 … 3 539 49 (9) 0.45 .43–.71 … 0.44 .30–.66 … 4 305 21 (7) 0.32 .30–.66 … 0.33 .19–.55 … 5 69 7 (10) 0.46 .19–.54 … 0.43 .18–.55 … 6 0 0 (0) … … … … … … <.001 <.001 Abbreviations: AL, artemether-lumefantrine; CI, confidence interval; LBW, low birth weight; OR, odds ratio; SP, sulfadoxine-pyrimethamine. Table 6. Anemia at Delivery, by Study Arm (All Countries) Characteristic No. Anemica, No. (%) OR 95% CI P Value Adjusted 95% CI P Value Arm Intervention 1989 861 (43) 1 … … 1 … … Control 2167 979 (45) 1.08 .92–1.28 .351 1.09 .92–1.28 .321 Seasonality Delivery in rainy season 2194 1089 (50) 1 … … 1 … … Delivery in dry season 1962 751 (38) 0.72 .63–.82 <.001 0.73 .64–.83 <.001 Gravida First or second pregnancy 1480 634 (43) 1 … … 1 … … >2 pregnancies 2674 1205 (45) 1.13 .99–1.30 .078 1.14 1.00–1.32 .052 No. of AL treatments given 0 3458 1623 (47) 1 … … 1 … … 1 505 147 (29) 0.85 .68–1.06 … 0.83 .66–1.04 … 2 146 56 (38) 1.54 1.07–2.21 … 1.58 1.09–2.28 … ≥3 (max 6) 47 14 (30) 1.23 .64–2.35 … 1.26 .65–2.41 … .029 .017 No. Downloaded from https://academic.oup.com/cid/article/68/4/586/5046692 by guest on 30 March 2021 % Anemia Hb <11 g/dL 861 43 979 45 1.09a .92–1.28 .351 Hb ≥11 g/dL 1128 57 1188 55 … … … Mean Hb (SD) 11.15 0.71 11.14 0.73 0.01b –.29 to .31 .944 Low birth weight (<2500 g) <2500 201 10 219 10 1.06a .82–1.38 .643 ≥2500 1749 90 1908 90 … … … Mean birth weight, kg (SD) 2.98 0.17 2.94 0.14 0.04b –.04 to .10 .254 Adverse pregnancy outcomes Congenital abnormalities 16 0.8 29 1.3 1.70a .92–3.15 .089 Miscarriage 16 0.7 16 0.7 0.98a .45–2.15 .959 Preterm birth 60 3 70 3 1.06a .68–1.67 .793 Stillbirth 39 2 47 2 1.05a .64–1.72 .855 Miscarriage, preterm, or stillbirth 99 4 116 5 1.10a .79–1.52 .582 Deaths Perinatal death 21 1 24 1 1.03a .57–1.85) .933 Maternal death 6 0.3 3 0.1 0.46a .12–1.86) .279 Perinatal death, miscarriage, preterm, or stillbirth 114 5 131 5 1.07a .79–1.45) .654 Abbreviations: CI, confidence interval; Hb, hemoglobin; OR, odds ratio; SD, standard deviation. aOR. bMean difference. bMean difference. IPTp-SP at each ANC, provided the doses are at least a month apart. The current trial data support this recommendation. RDT [26]. Therefore, measurements of the efficacy of intermit- tent screening and treatment may be limited by the sensitivity of current RDTs [10]. The decreased risk of low birth weight with increasing IPTp-SP doses and the borderline decrease of maternal anemia at delivery confirms the meta-analysis of 7 tri- als carried out in sub-Saharan Africa that reported a lower risk of low birth weight, maternal anemia, and placental malaria in women who received ≥3 IPTp-SP doses [27]. These results were used to support the WHOs’ recommendation of administering Pregnant women in the intervention arm had a higher risk of testing malaria positive, not because they had a higher risk of being infected, but rather because they were tested more frequently. Considering that most infections diagnosed in the intervention arm were treated with AL and that the risk of pla- centa; malaria increased with the number of AL treatments administered, these infections probably represented a small Table 9. Downloaded from https://academic.oup.com/cid/article/68/4/586/5046692 by guest on 30 March 2021 (%) OR 95% CI P Value Adjusted 95% CI P Value Arm Intervention 2127 219 (10) 1 … … 1 … … Control 1950 201 (10) 1.06 .82–1.38 .643 1.06 .81–1.38 .695 Seasonality Delivery in rainy season 2159 213 (10) 1 … … 1 … … Delivery in dry season 1918 207 (11) 1.09 .89–1.34 .417 1.13 .92–1.40 .247 Gravida First or second pregnancy 1456 212 (15) 1 … … 1 … … >2 pregnancies 2619 208 (8) 0.50 .40–.61 <.001 0.50 .40–.61 <.001 No. of AL treatments given 0 3393 334 (10) 1 … … 1 … … 1 493 62 (13) 1.36 1.00–1.85 … 1.24 .91–1.71 … 2 144 16 (11) 1.18 .68–2.04 … 0.88 .50–1.55 … ≥3 (max 6) 47 8 (17) 1.87 .84–4.12 … 1.34 .62–3.12 … .128 .429 No. of SP doses 1 731 111 (15) 1 … … 1 … … 2 2415 230 (10) 0.55 .28–6.96 … 0.54 .42–.70 … 3 539 49 (9) 0.45 .43–.71 … 0.44 .30–.66 … 4 305 21 (7) 0.32 .30–.66 … 0.33 .19–.55 … 5 69 7 (10) 0.46 .19–.54 … 0.43 .18–.55 … 6 0 0 (0) … … … … … … <.001 <.001 Abbreviations: AL, artemether-lumefantrine; CI, confidence interval; LBW, low birth weight; OR, odds ratio; SP, sulfadoxine-pyrimethamine. Table 7. Low Birth Weight, by Study Arm (All Countries) Malaria Treatment in Pregnancy • CID 2019:68 (15 February) • 593 Table 8. Adverse Outcome of Pregnancy, by Study Arm (All Countries) g y y y ( ) Outcome Control Intervention OR/Mean Difference 95% CI P Value No. % No. bbreviations: ANC, antenatal clinic; CI, confidence interval; IPTp-SP, intermittent preventive treatment in pregnancy with sulfadoxine-pyrimethamine tio; SD, standard deviation; SP, sulfadoxine-pyrimethamine. aIPTp-SP policies vary between countries; thus, data are not pooled. Downloaded from https://academic.oup.com/cid/article/68/4/586/5046692 by guest on 30 March 2021 Summary Estimates of the Effect of the Intervention on Antenatal Clinic Attendance and Intermittent Preventive Treatment With Sulfadoxine-Pyrimethamine Coverage, by Countrya Intervention Burkina Faso The Gambia Benin OR/IRR 95% CI P Value OR/IRR 95% CI P Value OR/IRR 95% CI P Value ANC visits At least 2 scheduled visits … … … 1.03b .66–1.62 .881 0.88b .63–1.24 .471 At least 4 scheduled visits 1.62b 1.02–2.59 .041 … … … … … … Mean No. of scheduled visits (SD) 1.08c 1.00–1.17 .045 1.00c .93–1.09 .912 0.98c .93–1.04 .484 Mean No. of unscheduled visits (SD) 0.90c .63–1.30 .589 1.34c .80–2.22 .266 1.07c .61–1.89 .81 Mean No. of any ANC visits (SD) 1.06c .96–1.16 .258 1.07c .90–1.27 .445 0.99c .91–1.07 .781 IPTp-SP coverage Mean No. of SP doses 1.04c .97–1.10 .294 1.01c .97–1.05 .611 0.98c .92–1.03 .387 At least 2 doses of SP 1.29b .96–1.73 .093 1.01b .70–1.46 .938 0.87b .62–1.22 .418 At least 4 doses of SP 1.14b .76–1.72 .517 … … … … … … Abb i i ANC l li i CI fid i l IPT SP i i i i i h lf d i i h i IRR i id i OR dd the Effect of the Intervention on Antenatal Clinic Attendance and Intermittent Preventive Treatment e Coverage, by Countrya aIPTp-SP policies vary between countries; thus, data are not pooled. bOR 594 • CID 2019:68 (15 February) • COSMIC Consortium proportion of all infections acquired during pregnancy and it was the undiagnosed infections that had a significant effect on the occurrence of placenta malaria. Receiving multiple AL treatments is probably a marker of a higher malaria risk. There is some controversy on the importance of low-density malaria infections during pregnancy, associated with anemia, lower mean hemoglobin, low birth weight, and premature births in some studies [28] but not in others [26]. Our results indicate that such infections are important and that, until better diag- nostic tests than standard RDTs become available, systematic treatment as many times as possible of all pregnant women until delivery is the best approach. the number of AL treatments administered was strongly asso- ciated with placental malaria, peripheral infection, and ane- mia, indicating that treated women had a higher risk of being reinfected over a relatively short period. Such a risk could have been lowered by a treatment with a much longer posttreatment prophylactic period. Notes Acknowledgments. We thank all the study participants of the COSMIC trial, in particular the pregnant women and the CHWs who carried out the intervention. We also thank the research and field staff of The Medical Research Council Unit, The Gambia, Unité de Recherche Clinique de Nanoro, Burkina Faso and Centre de Recherches Entomologiques de Cotonou, Benin and the Ministries of Health in the respective countries who supported the study. Staff of the Department of Pathology at the Edward Francis Small Teaching Hospital, The Gambia whose support we acknowledge, prepared the placental histology slides. The significant number of adverse pregnancy outcomes in both study arms highlights the poor access to timely and ade- quate care for women in rural and remote areas in sub-Saha- ran Africa. Both maternal anemia and low birth weight (45% and 10%, respectively) were below the West African regional estimates, 56% (95% CI, 46%–62%) [29] and 14% (https://data. unicef.org/topic/nutrition/low-birthweight/). Being in a trial may have influenced women’s healthcare-seeking behavior, which may explain these lower estimates. There were 45 perina- tal deaths and 9 maternal deaths with no difference in mortality between the study arms. Pooling the data together gives a peri- natal mortality rate of 10.77 per 1000 live births, and a maternal mortality rate of 215 per 100 000 live births. These remain high levels, but comparisons with other sources [30–32] should be interpreted with caution as they were estimated at time of deliv- ery while the neonatal and maternal mortality usually include days 28 and 42 postpartum, respectively. It is reassuring that no adverse events were associated with AL treatment, providing further safety data for AL treatment in the second and third tri- mester of pregnancy. g p p p gy Financial support. This work was supported by the European Community’s Seventh Framework Programme (grant agreement number 305662). Novartis kindly provided the study drug, Coartem.l Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the con- tent of the manuscript have been disclosed. COSMIC Consortium. Medical Research Council, The Gambia/London School of Hygiene and Tropical Medicine: Susana Scott, Umberto D’Alessandro, Lindsay Kendall, John Bradley, Kalifa Bojang, Simon Correa, Fanta Njie. Notes Institut de Recherche en Sciences de la Santé–Unité de Recherche Clinique de Nanoro, Burkina Faso: Halidou Tinto, Maminata Traore- Coulibaly, Hamtandi Magloire Natama, Ousmane Traoré, Innocent Valea. Centre de Recherches Entomologiques de Cotonou, Benin: Alain Nahum, Daniel Ahounou, Francis Bohissou, Gethaime Sondjo, Carine Agbowai. Academic Medical Centre, The Netherlands: Petra Mens, Esmée Ruizendaal, Henk Schallig. Institute of Tropical Medicine, Belgium: Susan Dierickx, Koen Peeters Grietens. Imperial College London, UK: Laetitia Duval, Lesong Conteh. Institut de Recherche en Sciences de la Santé, Burkina Faso: Maxime Drabo. Special Programme for Research and Training in Tropical Diseases, WHO, Switzerland: Jamie Guth, Franco Pagnoni. IST has been extensively evaluated as an alternative to IPTp-SP, with mixed results [8–10]. We decided to combine these 2 interventions, with the aim of providing additional pro- tection against malaria between ANC visits. Despite the sig- nificantly higher number of women treated for malaria in the intervention arm, none of the trial outcomes differed between study arms. Treatment with a long-acting artemisinin-based combination therapy such as dihydroartemisinin-piperaquine instead of AL would possibly have had a better outcome. Indeed, Supplementary Data Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author. Downloaded from https://academic.oup.com/cid/article/68/4/586/5046692 by guest on 30 March 2021 That the length of the prophylaxis period is more important than treating diagnosed infections is shown by the beneficial effect of increasing IPTp-SP doses on differ- ent pregnancy outcomes, including low birth weight. This may change when more sensitive diagnostic tests become available but, for now, increasing the number of IPTp-SP doses given during pregnancy is a priority. A major strength of this study was the selection of countries based on their varying malaria endemicity: low (The Gambia) vs high (Burkina Faso and Benin), and with varying degrees of SP resistance (high in Benin and moderate in The Gambia and Burkina Faso), which us enables to generalize study findings to West Africa and possibly other sub-Saharan African countries. The study was powered for individual countries for the overall primary outcome; thus, we were able to at least fully investigate the effect of the intervention on placental malaria for the study sites in The Gambia and Burkina Faso. Unfortunately, this was not possible for Benin. References 1. Rogerson SJ, Desai M, Mayor A, Sicuri E, Taylor SM, van Eijk AM. Burden, pathology, and costs of malaria in pregnancy: new developments for an old prob- lem. Lancet Infect Dis 2018; 18:e107–18. 1. Rogerson SJ, Desai M, Mayor A, Sicuri E, Taylor SM, van Eijk AM. Burden, pathology, and costs of malaria in pregnancy: new developments for an old prob- lem. Lancet Infect Dis 2018; 18:e107–18. 2. Hartman TK, Rogerson SJ, Fischer PR. The impact of maternal malaria on new- borns. Ann Trop Paediatr 2010; 30:271–82. 2. Hartman TK, Rogerson SJ, Fischer PR. The impact of maternal malaria on new- borns. Ann Trop Paediatr 2010; 30:271–82. 3. Dellicour S, Tatem AJ, Guerra CA, Snow RW, ter Kuile FO. Quantifying the num- ber of pregnancies at risk of malaria in 2007: a demographic study. PLoS Med 2010; 7:e1000221. 4. World Health Organization. Guidelines for the treatment of malaria. 3rd ed. Geneva, Switzerland: WHO, 2015. Malaria Treatment in Pregnancy • CID 2019:68 (15 February) • 595 5. van Eijk AM, Hill J, Larsen DA, et al. Coverage of intermittent preventive treat- ment and insecticide-treated nets for the control of malaria during pregnancy in sub-Saharan Africa: a synthesis and meta-analysis of national survey data, 2009–11. Lancet Infect Dis 2013; 13:1029–42. 19. Mubi M, Janson A, Warsame M, et al. Malaria rapid testing by community health workers is effective and safe for targeting malaria treatment: randomised cross- over trial in Tanzania. PLoS One 2011; 6:e19753. 20. Yeboah-Antwi K, Pilingana P, Macleod WB, et al. Community case management of fever due to malaria and pneumonia in children under five in Zambia: a cluster randomized controlled trial. PLoS Med 2010; 7:e1000340. 6. World Health Organization. World malaria report. Geneva, Switzerland: WHO, 2016. 7. Harrington WE, Mutabingwa TK, Kabyemela E, Fried M, Duffy PE. Intermittent treatment to prevent pregnancy malaria does not confer benefit in an area of widespread drug resistance. Clin Infect Dis 2011; 53:224–30. 21. Tine RC, Faye B, Ndour CT, et al. Impact of combining intermittent preventive treatment with home management of malaria in children less than 10 years in a rural area of Senegal: a cluster randomized trial. Malar J 2011; 10:358. 8. Tagbor H, Cairns M, Bojang K, et al. A non-inferiority, individually randomized trial of intermittent screening and treatment versus intermittent preventive treat- ment in the control of malaria in pregnancy. PLoS One 2015; 10:e0132247. 22. References Freeman PA, Schleiff M, Sacks E, Rassekh BM, Gupta S, Perry HB. Comprehensive review of the evidence regarding the effectiveness of community-based primary health care in improving maternal, neonatal and child health: 4. Child health findings. J Glob Health 2017; 7:010904. 9. Madanitsa M, Kalilani L, Mwapasa V, et al. Scheduled intermittent screening with rapid diagnostic tests and treatment with dihydroartemisinin-piperaquine versus intermittent preventive therapy with sulfadoxine-pyrimethamine for malaria in pregnancy in Malawi: an open-label randomized controlled trial. PLoS Med 2016; 13:e1002124. 23. Jennings MC, Pradhan S, Schleiff M, et al. Comprehensive review of the evidence regarding the effectiveness of community-based primary health care in improv- ing maternal, neonatal and child health: 2. Maternal health findings. J Glob Health 2017; 7:010902. 10. Desai M, Gutman J, L’lanziva A, et al. Intermittent screening and treatment or intermittent preventive treatment with dihydroartemisinin-piperaquine versus intermittent preventive treatment with sulfadoxine-pyrimethamine for the con- trol of malaria during pregnancy in western Kenya: an open-label, three-group, randomised controlled superiority trial. Lancet 2015; 386:2507–19. 24. Jaiteh F, Dierickx S, Gryseels C, et al. ‘Some anti-malarials are too strong for your body, they will harm you.’ Socio-cultural factors influencing pregnant women’s adherence to anti-malarial treatment in rural Gambia. Malar J 2016; 15:195. 25. World Health Organization. Malaria rapid diagnostic test performance: results of WHO product testing of malaria RDTs: round 6 (2014–2015) December 2015. Available at: http://www.who.int/malaria/publications/atoz/9789241510035/en/. Accessed 29 May 2018. 11. World Health Organization. The roll back malaria strategy for improving access to treatment through home management of malaria. Geneva, Switzerland: WHO, 2005. 12. Scott S, Mens PF, Tinto H, et al. Community-based scheduled screening and treatment of malaria in pregnancy for improved maternal and infant health in The Gambia, Burkina Faso and Benin: study protocol for a randomized controlled trial. Trials 2014; 15:340. 26. Williams JE, Cairns M, Njie F, et al. The performance of a rapid diagnostic test in detecting malaria infection in pregnant women and the impact of missed infec- tions. Clin Infect Dis 2016; 62:837–44. 27. Kayentao K, Garner P, van Eijk AM, et al. Intermittent preventive therapy for malaria during pregnancy using 2 vs 3 or more doses of sulfadoxine-pyrimeth- amine and risk of low birth weight in Africa: systematic review and meta-analysis. JAMA 2013; 309:594–604. 13. Ballard JL, Khoury JC, Wedig K, Wang L, Eilers-Walsman BL, Lipp R. New Ballard score, expanded to include extremely premature infants. J Pediatr 1991; 119:417–23. 596 • CID 2019:68 (15 February) • COSMIC Consortium References 14. Ruizendaal E, Schallig HDFH, Scott S, et al. Evaluation of malaria screening during pregnancy with rapid diagnostic tests performed by community health workers in Burkina Faso. Am J Trop Med Hyg 2017; 97:1190–7. 28. Cottrell G, Moussiliou A, Luty AJ, et al. Submicroscopic Plasmodium falciparum infections are associated with maternal anemia, premature births, and low birth weight. Clin Infect Dis 2015; 60:1481–8. 29. Stevens GA, Finucane MM, De-Regil LM, et al. Global, regional, and national trends in haemoglobin concentration and prevalence of total and severe anaemia in children and pregnant and non-pregnant women for 1995–2011: a systematic analysis of population-representative data. Lancet Glob Health 2013; 1:e16–25. 15. Hermsen CC, Telgt DS, Linders EH, et al. Detection of Plasmodium falciparum malaria parasites in vivo by real-time quantitative PCR. Mol Biochem Parasitol 2001; 118:247–51. 16. Bulmer JN, Rasheed FN, Francis N, Morrison L, Greenwood BM. Placental malaria. I. Pathological classification. Histopathology 1993; 22:211–8. 30. United Nations Children’s Fund, World Health Organization. Tracking progress towards universal coverage for reproductive, newborn and child health: the 2017 report. Washington, DC: UNICEF, 2017. 17. World Health Organization. WHO policy brief for the implementation of inter- mittent preventive treatment of malaria in pregnancy using sulfadoxine-pyri- methamine (IPTp-SP) April 2013 (rev. January 2014). Available at: http://www. who.int/malaria/publications/atoz/policy_brief_iptp_sp_policy_recommenda- tion/en/. Accessed 29 May 2018. 31. Wang H, Liddell CA, Coates MM, et al. Global, regional, and national levels of neonatal, infant, and under-5 mortality during 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 2014; 384:957–79. 18. Ruizendaal E, Dierickx S, Peeters Grietens K, Schallig HD, Pagnoni F, Mens PF. Success or failure of critical steps in community case management of malaria with rapid diagnostic tests: a systematic review. Malar J 2014; 13:229. 32. Kassebaum NJ, Bertozzi-Villa A, Coggeshall MS, et al. Global, regional, and national levels and causes of maternal mortality during 1990–2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet 2014; 384:980–1004.
https://openalex.org/W4223636193	https://arpi.unipi.it/bitstream/11568/1160250/1/cancers-14-01948.pdf	English	null	Evaluation of Browning Markers in Subcutaneous Adipose Tissue of Newly Diagnosed Gastrointestinal Cancer Patients with and without Cachexia	Cancers	2,022	cc-by	8,430	Citation: Molﬁno, A.; Belli, R.; Imbimbo, G.; Carletti, R.; Amabile, M.I.; Tambaro, F.; di Gioia, C.R.T.; Belloni, E.; Ferraro, E.; Nigri, G.; et al. Evaluation of Browning Markers in Subcutaneous Adipose Tissue of Newly Diagnosed Gastrointestinal Cancer Patients with and without Cachexia. Cancers 2022, 14, 1948. https://doi.org/10.3390/ cancers14081948 Academic Editor: Stephane Servais Received: 4 March 2022 Accepted: 8 April 2022 Published: 12 April 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- iations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). Citation: Molﬁno, A.; Belli, R.; Imbimbo, G.; Carletti, R.; Amabile, M.I.; Tambaro, F.; di Gioia, C.R.T.; Belloni, E.; Ferraro, E.; Nigri, G.; et al. Evaluation of Browning Markers in Subcutaneous Adipose Tissue of Newly Diagnosed Gastrointestinal Cancer Patients with and without Cachexia. Cancers 2022, 14, 1948. https://doi.org/10.3390/ cancers14081948 Abstract: We assessed the molecular phenotype of the browning of white adipose tissue in newly diagnosed cancer patients and controls undergoing surgery for gastrointestinal tumors and for non- malignant diseases, respectively. We collected subcutaneous adipose tissue (SAT) samples and using RT-PCR, we analyzed the expression of markers of browning and using Western blot the protein levels of UCP1 and PGC1α. The Ucp1 mRNA levels were lower in cancer patients vs. controls (p = 0.01), whereas Cidea and Tmem26 mRNA levels were higher in cancer patients. We found higher PGC1α protein levels in patients vs. controls, while no differences were seen for UCP1. The Ucp1 expression was lower in cachectic and non-cachectic patients vs. controls, whereas Cidea expression was higher in cachectic and non-cachectic patients vs. controls. Pgc1α mRNA levels were higher in cachectic vs. non-cachectic patients (p = 0.03) vs. controls (p = 0.016). According to type of tumors, we did not observe differences in Cidea expression, whereas Pgc1α was higher in pancreatic cancer vs. colorectal and vs. controls. We observed the lower expression of Ucp1 in pancreatic and colorectal cancer vs. controls. We documented higher UCP1 protein levels in pancreatic cancer patients vs. colorectal (p = 0.002) and vs. controls (p = 0.031). PGC1α protein levels were higher in pancreatic cancer patients vs. controls. Article Evaluation of Browning Markers in Subcutaneous Adipose Tissue of Newly Diagnosed Gastrointestinal Cancer Patients with and without Cachexia Alessio Molﬁno 1,, Roberta Belli 1, Giovanni Imbimbo 1, Raffaella Carletti 1, Maria Ida Amabile 2 , Federica Tambaro 1 , Cira R. T. di Gioia 3 , Elena Belloni 4, Elisabetta Ferraro 5 , Giuseppe Nigri 4 and Maurizio Muscaritoli 1 1 Department of Translational and Precision Medicine, Sapienza University of Rome, 00185 Rome, Italy; roberta.belli@uniroma1.it (R.B.); giovanni.imbimbo@uniroma1.it (G.I.); raffaella.carletti@uniroma1.it (R.C.); federica.tambaro@uniroma1.it (F.T.); maurizio.muscaritoli@uniroma1.it (M.M.) 1 Department of Translational and Precision Medicine, Sapienza University of Rome, 00185 Rome, Italy; roberta.belli@uniroma1.it (R.B.); giovanni.imbimbo@uniroma1.it (G.I.); raffaella.carletti@uniroma1.it (R.C.); federica.tambaro@uniroma1.it (F.T.); maurizio.muscaritoli@uniroma1.it (M.M.) 2 Department of Surgical Sciences, Sapienza University of Rome, 00161 Rome, Italy; mariaida.amabile@uniroma1.it 3 Department of Radiological, Oncological and Pathological Sciences, Sapienza University of Rome, 00161 Rome, Italy; cira.digioia@uniroma1.it y g 4 Department of Medical-Surgical Sciences and Translational Medicine, Sapienza University of Rome, 00189 Rome, Italy; elena.belloni@uniroma1.it (E.B.); giuseppe.nigri@uniroma1.it (G.N.) 5 4 Department of Medical-Surgical Sciences and Translational Medicine, Sapienza University o 00189 Rome, Italy; elena.belloni@uniroma1.it (E.B.); giuseppe.nigri@uniroma1.it (G.N.) 5 Department of Biology, University of Pisa, 56127 Pisa, Italy; elisabetta.ferraro@unipi.it y g pp g 5 Department of Biology, University of Pisa, 56127 Pisa, Italy; elisabetta.ferraro@unipi.it Department of Biology, University of Pisa, 56127 Pisa, Italy; elisabetta.ferraro@unipi.it Correspondence: alessio.molﬁno@uniroma1.it; Tel./Fax: +39-06-4997-2042 * Correspondence: alessio.molﬁno@uniroma1.it; Tel./Fax: +39-06-4997-2042 Simple Summary: Cachexia occurs frequently in cancer patients with deep metabolic derangements. The browning of adipose tissue promotes thermogenesis and energy expenditure and, in cancer, has been considered a major determinant of adipose tissue atrophy. We evaluated the molecular phenotype of this phenomenon in the subcutaneous adipose tissue (SAT) of newly diagnosed gas- trointestinal cancer patients compared to controls. We observed that the modulation of different markers of the browning of SAT in gastrointestinal cancer and, in particular, pancreatic cancer showed signiﬁcant changes in UCP1 and PGC1α; PGC1α was highly expressed in cachectic patients. Our study highlights the relevance of browning in patients with cancer, in particular in those with pancre- atic cancer. Understanding the browning phenomenon may allow us to counteract these metabolic alterations before the development of severe cachexia, which is characterized by deep adipose and muscle depletion, negatively affecting survival and quality of life. cancers cancers cancers Different markers of the browning of SAT are modulated, and pancreatic cancer showed changes in UCP1 and PGC1α; PGC1α was highly expressed in cachectic patients, with clinical implications that should be further clarified. Received: 4 March 2022 Accepted: 8 April 2022 Published: 12 April 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- iations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). Keywords: cachexia; cancer; browning; subcutaneous adipose tissue; wasting https://www.mdpi.com/journal/cancers Cancers 2022, 14, 1948. https://doi.org/10.3390/cancers14081948 2 of 14 Cancers 2022, 14, 1948 2.1. Participant’s Selection This was an observational, controlled study performed on gastrointestinal cancer patients eligible for surgery and on controls enrolled at the Department of Medical-Surgical Sciences and Translational Medicine, Sapienza University of Rome, Italy. The study was performed in accordance with the Declaration of Helsinki and approved by the local Ethics Committee (Sapienza University, Azienda Sant ’Andrea Hospital, Rome, Italy—prot. n. 167SA_2017). All patients provided written informed consent to participate in the study. We enrolled patients with a new diagnosis of pancreatic, gastric or colorectal cancer eligible for surgical tumor resection, naïve to any anticancer treatments, including adjuvant therapy, and subjects with non-malignant diseases undergoing abdominal surgery, serving as controls. We included patients with age ≥18 years and with the ability to give informed consent. In both groups, we excluded patients with acute or chronic conditions negatively affecting nutritional status, including infections, heart failure, liver cirrhosis, chronic kidney diseases and clear signs of malabsorption or intestinal occlusion. We also excluded patients with dysphagia. 1. Introduction Cachexia occurs frequently in cancer patients and is associated with metabolic al- terations involving central mechanisms and peripheral tissues, leading to body weight loss [1–3]. This implies deep changes, including wasting, mainly of adipose tissue and muscle mass [1,2,4] representing predictors of poor clinical outcomes [4–6]. Adipose tissue is considered pivotal in the development of cachexia in cancer patients, and different mechanisms of adipose atrophy were identiﬁed, including increased lipolysis and the enhanced browning of white adipose tissue [7–9]. Browning is a metabolic process characterized by the emergence of beige adipocytes in white adipose tissue (WAT), promoting thermogenesis and energy expenditure [10]. In cancer, browning has been considered a major determinant of adipose tissue atrophy, and the inhibition of speciﬁc pathways, such as that of parathyroid-hormone-related protein (PTHrP), improved cachexia in an experimental model [8]. Importantly, several biomarkers of browning were identiﬁed in previous animal studies that allowed a phenotypic charac- terization of the presence of beige/brown adipocytes in these settings [11,12]. In particular, animal and human experiments documented the modulation of speciﬁc molecules includ- ing Tbx1, Eva, Dio2, Tmem26 and Cidea, representing the signature of beige-selective and brown-selective genes [11,13]. g Petruzzelli et al. showed that the browning process may anticipate skeletal muscle wasting in cancer and is associated with enhanced uncoupling protein 1 (UCP1) expres- sion [13]. We recently showed that gastrointestinal cancer patients with cachexia showed de- creased adipocyte size, increased ﬁbrosis and inﬂammatory changes in subcutaneous adipose tissue and that histological changes reﬂected alterations in adiposity evaluated using CT scan [14]. In the present study, we aimed to (i) assess the gene expression and the protein levels of the markers of browning in the subcutaneous adipose tissue (SAT) of patients with gastrointestinal cancer with and without cachexia before any anticancer treatment, including surgery, compared to non-cancer, non-cachectic patients and to (ii) evaluate potential differences in these markers between different types of gastrointestinal cancer. 2.2. Clinical, Nutritional Status Evaluation and Adipose Tissue Biopsy In all the participants, we registered information on nutritional status including cur- rent weight, usual weight and involuntary body weight loss in the prior 6 months, and we calculated body mass index (BMI). In cancer patients, in accordance with international criteria routinely used in clinical studies [1,4], cachexia was diagnosed as a non-volitional Cancers 2022, 14, 1948 3 of 14 3 of 14 body weight loss >5% or BMI <20 kg/m2 and any degree of weight loss >2% [15], as- sessed during a study visit before surgery. During the enrollment in a fasting state, we collected blood samples in EDTA tubes, and after centrifugation, we measured albumin and C-reactive protein (CRP) circulating levels, and we tested hemoglobin concentration with standard automated techniques. Moreover, from patient clinical records, we collected data on the staging and histology of the cancer and the number and type of comorbidities. During the ﬁrst phase of the surgical procedure, in both cancer patients and controls, we collected the specimens of SAT (approximately 1 cm3) obtained anteriorly to the anterior sheath of the rectus abdominis. The samples were immediately frozen in liquid nitrogen and in part included in OCT, and stored at −80 ◦C. 2.4. Protein Isolation and Western Blot 2.4. Protein Isolation and Western Blot Subcutaneous adipose tissue (SAT) was lysed using ice-cold RIPA buffer (50 mM Tris/HCl pH 8.1% Triton X, 150 mM NaCl, 0.5% sodium-deoxycholate, 0.1% SDS) sup- plemented with Phosphatase Inhibitor Cocktail 2 and 3 (Sigma-Aldrich, Dorset, UK) and Halt Protease Inhibitor (Thermo Scientiﬁc, Waltham, MA, USA). A clear supernatant was obtained via the centrifugation of lysates at 13,000 rpm for 20 min at 4 ◦C. Protein concen- trations were measured using the Bradford protein assay (Bio-Rad). Western blot analysis was performed by loading and separating aliquots of the total lysate via SDS-PAGE using Miniprotean precast gels (Bio-Rad, Portland, ME, USA), and proteins were transferred to nitrocellulose membranes (Bio-Rad) using a Trans-Blot semidry electrophoretic sys- tem (Bio-Rad). Membranes were blocked for 1 h at RT with 5% non-fat milk in T-TBS (Tris-Buffered Saline with 0.05% Tween 20). Incubation with primary speciﬁc antibodies was performed in blocking solution overnight at 4 ◦C, and incubation with horseradish peroxidase-conjugated secondary antibodies or alkaline phosphatase (AP)-conjugate sec- ondary antibodies was performed in blocking solution for 1 h at RT. The following primary antibodies were used for detection: PGC1α (#3242, Millipore) 1:1000, UCP1 (#GTX112784, GeneTex) 1:1000, actin-β (#A2228, Sigma-Aldrich, Dorset, UK) 1:1000. Speciﬁc antibody signals were detected using appropriate horseradish peroxidase-conjugated secondary antibody anti-mouse (#1706516, Goat Anti-Mouse IgG antibody, Bio-Rad) 1:8000 or AP- conjugate secondary antibody anti-rabbit (#A16099, Invitrogen, Carlsbad, CA, USA) 1:3000. Immunoreactive bands were visualized using SuperSignal West Pico Plus Chemilumines- cent Substrate (Thermo Scientiﬁc, Waltham, MA, USA) and using the NBT/BCIP Color Development Substrate system (#S3771, Promega, Madison, WI, USA) in AP buffer (0.1 M TrisHCl pH 9.5, 100 mM NaCl, 5 mM MgCl2). Then, 0.5 M EDTA (pH 8.0) was used to block the reaction. Protein expression was normalized for β-Actin protein levels. The relative amounts of each band were quantiﬁed with densitometry using the ImageJ software. 2.5. Immunohistochemical Evaluation of UCP1 in Adipose Tissue UCP1 localization was evaluated with immunohistochemical stains performed on frozen adipose tissue sections (5 µm). Endogenous peroxidase activity was blocked using 3% hydrogen peroxide. The sections were incubated at room temperature for 1 h with anti-human antibody UCP1, (1:750, rabbit polyclonal antibody, GTX112784 Gene Tex, Irvine, CA, USA). The Universal Quick Kit, Peroxidase, R.T.U. Staining System (Vector laboratories, Burlingame, CA, USA) was used to label the primary antibody. The reaction product was visualized with 3,3′-diaminobenzidine (DAB) (Vector laboratories, Burlingame, CA, USA) and counterstaining with Mayer hematoxylin. A negative control was obtained by omitting the primary antibody. The immunostained slides of adipose tissue sections were viewed by two independent pathologists using a Leica microscope (Leitz Camera) [19]. 2.3. Quantitative Real-Time PCR We examined the expression of brown/beige selective genes of the adipose tissue, as well as those involved in the metabolic change of white adipocytes into beige cells (the browning of WAT) (Table 1), as previously described by others [11–13], to reveal potential differences between gastrointestinal cancer patients and controls. The primers we used, according to the available literature [16,17], are shown in Table 1. Table 1. Primers used for comparative real-time PCR. Cidea Fw 5′-GGAGCTCATCAGCAAGACTCTG-3′ Rv 5′-AACTCTTCTGTGTCCACCACG-3′ Ucp1 Fw 5′-GCAGGGAAAGAAACAGCACCT-3′ Rv 5′-ACTTTCACGACCTCTGTGGG-3′ Pgc1α Fw 5′-CCTGCATGAGTGTGTGCTCT-3′ Rv 5′-CAGCACACTCGATGTCACTCC-3′ Tbx1 Fw 5′-ACGACAACGGCCACATTATTC-3′ Rv 5′-CCTCGGCATATTTCTCGCTATCT-3′ Tmem26 Fw 5′-ATGGAGGGACTGGTCTTCCTT-3′ Rv 5′-CTTCACCTCGGTCACTCGC-3′ Eva1 Fw 5′-GGAATCCTGAGCGGTACGATG-3′ Rv 5′-CTGGCAGGTGTATGTCCCATT-3′ Dio2 Fw 5′-ATGCTGACCTCAGAGGGACT-3′ Rv 5′-ATCCTCACCCAATTTCACCTGT-3′ Pthr Fw 5′-CGTTAGTTTCCGTCTCCACCTT-3′ Rv 5′-GCAGAAATCCACACAGCTGA-3′ Pdk4 Fw 5′-GGAGCATTTCTCGCGCTACA-3′ Rv 5′-ACAGGCAATTCTTGTCGCAAA-3′ β-Actin Fw 5′-CCTGGCACCCAGCACAA-3′ Rv 5′-GGGCCGGACTCGTCATA-3′ Table 1. Primers used for comparative real-time PCR. Total RNA was extracted from subcutaneous white adipose tissue using an RNeasy Lipid Tissue Mini Kit (Qiagen, Germantown, MD, USA) according to the manufacturer’s instructions. cDNA was synthetized from 250 ng of total RNA using the High-Capacity cDNA Reverse Transcription Kits (Applied Biosystems, Thermo Fisher Scientiﬁc, Wilm- ington, DE, USA), according to the manufacturer’s instructions. Comparative real-time PCR was performed with GoTaq® qPCR Master Mix (Promega, Madison, WI, USA), using Applied Biosystem 7900HT Fast. Data were normalized to β-Actin (calibrator), used as the internal control. Resulting data were analyzed using SDS2.4 Software (Applied Biosys- tems, Bedford, MA, USA), and fold-change was determined by using the 2−∆∆CT [18]. All reactions were performed in duplicate. The primers we used are shown in Table 1. Total RNA was extracted from subcutaneous white adipose tissue using an RNeasy Lipid Tissue Mini Kit (Qiagen, Germantown, MD, USA) according to the manufacturer’s instructions. cDNA was synthetized from 250 ng of total RNA using the High-Capacity cDNA Reverse Transcription Kits (Applied Biosystems, Thermo Fisher Scientiﬁc, Wilm- ington, DE, USA), according to the manufacturer’s instructions. Comparative real-time PCR was performed with GoTaq® qPCR Master Mix (Promega, Madison, WI, USA), using Applied Biosystem 7900HT Fast. Data were normalized to β-Actin (calibrator), used as the internal control. Resulting data were analyzed using SDS2.4 Software (Applied Biosys- tems, Bedford, MA, USA), and fold-change was determined by using the 2−∆∆CT [18]. All reactions were performed in duplicate. The primers we used are shown in Table 1. 4 of 14 Cancers 2022, 14, 1948 2.6. Statistical Analyses Patients’ characteristics were described using mean ± SD and median (25th; 75th per- centiles) for non-normally distributed variables. Normal distribution was tested using the Shapiro–Wilk test. Categorical variables were described as number (%). To analyze poten- tial differences between cancer patients with cachexia, cancer patients without cachexia and controls, we performed the Analysis of Variance (ANOVA) and the Kruskal–Wallis test, as appropriate. According to the normal/non-normal distribution, we also assessed differences between groups using the two-tailed t-test or the Mann–Whitney test, respec- tively. The chi-square test was used to verify association(s) between categorical variables. A p value < 0.05 was considered statistically signiﬁcant. SPSS version 26 was used to perform statistical analyses. 5 of 14 Cancers 2022, 14, 1948 3.1. Patient’s Characteristics We enrolled 24 gastrointestinal cancer patients and 13 non-oncology patients who underwent surgery for non-malignant diseases (i.e., cholecystectomy for gallstones, ab- dominal wall surgery for hernia and surgery for cysts). The clinical characteristics of the participants are summarized in Table 2. In particular, gastrointestinal cancer patients (13 females and 11 males)—9 pancreatic, 7 gastric and 8 colorectal—presented with a me- dian age of 74 years (IQR 68; 80) and a mean BMI of 27.04 ± 3.29 kg/m2, whereas the control group (7 females and 6 males) presented with a median age of 63 years (IQR 55; 66) and a mean BMI of 28.18 ± 4.45 kg/m2 (Table 2). Table 2. Participants’ characteristics. Clinical Parameter Cancer Patients (N = 24) Controls (N = 13) p-Value Age, y 74 (68; 80) 63 (55; 66) 0.0006 Male, n (%) 11 (46) 6 (46) 0.985 BMI, kg/m2 27.04 ± 3.29 28.18 ± 4.45 0.190 Actual weight, kg 76.66 ± 12.83 80 ± 12.68 0.224 Body weight loss % 4.14 (2.88; 6.70) 0.00 (0.00; 0.00) <0.00001 Cachexia, yes (%) 9 (38) / Hemoglobin, g/dl 11.28 ± 2.50 14.07 ± 2.04 0.002 C-reactive protein , mg/dl 1.19 (0.24; 3.01) 0.30 (0.22; 0.50) 0.134 Albumin, g/dl 3.25 (2.98; 3.53) 4 (3.80; 4) 0.0006 Comorbidities Hypertension, n (%) 14 (58) 8 (61) 0.850 Diabetes, n (%) 7 (29) 4 (31) 0.919 Dyslipidemia, n (%) 7 (29) 6 (46) 0.301 Type of cancer Pancreas, n (%) 9 (38) / / Colorectal, n (%) 8 (33) / / Gastric, n (%) 7 (29) / / Stage I-II, n (%) 16 (67) / / Stage III-IV, n (%) 8 (33) / / Abbreviations. Body mass index, BMI. normal range: 0–0.5 mg/dL. Table 2. Participants’ characteristics. 3.2. Nutritional and Inﬂammatory Status 3.2. Nutritional and Inﬂammatory Status 3.2. Nutritional and Inﬂammatory Status 3.2. Nutritional and Inﬂammatory Status In gastrointestinal cancer patients, we documented a mean body weight loss of 4.85 ± 3.30%, and cachexia accounted for 38% (9/24) (Table 2). No modiﬁcations in body weight in the prior six months were recorded in the control group (Table 2). We did not observe differences in body weight loss (%) and BMI among the different types of cancer patients (p = 0.460 and p = 0.518, respectively). Although CRP serum levels were not different among patients and controls, only the cancer group showed median CRP values above the normal range (0–0.5 mg/dL) (Table 2). 3.3. Browning Genes Expression and Evaluation of UCP1 and PGC1a Protein Levels in SAT of Gastrointestinal Cancer Patients and Controls 3.3. Browning Genes Expression and Evaluation of UCP1 and PGC1a Protein Levels in SAT of Gastrointestinal Cancer Patients and Controls Using quantitative PCR (qPCR) analysis, gastrointestinal cancer patients showed a signiﬁcant decreased expression of Ucp1 compared to controls (median 0.56 IQR 0.32; 1.19 vs. 1.17 IQR 0.85; 1.67) (p = 0.01) (Figure 1A). 6 of 14 Cancers 2022, 14, 1948 Figure 1 igu e B control cancer 0.0 0.5 1.0 1.5 UCP1/ -Actin control cancer 0.0 0.5 1.0 1.5 2.0 2.5 PGC1 / -Actin & PGC1α/β‐Actin UCP1/β‐Ac0n p=0.005 Figure 1. Evaluation of browning genes and protein expression in white adipose tissue (WAT) of gastro-intestinal cancer patients. (A) The mRNA levels of Cell Death-Inducing DFFA-Like Effector A (Cidea), Iodothyronine Deiodinase 2 (Dio2), parathyroid hormone receptor (Pthr), peroxisome proliferator-activated receptor gamma coactivator-1-α (Pgc1α), Transmembrane protein 26 (Tmem26), T-box transcription factor 1 (Tbx1), uncoupling protein 1 (Ucp1), EVA1 and Pyruvate Dehydrogenase Kinase 4 (Pdk4) were analyzed using quantitative real-time PCR in SAT of gastrointestinal cancer patients (N = 24) and controls (N = 13). (B) Representative Western blot images and protein densitom- etry quantiﬁcation for PGC1α and UCP1 protein levels in SAT of a subset of cancer patients (N = 19) and controls (N = 8). β-Actin was used as loading control. B control cancer 0.0 0.5 1.0 1.5 UCP1/ -Actin control cancer 0.0 0.5 1.0 1.5 2.0 2.5 PGC1 / -Actin & PGC1α/β‐Actin UCP1/β‐Ac0n p=0.005 B Figure 1. Evaluation of browning genes and protein expression in white adipose tissue (WAT) of Figure 1. Evaluation of browning genes and protein expression in white adipose tissue (WAT) of gastro-intestinal cancer patients. 3.2. Nutritional and Inﬂammatory Status Evaluation of Browning Gene Expression, UCP1 and PGC1a Protein Levels in SAT of Gastrointestinal Cancer Patients with and without Cachexia and in Controls 3.4. Evaluation of Browning Gene Expression, UCP1 and PGC1a Protein Levels in SAT of Gastrointestinal Cancer Patients with and without Cachexia and in Controls As shown in Figure 2A, Ucp1 expression levels were signiﬁcantly decreased in cachectic (median 0.56 IQR 0.35; 1.39) and non-cachectic (median 0.61 IQR 0.26; 0.71) cancer patients compared to controls (median 1.17 IQR 0.85; 1.67) (p = 0.033 and p = 0.020, respectively). We also found a signiﬁcant increase in Cidea expression in cachectic patients vs. controls (1.40 IQR 0.94; 2.49 vs. 0.60 IQR 0.44; 0.86) (p = 0.002), as well as in non-cachectic patients (median 1.17 IQR 0.96; 2.47) vs. controls (p = 0.003) (Figure 2A). For both Ucp1 and Cidea expression, we did not ﬁnd differences between cachectic and non-cachectic patients (Figure 2A). As shown in Figure 2A, Ucp1 expression levels were signiﬁcantly decreased in cachectic (median 0.56 IQR 0.35; 1.39) and non-cachectic (median 0.61 IQR 0.26; 0.71) cancer patients compared to controls (median 1.17 IQR 0.85; 1.67) (p = 0.033 and p = 0.020, respectively) We also found a signiﬁcant increase in Cidea expression in cachectic patients vs. controls (1.40 IQR 0.94; 2.49 vs. 0.60 IQR 0.44; 0.86) (p = 0.002), as well as in non-cachectic patients (median 1.17 IQR 0.96; 2.47) vs. controls (p = 0.003) (Figure 2A). For both Ucp1 and Cidea expression, we did not ﬁnd differences between cachectic and non-cachectic patients (Figure 2A). Figure 2 B control non-cachectic cachectic 0.0 0.5 1.0 1.5 UCP1 / -Actin control non-cachectic cachectic 0 1 2 3 PGC1 / -Actin control non-cachectic cachectic PGC1α/β-Actin UCP1/β-Ac n Figure 2. Evaluation of browning genes and protein expression in SAT of cachectic and non-cachectic cancer patients.(A) The mRNA levels of Cidea, Dio2, Pthr, Pgc1α, Tmem26, Tbx1, Ucp1, Eva1 and Pdk4 ( g ) g B control chectic achectic 0.0 0.5 1.0 1.5 UCP1 / -Actin control hectic chectic 0 1 2 3 PGC1 / -Actin control non-cachectic cachectic PGC1α/β-Actin UCP1/β-Ac n B B control non-cachectic cachectic 0.0 0.5 1.0 1.5 UCP1 / -Actin control non-cachectic cachectic 0 1 2 3 PGC1 / -Actin control non-cachectic cachectic PGC1α/β-Actin UCP1/β-Ac n Figure 2. 3.2. Nutritional and Inﬂammatory Status (A) The mRNA levels of Cell Death-Inducing DFFA-Like Effector A (Cidea), Iodothyronine Deiodinase 2 (Dio2), parathyroid hormone receptor (Pthr), peroxisome proliferator-activated receptor gamma coactivator-1-α (Pgc1α), Transmembrane protein 26 (Tmem26), T-box transcription factor 1 (Tbx1), uncoupling protein 1 (Ucp1), EVA1 and Pyruvate Dehydrogenase Kinase 4 (Pdk4) were analyzed using quantitative real-time PCR in SAT of gastrointestinal cancer patients (N = 24) and controls (N = 13). (B) Representative Western blot images and protein densitom- etry quantiﬁcation for PGC1α and UCP1 protein levels in SAT of a subset of cancer patients (N = 19) and controls (N = 8). β-Actin was used as loading control. p p g ( g ), p ( ), T-box transcription factor 1 (Tbx1), uncoupling protein 1 (Ucp1), EVA1 and Pyruvate Dehydrogenase Kinase 4 (Pdk4) were analyzed using quantitative real-time PCR in SAT of gastrointestinal cancer patients (N = 24) and controls (N = 13). (B) Representative Western blot images and protein densitom- etry quantiﬁcation for PGC1α and UCP1 protein levels in SAT of a subset of cancer patients (N = 19) and controls (N = 8). β-Actin was used as loading control. Additionally, cancer patients showed higher Cidea levels with respect to controls (median 1.34 IQR 0.96; 2.41 vs. 0.60 IQR 0.44; 0.86) (p < 0.001) (Figure 1A). In parallel, Tmem26 levels were signiﬁcantly increased in cancer patients compared to controls (median 1.51 IQR 0.76; 2.03 vs. 0.77 IQR 0.41; 0.99) (p = 0.026) (Figure 1A). Additionally, Pdk4 levels tended to increase in the SAT of cancer patients with respect to controls (mean 2.09 ± 1.00 vs. 1.46 ± 0.90) (p = 0.066) (Figure 1A), whereas no changes in Pgc1α expression were detected in cancer patients with respect to controls (median 1.40 IQR 0.80; 2.45 vs. 1.17 IQR 0.68; 1.42) (p = 0.179). However, using Western blot analysis performed in a subset of samples, we found higher PGC1α levels in gastrointestinal cancer patients with respect to controls (1.50 IQR 0.97; 1.71 vs. 0.73 IQR 0.42; 1.12) (p = 0.005) (Figure 1B), whereas no Cancers 2022, 14, 1948 7 of 14 7 of 14 difference was observed between gastrointestinal cancer patients and controls in UCP1 protein levels (Figure 1B). difference was observed between gastrointestinal cancer patients and controls in UCP1 protein levels (Figure 1B). 3.4. 3.2. Nutritional and Inﬂammatory Status Evaluation of browning genes and protein expression in SAT of cachectic and non-cachectic cancer patients.(A) The mRNA levels of Cidea, Dio2, Pthr, Pgc1α, Tmem26, Tbx1, Ucp1, Eva1 and Pdk4 control non-cachectic cachectic 0.0 0.5 1.0 1.5 UCP1 / -Actin control non-cachectic cachectic 0 1 2 3 PGC1 / -Actin control non-cachectic cachectic PGC1α/β-Actin UCP1/β-Ac n Figure 2. Evaluation of browning genes and protein expression in SAT of cachectic and non-cachectic cancer patients.(A) The mRNA levels of Cidea, Dio2, Pthr, Pgc1α, Tmem26, Tbx1, Ucp1, Eva1 and Pdk4 control non-cachectic cachectic 0.0 0.5 1.0 1.5 UCP1 / -Actin control non-cachectic cachectic 0 1 2 3 PGC1 / -Actin PGC1α/β-Actin UCP1/β-Ac n control non-cachectic cachectic Figure 2. Evaluation of browning genes and protein expression in SAT of cachectic and non-cachectic cancer patients.(A) The mRNA levels of Cidea, Dio2, Pthr, Pgc1α, Tmem26, Tbx1, Ucp1, Eva1 and Pdk4 Cancers 2022, 14, 1948 8 of 14 were analyzed using quantitative real-time PCR in cachectic (N = 9) and non-cachectic (N = 15) gastrointestinal cancer patients and in control group (N = 13) (B) Representative Western blot images and protein densitometry quantiﬁcation for PGC1α and UCP1 protein levels in SAT of a subset in cachectic (N = 7) and non-cachectic (N = 12) patients and in control group (N = 8). β-Actin was used as loading control. were analyzed using quantitative real-time PCR in cachectic (N = 9) and non-cachectic (N = 15) gastrointestinal cancer patients and in control group (N = 13) (B) Representative Western blot images and protein densitometry quantiﬁcation for PGC1α and UCP1 protein levels in SAT of a subset in cachectic (N = 7) and non-cachectic (N = 12) patients and in control group (N = 8). β-Actin was used as loading control. Interestingly, cachectic patients showed higher expressions of Pgc1α (median 2.25 IQR 1.22; 3.47) compared to non-cachectic patients (median 1.04 IQR 0.69; 1.94) and to controls (median 1.17 IQR 0.68;1.42) (p = 0.030 and p = 0.016, respectively) (Figure 2A). Moreover, no differences between cachectic and non-cachectic patients were seen in Tmem26 mRNA expression, nor in Pdk4 (Figure 2A). In Western blotting analysis performed in a subset of samples, we did not ﬁnd differ- ences in PGC1α and UCP1 protein levels among the groups (cachectic and non-cachectic patients and controls) (Figure 2B). 3.5. Differences in Browning Gene Expression and Protein Levels according to the Type of Gastrointestinal Cancer 3.5. Differences in Browning Gene Expression and Protein Levels according to the Type of Gastrointestinal Cancer Analyzing Cidea, we observed increased expression in all the three types of cancer compared to controls: colorectal (median 1.71 IQR 1.00; 2.18, p = 0.004); gastric (median 1.12 IQR 0.91; 2.47, p = 0.018); pancreatic (median 1.40 IQR 0.86; 3.10, p = 0.003) (Figure 3A). No differences were detected in Cidea mRNA levels among the three different types of gastrointestinal cancer. Analyzing Pgc1α, we found higher expression in pancreatic cancer patients (median 2.08 IQR 1.43; 3.47) vs. controls (median 1.17 IQR 0.68; 1.42) (p = 0.004) and vs. colorectal cancer patients (median 0.73 IQR 0.49; 1.47) (p = 0.008), and we found a similar trend, although not signiﬁcant, vs. gastric patients (p = 0.159) (Figure 3A). g g g p p g Analyzing Ucp1, we found lower expression in pancreatic (median 0.63 IQR 0.35; 1.57) and colorectal (median 0.35 IQR 0.31; 1.07) cancer patients compared to controls (median 1.26 IQR 0.92; 1.81) (p = 0.017 and p = 0.004, respectively) (Figure 3A). In Western blot analysis performed in a subset of samples, we documented signiﬁcantly higher UCP1 protein levels in pancreatic cancer patients when compared to colorectal cancer patients (median 1.00 IQR 0.73; 1.11 vs. 0.36 IQR 0.31;0.41) (p = 0.002) and to controls (median 0.37 IQR 0.30; 1.00) (p = 0.031) (Figure 3B). Additionally, we described UCP1 localization in adipocytes among the different types of gastrointestinal cancers and controls, which is represented in Figure 4. Finally, PGC1α protein levels were signiﬁcantly higher in pancreatic cancer patients vs. controls (median 1.58 IQR 1.15; 2.06 vs. 0.73 IQR 0.42; 1.12, p = 0.004) (Figure 3B), whereas no differences were observed among the other cancer groups. UCP1 protein is presented as a brownish crescent around unilocular adipocytes. ncers 2022, 14, 1948 9 of 14 Figure 3 B control pancreatic cancer gastric cancer colorectal cancer 0.0 0.5 1.0 1.5 UCP1/ -Actin control pancreatic cancer gastric cancer colorectal cancer 0 1 2 3 4 PGC1 / -Actin * * $ control pancrea' c cancer gastric cancer colorectal cancer PGC1α/β‐Ac0n UCP1/β‐Actin p<0.005 p=0.031 p<0.005 Figure 3. Evaluation of browning genes and protein expression in white adipose tissue (WAT) of cancer patients divided by the three types of gastrointestinal cancer (colorectal, stomach and pancreas). 4. Discussion In the present study, we were able to document the involvement of several changes in browning markers among patients with gastrointestinal cancer undergoing surgery. In particular, unexpectedly, Ucp1 mRNA levels were shown to be decreased in our cohort of cancer patients with respect to controls, as well as among cachectic and non-cachectic patients compared to controls. UCP1 is located in the inner membrane of mitochondria and is considered central in the browning process due to its role in thermogenesis through the dissipation of pro- ton gradients, producing heat instead of ATP [20]. In particular, PTHrP induced UCP1 expression, promoting cachectic phenotypes in a Lewis lung carcinoma (LLC) model of cancer cachexia [8]. However, recent reports have shown contradictory results regard- ing UCP1 expression in adipose tissue in cancer. In fact, although Ucp1 gene expression and protein level were found to be increased in cancer in different animal and human studies [12,13], Michaelis K. et al. showed decreased UCP1 levels in a murine model of pancreatic-cancer-associated cachexia [21]. In particular, cachexia and low Ucp1 expression in brown and white adipose tissue were associated with increased inﬂammation in the central nervous system and in the periphery and with anemia and changes in circulating testosterone levels [21]. Importantly, Rohm et al. [22] showed that in cancer cachexia mice models, the ther- mogenesis mediated by UCP1 may be less pivotal in driving the wasting process with respect to the results obtained in previous studies [8,13,23]. In particular, data showed that an AMP-activated protein kinase (Ampk)-stabilizing peptide determined the ameliora- tion of adipose tissue wasting interfering with CIDEA pathways independently of UCP1; moreover, the absence of UCP1 did not protect against the development of cachexia in mice [23]. Interestingly, different thermogenic pathways independent of UCP1 were described to play a potential role in cancer cachexia [24]. However, data on independent thermogenic pathways of UCP1 are lacking in this setting, especially in humans. Our data suggest negative modulation in the Ucp1 gene expression, and although not conﬁrmed at the protein level, the data represent results that are, at least in part, in line with previous reports [21,22]. On the other hand, UCP1 protein levels were increased in pancreatic cancer patients, suggesting potential translational regulation according to the different cancer type. 3.5. Differences in Browning Gene Expression and Protein Levels according to the Type of Gastrointestinal Cancer (A) The mRNA levels of Cidea, Dio2, Pthr, Pgc1α, Tmem26, Tbx1, Ucp1, Eva1 and Pdk4 were analyzed using quantitative real-time PCR in SAT of gastrointestinal cancer patients according to the different type of tumor (gastric N = 7, colorectal N = 8, pancreas N = 9) and in control group (N = 13). (B) Representative Western blot images and protein densitometry quantiﬁcation for PGC1α and UCP1 in a subset of pancreatic (N = 7), gastric (N = 5) and colorectal (N = 7) cancer patients, and in control group (N = 8). β-Actin was used as loading control. 9 of 14 Cancers 2022, 14, 1948 Figure 3 B control pancreatic cancer gastric cancer colorectal cancer 0.0 0.5 1.0 1.5 UCP1/ -Actin control pancreatic cancer gastric cancer colorectal cancer 0 1 2 3 4 PGC1 / -Actin * * $ control pancrea' c cancer gastric cancer colorectal cancer PGC1α/β‐Ac0n UCP1/β‐Actin p<0.005 p=0.031 p<0.005 B B Figure 3. Evaluation of browning genes and protein expression in white adipose tissue (WAT) Figure 3. Evaluation of browning genes and protein expression in white adipose tissue (WAT) of cancer patients divided by the three types of gastrointestinal cancer (colorectal, stomach and pancreas). (A) The mRNA levels of Cidea, Dio2, Pthr, Pgc1α, Tmem26, Tbx1, Ucp1, Eva1 and Pdk4 were analyzed using quantitative real-time PCR in SAT of gastrointestinal cancer patients according to the different type of tumor (gastric N = 7, colorectal N = 8, pancreas N = 9) and in control group (N = 13). (B) Representative Western blot images and protein densitometry quantiﬁcation for PGC1α and UCP1 in a subset of pancreatic (N = 7), gastric (N = 5) and colorectal (N = 7) cancer patients, and in control group (N = 8). β-Actin was used as loading control. Cancers 2022, 14, 1948 10 of 14 10 of 14 Figure 4. Immunohistochemical cytoplasmatic localization of UCP1 protein in SAT from control and patients with pancreatic, gastric and colorectal cancer. The scale bar used is 200 µm. Figure 4. Immunohistochemical cytoplasmatic localization of UCP1 protein in SAT from control and patients with pancreatic, gastric and colorectal cancer. The scale bar used is 200 µm. 4. Discussion In our analyses, Cidea and Tmem26 mRNA levels and PGC1α protein levels were increased in the SAT of cancer patients with respect to controls, whereas Pdk4 expression was only slightly increased with respect to controls. In particular, Cidea expression was higher among cachectic and non-cachectic patients vs. controls, whereas no difference was found between the cachectic and non-cachectic Cancers 2022, 14, 1948 11 of 14 11 of 14 groups. CIDEA was recently indicated as a crucial factor for the development of adipose tissue wasting via the promotion of browning and increased lipolysis [23,25]. In particular, robust data showed that CIDEA regulates UCP1 for britening and thermogenesis in human adipocytes [25]. CIDEA expression was particularly increased in the adipocytes of cachectic cancer patients likely due to their reduced body size and low adiposity, determined by the presence of a poor nutrition status [26]. This phenomenon might not have been detectable in our cohort among cachectic and non-cachectic patients considering the relatively early stage of cancer disease (new diagnosis) and the limited percent of body weight loss experienced in the prior months. However, Cidea represents an important marker of browning contributing to adipose loss with mechanism(s) that may differ according to the type of cancer and/or patients’ characteristics (e.g., percent of body weight loss), as well as Tmem26, which was modulated in cancer but not according to the presence/absence of cachexia. Interestingly, regarding the differences between cachectic and non-cachectic cancer patients, we found that Pgc1α gene expression was higher in the ﬁrst group, whereas no difference was found at the protein level. PGC1α is a key regulator of mitochondrial biogenesis. Interestingly, the irisin–PGC1α pathway was recently considered as a promoter of the browning process in adipose tissue [27]. In fact, PGC1α was initially suggested as a co-activator of PPAR- γ that regulates the expression of UCP1 and thermogenesis in brown adipocytes [28]. Lately, data showed that it stimulates the secretion of irisin from muscle able to interact with other tissues, including adipose [27,29]. PGC1α also represents a target of adaptative thermogenesis in brown adipose tissue by the activation of adrenergic receptor [30,31]. Studying PGC1α mechanisms may elucidate the metabolic regulation of adipose tissue in catabolic states, such as cancer. 4. Discussion We were not able to document signiﬁcant changes in other makers of browning, including Tbx1, Pthr, Eva1 and Dio2, in the SAT of our cohort patients, although pre- vious experiments documented the modulation of these markers during the browning process [8,11–13]. In light of this, additional evidence is needed, especially in humans, to clarify their role. y Furthermore, our study allows us to speculate on the potential differences in browning markers according to the type of gastrointestinal cancer. In fact, we found that Ucp1 mRNA levels were lower in colorectal and pancreatic cancer patients vs. controls, and the UCP1 protein level was higher in pancreatic cancer patients vs. controls and vs. colorectal cancer patients. This behavior was not observed for Cidea (no difference according to the type of tumor) but was similar for PGC1α, which showed signiﬁcant modulation in pancreatic but not in other gastrointestinal cancers. g Our observations appear to be clinically relevant considering that nutritional and metabolic alterations have a negative impact on prognosis in patients with gastrointestinal tumors, in particular in those with pancreatic cancer [32]. This is particularly important when taking into account that pancreatic cancer is a disease more frequently complicated by cachexia with respect to other gastrointestinal cancers [1,33]. The most common conditions complained about by patients are the high percentage of unvoluntary body weight loss and low appetite [4,34,35]. Recent evidence showed that approximately 70% of pancreatic cancer patients were affected by moderate or severe protein-energy malnutrition, conditioning a poor functional status [32]. In a model of cancer cachexia, low food intake was not the main reason determining weight loss; rather, changes in the expression of factors controlling lipid metabolism and thermogenesis in brown adipose tissue were involved in the adipose tissue catabolism in disease-associated cachexia [36]. Recent clinical observations indicate that patients with advanced pancreatic cancer lose a large proportion of visceral and subcutaneous adipose tissue rapidly (approximately in 4 weeks), and this was associated with poor survival [37]. We believe that the early identiﬁcation of molecular targets involved in adipose tissue depletion, in particular the browning phenomenon, may allow us to identify the metabolic Cancers 2022, 14, 1948 12 of 14 12 of 14 alterations observed in cachexia before developing a severe clinical picture represented by the deep adipose depletion observed in body composition analysis, which also negatively impinges on quality of life [1,12]. 4. Discussion Our study presents several limitations, including the relatively reduced number of patients with each cancer type, the stage of cancer disease (i.e., ﬁrst cancer diagnosis), which may have limited the presence of advanced cachexia likely associated with more evident adipose tissue depletion. However, this allowed us to avoid interference with other factors, in particular with anti-cancer treatments, on the catabolic state and to identify the role of the browning process in cachexia due to the early stage of cancer disease. Some of the browning markers showed non-univocal behavior in the gene expression and at the protein level, although this may be commonly observed in such studies. A second PCR primer or a different control would have been helpful to rule out potential experimental biases, although here, this was not performed. We did not assess browning markers in serum/plasma but speciﬁcally in SAT. Additionally, we did not perform the immunohistochemistry of PGC1α. To interpret the role of the browning markers in cachexia in a more complete manner, we believe that the study of the lipolytic pathways might have clariﬁed the interaction with the browning process in the development of adipose tissue depletion. 5. Conclusions In conclusion, we found that in patients with gastrointestinal cancer, different mark- ers of browning are modulated and, in particular, pancreatic cancer showed signiﬁcant changes in terms of UCP1 and PGC1α. This suggests that thermogenesis abnormalities are clinically relevant in cancer cachexia and should be promptly investigated to counteract the development of severe nutritional and metabolic derangements occurring in cancer, which negatively impact patients’ prognoses. Author Contributions: Conceptualization, A.M., R.B. and M.M.; data curation, A.M., R.B., G.I. and E.B.; formal analysis, R.B., G.I. and R.C.; funding acquisition, A.M. and M.M.; investigation, A.M., M.I.A., F.T., E.B. and G.N.; methodology, A.M., R.C. and G.N.; project administration, A.M., M.I.A. and G.N.; supervision, C.R.T.d.G. and E.F.; validation, A.M., C.R.T.d.G. and E.F.; writing—original draft, A.M. and G.I.; writing—review and editing, A.M. and M.M. All authors have read and agreed to the published version of the manuscript. Author Contributions: Conceptualization, A.M., R.B. and M.M.; data curation, A.M., R.B., G.I. and E.B.; formal analysis, R.B., G.I. and R.C.; funding acquisition, A.M. and M.M.; investigation, A.M., M.I.A., F.T., E.B. and G.N.; methodology, A.M., R.C. and G.N.; project administration, A.M., M.I.A. and G.N.; supervision, C.R.T.d.G. and E.F.; validation, A.M., C.R.T.d.G. and E.F.; writing—original draft, A.M. and G.I.; writing—review and editing, A.M. and M.M. All authors have read and agreed to the published version of the manuscript. Funding: For this study, A. Molﬁno received a research grant as Principal Investigator provided by Sapienza University of Rome (Grant number RG11816427B021CF). Institutional Review Board Statement: The study was performed in accordance with the Declaration of Helsinki and approved by the local Ethics Committee (Sapienza University, Azienda Sant’Andrea Hospital, Rome, Italy—prot. n. 167SA_2017). Institutional Review Board Statement: The study was performed in accordance with the Declaration of Helsinki and approved by the local Ethics Committee (Sapienza University, Azienda Sant’Andrea Hospital, Rome, Italy—prot. n. 167SA_2017). Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: The data presented in this study are available on request from the corresponding author. Acknowledgments: R. Belli and R. Carletti contributed to this study as recipients of the PhD program in Innovative Biomedical Technologies in Clinical Medicine, Sapienza University of Rome. Conﬂicts of Interest: The authors declare no conﬂict of interest. 3. Martin, L.; Muscaritoli, M.; Bourdel-Marchasson, I.; Kubrak, C.; Laird, B.; Gagnon, B.; Chasen, M.; Gioulbasanis, I.; Wallengren, O.; Voss, A.C.; et al. Diagnostic criteria for cancer cachexia: Reduced food intake and inﬂammation predict weight loss and survival in an international, multi-cohort analysis. J. Cachexia Sarcopenia Muscle 2021, 12, 1189–1202. [CrossRef] [PubMed] 1. Baracos, V.E.; Martin, L.; Korc, M.; Guttridge, D.C.; Fearon, K.C.H. Cancer-associated cachexia. Nat. Rev. Dis. Primers 2018, 4, 17105. [CrossRef] [PubMed] References [CrossRef] [ ] 10. Bartelt, A.; Heeren, J. Adipose tissue browning and metabolic health. Nat. Rev. Endocrinol. 2014, 10, 24 eeren, J. Adipose tissue browning and metabolic health. Nat. Rev. Endocrinol. 2014, 10, 24–36. [CrossRef] 11. Wu, J.; Boström, P.; Sparks, L.M.; Ye, L.; Choi, J.H.; Giang, A.H.; Khandekar, M.; Virtanen, K.A.; Nuutila, P.; Schaart, G.; et al. Beige adipocytes are a distinct type of thermogenic fat cell in mouse and human. Cell 2012, 150, 366–376. [CrossRef] 12. Jespersen, N.Z.; Larsen, T.J.; Peijs, L.; Daugaard, S.; Homøe, P.; Loft, A.; de Jong, J.; Mathur, N.; Cannon, B.; Nedergaard, J.; et al. A classical brown adipose tissue mRNA signature partly overlaps with brite in the supraclavicular region of adult humans. Cell Metab. 2013, 17, 798–805. [CrossRef] 13. Petruzzelli, M.; Schweiger, M.; Schreiber, R.; Campos-Olivas, R.; Tsoli, M.; Allen, J.; Swarbrick, M.; Rose-John, S.; Rincon, M.; Robertson, G.; et al. A switch from white to brown fat increases energy expenditure in cancer-associated cachexia. Cell Metab. 2014, 20, 433–447. [CrossRef] 14. Molﬁno, A.; Carletti, R.; Imbimbo, G.; Amabile, M.I.; Belli, R.; di Gioia, C.R.T.; Belloni, E.; Spinelli, F.; Rizzo, V.; Catalano, C.; et al. Histomorphological and inﬂammatory changes of white adipose tissue in gastrointestinal cancer patients with and without cachexia. J. Cachexia Sarcopenia Muscle 2021, 13, 333–342. [CrossRef] p 15. Fearon, K.; Strasser, F.; Anker, S.D.; Bosaeus, I.; Bruera, E.; Fainsinger, R.L.; Jatoi, A.; Loprinzi, C.; MacD et al. Deﬁnition and classiﬁcation of cancer cachexia: An international consensus. Lancet Oncol. 2011, 1 15. Fearon, K.; Strasser, F.; Anker, S.D.; Bosaeus, I.; Bruera, E.; Fainsinger, R.L.; Jatoi, A.; Loprinzi, C.; MacDonald, N.; Mantovani, G.; et al. Deﬁnition and classiﬁcation of cancer cachexia: An international consensus. Lancet Oncol. 2011, 12, 489–495. [CrossRef] 16 Kir S ; Komaba H ; Garcia A P; Economopoulos K P; Liu W ; Lanske B ; Hodin R A ; Spiegelman B M PTH/PTHrP Receptor 16. Kir, S.; Komaba, H.; Garcia, A.P.; Economopoulos, K.P.; Liu, W.; Lanske, B.; Hodin, R.A.; Spiegelman, B.M Mediates Cachexia in Models of Kidney Failure and Cancer. Cell Metab. 2016, 23, 315–323. [CrossRef] 17. Han, J.; Meng, Q.; Shen, L.; Wu, G. Interleukin-6 induces fat loss in cancer cachexia by promoting white adipose tissue lipolysis and browning. Lipids Health Dis. 2018, 17, 14. [CrossRef] 18. Livak, K.J.; Schmittgen, T.D. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. References Methods 2001, 25, 402–408. [CrossRef] 19. Finlin, B.S.; Memetimin, H.; Conﬁdes, A.L.; Kasza, I.; Zhu, B.; Vekaria, H.J.; Harfmann, B.; Jones, K.A.; Johnson, Z.R.; West- gate, P.M.; et al. Human adipose beiging in response to cold and mirabegron. JCI Insight 2018, 3, e121510. [CrossRef] g p g g p g g 20. Ikeda, K.; Yamada, T. UCP1 Dependent and Independent Thermogenesis in Brown and Beige Adipocytes. Front. Endocrinol. 2020, 11, 498. [CrossRef] A.; Zhu, X.; Burfeind, K.G.; Krasnow, S.M.; Levasseur, P.R.; Morgan, T.K.; Marks, D.L. Establishment and cha ovel murine model of pancreatic cancer cachexia. J. Cachexia Sarcopenia Muscle 2017, 8, 824–838. [CrossRef] Ü 21. Michaelis, K.A.; Zhu, X.; Burfeind, K.G.; Krasnow, S.M.; Levasseur, P.R.; Morgan, T.K.; Marks, D.L. E zation of a novel murine model of pancreatic cancer cachexia. J. Cachexia Sarcopenia Muscle 2017, 8, 22. Rohm, M.; Schäfer, M.; Laurent, V.; Üstünel, B.E.; Niopek, K.; Algire, C.; Hautzinger, O.; Sijmonsma, T.P.; Zota, A.; Medrikova, D.; et al. An AMP-activated protein kinase-stabilizing peptide ameliorates adipose tissue wasting in cancer cachexia in mice. Nat. Med. 2016, 22, 1120–1130. [CrossRef] eigerer, A.; Machado, J.; Herzig, S. Energy metabolism in cachexia. EMBO Rep. 2019, 20, e47258. [CrossRef] 23. Rohm, M.; Zeigerer, A.; Machado, J.; Herzig, S. Energy metabolism in cachexia. EMBO Rep. 2019, 20, e 24. Ikeda, K.; Kang, Q.; Yoneshiro, T.; Camporez, J.P.; Maki, H.; Homma, M.; Shinoda, K.; Chen, Y.; Lu, X.; Maretich, P.; et al. UCP1-independent signaling involving SERCA2b-mediated calcium cycling regulates beige fat thermogenesis and systemic glucose homeostasis. Nat. Med. 2017, 23, 1454–1465. [CrossRef] g 25. Jash, S.; Banerjee, S.; Lee, M.J.; Farmer, S.R.; Puri, V. CIDEA Transcriptionally Regulates UCP1 for Britening and Thermogenesis in Human Fat Cells. iScience 2019, 20, 73–89. [CrossRef] 26. Laurencikiene, J.; Stenson, B.M.; Arvidsson Nordström, E.; Nordström, E.; Agustsson, T.; Langin, D.; Isaksson, B.; Permert, J.; Rydén, M.; Arner, P. Evidence for an important role of CIDEA in human cancer cachexia. Cancer Res. 2008, 68, 9247–9254. [CrossRef] 27. Boström, P.; Wu, J.; Jedrychowski, M.P.; Korde, A.; Ye, L.; Lo, J.C.; Rasbach, K.A.; Boström, E.A.; Choi, J.H.; Long, J.Z.; et al. A PGC1-α-dependent myokine that drives brown-fat-like development of white fat and thermogenesis. Nature 2012, 481, 463–468. [CrossRef] [ ] 28. Puigserver, P.; Wu, Z.; Park, C.W.; Graves, R.; Wright, M.; Spiegelman, B.M. A cold-inducible coactiv linked to adaptive thermogenesis. Cell 1998, 92, 829–839. [CrossRef] 29. Ji, L.L.; Kang, C. References 1. Baracos, V.E.; Martin, L.; Korc, M.; Guttridge, D.C.; Fearon, K.C.H. Cancer-associated cachexia. Nat. Rev. Dis. Primers 2018, 4, 17105. [CrossRef] [PubMed] Links with sarcopenia, anorexia and asthenia. J. Cachexia Sarcopenia Muscle 2020, 11, 619–635. [CrossRef] [PubMed] 3. Martin, L.; Muscaritoli, M.; Bourdel-Marchasson, I.; Kubrak, C.; Laird, B.; Gagnon, B.; Chasen, M.; Gioulbasanis, I.; Wallengren, O.; Voss, A.C.; et al. Diagnostic criteria for cancer cachexia: Reduced food intake and inﬂammation predict weight loss and survival in an international, multi-cohort analysis. J. Cachexia Sarcopenia Muscle 2021, 12, 1189–1202. [CrossRef] [PubMed] 13 of 14 13 of 14 Cancers 2022, 14, 1948 4. Martin, L.; Senesse, P.; Gioulbasanis, I.; Antoun, S.; Bozzetti, F.; Deans, C.; Strasser, F.; Thoresen, L.; Jagoe, R.T.; Chasen, M.; et al. Diagnostic criteria for the classiﬁcation of cancer-associated weight loss. J. Clin. Oncol. 2015, 33, 90–99. [CrossRef] [PubMed] g g 5. Da Rocha, I.M.G.; Marcadenti, A.; de Medeiros, G.O.C.; Bezerra, R.A.; Rego, J.F.M.; Gonzalez, M.C.; Fayh, A.P.T. Is cachexia associated with chemotherapy toxicities in gastrointestinal cancer patients? A prospective study. J. Cachexia Sarcopenia Muscle 2019, 10, 445–454. [CrossRef] [PubMed] [ ] [ ] 6. Laviano, A.; Meguid, M.M.; Guijarro, A.; Muscaritoli, M.; Cascino, A.; Preziosa, I.; Molﬁno, A.; Fanelli, F.R. Antimyopathic effects of carnitine and nicotine. Curr. Opin. Clin. Nutr. Metab. Care 2006, 9, 442–448. [CrossRef] [PubMed] p 7. Das, S.K.; Eder, S.; Schauer, S.; Diwoky, C.; Temmel, H.; Guertl, B.; Gorkiewicz, G.; Tamilarasan, K.P.; Kumari, P.; Trauner, M.; et al. Adipose triglyceride lipase contributes to cancer-associated cachexia. Science 2011, 333, 233–238. [CrossRef] p 7. Das, S.K.; Eder, S.; Schauer, S.; Diwoky, C.; Temmel, H.; Guertl, B.; Gorkiewicz, G.; Tamilarasan, K.P.; Kumari, P.; Trauner, M.; et al. Adi t i l id li t ib t t i t d h i S i 2011 333 233 238 [C R f] Adipose triglyceride lipase contributes to cancer-associated cachexia. Science 2011, 333, 233–238. [CrossRef] 8. Kir, S.; White, J.P.; Kleiner, S.; Kazak, L.; Cohen, P.; Baracos, V.E.; Spiegelman, B.M. Tumour-derived PTH-related protein triggers adipose tissue browning and cancer cachexia. Nature 2014, 513, 100–104. [CrossRef] p g y p 8. Kir, S.; White, J.P.; Kleiner, S.; Kazak, L.; Cohen, P.; Baracos, V.E.; Spiegelman, B.M. Tumour-derived PTH-related protein triggers adipose tissue browning and cancer cachexia. Nature 2014, 513, 100–104. [CrossRef] 9. Molﬁno, A.; Imbimbo, G.; Muscaritoli, M. Role of metabolic changes of adiposity in cancer. Trends Endocrinol. Metab. 2021, 32, 957. References Role of PGC-1α in sarcopenia: Etiology and potential intervention—A mini-review. Gerontology 2015, 61, 139–148. [CrossRef] [PubMed] Cancers 2022, 14, 1948 14 of 14 14 of 14 30. Cannon, B.; Jacobsson, A.; Rehnmark, S.; Nedergaard, J. Signal transduction in brown adipose tissue recruitment: Noradrenaline and beyond. Int. J. Obes. Relat. Metab. Disord. 1996, 20 (Suppl. S3), S36–S42. [PubMed] y J , ( pp ), [ ] 31. Ricquier, D. Molecular biology of brown adipose tissue. Proc. Nutr. Soc. 1989, 48, 183–187. [CrossRef] [PubMed] q gy p [ ] [ Santos, I.; Mendes, L.; Mansinho, H.; Santos, C.A. Nutritional status and functional status of the pancreat the impact of adjacent symptoms. Clin. Nutr. 2021, 40, 5486–5493. [CrossRef] [PubMed] p j y p 33. Muscaritoli, M.; Lucia, S.; Farcomeni, A.; Lorusso, V.; Saracino, V.; Barone, C.; Plastino, F.; Gori, S.; Magarotto, R.; Carteni, G.; et al. Prevalence of malnutrition in patients at ﬁrst medical oncology visit: The PreMiO study. Oncotarget 2017, 8, 79884–79896. [CrossRef] [PubMed] 34. Molﬁno, A.; de van der Schueren, M.A.E.; Sánchez-Lara, K.; Milke, P.; Amabile, M.I.; Imbimbo, G.; Di Lazzaro, L.; Cavuto, S.; Ronzani, G.; Snegovoy, A.; et al. Cancer-associated anorexia: Validity and performance overtime of different appetite tools among patients at their ﬁrst cancer diagnosis. Clin. Nutr. 2021, 40, 4037–4042. [CrossRef] 35. Molﬁno, A.; Laviano, A.; Rossi Fanelli, F. Contribution of anorexia to tissue wasting in cachexia. Curr. Opin. Support Palliat. Care. 2010, 4, 249–253. [CrossRef] 36. Tsoli, M.; Moore, M.; Burg, D.; Painter, A.; Taylor, R.; Lockie, S.H.; Turner, N.; Warren, A.; Cooney, G.; Oldﬁeld, B.; et al. Activation of thermogenesis in brown adipose tissue and dysregulated lipid metabolism associated with cancer cachexia in mice. Cancer Res. 2012, 72, 4372–4382. [CrossRef] 37. Nakano, O.; Kawai, H.; Kobayashi, T.; Kohisa, J.; Ikarashi, S.; Hayashi, K.; Yokoyama, J.; Terai, S. Rapid decline in visceral adipose tissue over 1 month is associated with poor prognosis in patients with unresectable pancreatic cancer. Cancer Med. 2021, 10, 4291–4301. [CrossRef]
https://openalex.org/W2061700308	https://zenodo.org/records/2311597/files/article.pdf	English	null	Efficiency and a Nine Hour Day	Journal of the American Pharmaceutical Association/The Journal of the American Pharmaceutical Association	1,915	public-domain	2,396	320 320 THB JOUBNAL OF THH We were careful to keep the moisture in the case between 60-65 degrees, keep- ing the case tightly closed when not in use, consequently the stock is always in perfect condition and now we have the reputation of being the only store in town, whose stock is not all dried out during the winter time, this, of course, has helped the trade greatly and the cigar department has grown until it is the best paying side line in the store, doing nearly three times the business it did before. During the winter months, when your store is heated by artificial heat, cigars and tobacco must be watched very carefully and a certain degree of moisture (60- 65) maintained in the case constantly, lest your stock dry out, lose its flavor and your trade go elsewhere, where the stock is properly kept. y g p p y p I figure that my cigar department pays my rent, light and heat, and that is doing pretty well in a town of 1600 with 15 places selling cigars and one cigar manufacturer. Too little attention is paid to this department in the average store, the stock is purchased, placed in the case and no further attention given it except to hand out the goods when called for. .Cigars are very sensitive to artificial heat and quickly dry out and lose their flavor and can not be brought back to their original condition. A little thought and attention will prevent this. In the first place get a good tight case, with plenty of moistening surface, place a hygrometer in it, and then see to it that the moisteners are filled at least once a week or as often as the hygrometer falls be- low 60 degrees, and keep the case tightly closed when not selling from it. g p g y g Have variety enough to suit all classes of customers, and endeavor to learn the likes and dislikes of your customers. When you put in a new brand show it up and pass out enough samples to start a demand for it. 320 In a community such as the one I live in, special sales and premiums do not pay, but by careful attention to keeping stock, giving good value, and a pleasant word to the customer any one can largely increase the trade in this department, without adding to the expense account. C. A. \VE.\\‘ER, PH. C. In taking up this subject of shorter hours for the Retail Druggist, I have been led to believe, from an experience of some twenty-eight years, behind the counter, that the working of long hours was an unnecessary hardship that could be easily avoided. This paper is written from the view-point of the so-called “two-man” drug store, comprising a registered proprietor, and one registered clerk, which I be- lieve, constitutes about 60 per cent. of all the drug stores of the United States- stores doing a business of from fifteen to thirty thousand dollars a year. The smaller amount representing the two-man store, with the ‘limited territory and 321 AMEBICAN PHARMACEUTICAL ASSOCIATI little effort, and the twenty to thirty thousand dollars, the busy, “get-there” store, with a larger field, and a more pronounced effort to build up trade. little effort, and the twenty to thirty thousand dollars, the busy, “get-there” store, with a larger field, and a more pronounced effort to build up trade. Starting some twenty-eight years ago, as errand boy in a drug store, I nat- tirally inherited the then prevailing hours of 7 a. m. to 10 p. m., with one night and one afternoon or two nights and part of every Sunday off. It seemed only natural to fall into this rut, as all drug stores were conducted on very much the same plan and any scheme for shorter hours at that time would have been pure heresy. My experience of the first eighteen years resulted in my coming to the follow- ing conclusion, that the drug business with its long, nerve-wracking hours, was not worth its remuneration, especially if a probable loss of health were considered. It was quite natural then, at this time that I decided to forsake the retail for the manufacturing end of the business, the latter offering every opportunity for the continuation along lines of work, for which my education had prepared me, at the same time diminishing the hours of labor to a point where rest and recreation, so absolutely necessary to a man’s health and efficiency were in evidence. C. A. \VE.\\‘ER, PH. C. Despite the fact that the relief from long hours was fully obtained, after one year I accepted a position as manager of the store in which I had formerly clerked, having come to the conclusion that the retail drug ‘business offered a greater opportunity, over the manufacturing, for the individual, and with the firm conviction that a scheme of hours could be evolved in which greater results’ would be obtained by concentrated effort, over a period of eight or nine, rather than a weakened effort extended over fifteen or sixteen hours, as formerly. At this time I was fortunate in having the services of a very good clerk, who felf very much as I did in regard to the long hours devoted to the accomplish- ment of a given amount of work. So after considering the requirements of ouf- store, we decided the work could be done to better advantage in shifts averaging about nine hours each, sometimes a little more, sometimes a little less. One week the clerk opening the store at 7 a. m., having one and a half hours for dinner, and leaving at 6 p. m., myself coming on at twelve, noon, having one hour for s u p per and closing at ten p. m. We also worked but six days a week-one week having all day Saturday and the next week, all day Sunday off. This was the only hardship that went with the arrangement, as it necessitated having our meals in the store one day each week, but as it gave us one whole day for recreation, we were. glad to overlook this slight discomfort. This Saturday and Sunday work has been varied and changed from time to time up to the present moment. This scheme of hours was continued for some seven years, and still appeals to me as the more desirable one, as the hours of labor are continuous. Individual requirements, however, necessitate various modifications, for instance at the present time, we are working in divided shifts-one man working in the forenoon and evening of one day and the morning and afternoon of the next day, which gives each man every other afternoon and every other evening off. This plan; like the preceding one, provides for about nine hours in the store each day and has been in operation the last three years. C. A. \VE.\\‘ER, PH. C. It was a pleasure to see that clerk come in at noon bf the days he was to work the late To say the,oi-iginal scheme was a success is putting it mildly. 322 THE JOURNAL OF THE shift-to see him get into his store coat, and then notice the work fly. It dis- appeared like magic. This man was not tired out, he was not filled with the toxins of fatigue-he was “Johnny-on-the-Spot”-he had a new toy-he had the pleasure of labor, with the time for play coming, also the time for rest, and plenty of it in sight. The clerk and I knowing it was a test case, naturally started out to make it a success, and we did. The work of the store was never so well or quickly done. Detroit at this time was just on the rise, and our business grew by leaps and bounds, without seemingly to disturb or very much overwork us at any time. Efficiency is the order of the day. It is not how long we work, but how much we accomplish in a given time. I believe a man can do about so much work in a day, and from experience, believe this amount can be accomplished in eight or nine hours of industrious labor. If your labor be extended over a much longer period, a slowing down, a lack of “pep” and a general loss of efficiency will fol- low. For a time the work may be done, and possibly as well, but at what a foolish cost, when at least one-third of the time might have been spent away from the store. To-day, when I hear a druggist complain of the confinement and long hours of his business, I know he must be at fault personally. He either does not work long hours, or applies himself only hal’f-heartedly, at the time he is supposed to be on duty. The latter is undoubtedly the main trouble. He has no regular hours for himself. He comes and goes when he pleases and is only governed by the clerk’s time off. The store is on his mind when he is not there and he thinks he is working. C. A. \VE.\\‘ER, PH. C. Let him get right down to “brass tacks” for he is not “delivering the goods.” He could not work for anyone else in the slip-shod way he works for himself, and what excuse a man can offer for giving himself poorer service than he would render others, is “too deep” for me. Give yourself and your clerk the benefit of shorter hours and then see that you devote those hours entirely to work. You are the one to set the example for your clerk-you are the measure by which he plans his work.-Set him an ex- ample by being prompt and industrious in your hours of labor. Visiting, stand- ing around unoccupied, reading the newspapers, etc., never brought a man any- where in this world of business, with its keen competition demanding your atten- tion every minute. Hustle a little-get all the work finished for once and strive to keep it so, by systematic effort. The appearance of your store resulting in more and better business, will repay you beyond your fondest expectations. That the unfavorable impression too often given customers by a clerk or even a proprietor, in grudgingly leaving a story or magazine, to half-heartedly wait on them, should be eliminated entirely from our stores, there can be no question, and undoubtedly shorter hours is the solution. Read more, play more, rest more away from business. but work more, strive more, accomplish more during your hours of labor. short hour scheme out of the question, but my contentions to the contrary are based on the fact that ,my personal experience covered two neighborhoods vastly different from each other, and in each the resulting success was the same. Druggists may argue that locations of their businesses, etc., may make this - 323 AMERICAN PHARMACEUTICAL ASSOCIATI AMERICAN PHARMACEUTICAL ASSOCIATI Many imaginary obstacles may arise, in the minds of the individual, as he en- deavors to apply the shorter hours to his own business, but experiment will con- vince the most skeptical that this is one of the problems of the drug business that can be most easily solved. In summing up, I would have you thoroughly appreciate the fact, that this is not merely a suggestion or theory, but my personal experience of short hours in the past ten years of my business life. Any druggist can denzonstratc the practicability of shorter hours to his entire satisfaction, if he will but make a determined start, necessarily with the co-opera- tion of his clerk and help in general, along the lines of systematic labor. Isn’t it really worth the while when a better-kept, and “up-to-the-minute” store naturally producing more business, is the result, while at the same time, health and recreation are not sacrificed, but on the contrary, are a direct outcome of the whole scheme? Surely you will find that much of the weariness and dissatisfaction with our w.ork, which we as employers experience, and much of the unrest and shifting around of our clerks will disappear. TREATMENT OF IL‘OUNDS IN WAR. A striking fact observed in the treatment of wounded in the present war, whether on land or sea, is the great prevalence of sepsis. In the author’s own experience, all the wounds he had seen so far were septic, some of them very badly so. Tetanus and acute spreading gangrene are also common, though tetanus has not yet been seen in the naval wounded, and this can be understood when it is remembered that the tetanus bacillus reaches wounds from the soil. The prevalence of sepsis in the large wounds is possibly to be accounted for by the length of time which may elapse after the injury before the patient comes under treatment. As to the treatment of this class of wounds, the author lays it down as an axiom of practice that if the treatment can be carried out within the first twenty-four hours (and, in the case of wounds soiled with earth, forty-eight hours) an attempt should be made to kill the organisms which have entered the wound. For this purpose, chemical antiseptics are probably the only means available, and in this connection there are two important points to be taken into consideration-namely, to kill actively-growing bacteria, and also the spores of bacilli. A saturated aqueous solution of carbolic acid (1 in 20) will kill naked, actively growing bacteria in a few seconds, but will not kill spores with certainty under twelve to fifteen hours. Liquefied carbolic acid, however, will kill spores in a very few minutes, and this must be used for wounds soiled with earth. Carbolic acid is an anasthetic, and its application causes very slight pain, which subsides almost immediately. But iodine, though prabtically of the same anti- septic power as carbolic acid, has a number of disadvantages, one of them being the great pain it causes, this lasting for a considerable time. Therefore, the author prefers carbolic acid to iodine not only for the disinfection of the skin, but also, and more especially, as a means of destroying the bacteria which have already entered wounds before they come under the care of the surgeon.-Sir W. Myatson Cltcync, Bart. (Brit. dlcd. Jaunt., Noveinber 21. 1914, 865).
https://openalex.org/W2896759511	https://discovery.ucl.ac.uk/10058727/1/GEC_2018_426_Original_V0.pdf	English	null	Social cost of carbon under shared socioeconomic pathways	Global environmental change	2,018	cc-by	7,077	Social Cost of Carbon under Shared Socioeconomic Pathways Social Cost of Carbon under Shared Socioeconomic Pathways Research Paper Abstract The Social Carbon Cost (SCC) represents the economic damage caused by an additional ton of carbon emissions and is widely used by governments to price carbon. Because the SCC is defined by social welfare, its estimation is necessarily dependent on future assumptions that are difficult to project. Many approaches consider the impact of population or economic growth on the SCC, but these socioeconomic factors must be grounded on solid assumptions concerning political, technological and environmental developments. Over the past seven years, the climate change research community has established five plausible socioeconomic narratives, called ‘Shared Socioeconomic Pathways’ (SSPs), numbered SSP1–SSP5. These scenarios provide descriptions of how the future might unfold in several key areas. To this end, we use the China Climate Change integrated assessment model (C3IAM) and the Dynamic integrated model of Climate and the Economy (DICE) to update the SCC under the five socioeconomic pathways, while also considering alternative damage functions and the social welfare discount rate to address uncertainty. The results show that, in a world developing towards regional rivalry (SSP3), the average SCC today will likely double compared with other scenarios. If additional developing countries emerge that follow the same path as previous industrializations (SSP5), the SCC will experience a rapid increase after 2060. Inequality (SSP4) will experience low mitigation pressure under a sustainable development scenario (SSP1), while the historical development pattern (SSP2) will have a moderate SCC with higher uncertainty. The results can provide carbon price benchmarks for policy makers who hold different attitudes towards the future and can help address the need to avoid regional rivalries and fossil-fueled development, which may counteract mitigation efforts. Manuscript Details Manuscript Details Manuscript number Title Article type GEC_2018_426 GEC_2018_426 Keywords Corresponding Author Corresponding Author's Institution Order of Authors Suggested reviewers Pu Yang, Yun-Fei Yao, Zhifu Mi, Yun-Fei Cao, Hua Liao, Biying Yu, Qiao-Mei Liang, D'Maris Coffman, Y-Ming Wei BIN SU, Xunpeng Shi, Rong-Gang Cong, Jing Meng Pu Yang, Yun-Fei Yao, Zhifu Mi, Yun-Fei Cao, Hua Liao, Biying Yu, Qiao-Mei Liang, D'Maris Coffman, Y-Ming Wei Submission Files Included in this PDF File Name [File Type] Cover letter.docx [Cover Letter] Highlights.docx [Highlights] Title page.docx [Title Page (with Author Details)] 20180419 Manuscript.docx [Manuscript (without Author Details)] To view all the submission files, including those not included in the PDF, click on the manuscript title on your EVISE Homepage, then click 'Download zip file'. Submission Files Included in this PDF To view all the submission files, including those not included in the PDF, click on the manuscript title on your EVISE Homepage, then click 'Download zip file'. The Bartlett School of Construction & Project Management The Bartlett School of Construction & Project Management 19th April 2018 19th April 2018 Dear Editors, I would like to submit this paper entitled of ‘Social Cost of Carbon under Shared Socioeconomic Pathways’ for your consideration of potential publication in Global Environmental Change – Human and Policy Dimensions. I would like to submit this paper entitled of ‘Social Cost of Carbon under Shared Socioeconomic Pathways’ for your consideration of potential publication in Global Environmental Change – Human and Policy Dimensions. The Social Carbon Cost (SCC) represents the economic damage caused by an additional ton of carbon emissions and is widely used by governments to price carbon. Over the past seven years, the climate change research community has established five plausible socioeconomic narratives, called ‘Shared Socioeconomic Pathways’ (SSPs), numbered SSP1–SSP5. We use the China Climate Change integrated assessment model (C3IAM) and the Dynamic integrated model of Climate and the Economy (DICE) to update the SCC under the five socioeconomic pathways, while also considering alternative damage functions and the social welfare discount rate to address uncertainty. In a world developing towards regional rivalry (SSP3), the average SCC today will likely double compared with other scenarios. If additional developing countries emerge that follow the same path as previous industrializations (SSP5), the SCC will experience a rapid increase after 2060. Inequality (SSP4) will experience low mitigation pressure under a sustainable development scenario (SSP1), while the historical development pattern (SSP2) will have a moderate SCC with higher uncertainty. The results can provide carbon price benchmarks for policy makers who hold different attitudes towards the future and can help address the need to avoid regional rivalries and fossil-fueled development, which may counteract mitigation efforts. We confirm that there are not conflicts of interest for this work. Sincerely yours, Zhifu Mi, PhD Permanent Research Fellow in Climate Change Economics The Bartlett School of Construction and Project Management University College London (UCL) Email: z.mi@ucl.ac.uk Tel: +44 (0)7598468210 Zhifu Mi, PhD The Bartlett School of Construction and Project Management The Bartlett UCL Faculty of the Built Environment 2nd Floor 1-19 Torrington Place London WC1E 7HB T+44 (0)20 7679 8263 www.bartlett.ucl.ac.uk 1Corresponding authors: wei@bit.edu.cn (Y.-M. Wei) and Z.Mi@uea.ac.uk (Z. Mi). Highlights  We update the social carbon cost under five shared socioeconomic pathways.  This is the first article to address the socioeconomic impact of social carbon cost.  The development of regional rivalries will double the present social carbon cost.  After 2060, social carbon costs will rise to unbearable levels with continued fossil- fueled development.  We update the social carbon cost under five shared socioeconomic pathways.  This is the first article to address the socioeconomic impact of social carbon cost.  The development of regional rivalries will double the present social carbon cost.  This is the first article to address the socioeconomic impact of social carbon cost.  The development of regional rivalries will double the present social carbon cost.  After 2060, social carbon costs will rise to unbearable levels with continued fossil- fueled development.  After 2060, social carbon costs will rise to unbearable levels with continued fossil- fueled development.  After 2060, social carbon costs will rise to unbearable levels with continued fossil- fueled development. Social Cost of Carbon under Shared Socioeconomic Pathways Pu Yang a,b,c, Yun-Fei Yaoc, Zhifu Mi e , Yun-Fei Cao a,b,c,d , Hua Liao a,b,c,d, Bi- Ying Yu a,b,c,d, Qiao-Mei Liang a,b,c,d, D'Maris Coffman e ,Yi-Ming Wei a,b,c,d,1 a Center for Energy and Environmental Policy Research, Beijing Institute of Technology, Beijing 100081, China b School of Management and Economics, Beijing Institute of Technology, Beijing 100081, China c Beijing Key Laboratory of Energy Economics and Environmental Management, Beijing 100081, China d Sustainable Development Research Institute for Economy and Society of Beijing, Beijing 100081, China e The Bartlett Sch of Const & Proj Mgt,Faculty of the Built Environment, University College London, London WC1E 6BT,UK Keywords: Climate Change; Integrated Assessment; Social cost of Carbon; Shared Socioeconomic Pathways; C3IAM; DICE Keywords: Climate Change; Integrated Assessment; Social cost of Carbon; Shared Socioeconomic Pathways; C3IAM; DICE Abstract The Social Carbon Cost (SCC) represents the economic damage caused by an additional ton of carbon emissions and is widely used by governments to price carbon. Because the SCC is defined by social welfare, its estimation is necessarily dependent on future assumptions that are difficult to project. Many approaches consider the impact of population or economic growth on the SCC, but these socioeconomic factors must be grounded on solid assumptions concerning political, technological and environmental developments. Over the past seven years, the climate change research community has established five plausible socioeconomic narratives, called ‘Shared Socioeconomic Pathways’ (SSPs), numbered SSP1–SSP5. These scenarios provide descriptions of how the future might unfold in several key areas. To this end, we use the China Climate Change integrated assessment model (C3IAM) and the Dynamic integrated model of Climate and the Economy (DICE) to update the SCC under the five socioeconomic pathways, while also considering alternative damage functions and the social welfare discount rate to address uncertainty. The results show that, in a world developing towards regional rivalry (SSP3), the average SCC today will likely double compared with other scenarios. If additional developing countries emerge that follow the same path as previous industrializations (SSP5), the SCC will experience a rapid increase after 2060. Inequality (SSP4) will experience low mitigation pressure under a sustainable development scenario (SSP1), while the historical development pattern (SSP2) will have a moderate SCC with higher uncertainty. The results can provide carbon price benchmarks for policy makers who hold different attitudes towards the future and can help address the need to avoid regional rivalries and fossil-fueled development, which may counteract mitigation efforts. Social Cost of Carbon under Shared Socioeconomic Pathways Pu Yang a,b,c, Yun-Fei Yaoc, Zhifu Mi e , Yun-Fei Cao a,b,c,d , Hua Liao a,b,c,d, Bi- Ying Yu a,b,c,d, Qiao-Mei Liang a,b,c,d, D'Maris Coffman e ,Yi-Ming Wei a,b,c,d,1 c Beijing Key Laboratory of Energy Economics and Environmental Management, Beijing 100081, China d Sustainable Development Research Institute for Economy and Society of Beijing, Beijing 100081, China e The Bartlett Sch of Const & Proj Mgt,Faculty of the Built Environment, University College London, London WC1E 6BT,UK 1. Introduction After the Paris Agreement, countries have increasingly taken actions to address climate change. However, the policy costs vary among countries and sectors, making it difficult to select the most worthwhile policies. This problem can be addressed by calculating the social cost of carbon (SCC), which balances the social costs resulting from emission reductions with the incremental costs of regulation policy. The US government has relied on the SCC estimates provided by the Interagency Working Group (IWG) as a basis for taxing and implementing regulation policies (Revesz et al., 2017). The IWG SCC estimates started in 2010 and were updated with new scientific developments in 2013 and 2016, resulting in policy benefits of more than $1 trillion (Nordhaus, 2017). The SCC is also increasingly being adopted for regulations at the state level, resulting in regulatory policies in California, New York and Minnesota (California, 2016; Larson, 2016; Minnesota, 2016). Given the wide range of social and climate interactions included in the calculation, SCC estimation is necessarily complex and highly uncertain (Pindyck, 2013). Damage functions and social welfare discounts are considered the two major contributors to this uncertainty (Cai et al., 2016; Diaz and Moore, 2017; Heal and Millner, 2014; Howarth et al., 2014; Pycroft et al., 2014); however, any discussion of these issues is necessarily based on the underlying socioeconomic assumptions. Economic development can alter emission flow patterns (Mi et al., 2017), and—because the SCC is defined by social welfare—population and economic projections are fundamental determinants in its estimation. Scovronick et al. (2017) investigated the influence of future population growth on the SCC, Dietz and Stern (2015) and Moore and Diaz (2015) considered the impacts of climate on economic growth as the drivers of uncertainty. However, the democratic and economic assumptions are only two aspects of the socioeconomic assumption, which may be associated with a wide range of political, technological and environmental contingencies. If the recently imposed steel tariff continues developing and becomes a regional rivalry, it may well alter the historical development path of economic and policy characters, resulting in different SSC patterns. The SSP framework was initially proposed by Moss et al. (2010) and Van Vuuren et al. (2012), but the quantified and qualified version was published seven years later by Riahi et al. (2017). 1. Introduction The five SSPs characterize societal futures that present unique combinations of challenges to carbon mitigation and adaptation, including six broad projected categories, namely, demographics, economy and lifestyles, human development, policies, technology and natural resources. The SSP framework greatly facilitates integrated analyses of mitigation and adaptation. Pizer et al. (2014) revealed the importance of considering the new SSP framework into SCC estimates. Our paper estimates the SCC under the five SSP scenarios; we also extend our research by considering the uncertainty caused by damage functions and the social welfare discount rate. The results demonstrate the need to avoid regional rivalries and fossil-fueled development, which can raise the current SCC cost or induce much heavier mitigation pressures by the end of this century. The SCC value provides a carbon price benchmark for policy makers who hold different attitudes towards the future and is an important reference for future research under the various SSPs. 2.1 Overview of the methodology We use the C3IAM model to characterize the SSP and the DICE model to calculate the SCC. The DICE model is one of the three models used by the U.S. government and has been widely used for SCC estimation by scholars (e.g., (Crost and Traeger, 2014; Moore and Diaz, 2015; Scovronick et al., 2017)). Four variables, namely, the population, gross output, carbon intensity, and adaptation functions may change under different SSPs. However, many serve as exogenous variables in the model, which requires a model that includes detailed descriptions of socioeconomic factors to update these factors. The C3IAM, developed by Center for Energy & Environmental Policy Research, Beijing Institute of Technology, is an integrated assessment model that is theoretically based on the Computable General Equilibrium (CGE) and long term growth theory (the theoretical basis and structure of C3IAM are described in the Supporting Information). The C3IAM model can estimate policy costs under different socioeconomic assumptions and provide an aggregate emissions cost from all sectors. We use the C3IAM result to update the adaptation function and carbon intensity in the DICE model to embody the differences between SSPs, as described in Section 2.2. Because the SCC is also sensitive to alternative damage functions and to social welfare discounts, we extend our research by considering the uncertainty that arises from these two aspects. The damage functions and alternative social welfare discounts selected in this study are discussed in Section 2.3. An outline of our research is shown in Figure 1. Figure 1. Research Framework. L: population, TFP: total factor productivity, c: consumption per capita, MAT: atmospheric concentration of CO2, MUP: upper ocean/biosphere concentration of CO2, MLO: deep ocean concentration of CO2, TAT: atmospheric temperature, TLO: lower ocean temperature Figure 1. Research Framework. L: population, TFP: total factor productivity, c: consumption per capita, MAT: atmospheric concentration of CO2, MUP: upper ocean/biosphere concentration of CO2, MLO: deep ocean concentration of CO2, TAT: atmospheric temperature, TLO: lower ocean temperature 2.2.2 Qualify the SSPs in the DICE model For the quantitative factors, the population can be directly set as exogenous in the DICE model. Using the gross output driven by the endogenous capital formation, we chose the total factor productivity (TFP) to reflect different growth rates under different SSPs. The factor is derived using the Ramsey model. The population and output projection methods of the SSPs are described in (Leimbach et al., 2017; Samir and Lutz, 2017). For the qualitative factors, carbon intensity and the adaptation function are two other differentiating factors for SSPs. However, as they are highly related to the energy and sectoral assumptions, which cannot be directly reflected in DICE, C3IAM provides the SSP baseline and combines the Representative Concentration Pathways (RCPs) with the SSPs (Wei et al., 2018). The baseline emission is aggregated from 27 sectors and can be input as the carbon intensity in DICE. The RCPs can be considered as a set of policy shocks to the CGE model, which results in a set of emission reduction percentages accompanied by their policy costs. We use the result to fit the adaptation function in DICE, and the regression results are listed in the Supporting Information. 2.2.1 Characteristics of the five SSPs The SSPs consist of a set of quantitative projections and qualitative descriptions. The quantitative projections include population and economic growth. These factors are exogenous to most integrated assessment models that form the fundamental characteristics of each SSP. The qualitative narratives are thoroughly described in (O’Neill et al., 2017), including the aspects that are difficult to project quantitatively. To summarize, SSP1 represents the sustainable development path, SSP2 implies a development pathway consistent with typical historical patterns, and SSP3 is characterized by international fragmentation and regional rivalry. Low challenges to mitigation but high challenges to adaptation are observed in SSP4, which emphasizes extreme inequality. In contrast, SSP5 represents high challenges to mitigation and low challenges to adaptation, forecasting economic successes for both industrialized and emerging economies. 2.3 Alternative damage functions and social welfare discount rate The SCC is very sensitive to the damage function and social welfare discount rate. However, with our limited knowledge about the mechanisms of climate change, the accuracy of damage functions is unknown. The social welfare discount is valued not only as an economic term but also considered as an ethical primitive. Therefore, we provide the social carbon cost under nine damage functions and discuss the SCC under six alternative social welfare discount rates. The damage function in DICE-2016R has been used to provide SCC estimations for the U.S. government. To consider the uncertainty of damage functions, we tested eight additional functions based on the meta-analysis by Richard Tol (Tol, 2018), which includes 27 published estimates of the economic impact of climate change. The piecewise linear function provides the best fit with the lowest standard error; however, Tol also used seven other forms to fit the data. Although some functions have a higher standard error of regression, we still include the results as possibilities. Together with the damage function in DICE-2016R, we estimate the SCC under 9 damage functions to consider all the possibilities (as shown in Table 1). Table 1. Damage Function Based on Meta-analysis by Richard Tol (Tol, 2018) Specification Proposer Standard Error of Regression 0.236 T2 DICE-2016R (-0.74 T) IT<1.01 + (1.41 T-2.18) IT≥1.01 Meta-analysis 1.12 0.12 T + 0.16 T2 Tol (2009) 1.17 0.19 T2 Nordhaus 1.25 0.71 T Hope 1.34 0.02 exp(T) – 0.02 Karp; Van der Ploeg 1.71 4.210-175 exp(exp(T)) – 1.110-174 Golosov 2.10 1.610-4 T2-0.36 T2 Weitzman 2.69 2.610-5 T2-0.35 T2 Weitzman 2.73 Table 1. Damage Function Based on Meta-analysis by Richard Tol (Tol, 2018) Although the social welfare discount can be defined as an economic concept, many argue that the choice of discount is also an ethical primitive. Stern recommended a value of 0.1% (Stern, 2006). Nordhaus valued it in the Ramsey equation, resulting in an estimate of 1.5% (Nordhaus, 2017). The IWG provided evaluations using discounts of 2.5%, 3% and 5%. Thus far, however, the social welfare discount concept has not converged to a single value in the literature. The SCC is highly sensitive to the discount rate(Heal and Millner, 2014). To better illustrate the uncertainty caused by socioeconomic assumptions and damage functions, we chose the 1.5% economic discount rate for discussion. The alternative discounts, ranging from 0% to 5% SI, are discussed in Section 3.3.3, 3.3.3 Alternative Social Welfare Discounts. 3.1 Evaluating the SSP outcomes in DICE As shown in Figure 1, the socioeconomic assumption is accompanied by a particular emission trajectory. The emission patterns differentiate under each SSP, leading to increases in atmospheric concentrations, which indicate the long-term temperature trends. Temperature is the direct indicator of climate change and produces different degrees of climate damage, which further determine the SCC. Therefore, we chose emission, concentration and temperature to illustrate the major outcome of SSP in the DICE model (Figure 2). Compared with the five SSP marker scenarios(Calvin et al., 2017; Fricko et al., 2017; Fujimori et al., 2017; Kriegler et al., 2017; van Vuuren et al., 2017), the DICE model considers the optimal emission reduction strategy under a cost- benefit analysis. The results are slightly lower than the baseline scenarios, but the relationships accord with the general narratives. Under SSP5, industrial emissions are markedly higher than other scenarios because of industry’s reliance on fossil fuels. Thus, SSP5 results in 130 GtCO2 emissions in 2100. The emission trajectories of SSP1 to SSP4 diverge after 2050. SSP2 and SSP3 exhibit an increasing emissions trend until the end of this century; in 2100, their values are 65 GtCO2 and 70 GtCO2, respectively. With low challenges to mitigation, SSP1 and SSP4 both reach their emissions peaks in the middle of the century. Under SSP1, emissions reach 47 GtCO2 in 2050 but decrease to 42 GtCO2 by 2100. Under SSP4, the peak emission is higher at 48 GtCO2 in 2055 and decreases more slowly to 45 GtCO2 at the end of this century. Under SSP5, higher emissions will magnify the uncertainty of climate damage, resulting in a wider range of emissions trajectories. Under some damage functions, the optimal emission control rate results in an emissions decrease in SSP5, but under most scenarios, the emissions generally increase. Figure 2. Industrial emissions and their influences on atmosphere concentration and temperature under the five SSPs: (a) industrial emissions. (b) atmospheric concentration. (c) atmospheric temperature. The gray lines indicate the results from nine different damage functions; the colored lines show the smoothed conditional means of the results under the five SSPs. The difference in emissions is directly reflected by the atmospheric concentration trend, which determines the long-term growth of temperature. Under SSP 1 and SSP4, the concentration nearly stabilizes by the end of this century, reaching average levels of approximately 720 ppm. 3.1 Evaluating the SSP outcomes in DICE The concentrations in SSP2 and SSP3 continue increasing to 760 ppm and 799 ppm, respectively, by the end of this century. With high dependence on fossil fuels, in SSP5, the concentration rises to 1019 ppm by 2100. According to the IPCC Fifth Assessment Report (Pachauri et al., 2014), the concentration is likely (>66%) to cause temperature increases of up to 4°C by 2100 under SSP1–SSP4. At concentrations above 1000ppm, the SSP5 temperture is unlikely (<33%) to remain at 4°Cin 2100 and will continue to rise according to the concentration trend. The result from the DICE model agrees with the IPCC result. Because the climate cycle is a long-term process, the temperature increases are quite similar among the scenarios. SSP5 has the largest temperature increase—4.6°C compared to the preindustral level. In SSP1 to SSP4, temperatures increase by approximately 4°C, to 3.8°C, 3.9°C, 4.0°C and 3.9°C, respectively. 3.2.1 Impact of Socioeconomic Assumptions on Social Carbon Cost Socioeconomic assumptions greatly affect the levels and trends of the SCC. Under the five SSP narratives, the SCC calculated under nine damage functions result in different uncertainty extents as shown in Figure 3. Figure 3. Social carbon cost ($/tCO2) under five SSPs. The gray lines show the SCC calculated under nine damage functions; the colored lines show the smoothed conditional means of the SCC. Figure 3. Social carbon cost ($/tCO2) under five SSPs. The gray lines show the SCC calculated under nine damage functions; the colored lines show the smoothed conditional means of the SCC. Figure 3. Social carbon cost ($/tCO2) under five SSPs. The gray lines show the SCC calculated under nine damage functions; the colored lines show the smoothed conditional means of the SCC. In 2020, the average SCC estimations under SSP1, SSP2, SSP4 and SSP5 are 10$/tCO2, 19$/tCO2, 18$/tCO2 and 12$/tCO2, respectively. The SSP3, which represents high mitigation and adaptation challenges, has the highest SCC early in this century, reaching 45$/tCO2 in 2020 and increasing to 108$/tCO2 by 2050. This level is remarkably high compared with other scenarios, and it suggests that if the world socioeconomic conditions increasingly develop into regional rivalries, the SCC will undergo a significant increase. Under the benefit-cost framework, the SCC equals the carbon price under a tax or trade instrument (Nordhaus, 2013). In 2017, 19 carbon trading markets were in place with an average price of 9.1$/tCO2, and 22 nations/sectors have implemented a carbon tax, which averages 29.9$/tCO2 (World Bank, 2017). Under all the scenarios, the SCC is higher than the quota price but quite similar to the carbon tax. This result indicates that, thus far, the carbon trading system is not efficiently reflecting the social cost of emissions. However, as carbon taxes are implemented by governments, they can be targeted to maximize public welfare. Therefore, a carbon tax can better reflect the social costs of additional emissions. Different socioeconomic assumptions can also alter the SCC trend, especially after 2050. SSP3 features a slow growth of SCC at high levels; its annual growth rate is 3% from 2015 to 2050 but the level is the highest among all scenarios. Social costs undergo rapid growth in SSP5, with an annual growth rate of 5% from 2015 to 2050 and continued increases thereafter at 4% annually until the end of this century. 3.2.1 Impact of Socioeconomic Assumptions on Social Carbon Cost SSP1, SSP2 and SSP4 are characterized by medium growth throughout the century, with annual growth rates of 4% from 2015 to 2050. The SCC trend can be an important indicator of policy section among price and quantity instruments(Weitzman, 1974). Therefore, different socioeconomic developments may affect the choice of policy instruments. Varying levels of uncertainty can be witnessed within the socioeconomic scenarios due to the impact of damage functions. Higher emissions magnify the uncertainty from climate damage, thereby resulting in a wider SCC range in SSP2, SSP3 and SSP5. 3.2.2 Impacts of Damage Functions on Social Carbon Costs The SCC values under the nine damage functions can be divided into two groups, indicating different expectations for climate change. However, the choice of damage function will not reverse the SCC relationships under the five SSPs. As shown in Figure 4, the SCC under the nine damage functions can be classified into ‘moderate’ estimation and ‘sharp change’ estimation. Moderate estimations include the five functions proposed by Hope, Tol and Nordhaus. Using these, the estimated SCC never exceeds 300$/tCO2 in this century, and it reaches an average level of 157$/tCO2 by 2100 under the five SSPs. In contrast, applying the damage functions provided by Weitzman, Karp and Golosov results in a sharp increase of SCC by the end of this century, reaching an extremely high average level of 864$/tCO2 by 2100. The moderate estimation is mainly extrapolated from observation, while the sharp change group suggests that several of the climate system elements could be tipped into a different state by the temperature increase. According to the two damage functions from Weitzman, a modest increase of temperature will initially benefit the economics but then abruptly decrease after the tipping point. These functions result in an initially negative SCC, which indicates that the additional emissions will provide positive effects and a social welfare gain under a moderate temperature increase. Under SSP1 and SSP4, with low mitigation challenges, the negative SCC will continue until 2060 to 2075, while under SSP3, with high mitigation challenges, the SCC becomes positive between 2015 and 2020. Under the damage function from Golosov, the SCC increases from 13$/tCO2 in 2015 to 1192$/tCO2 in 2100, leading to enormous mitigation and adaptation pressures by the end of this century. Figure 4. Social carbon cost ($/tCO2) under nine damage functions. Figure 4. Social carbon cost ($/tCO2) under nine damage functions. Regardless of the damage function used, the SCC relationships under the five SSPs are not reversed. The sustainable development scenario (SSP1) always ranks the lowest under all damage functions. SSP5 is characterized as a rapid increase of SCC by the end of this century. SCC under SSP3 is initially high but has a low growth rate over the century. Moderate growth is also observed in SSP2, but the initial level is lower than in SSP3. 3.3.3 Alternative Social Welfare Discount The social cost of carbon is also highly sensitive to the social welfare discount rate. Because climate damage mainly accrues over the long term, the discount rate affects how the prospect of future damage should be addressed today. A high discount rate will significantly reduce the present perception of future climate damage, which results in a low SCC. In contrast, when the climate damage has no future discount, so that people today are as concerned with their descendants’ welfare as with their own well-being, ambitious climate actions should be taken immediately under high SCC. A near-zero discount rate highlights the ethical issues of climate policy, while the Ramsey discount emphasizes the economic benefits of adaptation. This section provides alternative SCC estimations under discounts of 0% to 5% and compares them with the carbon tax and quota price in 2017 to reflect the present policy intensity. Figure 5. The social cost of carbon in 2020 under a 0%–5% social welfare discount compared with the carbon price in 2017. Average SCC decrease exponentially from 0% to 5% (as shown in Figure 5). In 2020, the average SCC values under discounts from 0% to 5% SI are 146$/tCO2, 37$/tCO2, 12$/tCO2, 5$/tCO2, 3$/tCO2 and 2$/tCO2, respectively. A low discount rate substantially magnifies the uncertainty of impacts from emissions, temperature and climate, resulting in a wide range of SCC estimations—from -25$/tCO2 to 501$/tCO2 in 2020 under a 0% discount. When the discount rate exceeds 3%, the wellbeing of future generations has less influence on today’s policy decisions, resulting in SCC l b l 5$/tCO Figure 5. The social cost of carbon in 2020 under a 0%–5% social welfare discount compared with the carbon price in 2017. Figure 5. The social cost of carbon in 2020 under a 0%–5% social welfare discount compared with the carbon price in 2017. Average SCC decrease exponentially from 0% to 5% (as shown in Figure 5). In 2020, the average SCC values under discounts from 0% to 5% SI are 146$/tCO2, 37$/tCO2, 12$/tCO2, 5$/tCO2, 3$/tCO2 and 2$/tCO2, respectively. A low discount rate substantially magnifies the uncertainty of impacts from emissions, temperature and climate, resulting in a wide range of SCC estimations—from -25$/tCO2 to 501$/tCO2 in 2020 under a 0% discount. When the discount rate exceeds 3%, the wellbeing of future generations has less influence on today’s policy decisions, resulting in SCC values below 5$/tCO2. 3.3.3 Alternative Social Welfare Discount The SCC provides a basis for pricing carbon; in 2017, the average quota price was 9.1$/tCO2, while the carbon tax averaged 29.9$/tCO2. Comparing these figures with our SCC estimates, the carbon tax indicates a discount rate preference of 1%, and the carbon quota price indicates a discount rate of 2% to 3%. However, although the average carbon tax can be as high as 139.58$/tCO2 in Sweden, the mean distribution of the tax is still quite low. 4. Conclusions As more countries begin to implement climate policy, estimating the SCC under a cost-benefit analysis is necessary to provide a pricing benchmark. The term has been used for carbon tax, tradable obligations or renewable portfolio standards(Burke, 2016). However, SCC estimation relies heavily on future assumptions (e.g., mitigation and adaptation challenges, population growth and economic development), a reliance that has not previously been recognized. Previous studies have discussed the population and economic impacts separately; however, these factors have a synergistic effect on all aspects. This paper is based on the five plausible future descriptions established by the climate change research community, and it provides the future social costs of emissions under different development pathways. We found that the scenario representing extreme regional rivalry (SSP3) will cause substantial increases in the SCC in the near term, indicating that if more trade tariffs are implemented due to increasing regional conflicts, the social carbon cost today will be underestimated. Under SSP5, where developing countries emerge by exploiting abundant fossil fuels, the pressures for mitigation will become unbearable by the end of this century. The SCC is initially at a relatively low level. Then, it undergoes a rapid increase after 2060 and reaches an average level of 471$/tCO2 by 2100. The damage this growing trend will cause is unstoppable, according to the atmospheric concentration; thus, it demands an increase in attention to the clean development of emerging economies. SCC under increasing inequality (SSP4) is similar to the sustainable development pathway (SSP1) and maintains a low growth rate at a moderate level. The scenario that follows the historical development patterns (SSP2) experiences the same annual growth rate as SSP1 but at a higher social cost and with more uncertainty. The results of this study highlight the importance of avoiding regional rivalries and expending efforts to ensure the green development of emerging economies. Our results also provide a breakeven carbon price for policy makers who hold different attitudes concerning the future and they facilitate mitigation and adaptation analysis under the SSPs. Because the SCC is still under discussion and difficult to explain even at the domestic level (Fraas et al., 2016; Guivarch et al., 2016), we defer a discussion of regional SCC for the future. Acknowledgement The authors gratefully acknowledge the support from the National Key R & D Program (Grant No. 2016YFA0602603), the National Natural Science Foundation of China (Grant Nos. 71521002, 71642004, 71673026). The paper also benefitted from the participants at a seminar at Beijing Institute of Technology. Guivarch, C., Mejean, A., Pottier, A., Fleurbaey, M. (2016) Social cost of carbon: Global duty. Guivarch, C., Mejean, A., Pottier, A., Fleurbaey, M. (2016) Social cost of carbon: Global duty. plants/. Leimbach, M., Kriegler, E., Roming, N., Schwanitz, J. (2017) Future growth patterns of world regions – A GDP scenario approach. Global Environmental Change 42, 215-225. Leimbach, M., Kriegler, E., Roming, N., Schwanitz, J. (2017) Future growth patterns of world regions – A GDP scenario approach. Global Environmental Change 42, 215-225. Mi, Z., Meng, J., Guan, D., Shan, Y., Song, M., Wei, Y.-M., Liu, Z., Hubacek, K. (2017) Chinese Leimbach, M., Kriegler, E., Roming, N., Schwanitz, J. (2017) Future growth patterns of world regions – A GDP scenario approach. Global Environmental Change 42, 215-225. Mi, Z., Meng, J., Guan, D., Shan, Y., Song, M., Wei, Y.-M., Liu, Z., Hubacek, K. (2017) Chinese CO(2) emission flows have reversed since the global financial crisis. Nature Communications 8, 1712. Mi, Z., Meng, J., Guan, D., Shan, Y., Song, M., Wei, Y.-M., Liu, Z., Hubacek, K. (2017) Chinese CO(2) emission flows have reversed since the global financial crisis. Nature Communications 8, 1712. Minnesota, S.o., (2016) Findings of Fact, Conclusions, and Recommendations: Carbon Dioxide Values., in: Office of Administrative HearingsSt Paul, M. (Ed.). State of Minnesota, Office of Administrative HearingsSt Paul, MN, State of Minnesota. Minnesota, S.o., (2016) Findings of Fact, Conclusions, and Recommendations: Carbon Dioxide Values., in: Office of Administrative HearingsSt Paul, M. (Ed.). State of Minnesota, Office of Administrative HearingsSt Paul, MN, State of Minnesota. Moore, F.C., Diaz, D.B. (2015) Temperature impacts on economic growth warrant stringent mitigation policy. Nature Climate Change 5, 127-131. Moss, R.H., Edmonds, J.A., Hibbard, K.A., Manning, M.R., Rose, S.K., Van Vuuren, D.P., Carter, T.R., Emori, S., Kainuma, M., Kram, T. (2010) The next generation of scenarios for climate change research and assessment. Nature 463, 747. Nordhaus, W.D. (2013) The climate casino: Risk, uncertainty, and economics for a warming world. Yale University Press. Nordhaus, W.D. (2017) Revisiting the social cost of carbon. Proceedings of the National Academy of Sciences of the United States of America 114, 1518-1523. O’Neill, B.C., Kriegler, E., Ebi, K.L., Kemp-Benedict, E., Riahi, K., Rothman, D.S., van Ruijven, B.J., van Vuuren, D.P., Birkmann, J., Kok, K. (2017) The roads ahead: narratives for shared socioeconomic pathways describing world futures in the 21st century. Global Environmental Change 42, 169-180. Pachauri, R.K., Allen, M.R., Barros, V.R., Broome, J., Cramer, W., Christ, R., Church, J.A., Clarke, L., Dahe, Q., Dasgupta, P. (2014) Climate change 2014: synthesis report. References Burke, M. (2016) Opportunities for advances in climate change economics. Science 352. Burke, M. (2016) Opportunities for advances in climate change economics. Science 352. Cai, Y.Y., Lenton, T.M., Lontzek, T.S. (2016) Risk of multiple interacting tipping points should encourage rapid CO2 emission reduction. Nature Climate Change 6, 520-+. California, S.o., (2016) in: California, S.o. (Ed.), Assembly Bill 197 California. California, S.o., (2016) in: California, S.o. (Ed.), Assembly Bill 197 California. Calvin, K., Bond-Lamberty, B., Clarke, L., Edmonds, J., Eom, J., Hartin, C., Kim, S., Kyle, P., Link, R., Moss, R. (2017) The SSP4: A world of deepening inequality. Global Environmental Change 42, 284-296. Calvin, K., Bond-Lamberty, B., Clarke, L., Edmonds, J., Eom, J., Hartin, C., Kim, S., Kyle, P., Link, R., Moss, R. (2017) The SSP4: A world of deepening inequality. Global Environmental Change 42, 284-296. Calvin, K., Bond-Lamberty, B., Clarke, L., Edmonds, J., Eom, J., Hartin, C., Kim, S., Kyle, P., Link, R., Moss, R. (2017) The SSP4: A world of deepening inequality. Global Environmental Change 42, 284-296. Crost, B., Traeger, C.P. (2014) Optimal CO2 mitigation under damage risk valuation. Nature Climate Change 4, 631-636. Crost, B., Traeger, C.P. (2014) Optimal CO2 mitigation under damage risk valuation. Nature Climate Change 4, 631-636. Diaz, D., Moore, F. (2017) Quantifying the economic risks of climate change. Nature Climate Change 7, 774-782. Diaz, D., Moore, F. (2017) Quantifying the economic risks of climate change. Nature Climate Change 7, 774-782. Dietz, S., Stern, N. (2015) Endogenous Growth, Convexity of Damage and Climate Risk: How Nordhaus' Framework Supports Deep Cuts in Carbon Emissions. The Economic Journal 125, 574-620. Fraas, A., Lutter, R., Dudley, S., Gayer, T., Graham, J., Shogren, J.F., Viscusi, W.K. (2016) Social cost of carbon: Domestic duty. Science 351, 569-569. Fricko, O., Havlik, P., Rogelj, J., Klimont, Z., Gusti, M., Johnson, N., Kolp, P., Strubegger, M., Valin, H., Amann, M. (2017) The marker quantification of the Shared Socioeconomic Pathway 2: a middle-of-the-road scenario for the 21st century. Global Environmental Change 42, 251-267. Fujimori, S., Hasegawa, T., Masui, T., Takahashi, K., Herran, D.S., Dai, H., Hijioka, Y., Kainuma, M. (2017) SSP3: AIM implementation of shared socioeconomic pathways. Global Environmental Change 42, 268-283. Science 351, 1160-1161. Heal, G.M., Millner, A. (2014) Agreeing to disagree on climate policy. Proceedings of the National Academy of Sciences of the United States of America 111, 3695-3698. Howarth, R.B., Gerst, M.D., Borsuk, M.E. References (2014) Risk mitigation and the social cost of carbon. Global Environmental Change-Human and Policy Dimensions 24, 123-131. Kriegler, E., Bauer, N., Popp, A., Humpenöder, F., Leimbach, M., Strefler, J., Baumstark, L., Bodirsky, B.L., Hilaire, J., Klein, D. (2017) Fossil-fueled development (SSP5): an energy and resource intensive scenario for the 21st century. Global Environmental Change 42, 297-315. Larson, A., (2016) Subsidies Proposed for New York’s Upstate Power Plants, Power, Available Kriegler, E., Bauer, N., Popp, A., Humpenöder, F., Leimbach, M., Strefler, J., Baumstark, L., Bodirsky, B.L., Hilaire, J., Klein, D. (2017) Fossil-fueled development (SSP5): an energy and resource intensive scenario for the 21st century. Global Environmental Change 42, 297-315. Larson, A., (2016) Subsidies Proposed for New York’s Upstate Power Plants, Power, Available at: http://www.powermag.com/subsidies-proposed-for-new-yorks-upstate-nuclear-power- plants/. World Bank, Ecofys, Vivid Economics (2017) State and Trends of Carbon Pricing 2017. World Bank, Washington, DC. plants/. Contribution of Working Groups I, II and III to the fifth assessment report of the Intergovernmental Panel on Climate Change. IPCC. Pindyck, R.S. (2013) Climate Change Policy: What Do the Models Tell Us? Journal of Economic Literature 51, 860-872. Pizer, W., Adler, M., Aldy, J., Anthoff, D., Cropper, M., Gillingham, K., Greenstone, M., Murray, B., Newell, R., Richels, R., Rowell, A., Waldhoff, S., Wiener, J. (2014) Using and improving the social cost of carbon. Science 346, 1189-1190. Pycroft, J., Vergano, L., Hope, C. (2014) The economic impact of extreme sea-level rise: Ice sheet vulnerability and the social cost of carbon dioxide. Global Environmental Change-Human and Policy Dimensions 24, 99-107. sheet vulnerability and the social cost of carbon dioxide. Global Environmental Change-Human and Policy Dimensions 24, 99-107. Revesz, R., Greenstone, M., Hanemann, M., Livermore, M., Sterner, T., Grab, D., Howard, P., Revesz, R., Greenstone, M., Hanemann, M., Livermore, M., Sterner, T., Grab, D., Howard, P., Schwartz, J. (2017) Best cost estimate of greenhouse gases. Science 357, 655-655. Schwartz, J. (2017) Best cost estimate of greenhouse gases. Science 357, 655-655. Riahi, K., Van Vuuren, D.P., Kriegler, E., Edmonds, J., O’neill, B.C., Fujimori, S., Bauer, N., Calvin, K., Dellink, R., Fricko, O. (2017) The shared socioeconomic pathways and their energy, land use, and greenhouse gas emissions implications: an overview. Global Environmental Change 42, 153-168. Samir, K., Lutz, W. (2017) The human core of the shared socioeconomic pathways: Population scenarios by age, sex and level of education for all countries to 2100. Global Environmental Change 42, 181-192. Scovronick, N., Budolfson, M.B., Dennig, F., Fleurbaey, M., Siebert, A., Socolow, R.H., Spears, D., Wagner, F. (2017) Impact of population growth and population ethics on climate change mitigation policy. Proceedings of the National Academy of Sciences of the United States of America 114, 12338-12343. Stern, N. (2006) Stern review report on the economics of climate change. Tol, R.S.J. (2018) The Economic Impacts of Climate Change. Review of Environmental Economics and Policy, rex027-rex027. Van Vuuren, D.P., Riahi, K., Moss, R., Edmonds, J., Thomson, A., Nakicenovic, N., Kram, T., Berkhout, F., Swart, R., Janetos, A. (2012) A proposal for a new scenario framework to support research and assessment in different climate research communities. Global Environmental Change 22, 21-35. Change 22, 21 35. plants/. van Vuuren, D.P., Stehfest, E., Gernaat, D.E.H.J., Doelman, J.C., van den Berg, M., Harmsen, M., de Boer, H.S., Bouwman, L.F., Daioglou, V., Edelenbosch, O.Y., Girod, B., Kram, T., Lassaletta, L., Lucas, P.L., van Meijl, H., Müller, C., van Ruijven, B.J., van der Sluis, S., Tabeau, A. (2017) Energy, land-use and greenhouse gas emissions trajectories under a green growth paradigm. Global Environmental Change 42, 237-250. Wei, Y.-M., Han, R., Liang, Q.-M., Yu, B.-Y., Yao, Y.-F., Xue, M.-M., Zhang, K., Liu, L.-J., Peng, J., Yang, P., Mi, Z.-F., Du, Y.-F., Wang, C., Chang, J.-J., Yang, Q.-R., Yang, Z., Shi, X., Xie, W., Liu, C., Ma, Z., Tan, J., Wang, W., Tang, B.-J., Cao, Y.-F., Wang, M., Wang, J.-W., Kang, J.-N., Wang, K., Liao, H. (2018) An integrated assessment of INDCs under SSPs: An implementation of China's Climate Change Integrated Assessment Model. Natural Hazards (in press). Weitzman, M.L. (1974) Prices vs. quantities. The review of economic studies 41, 477-491. World Bank, Ecofys, Vivid Economics (2017) State and Trends of Carbon Pricing 2017. World Bank, Washington, DC. van Vuuren, D.P., Stehfest, E., Gernaat, D.E.H.J., Doelman, J.C., van den Berg, M., Harmsen, M., de Boer, H.S., Bouwman, L.F., Daioglou, V., Edelenbosch, O.Y., Girod, B., Kram, T., Lassaletta, L., Lucas, P.L., van Meijl, H., Müller, C., van Ruijven, B.J., van der Sluis, S., Tabeau, A. (2017) Energy, land-use and greenhouse gas emissions trajectories under a green growth paradigm. Global Environmental Change 42, 237-250. Wei, Y.-M., Han, R., Liang, Q.-M., Yu, B.-Y., Yao, Y.-F., Xue, M.-M., Zhang, K., Liu, L.-J., Peng, J., Yang, P., Mi, Z.-F., Du, Y.-F., Wang, C., Chang, J.-J., Yang, Q.-R., Yang, Z., Shi, X., Xie, W., Liu, C., Ma, Z., Tan, J., Wang, W., Tang, B.-J., Cao, Y.-F., Wang, M., Wang, J.-W., Kang, J.-N., Wang, K., Liao, H. (2018) An integrated assessment of INDCs under SSPs: An implementation of China's Climate Change Integrated Assessment Model. Natural Hazards (in press). Weitzman, M.L. (1974) Prices vs. quantities. The review of economic studies 41, 477-491. Weitzman, M.L. (1974) Prices vs. quantities. The review of economic studies 41, 477-491. World Bank, Ecofys, Vivid Economics (2017) State and Trends of Carbon Pricing 2017. World Bank, Washington, DC.
https://openalex.org/W2246847192	http://pdf.blucher.com.br/chemicalengineeringproceedings/cobeq2014/0805-23684-140090.pdf	Portuguese	null	ESTUDO CINÉTICO DO CRESCIMENTO DO CLOSTRIDIUM ACETOBUTYLICUM ATCC 4259 UTILIZANDO GLICEROL P.A. COMO SUBSTRATO	null	2,015	cc-by	3,007	ESTUDO CINÉTICO DO CRESCIMENTO DO CLOSTRIDIUM ACETOBUTYLICUM ATCC 4259 UTILIZANDO GLICEROL P.A. COMO SUBSTRATO C. E. de O. LACERDA1, G. M. VINHAS1, Y. M. B. de ALMEIDA1 1 Universidade Federal de Pernambuco, Departamento de Engenharia Química E-mail para contato: carlos.olacerda@hotmail.com RESUMO – A glicerina, subproduto da produção de biodiesel, vem sendo investigada como fonte de carbono em processos microbianos para a obtenção de bioprodutos com alto valor agregado. A avaliação do crescimento do micro- organismo responsável por essa bioconversão é uma importante etapa desse processo. O presente trabalho consistiu em avaliar a cinética de crescimento do Clostridium acetobutylicum ATCC 4259 através de uma fermentação em batelada utilizando glicerol P.A. como fonte de carbono. A técnica de peso seco foi utilizada para o calculo da concentração do micro-organismo presente no meio de cultura após a fermentação. A cinética microbiana mostrou que, para as condições experimentais da pesquisa, a fase exponencial de crescimento ocorreu até o tempo de 14 horas, e uma produtividade celular (P) de 0,0405 g/L.h, uma velocidade máxima de crescimento (µmáx) de 0,0403 h-1 e um tempo de geração (G) de 15,7 h foram alcançados. 1. INTRODUÇÃO O biodiesel é um combustível renovável, biodegradável e não tóxico. Pode ser definido como sendo um mono-alquil éster de ácidos graxos derivados de fontes renováveis (óleos vegetais e gordura animal) com álcool na presença de um catalisador, obtido através de um processo de transesterificação, no qual ocorre a transformação de triglicerídeos em moléculas menores de ésteres de ácidos graxos e tendo como subproduto principal, a glicerina (glicerol mais impurezas) (Fukuda et al., 2009; Leoneti et al., 2012). Nos primeiros anos de produção do biodiesel, a alta percentagem de glicerina obtida era considerada como um aspecto positivo, e com sua venda, o aumento da competitividade econômica global da fabricação de biodiesel parecia seguro, uma vez que a glicerina gerada na produção corresponde a aproximadamente 10 % do volume total de biodiesel produzido (Yazdani & Gonzalez, 2007). Em 2010, a produção de biodiesel no Brasil foi de aproximadamente 2,4 bilhões de litros, o que gerou aproximadamente 240 milhões de litros de glicerina (Leoneti et al., 2012). O acúmulo da oferta ocasionado pelo crescimento das indústrias de biocombustíveis e a quantidade de impurezas encontradas na glicerina bruta, fizeram com que seu preço diminuísse com o passar dos anos (Yazdani & Gonzalez, 2007; Asad-Ur-Rehman et al., 2008). Desta forma, essa glicerina oriunda da produção de biodiesel passou de um produto apreciado a um problema de eliminação de resíduo, e existe um grande interesse em sua 1 Área temática: Processos Biotecnológicos purificação ou no seu reaproveitamento direto sem tratamento, proporcionando ao processo de produção de biodiesel maior competitividade e valorização crescente no mercado de biocombustíveis (Kaur et al., 2012; Clomburg et al., 2013). purificação ou no seu reaproveitamento direto sem tratamento, proporcionando ao processo de produção de biodiesel maior competitividade e valorização crescente no mercado de biocombustíveis (Kaur et al., 2012; Clomburg et al., 2013). Sendo assim, a glicerina bruta não purificada ou com purificação parcial vem sendo investigada como fonte de carbono em processos microbianos para obtenção de bioprodutos de alto valor agregado como 1,3-propanodiol, etanol, butanol, acetona, ácido acético, ácido butírico, ácido cítrico, ácido fórmico, ácido lático, ácido succínico, CO2 e H2. (Anand et al., 2011; Clomburg et al., 2013). O estudo cinético do crescimento do micro-organismo responsável por essa bioconversão é uma importante etapa para compreender todo o mecanismo e otimizar o processo de fermentação. (Sun et al., 2008). 1. INTRODUÇÃO A formação de biomassa no processo fermentativo depende de duas moléculas transportadoras de energia, o NAD+ (dinucleotídeo de adenina nicotinamida) e o ATP (adenosina trifosfato) (Kaur et al., 2012). O crescimento de um micro-organismo está dividido em quatro fases: a fase lag, conhecida também como fase de adaptação, em que nenhum crescimento aparente ocorre, uma vez que as células estão se adaptando ao meio de cultivo; tem-se a fase exponencial, também conhecida como fase logarítmica ou fase log, em que ocorre um aumento exponencial do número de células, atingindo a sua velocidade máxima; a fase estacionária, onde a população atinge seu máximo e constante crescimento; e a fase da morte, onde, devido ao acúmulo de metabólitos e limitação de nutrientes, eventualmente poderá ocorrer um declínio do número de células, característico dessa fase (Bastos, 2010). O crescimento de um micro-organismo está dividido em quatro fases: a fase lag, conhecida também como fase de adaptação, em que nenhum crescimento aparente ocorre, uma vez que as células estão se adaptando ao meio de cultivo; tem-se a fase exponencial, também conhecida como fase logarítmica ou fase log, em que ocorre um aumento exponencial do número de células, atingindo a sua velocidade máxima; a fase estacionária, onde a população atinge seu máximo e constante crescimento; e a fase da morte, onde, devido ao acúmulo de metabólitos e limitação de nutrientes, eventualmente poderá ocorrer um declínio do número de células, característico dessa fase (Bastos, 2010). O objetivo do presente trabalho foi avaliar a cinética de crescimento do Clostridium acetobutylicum ATCC 4259 através de uma fermentação em batelada utilizando glicerol P.A. como fonte de carbono. 2.1. Materiais Micro-organismo: O micro-organismo utilizado no trabalho foi o Clostridium acetobutylicum ATCC 4259. Essa linhagem foi cedida pela coleção de micro-organismos UFPEDA do Departamento de Antibióticos da UFPE. Foi adquirida em tubos contendo meio RCM (Caldo Clostridial Reforçado) semissólido, e foi mantida em estufa a 37 °C até ser inoculada e/ou repicada. Soluções Utilizadas: Meio de tioglicolato fluido (marca DIFCO), utilizado para manutenção e crescimento do micro-organismo e solução salina para diluição (Tabela 1). 2 2 Área temática: Processos Biotecnológicos Tabela 1 – Composição da solução salina para diluição Solução A KH2PO4 34,0 g Água destilada 1000,0 mL Solução B MgSO4.7H2O 50,0 g Água destilada 1000,0 mL Solução Salina Solução A 1,25 mL Solução B 5,0 mL Água destilada 1000,0 mL A pré-cultura utilizada para adaptar o micro-organismo aos componentes do meio de cultura para fermentação tem a seguinte composição (Günzel et al., 1991): K2HPO4 (3,4 g/L), KH2PO4 (1,3 g/L), (NH4)2SO4 (2,0 g/L), MgSO4.7H2O (0,2 g/L), CaCl2.2H2O (0,02 g/L), CaCO3 (2,0 g/L), extrato de levedura (1,0 g/L), glicerol P.A. (20 g/L), solução de elementos traços (1,0 mL/L) (Tabela 2) e solução de ferro (2,0 mL/L) (Tabela 3). Tabela 1 – Composição da solução salina para diluição Solução A KH2PO4 34,0 g Água destilada 1000,0 mL Solução B MgSO4.7H2O 50,0 g Água destilada 1000,0 mL Solução Salina Solução A 1,25 mL Solução B 5,0 mL Água destilada 1000,0 mL Tabela 1 – Composição da solução salina para diluição Tabela 1 – Composição da solução salina para diluição Solução A KH2PO4 34,0 g Água destilada 1000,0 mL Solução B MgSO4.7H2O 50,0 g Água destilada 1000,0 mL Solução Salina Solução A 1,25 mL Solução B 5,0 mL Água destilada 1000,0 mL A pré-cultura utilizada para adaptar o micro-organismo aos componentes do meio de cultura para fermentação tem a seguinte composição (Günzel et al., 1991): K2HPO4 (3,4 g/L), KH2PO4 (1,3 g/L), (NH4)2SO4 (2,0 g/L), MgSO4.7H2O (0,2 g/L), CaCl2.2H2O (0,02 g/L), CaCO3 (2,0 g/L), extrato de levedura (1,0 g/L), glicerol P.A. (20 g/L), solução de elementos traços (1,0 mL/L) (Tabela 2) e solução de ferro (2,0 mL/L) (Tabela 3). 2.2. Metodologia Experimental Preparação dos meios de cultura: Os meios de cultura foram preparados de acordo com as seguintes etapas: os nutrientes de cada meio foram pesados e diluídos em água destilada; após diluição, foram insuflados com gás nitrogênio e aquecidos a T = 55 ºC durante 20 minutos; os meios de cultura então foram transferidos para tubos de penicilina e para o frasco biorreator, selados e lacrados; seguiram então para a esterilização em autoclave, a 121 °C por 15 minutos. Preparo da cinética: Utilizando uma seringa (c/ agulha) estéril, o C. acetobutylicum ATCC 4259 foi inoculado em um tubo de penicilina de 100 mL contendo 90 mL de meio de tioglicolato fluido na proporção de 10 % v/v e incubado a 35 °C. Após 24 horas de crescimento o micro-organismo foi inoculado, também na proporção de 10 % v/v, em outro tubo de penicilina de 100 mL contendo 90 mL de pré-cultura. Após as 24 horas, o micro- organismo adaptado na pré-cultura foi inoculado no meio de fermentação para realização da cinética. A cinética foi conduzida em um frasco biorreator de 500 mL, sem controle de pH, temperatura e agitação, contendo 300 mL de meio de cultura para fermentação que foi inoculado com 34 mL de inóculo vindo da pré-cultura (correspondendo a 10 % v/v). O frasco seguiu para a incubadora a T = 35 ºC durante 20 h, tempo que durou a fermentação. Avaliação da biomassa: Para avaliar o crescimento do micro-organismo foi utilizada a técnica do peso seco. Em intervalos de 1 hora, amostras de 10 mL foram retiradas do frasco biorreator e filtradas com o auxílio de um conjunto para filtração em membranas de 2 µm previamente taradas, utilizando água de diluição (Tabela 1) para lavagem. Antes de cada filtração, o frasco foi levemente agitado para homogeneizar o meio e a retirada das amostras foi realizada de maneira estéril, com o auxílio da chama do bico de Bunsen As membranas permaneceram 2 horas na estufa a T = 80 ºC para secar o filtrado. A diferença entre o peso inicial da membrana e o peso após a secagem determina a biomassa que cresceu durante a fermentação. Cálculo dos parâmetros cinéticos: Para calcular os parâmetros importantes da cinética microbiana, têm-se as Equações 1, 2 e 3. A concentração dos nutrientes do meio de cultura para fermentação é igual à concentração dos nutrientes da pré-cultura, com exceção de: K2HPO4 (1,0 g/L), KH2PO4 (0,5 g/L), (NH4)2SO4 (1,0 g/L) e solução de ferro (1,0 mL/L). 2.1. Materiais Tabela 1 – Composição da solução salina para diluiçã Solução A KH2PO4 34,0 g Água destilada 1000,0 mL Solução B MgSO4.7H2O 50,0 g Água destilada 1000,0 mL Solução Salina Solução A 1,25 mL Solução B 5,0 mL Água destilada 1000,0 mL A pré-cultura utilizada para adaptar o micro-organismo aos componentes do meio de cultura para fermentação tem a seguinte composição (Günzel et al., 1991): K2HPO4 (3,4 g/L), KH2PO4 (1,3 g/L), (NH4)2SO4 (2,0 g/L), MgSO4.7H2O (0,2 g/L), CaCl2.2H2O (0,02 g/L), CaCO3 (2,0 g/L), extrato de levedura (1,0 g/L), glicerol P.A. (20 g/L), solução de elementos traços (1,0 mL/L) (Tabela 2) e solução de ferro (2,0 mL/L) (Tabela 3). Tabela 2 – Composição da solução de elementos traços Componentes ZnCl2 0,070 g/L MnCl2.4H2O 0,100 g/L H3BO3 0,060 g/L CoCl2.6H2O 0,200 g/L CuCl2.2H2O 0,020 g/L NiCl2.6H2O 0,025 g/L Na2MoO4.H2O 0,035 g/L HCl 37% 0,900 mL/L Tabela 3 – Composição da solução de Ferro Componentes FeSO4.7H2O 5,0 g/L HCl 37% 4,0 mL/L Tabela 2 – Composição da solução de elementos traços Componentes ZnCl2 0,070 g/L MnCl2.4H2O 0,100 g/L H3BO3 0,060 g/L CoCl2.6H2O 0,200 g/L CuCl2.2H2O 0,020 g/L NiCl2.6H2O 0,025 g/L Na2MoO4.H2O 0,035 g/L HCl 37% 0,900 mL/L Tabela 3 – Composição da solução de Ferro Componentes FeSO4.7H2O 5,0 g/L HCl 37% 4,0 mL/L Tabela 2 – Composição da solução de elementos traços Tabela 2 – Composição da solução de elementos traços Componentes ZnCl2 0,070 g/L MnCl2.4H2O 0,100 g/L H3BO3 0,060 g/L CoCl2.6H2O 0,200 g/L CuCl2.2H2O 0,020 g/L NiCl2.6H2O 0,025 g/L Na2MoO4.H2O 0,035 g/L HCl 37% 0,900 mL/L Tabela 3 – Composição da solução de Ferro Componentes FeSO4.7H2O 5,0 g/L HCl 37% 4,0 mL/L 3 Área temática: Processos Biotecnológicos 2.2. Metodologia Experimental A Equação 1 determina a produtividade celular (P) que é a relação entre a variação da concentração celular pela variação do tempo de cultivo englobando todas as fases de crescimento microbiano (Borzani et al., 2001). P = ( ) ( ) (1) (1) A Equação 2 mostra a velocidade máxima de crescimento do micro-organismo (µmáx). O intervalo de tempo t-t0 corresponde apenas à fase exponencial de crescimento; X0 é a Área temática: Processos Biotecnológicos 4 concentração celular no início da fase exponencial de crescimento, enquanto X é a concentração celular no final dessa fase (Bastos, 2010). μá = (/) (2) (2) A Equação 3 mostra o tempo de geração (G), ou seja, o tempo que o micro-organismo leva para duplicar a sua biomassa. O tempo de geração também é calculado apenas na fase exponencial de crescimento. Esse parâmetro é importante na interpretação do crescimento celular, uma vez que dá uma ideia mais prática de quão rápido ocorre o crescimento do micro- organismo (Bastos, 2010). G = á (3) G = á (3) 3.1. Resultados da cinética microbiana Os resultados da cinética microbiana encontram-se na Tabela 4. Os resultados da cinética microbiana encontram-se na Tabela 4. Os resultados da cinética microbiana encontram-se na Tabela 4. Tabela 4 – Construção da curva de crescimento por peso seco Tempo (h) Concentração (g/L) Tempo (h) Concentração (g/L) 1 0,92 11 1,37 2 0,88 12 0,20 3 1,18 13 1,22 4 0,16 14 1,60 5 1,25 15 1,49 6 1,37 16 1,45 7 0,14 17 1,43 8 1,31 18 1,61 9 1,51 19 1,66 10 1,18 20 1,69 Tabela 4 – Construção da curva de crescimento por peso seco Com os valores encontrados na Tabela 4, foi calculada a produtividade celular P = 0,0405 g/L.h. Esse valor representa diretamente o crescimento celular em g/L a cada hora de experimento. Na Figura 1 encontra-se a regressão quadrática do crescimento celular em função do tempo. 5 Área temática: Processos Biotecnológicos Figura 1 – Crescimento celular ajustado Figura 1 – Crescimento celular ajustado Figura 1 – Crescimento celular ajustado Incialmente observa-se através da Figura 1 que a fase lag de crescimento não é visualizada, o que mostra que o Clostridium acetobutylicum ATCC 4259 se adaptou bem ao meio de cultura. Pode-se identificar a fase de crescimento exponencial para as nossas condições experimentais. Essa fase exponencial ocorre até o ponto de 14 horas. Além disso, temos a fase estacionária de crescimento a partir de 14 horas. Podemos encontrar a velocidade máxima de crescimento (µmáx) através da Equação 2, calculando o logaritmo neperiano de X/X0 e construindo o gráfico da Figura 2. Na Tabela 5 encontram-se os resultados desses cálculos. Tabela 5 – Valores de ln (X/X0) para cálculo da velocidade máxima de crescimento Tempo (h) X (g/L) ln (X/X0) Tempo (h) X (g/L) ln (X/X0) 1 0,92 0 8 1,31 0,353409 2 0,88 -0,04445 9 1,51 0,495491 3 1,06 0,141651 10 1,47 0,468644 4 1,14 0,21441 11 1,51 0,495491 5 1,25 0,306525 12 1,55 0,521637 6 1,37 0,398192 13 1,57 0,534457 7 1,33 0,368561 14 1,60 0,553385 Tabela 5 – Valores de ln (X/X0) para cálculo da velocidade máxima de crescimento T (h) X ( /L) l (X/X ) T (h) X ( /L) l (X/X ) 6 Área temática: Processos Biotecnológicos Figura 2 – Caracterização do µmáx para o experimento Figura 2 – Caracterização do µmáx para o experimento Figura 2 – Caracterização do µmáx para o experimento O coeficiente angular da equação da reta do gráfico da Figura 2 representa a velocidade máxima de crescimento celular, portanto µmáx = 0,0403 h-1. Os resultados da cinética microbiana encontram-se na Tabela 4. De posse do valor da velocidade máxima, podemos calcular G, o tempo de geração, através da Equação 3. Então, temos que G = 15,7 h. Esse valor representa o tempo necessário para que o micro-organismo duplique a sua biomassa. Então, quanto menor o valor de G, mais rápido é o crescimento. 4. CONCLUSÕES As condições operacionais para manutenção e crescimento do Clostridium acetobutylicum ATCC 4259 foram alcançadas, produzindo inóculos com boa viabilidade e respostas satisfatórias para o crescimento do micro-organismo utilizando glicerol P.A. como principal fonte de carbono em um processo descontínuo. Foi possível observar que o Clostridium acetobutylicum ATCC 4259 alcançou o fim da fase exponencial de crescimento em 14 horas de fermentação. Então, estudando apenas essa fase, dados cinéticos que caracterizam o crescimento do micro-organismo foram coletados. Os dados serviram para compreender o seu comportamento diante do meio de cultura próprio para a produção de 1,3-propanodiol. O primeiro dado importante foi o tempo de geração (G = 15,7 horas). Através desse valor conclui-se que o Clostridium acetobutylicum ATCC 4259 leva aproximadamente toda a fase exponencial de crescimento para duplicar a sua biomassa. Fato esse que pôde ser observado também pela baixa velocidade de crescimento máxima (µmáx = 0,0403 h-1), característica do gênero Clostridium. 7 7 Área temática: Processos Biotecnológicos 5. REFERÊNCIAS ANAND, P.; SAXENA, R. K.; MARWAH, R. G. A novel downstream process for 1,3- propanediol from glycerol-based fermentation. Appl. Microbiol. Biotechnol., v. 90, n. 4, p. 1267-1276, 2011. ASAD-UR-REHMAN; MATSUMURA, M.; NOMURA, N.; SATO, S. Growth and 1,3- propanediol production on pre-treated sunflower oil biodiesel raw glycerol using a strict anaerobe Clostridium butyricum. Curr. Res. Bacteriol., v. 1, n. 1, p. 7-16, 2008. BASTOS, R. G. Tecnologia das fermentações: fundamentos de bioprocessos. São Carlos: EdUFSCar, 2010. BORZANI, W.; SCHMIDELL, W.; LIMA, A. U.; AQUARONE, E. Biotecnologia Industrial. Volume 2, São Paulo: Edgar Bülcher, 2001. CLOMBURG, J. M.; GONZALEZ, R. Anaerobic fermentation of glyrcerol: a plataform for renewable fuels and chemicals. Trends Biotecnol., v. 31, p. 20-28, 2013. FUKUDA, H.; KONDO, A.; TAMALAMPUDI, S. Bioenergy: sustainable fuels from biomass by yeast and fungal whole-cell biocatalysts. Biochem, Eng. J., v. 44, p. 2-12, 2009. GÜNZEL B.; YONSEL, S.; DECKWER, W. D. Fermentative production of 1,3-propanediol from glycerol by Clostridium butyricum up to a scale of 2m3. Appl. Microbiol., v. 36, p. 289-294, 1991. KAUR, G.; SRIVASTAVA, A. K.; CHAND, S. Advances in biotechnological production of 1,3-propanediol. Biochem. Eng. J., v. 64, p.106-118, 2012. LEONETI, A. B.; ARAGÃO-LEONETI, V.; DE OLIVEIRA, S. V. W. B. Glycerol as a by- product of biodiesel production in Brazil: alternatives for the use of unrefined glycerol. Renew. Energ., v. 45, p. 138-145, 2012. SUN, Q. Y.; QI, T. W.; TENG, H.; XIU, L. Z.; ZENG, P. A. Mathematical modeling of glycerol fermentation by Klebsiella pneumonia: concerning enzyme-catalytic reductive pahway and transport of glycerol and 1,3-propanediol across cell membrane. Biochem. Eng. J., v. 38, p. 22-32, 2008. YAZDANI, S. S.; GONZALEZ, R. Anaerobic fermentation of glycerol: a path economic viability for the biofuels industry. Curr. Opin. Biotech., v. 18, p. 213-219, 2007. 8 Área temática: Processos Biotecnológicos
https://openalex.org/W3006116586	https://scindeks-clanci.ceon.rs/data/pdf/0353-9008/2019/0353-90081947287Z.pdf	Serbian	null	Sultan's journey through old Serbia and Macedonia	Baština	2,019	cc-by	9,070	БАШТИНА, Приштина – Лепосавић, св. 47, 2019 БАШТИНА, Приштина – Лепосавић, св. 47, 2019 УДК 94(497.115-89)"1911" 323(497.115-89)"1911" 94(560)"1911" 321.61:929 Мехмед Решад V"1911" doi:10.5937/bastina1947287Z Весна С. ЗАРКОВИЋ* Институт за српску културу – Приштина, Лепосавић Весна С. ЗАРКОВИЋ* Институт за српску културу – Приштина, Лепосавић у р g ** Рад је написан у оквиру пројекта Материјална и духовна култура Косова и Метохије (ев. бр. 178028), који је одобрило и финансира Министарство просвете, науке и тех- нолошког развоја Републике Србије. * научни сарадник, vesna.zarkovic07@gmail.com СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Апстракт: У раду се говори о султановим припремама и посети Старој Ср- бији и Македонији која представља дело младотурског комитета. Његово седамна- естодневно путовање током којег је посетио Скопље, Приштину, Битољ и Солун имало је за циљ да умири Арбанасе и придобије их за себе. На том путу је органи- зована и заједничка молитва одржана приликом посете Муратовог турбета, где је прочитан указ о амнестији свих политичких криваца у земљи и ван ње. Овај и други слични уступци нису у потпуности задовољили арбанашке зликовце који су наста- вили са својим „старим занатом“, чије су последице најбоље осетили Срби. Арба- нашки захтеви су се свакодневно умножавали и водили ка аутономији, што ће, уз залагање Аустроугарске и Италије, резултирати стварањем Албаније 1912. године. Кључне речи: Османско царство, Стара Србија, Македонија, султан, путо- вање, Срби, Арбанаси. Дугогодишња криза која је захватила Османско царство кулминирала је крајем прве деценије XX века. Почетком јула 1908. године дошло је до бројних убистава људи оданих султану, што је проузроковало слање трупа од 18.000 војника из Анадолије у Македонију да се обрачунају са побуњени- цима. Уместо да се боре, они су прешли на супротну страну, a такав неоче- кивани обрт навео је султана да им се приклони и 22. јула за великог везира постави Саид-пашу. Већ наредног дана султан је објавио царску ираду којом је враћен Устав из 1876. године, најавио расписивање избора и сазивање Скупштине која се није одржавала пуних тридесет година (Мантран 2002: 696–697). Априла месеца 1909. године у Цариграду је дошло до неуспелог државног удара. Безуспешни покушај султана Абдул Хамида да свргне мла- дотурке довео је до промене на престолу, а за новог султана је проглашен 288 Весна С. Зарковић његов брат Махмуд V Решад (Батаковић 1989: 274). Кад су младотурци до- шли на власт, први и основни задатак им је био да у пракси спроведу прву реч из свог слогана – „уједињење“ (Мантран 2002: 718). Код младотурских вођа преовладала је паносманска струја која је све поданике прогласила једним недељивим османским народом. Власт, која је била централизована, погађала је на простору Старе Србије и Македо- није и Србе и Арбанасе. Прве зато што је било забрањено деловање српских друштава, одузимана је црквена и манастирска земља и ометан рад школских и верских одбора. Српско становништво је највише погодила мера која се односила на одузимање земље и то на основу Закона о замени тапија и убаш- тињењу имања. 1 Документи о спољној политици Краљевине Србије 1903–1914 (=ДСПКС), књ. IV, св. 3/1, Београд 2009, док. 152, 408. . у ј р 3/1, Београд 2009, док. 152, 408. . СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Долазак младотурака на власт изазвао је узнемирење код Арбанаса, који су се, навикнути на привилегије добијене од ранијег султана Абдул Хамида, противили увођењу редовне војне обавезе и новој пореској политици. Они су на годишњицу револуције, 1909. године, у Дебру одржа- ли конгрес на којем су одбацили захтев за редовно служење војног рока и покренули питање стварања посебне аутономне области која би обухватала све територије на којима живе Арбанаси (Батаковић 1989: 275; Богдано- вић1985: 159–160). Упорна настојања младотурака да своје замисли спроведу у дело изазва- ла су додатну нетрпељивост код Арбанаса, па је зато 1910. и 1911. године на простору Старе Србије и Македоније дошло до побуна и отворених сукоба са турским властима. Резултат тих сукоба представља велики број мртвих и рањених Арбанаса, оштећених кућа, процесуираних побуњеника и одузи- мање оружја (Зарковић–Савић 2018: 291–313; Зарковић 2018: 213–229). Прилике у западном делу Османског царства су биле лоше, а ситуација је претила да угрози опстанак државе. Знајући да Арбанаси представљају битан фактор у очувању Царства, у Цариграду, у султановим окружењем, почели су да размишљају о његовој посети тим крајевима. По њима, султа- нова посета Старој Србији и Македонији је донекле могла да утиче на даљи развој догађаја и допринесе јачању државе и опстанку уопште. Гласине о таквом путовању су почеле да се шире, а неки листови и да их објављују. Тим вестима у почетку није придавана нека значајнија пажња, али од оног момента када су потврђене од стране првог секретара Зија-беја, постале су главна тема не само у дипломатским и војним круговима, него и међу обич- ним становништвом. По речима првог секретара, посета је била планирана за мај 1911. године и предвиђено је било да султан обиђе Солун, Скопље и Приштину. Такође, било је замишљено и полагање темеља за изградњу мед- ресе у близини места погибије султана Мурата I.1 Султаново путовање по Старој Србији и Македонији 289 План је захтевао и бројне припреме, након којих је био направљен конач- ни програм који је предвиђао да султан након тродневног боравка у Солуну посети Скопље, а затим и Приштину, одакле би обишао Муратово турбе. Из Приштине би отишао у Солун, а затим у Битољ. Иако је раније било планира- но да султан из Битоља обиђе Албанију и посети три највеће вароши, од ове идеје се одустало због побуне у северном делу. Д , , , д , 3 ДСПКС, књ. IV, св. 3/1, док. 232, 542. 2 ДСПКС, књ. IV, св. 3/1, док. 181, 460. СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Као претходница пут Скопља, а касније и Приштине, требало је да отпутује министар војни Махмуд Шефкет-паша.5 Вести о планираним протестима Арбанаса и Бугара, које су долазиле до Ца- риграда, нису битно утицале на план султановог путовања. Идеја о путовању није напуштана, а ширење информација о демонстрацијама само је могло да користи младотурском режиму, који је и раније намеравао да концентрише већи број војника на простору Старе Србије и Македоније. Ипак, коначана одлука о повећању војске зависила је од извештаја Махмуд Шефкет-паше који је имао задатак да процени ситуацију на терену.6 султаново путовање, па су мобилисале две дивизије из Мале Азије, једну из Сиваса, а другу из Амасије.4 Планом је било предвиђено да војска крене пре- ма Скопљу и учествује у дочеку. Као претходница пут Скопља, а касније и Приштине, требало је да отпутује министар војни Махмуд Шефкет-паша.5 Вести о планираним протестима Арбанаса и Бугара, које су долазиле до Ца- риграда, нису битно утицале на план султановог путовања. Идеја о путовању није напуштана, а ширење информација о демонстрацијама само је могло да користи младотурском режиму, који је и раније намеравао да концентрише већи број војника на простору Старе Србије и Македоније. Ипак, коначана одлука о повећању војске зависила је од извештаја Махмуд Шефкет-паше који је имао задатак да процени ситуацију на терену.6 Султанова посета је оживела поједине крајеве. У Приштини и Фе- ризовићу се почело са подизањем нових касарни, а у Бресју, недалеко од Приштине, радило се на изградњи нове железничке станице. За разлику од ових места, остале крајеве су обележили сиромаштво и беда праћени разним зулумима (Политика, 02. 04. 1911: 2). у у Припреме за дочек султана су највише биле приметне у самој Приштини. Највећи радови су били на згради хућумата у којој је султан требало да одсед- не, као и на улици којом би ушао у варош. Међутим, поједини радови, попут скидања покривене чаршије и сечење стреха изазивали су негодовање код ста- новника Приштине који нису са одушевљењем прихватали султанов долазак. Све послове у самој вароши и око ње контролисали су и надзирали чиновни- ци, чланови главног одбора за дочек. 4 Архив Србије (=АС), Министарство иностраних дела (=МИД) Политичко-просветно одељење (=ППО), 1911, ред 373, ПП Бр. 83, Конзул Живојин Балугџић – Милова- ну Ђ. Миловановићу, министру иностраних дела, Солун 8/21. март 1911; За разли- ку од извештаја генералног конзула Краљевине Србије у Солуну Живојина Балугџића у којем наводи да ће бити мобилисане две дивизије и упућене у Скопље и Приштину, Политика пише да ће бити присутне четири дивизије из Мале Азије (Политика, 02. 04. 1911: 2). 5 АС, МИД ППО, 1911, ред 373, ПП Бр. 83, Конзул Живојин Балугџић – Миловану Ђ. Миловановићу, министру иностраних дела, Солун 8/21. март 1911 6 ДСПКС, књ. IV, св. 3/1, док. 244, 555. 7 ДСПКС, књ. IV, св. 3/1, док. 336, 714–715. Д , , , , 7 ДСПКС, књ. IV, св. 3/1, док. 336, 714–715. 5 АС, МИД ППО, 1911, ред 373, ПП Бр. 83, Конзул Живојин Балугџић – Миловану Ђ. Миловановићу, министру иностраних дела, Солун 8/21. март 1911 6 ДСПКС, књ. IV, св. 3/1, док. 244, 555. СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Такође, било је предвиђено да на том путу султана прате бројне познате личности, међу којима су и двојица његових синова, велики везир Хаки-паша, министар војни Махмуд Шефкет- паша, министар унутрашњих дела и министар марине, први секретар султанов, први дворски маршал и још многи други цивилни и војни великодостојници. Присуство престолонаследника Јусуф Изедин-ефендије није ни разматрано, због одласка у Лондон на крунисање Ђорђа V (Политика, 02. 04. 1911: 1). Султаново путовање по Старој Србији и Македонији било је непозна- ница за многе представнике страних земаља у Османском царству, па и за Живојина Балугџића, генералног конзула Краљевине Србије у Солуну. Он је сматрао да путовање представља реализацију идеје, осмишљене од стране Хилми и Шефкет-паше, која би имала за циљ да умири арбанашке прваке и придобије их за себе. Овакво Балугџићево мишљење заснивало се на саз- нању да је Парламент одобрио два милиона гроша за подизање арбанашких кућа које су уништене приликом гушења побуне од стране војске и званич- них турских власти. Арбанашко становништво у Старој Србији и Македо- нији је било прилично непријатељски расположено према султану и новом режиму, уопште. Такве тврдње је износио и аустроугарски конзул Пара, који је располагао информацијама о могућим арбанашким протестима, што потврђује да су представници Двојне монархије у Османском царству с ве- ликим интересовањем пратили догађаје и неретко били директно или ин- директно укључени у збивања на терену.2 За разлику од Арбанаса, Турци су веровали да би им султанова посета Старој Србији и Македонији донела неке позитивне резултате, пре свега умирила арбанашко становништво у тим крајевима. Међутим, ситуација на терену је била врло компликована, нарочито „с оне стране Качаника“. Као илустрација расположења на том простору може нам послужити легенда, потекла од католичких свештеника, која се све чешће могла чути. „Кад је први Султан дошао у Арбанију он ју је добио и умро, а кад други дође у њу изгубиће је а остаће жив“.3 Поред Арбанаса, незадовољство су испољавали и Бугари који су намера- вали да приликом султанове посете организују демонстрације и употребе их као средство за наметање свог питања пред Европом. Турске власти, упозна- те са ситуацијом у Старој Србији и Македонији, пажљиво су планирале 290 Весна С. Зарковић султаново путовање, па су мобилисале две дивизије из Мале Азије, једну из Сиваса, а другу из Амасије.4 Планом је било предвиђено да војска крене пре- ма Скопљу и учествује у дочеку. 4 Архив Србије (=АС), Министарство иностраних дела (=МИД) Политичко-просветно одељење (=ППО), 1911, ред 373, ПП Бр. 83, Конзул Живојин Балугџић – Милова- ну Ђ. Миловановићу, министру иностраних дела, Солун 8/21. март 1911; За разли- ку од извештаја генералног конзула Краљевине Србије у Солуну Живојина Балугџића у којем наводи да ће бити мобилисане две дивизије и упућене у Скопље и Приштину, Политика пише да ће бити присутне четири дивизије из Мале Азије (Политика, 02. 04. 1911: 2). 5 АС, МИД ППО, 1911, ред 373, ПП Бр. 83, Конзул Живојин Балугџић – Миловану Ђ. Миловановићу, министру иностраних дела, Солун 8/21. март 1911 6 ДСПКС, књ. IV, св. 3/1, док. 244, 555. 7 ДСПКС, књ. IV, св. 3/1, док. 336, 714–715. у, ру р 6 ДСПКС, књ. IV, св. 3/1, док. 244, 555. 8 АС, МИД ППО, 1911, ред 373, ПП Бр. 257, Конзул Милан Ракић – Министарству иностраних дела у Београду, Приштина 27. април/10. мај 1911. 9 АС, МИД ППО, 1911, ред 373, ПП Бр. 966, Конзул Јов. М. Јовановић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 28. април/11. мај 1911. 10 АС, МИД ППО, 1911, ред 373, ПП Бр. 312, Посланик Ј. М. Ненадовић – Министар- ству иностраних дела Краљевине Србије, Цариград 3/16. мај 1911. 11 АС, МИД ППО, 1911, ред 373, ПП Бр. 966, Министар иностраних дела М. Ђ. Миловановић – Конзулатима у Скопљу, Битољу, Солуну и Приштини, 5/20. мај 1911. 12 АС, МИД ППО, 1911, ред 373, ПП Бр. 548, Конзул Љуб. М. Михаиловић – Министар- ству иностраних дела Краљевине Србије, Битољ 3/16. мај 1911. СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Они су сазвали конференцију на којој су представницима српске општине дали посебна упутства која су подразумевала: да своје домове украсе заставама и увече осветле, да деца буду прикладно и све- чано одевена за ту прилику, с посебним нагласком да приликом одабира боје ђачких униформи буде изузета плава боја, која је могла да утиче на формирање српске тробојке.7 Радило се и на изградњи новог пута који је повезивао Гази- местан и Муратово турбе са друмом Приштина–Глободерица (данашњи Оби- лић) (Микић 1988: 279). Иначе, преко железничке станице у Глободерици је требало да стигне велики број војника, планираних за ову прилику. Султаново путовање по Старој Србији и Македонији 291 У припреме за султанов долазак су били укључени и представници срп- ског народа. Тим поводом архимандрит Сава је разговарао са приштинским мутесарифом који је исказао потребу за ангажовањем српских ђака и изра- зио жељу за учешће оних из Призренске богословије.8 Срби су планирали да султаново присуство искористе као згодну прилику која би им послужи- ла да поједина питања, битна за њихов опстанак на тим просторима, изнесу пред њега. Генерални конзул Краљевине Србије у Скопљу Јован Јовановић предлагао је да тај посао буде поверен архимандриту Сави који би одредио људе из свих крајева рашко-призренске епархије да пред султана усмено или у писаној форми изнесу жалбе. То је подразумевало да се акценат стави на многобројна насиља вршена над српским живљем, као и на одузимање мана- стирских имања и протеривање чифчија са земље.9 Турске власти су изражавале жељу да приликом султановог боравка у Скопљу и Битољу буду, у што већем броју, заступљени представници срп- ског народа. Њихово бројно присуство би ослабило ефекат истицања бугар- ског елемента.10 О овим жељама турских власти посланик Србије у Царигра- ду Ј. М. Ненадовић је обавестио Владу у Београду, која је дописом захтевала од конзулата у Скопљу, Битољу, Солуну и Приштини да узму активно учешће у збивањима на терену.11 Конзул Краљевине Србије у Битољу имао је већ разрађени план, који је подразумевао присуство епископа Варнаве и десетине свештеника из његове епархије, присуство делегације од два до три члана са управником из свих школских управа, присуство до стотину сељака из најближих села, затим присуство великог броја ђака, од којих би 170 било из битољске гиманзије. Конзул је планирао и уручење пригодног поклона за чију израду би био задужен директор гимназије Јотић. у у у, у, у у р , / ј 12 АС, МИД ППО, 1911, ред 373, ПП Бр. 548, Конзул Љуб. М. Михаиловић – Минист ству иностраних дела Краљевине Србије, Битољ 3/16. мај 1911. СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Замишљено је било да се направи један албум српских и турских школа у Битољу који би садржао око 50 фотографија и чија би израда износила 250–300 динара. Трошкови пута ангажованих, гардероба за ученике, израда декорације и поклона изис- кивали су одређену суму новца, па је зато од Министарства иностраних дела затражено 6.000 динара за ту намену.12 Конзуларно одељење Министарства иностраних дела у Београду је рас- полагало информацијом да је султанов полазак из Цариграда био планиран 292 Весна С. Зарковић за 16. мај.13 Зато су министру иностраних дела дате инструкције којима се захте- вало учешће српског народа у том догађају и то преко мешовитих делегација, школских и црквених власти. Држање српског народа требало је да, с једне стране, искаже лојалност, а с друге, искористи прилику за упућивање молби и жалби, које би биле дефинисане у форми присталица новог режима. Планира- но је било да у молбама и жалбама буду садржана општа и питања појединаца, а не и она велика, која су се односила на питање о народности, независној цркви, школска и многа друга која су се решавала на вишем нивоу. Општа питања су подразумевала помиловање затвореника, поступање са чифчијама, неправилан рад за оцењивање и класификацију земљишта, заузимање цркава, манастира, њихових имања и парнице око њих, питање о дозволама за школе по епархија- ма ван рашко-призренске епархије, као и незаконитост и самовољу админи- стративних органа. У овим инструкцијама је било наглашено да се у жалбама о арбанашким зулумима избегава директно оптуживање Арбанаса, као и да се не тражи помоћ у спорним питањима између Патријаршије и Срба. Такође, било је наглашено да ће се сви трошкови рефундирати из поверљивог буџета конзу- лата Краљевине Србије у тим деловима Османског царства.14 На седници Синода одржаној 6. маја било је одлучено да митрополити заједно са општинарима и народним првацима дочекају султана. Према овој одлуци требало је да, поред митрополита, и све школе учествују у дочеку. По- ред тога, у Патријаршији су размишљали и о месту које би заузели митропо- лити, народни прваци и ученици и одлучили да то, као и до тада, буде одмах до турских представника народности, вере и просвете. Сходно одлукама до- нетим у Патријаршији, конзул из Скопља Јован Јовановић је предложио да се оснује одбор за дочек који би чинили: митрополит (председник) и чланови Глигорије Глиша Елезовић, Ал. Станишић, Ђр. Х. Костић, Петар Лекић.15 Са одлукама из Патријаршије био је упознат и архимандрит Сава. 13 Сви датуми у тексту су навођени по старом календару. 14 АС, МИД ППО, 1911, ред 373, ПП Бр. 2771, Концепт Конзуларног одељења Мини- старства иностраних дела – Министру иностраних дела, Београд 5/18. мај 1911. 15 ДСПКС, књ. IV, св. 3/II, Београд 2009, док. 485, 932–933. 16 АС, МИД ППО, 1911, ред 373, ПП Бр. 337, Конзул Милан Ракић – Министарству иностраних дела Краљевине Србије, Приштина 20. мај/2. јун 1911. 5 Д , , 3/ , р д 9, д 4 5, 93 933 16 АС, МИД ППО, 1911, ред 373, ПП Бр. 337, Конзул Милан Ракић – Министарству иностраних дела Краљевине Србије, Приштина 20. мај/2. јун 1911. 13 Сви датуми у тексту су навођени по старом календару. у у у у р ру 14 АС, МИД ППО, 1911, ред 373, ПП Бр. 2771, Концепт Конзуларног одељења Мин старства иностраних дела – Министру иностраних дела, Београд 5/18. мај 1911. СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Поред наве- дених учесника, архијерејима, члановима општина и ученицима је било нагла- шено обавезно присуство на железничким станицама у местима у којима она постоји, а тамо где не постоји, архијереј је био дужан дужан да иде у валилук и заједно са месним и осталим архијерејима у епархији дочекају султана.16 За разлику од Срба који су активно учествовали у припремама за дочек, Турци су невољно прилазили том послу. Иако је постојала разлика и јаз из- међу присталица старог и новог режима, били су јединствени када је реч о ис- тицању муслиманства и отоманства. У појединим ситуацијама су младоруци Султаново путовање по Старој Србији и Македонији 293 предњачили у односу на своје представнике. На то нас упућују и наредбе из Цариграда које су се односиле на фес, заставе и учешће турских жена у султа- новом дочеку. Ношење феса је било обавезно, а свака друга капа „којом би се могла изазвати сумња да све што је изашло да поздрави љубљеног владара није мусломан“ била је забрањена. Иако је на први поглед ова наредба изазивала подсмех, изазвала је позитивне коментаре присталица старог режима, којима се није допадало ношење других капа код немуслиманског становништва, по- себно у градовима. Поред забране ношења капа, била је забрањена и употреба застава других држава, осим турске. Са оваквом одлуком нису били сагласни страни представници, пре свега Италије и Француске, чије су школе учество- вале у дочеку. Они су успели да преко амбасада издејствују употребу својих застава, поред турске. Разматрано је било и питање о учешћу турских жена, посебно донма, удатих за потурчене Јевреје. За разлику од осталих Туркиња које су прекривале лице, донме нису носиле јашмаке. То је навело валију у Скопљу да одржи састанак са представницама донми и предложи им прикри- вање лица за ту прилику. Валијин предлог је изазвао противљење јер је био у супротности са њиховим, већ утврђеним, обичајима. Ови и слични предлози имали су за циљ да што више истакну муслиманство, чак и у оним случајевима где су поједини народи имали своје представнике.17 Колико су турске власти водиле рачуна о истицању муслиманства говори нам и чињеница да је у Ско- пљу програмом било предвиђено груписање свих школа заједно, по степену и струци, а не једне народности одвојено. Ношење феса било је изричито, а по- себно се ова одлука односила на хришћане који су били заступљени у служби, попут учитеља, наставника, лекара, инжењера и осталих.18 Султанов воз је стигао 29. 17 ДСПКС, књ. IV, св. 3/II, док. 512, 969–971. 18 АС, МИД ППО, 1911, ред 373, ПП Бр. 1178, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 24. мај/6. јун 1911. 19 Соколско друштво је основано почетком 1909. године на иницијативу конзула Ми- лана Ракића. Главни организатор је био Рајко Караклајић, учитељ гимнастике српске гимназије у Скопљу, а први начелник учитељ Димитрије Синадиновић, родом из При- зрена (Поповић 2007: 343). 19 Соколско друштво је основано почетком 1909. године на иницијативу конзула Ми- лана Ракића. Главни организатор је био Рајко Караклајић, учитељ гимнастике српске гимназије у Скопљу, а први начелник учитељ Димитрије Синадиновић, родом из При- зрена (Поповић 2007: 343). 18 АС, МИД ППО, 1911, ред 373, ПП Бр. 1178, Конзул Јован Ј. Вучковић – Миловану Миловановићу, министру иностраних дела, Скопље 24. мај/6. јун 1911. 17 ДСПКС, књ. IV, св. 3/II, док. 512, 969–971. СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ маја, тачно у 4 сата после подне на станицу у Скопљу, на којој су већ биле присутне царска гарда, коњица и један шпалир пешадије (Политика, 31. 05. 1911: 2). Поред њих, дочеку су присуствовали и представници турских школа, српске гимназије, учитељске школе, сокол- ског друштва,19 Женске раденичке школе, грчке и бугарске школе. Распоре- дом је било предвиђено да на перону добију своја места и чланови конзулар- ног кора и то најпре из Скопља са својим колегама из Митровице, Призрена и Приштине. Из Призрена су били присутни аустроугарски конзул Проха- ска и италијански вицеконзул Цукулин, док руски вицеконзул Зујев није до- шао. Из Митровице су допутовали руски конзул Лобачев и аустроугарски 294 Весна С. Зарковић вицеконзул Тахи, а из Приштине конзул Краљевине Србије Милан Ракић. Иза њих су стајали верски представници, на првом месту муфтија, до њега митрополит и архимандрит Сава из Приштине, затим грчки архимандрит, бугарски Неофит и католички бискуп. Они су претходили изасланицима меџлис идаре из каза. Војска је била распоређена по улицама којима је сул- тан пролазио и на којима су се налазили многобројни људи, обучени у раз- на одела. Након бројних поздрава и клицања, султан је од железничке ста- нице до идадије (своје палате) стигао у пратњи двојице синова, Зије Етин и Омера Хилми, великог везира, министра унутрашњих послова, просвете и неколико чиновника двора. Поред њих у пратњи су били Нијази-беј, ре- сански јунак, Хаџи Адил-беј и секретар Централног Одбора „Јединство и Напредак“ Галиб-паша, ранији шеф полиције у Цариграду.20 Иначе, про- грамом је било превиђено да конзули из Призрена, Митровице и Пришти- не отпутују у Скопље, где је и био планиран њихов пријем код султана. Ова одлука се односила и на конзула Ракића, који је о томе благовреме- но био обавештен од косовског валије (Перуничић 1989: 531). Поједини новинарски кругови су пласирали информацију да ће српски престолона- следник Александар доћи у Скопље и поздравити султана. Таква вест је изазивала одушевљење код српског народа у Старој Србији, који је при- жељкивао његову посету и Приштини.21 Међутим, овај пут краља Алексан- дра није реализован. вицеконзул Тахи, а из Приштине конзул Краљевине Србије Милан Ракић. Иза њих су стајали верски представници, на првом месту муфтија, до њега митрополит и архимандрит Сава из Приштине, затим грчки архимандрит, бугарски Неофит и католички бискуп. Они су претходили изасланицима меџлис идаре из каза. 20 АС, МИД ППО, 1911, ред 373, ПП Бр. 1258, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 3/16. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 575, 1061–1062. 20 АС, МИД ППО, 1911, ред 373, ПП Бр. 1258, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 3/16. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 575, 1061–1062. 21 АС, МИД ППО, 1911, ред 373, ПП Бр. 2906, Телеграм конзула Милана Ракића –Ми- нистарству иностраних дела, 14/27. мај 1911; ДСПКС, књ. IV, св. 3/II, док. 492, 941. 22 АС, МИД ППО, 1911, ред 373, ПП Бр. 1246, Конзул Јован Ј. Вучковић – Министар- ству иностраних дела Краљњевине Србије, Скопље 31. мај/13. јун 1911. р у р , ј ; Д , , , , 22 АС, МИД ППО, 1911, ред 373, ПП Бр. 1246, Конзул Јован Ј. Вучковић – Министар- ству иностраних дела Краљњевине Србије, Скопље 31. мај/13. јун 1911. 21 АС, МИД ППО, 1911, ред 373, ПП Бр. 2906, Телеграм конзула Милана Ракића –Ми- нистарству иностраних дела, 14/27. мај 1911; ДСПКС, књ. IV, св. 3/II, док. 492, 941. / , д , 21 АС, МИД ППО, 1911, ред 373, ПП Бр. 2906, Телеграм конзула Милана Ракића –Ми- нистарству иностраних дела, 14/27. мај 1911; ДСПКС, књ. IV, св. 3/II, док. 492, 941. 22 АС, МИД ППО, 1911, ред 373, ПП Бр. 1246, Конзул Јован Ј. Вучковић – Министар- ству иностраних дела Краљњевине Србије, Скопље 31. мај/13. јун 1911. Миловановићу, министру иностраних дела, Скопље 3/16. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 575, 1061–1062. 21 АС, МИД ППО, 1911, ред 373, ПП Бр. 2906, Телеграм конзула Милана Ракића –Ми- нистарству иностраних дела, 14/27. мај 1911; ДСПКС, књ. IV, св. 3/II, док. 492, 941. 22 АС МИД ППО 1911 373 ПП Б 1246 К Ј Ј В ћ М СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Војска је била распоређена по улицама којима је сул- тан пролазио и на којима су се налазили многобројни људи, обучени у раз- на одела. Након бројних поздрава и клицања, султан је од железничке ста- нице до идадије (своје палате) стигао у пратњи двојице синова, Зије Етин и Омера Хилми, великог везира, министра унутрашњих послова, просвете и неколико чиновника двора. Поред њих у пратњи су били Нијази-беј, ре- сански јунак, Хаџи Адил-беј и секретар Централног Одбора „Јединство и Напредак“ Галиб-паша, ранији шеф полиције у Цариграду.20 Иначе, про- грамом је било превиђено да конзули из Призрена, Митровице и Пришти- не отпутују у Скопље, где је и био планиран њихов пријем код султана. Ова одлука се односила и на конзула Ракића, који је о томе благовреме- но био обавештен од косовског валије (Перуничић 1989: 531). Поједини новинарски кругови су пласирали информацију да ће српски престолона- следник Александар доћи у Скопље и поздравити султана. Таква вест је изазивала одушевљење код српског народа у Старој Србији, који је при- жељкивао његову посету и Приштини.21 Међутим, овај пут краља Алексан- дра није реализован. Султанов боравак у Скопљу обележило је присуство осамнаест клубо- ва младотурака из румелијских вилајета, међу којима су били најбројнији из Косовског. Поред њих били су присутни и изасланици из Смирне, Ангоре, Ерзерума, Самсуна, Трапезунта, Магнезије и Курдистана, међу којима су већину чинили официри. Током вечери је у беледијском позоришту одржана конференција коју је отворио неки хоџа из Курдистана. Међу говорницима су били бег из Ерзерума и Омер Наџи, секретар солунског комитета. Суштина оба говора је указивала на препород Османског царства, његових циљева и резултата, као и значај уједињења Турака из Мале Азије и оних из Европе. За разлику од муслимана, хришћани су били присутни у малом броју и то углав- ном попови. Значај ове конференције је у томе што је на њој присуствовао и велики везир.22 Султан је током боравка у Скопљу примао бројне делегације и то најпре народне посланике из Косовског вилајета које је предводио Хоџа Султаново путовање по Старој Србији и Македонији 295 Саид, посланик скопски. За њима су на аудијенцију били примљени: шефови одељења скопског хућумата, виши официри, представници беледије, шеици и улеме, изасланици џемијета Косовског вилајета на челу са Хаџи Али-бејом, изасланства из каза, међу којима су били митрополит и архимандрит Сава са свим намесницима, изасланство из Пљеваља и представници конзуларног кора. За разлику од њих, поглавари вера били су поздрављени од стране ве- ликог везира. 23 АС, МИД ППО, 1911, ред 373, ПП Бр. 1258, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 3/16. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 575, 1062–1063. 25 АС, МИД ППО, 1911, ред 373, ПП Бр. 1258, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 3/16. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 575, 1063. 24 АС, МИД ППО, 1911, ред 373, ПП Бр. 1255, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 2/15. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 570, 1050–1051. 23 АС, МИД ППО, 1911, ред 373, ПП Бр. 1258, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 3/16. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 575, 1062–1063. 24 АС, МИД ППО, 1911, ред 373, ПП Бр. 1255, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 2/15. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 570, 1050–1051. 25 АС, МИД ППО, 1911, ред 373, ПП Бр. 1258, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 3/16. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 575, 1063. Миловановићу, министру иностраних дела, Скопље 3/16. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 575, 1062–1063. 24 АС, МИД ППО, 1911, ред 373, ПП Бр. 1255, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 2/15. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 570, 1050–1051. СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Током пријема султан је поделио разне поклоне и то: турским посланицима златне сатове, а хришћанским накит за жене. Сава Стојано- вић је добио брош за жену, а бугарски Павлов дугмад за манжетне. Бугар- ски и српски митрополит и архимандрит грчки су добили по златни сат са ланцем. Неколико одабраних Арбанаса добило је по једну лиру на којој је писало „Косово“.23 Међу виђенијим арбанашким првацима који су дошли да поздраве султа- на налазили су: Сулејман Батуша, Хасан Феровић и посланици Неџиб Драга, вучитрнски Хасан-бег и призренски Шериф-ефендија. Сусрет са њима и срда- чан пријем од стране султана наговестио је општу амнестију, која је требало да претходи помирењу са Арбанасима. У ствари, султаново путовање по Старој Србији и Македонији је имало чисто муслимански карактер. О томе се говори- ло и међу простим Турцима који су понављали: „Ове земље још мало па су биле изгубљене за Турску, и сад падишах иде на Косово да их сасвим освоји и сачува Турцима“. Можда су ова тумачења била заснована на сазнању да је мерћез кај- макам, уочи султановог доласка, пред бројним виђенијим Турцима и представ- ницима других народности на једној конференцији изговорио: „Овај султанов пут значи ново освојење ових земаља“.24 Овакво тумачење потврђују и речи са- мог султана који је као циљ путовања наводио зближење разних народности, нарочито Арбанаса и истицао: „Срећан ћу бити ако у томе успем“.25Колико је султлану било битно међусобно зближавање припадника разних народности, најбоље илуструје пример ђака од којих је захтевао да један другом пруже руку, као знак уједињења, љубави и слоге. Након тог чина даривао је четворицу уче- ника са по пола турске лире. Током боравка у Скопљу султан је, поред ученика, даривао и школе. Тако је на име поклона турској занатској школи уручио 300 турских лира, разним народносним школама 500, медресама и текијама 150, муслиманским и немуслиманским добротворним установама 200, сиротињи 296 Весна С. Зарковић у Скопљу 500, болници 300, текији дервиша из секте „мевлеви“ 200 и Комитету „Јединство и Напредак“ 5.000, у чије име је новац примио Хаџи Адил-беј.26 у Скопљу 500, болници 300, текији дервиша из секте „мевлеви“ 200 и Комитету „Јединство и Напредак“ 5.000, у чије име је новац примио Хаџи Адил-беј.26 Султанов дочек у Скопљу, разне свечаности, манифестације и изложбе које су протекле без иједног инцидента, пратили су бројни дописници стра- них агенција. Били су присутни новинари Тајмса, Тана, Руског Слова, Дејли Телеграфа и других. СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Карактеристично је да међу њима није било представ- ника ниједног листа из Србије.27 Посета султана седишту вилајета није импресионирала тамошње стано- вништво, нарочито хришћане. Он је, окружен својом пратњом, након Ско- пља кренуо пут Приштине. Тим поводом је у правцу Косова кренуло више од 2.000 људи, а о овом броју се у самом граду говорило да представља ре- зултат притиска, вршеног од стране клуба „Јединство и Напредак“.28 Султана је у Приштини дочекало више од 20.000 људи.29 Међу њима је било житеља Приштине, војске, званичних делегација, сељака из околних каза, Срба и Арбанаса. Било је присутно 4.000 Арбанаса из Дренице, пред- вођених Хасан-бегом, послаником из Вучитрна. Поред њих, дочеку су у ве- ликом броју присуствовали и Срби и Арбанаси из моравског краја, док су Арбанаси из Лаба били малобројни. Међу присутнима није било представ- ника из Призрена, Пећи и Ђаковице.30 Чињеница која пада у очи је мали број присутних из Лаба, с обзиром на близину, као и изостајање представника из Пећи и Ђаковице, у којима су сукоби Арбанаса и турских власти били нај- жешћи у претходној години.31 Срби су се окупљали у митрополији, одакле су, предвођени призренским богословима, ишли према хућумату. Окупље- не богословске ђаке, којих је било 46, предводио је ректор Призренске 27 АС, МИД ППО, 1911, ред 373, К ПП Бр. 1258, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 4/17. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 579, 1068. 27 АС, МИД ППО, 1911, ред 373, К ПП Бр. 1258, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 4/17. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 579, 1068. 28 АС, МИД ППО, 1911, ред 373, К ПП Бр. 1258, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 4/17. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 579, 1069. 28 АС, МИД ППО, 1911, ред 373, К ПП Бр. 1258, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 4/17. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 579, 1069. 29 Број Арбанаса који су дочекали султана се разликује у зависности од извештаја. Поли- тика је у два наврата писала да је султана дочекало 100.000 Арбанаса (Политика, 04. 06. 1911: 2; Политика, 01. 06. 1911: 2). 26 АС, МИД ППО, 1911, ред 373, К ПП Бр. 1258, Конзул Јован Ј. Вучковић – Миловану Ђ. Миловановићу, министру иностраних дела, Скопље 4/17. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 579, 1066–1067. 29 Број Арбанаса који су дочекали султана се разликује у зависности од извештаја. Поли- тика је у два наврата писала да је султана дочекало 100.000 Арбанаса (Политика, 04. 06. 1911: 2; Политика, 01. 06. 1911: 2). Сматрамо је да је овај број претеран у односу на извештај конзула из Приштине Милана Ракића који говори о присуству 20.000 АС, МИД ППО, 1911, ред 373, ПП Бр. 377, Конзул Милан Ракић – Министарству ино- страних дела Краљевине Србије, Приштина 4/17. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 580, 1069; (Ракић 1985: 245–246). СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Сматрамо је да је овај број претеран у односу на извештај конзула из Приштине Милана Ракића који говори о присуству 20.000 АС, МИД ППО, 1911, ред 373, ПП Бр. 377, Конзул Милан Ракић – Министарству ино- страних дела Краљевине Србије, Приштина 4/17. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 580, 1069; (Ракић 1985: 245–246). 30 АС, МИД ППО, 1911, ред 373, ПП Бр. 377, Конзул Милан Ракић – Министарству иностраних дела Краљевине Србије, Приштина 4/17. јун 1911; ДСПКС, књ. IV, св. 3/ II, док. 580, 1069; (Ракић 1985: 245–246). 31 Детаљније о арбанашким побунама видети: Зарковић–Савић 2018: 291–313. Султаново путовање по Старој Србији и Македонији 297 богословије Стева Димитријевић. Српској делегацији се придружио и ар- химандртит Сава Протић који је искористио пријем код великог везира и упутио му молбе које су се односиле на железницу и Пећку патријаршију, а усмено изложио проблеме настале решавањем чифчијског питања. Поред тога, архимандрит је и министру просвете предочио значај и упутио молбу за отварање пољопривредне и других занатских школа и истовремено ука- зао на недостатак поједних ствари, неопходних за српске народне школе.32 Током вечери архимандрит Сава је био примљен од стране султана који га је дариивао златном табакером на којој је на француском било исписано ње- гово име Решет у брилијантима.33 Боравак султанов у Приштини је био веома кратак, свега дан и по. Али, с обрзиром на даљи развој догађаја, с правом се може тврдити да је ова по- сета имала превасходно муслимански карактер. Од главног циља, посете Муратовог турбета, се није одустало. Султан је од Приштине до Мурато- вог гроба пролазио улицама запоседнутим војском, пошто је грађанству био забрањен излазак. Након разгледања Муратовог гроба уследила је мо- литва, којој је према информацијама конзула Милана Ракића присуствова- ло око 30.000 људи. Ракић је сматрао да то и није велики број, с обзиром на чињеницу да је у тим крајевима живео велики број Арбанаса. После молит- ве присутнима се обратио велики везир и истакао значај султанове посете, која је имала за циљ да оповргне разне гласине о Арбанасима као бунтовни- цима и одметницима од власти, подложним страном утицају. Он је нагла- сио да се султан уверио у супротно и као доказ љубави према Арбанасима прочитао ферман о амнестији свих политичких криваца, у земљи и ван ње. Саопштио је и да је султан из своје приватне касе издвојио 30.000 турских лира за мирење међу Арбанасима. 32 Архив Југославије (=АЈ), Збирка Јована Јовановића Пижона, 80, кут. 48, с 420–422а. 33 АЈ, 80, кут. 48, с 420–422а. 34 АС, МИД ППО, 1911, ред 373, ПП Бр. 380, Конзул Милан Ракић – Миловану Ђ. Миловановићу, министру иностраних дела Краљевине Србије, Приштина 4/17. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 582, 1074; (Ракић 1985: 248). 34 АС, МИД ППО, 1911, ред 373, ПП Бр. 380, Конзул Милан Ракић – Миловану Ђ. Миловановићу, министру иностраних дела Краљевине Србије, Приштина 4/17. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 582, 1074; (Ракић 1985: 248). 32 Архив Југославије (=АЈ), Збирка Јована Јовановића Пижона, 80, кут. 48, с 420–422а. 33 АЈ, 80, кут. 48, с 420–422а. СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Истовремено, није пропустио прилику да подсети Арбанасе на султанове речи изговорене у парламенту годину дана раније: „Да су они најлепши дијамант у круни отоманској. Отаџбина гледа у њих као најбоље синове своје и све су очи упрте у њих. С тога они треба да се окану међусобних зађевица и убистава, па да, сложни и снажни, стану на браник отаџбине“.34 Знајући да Арбанаси доста значаја придају ордењу и одликовањима, било је дозвољено подношење спискова у којима су се налазила имена од преко хиљаду предложених. Зато је било одлучено да се формира посеб- на комисија са задатком да поброји све учеснике ратова 1876–1878. годи- не против Србије и одликује их за храброст и војничке врлине. О посети 298 Весна С. Зарковић Муратовог турбета телеграфским путем су из Приштине послате депеше у све делове света у којима је било муслимана.35 Муратовог турбета телеграфским путем су из Приштине послате депеше у све делове света у којима је било муслимана.35 Овом, за Османско царство значајаном догађају, када је султан први пут посетио простор Старе Србије и Македоније, код Муратовог турбета су при- суствовали бројни арбанашки прваци. Међу њима су били Хасан Феровић, Сулејман-ага Батуш, Асан Хусеин са друговима из Лесковца и синови Исе Бољетинца. Упркос напору Турака, Идриз Сефер и Иса се нису одазвали по- зиву.36 Иса Бољетинац је био један од истакнутијих Арбанаса који се настањен у својој кули, у близини Митровице, није предао, пркосио је и у арбанашком духу исмевао султана поредећи га са „мечком коју воде тамо амо и непрестано мувају, а јадна мечка, само урличе и мумла“ (Перуничић 1988: 451). Султанова посета Приштини је трајала краће од предвиђеног и он је 4. јуна кренуо према Солуну. Конзул Краљевине Србије Милан Ракић је располагао информацијама да је на такву одлуку утицала вест о могућим арбанашким протестима којима би се осудило држање Џемијета и његово мешање у административне и судске послове. Сматрало се да су на промену плана путовања утицали Нази-беј и Омер Наџи, који су у томе видели начин да се избегну могуће непријатности.37 у у р ј На путу према Солуну султан се задржао у Битољу, у којем је био при- ређен свечани дочек. Улицама којима је пролазио било је подигнуто осам тријумфалних капија, од којих су три припадале вароши, а осталих пет је било урађено на рачун Бугара, Румуна, Грка, Срба и Јевреја. 35 Исто. 36 АС, МИД ППО, 1911, ред 373, ПП Бр. 380, Конзул Милан Ракић – Министарству иностраних дела Краљевине Србије, Приштина 4/17. јун 1911; ДСПКС, књ. IV, св. 3/ II, док. 582, 1076. 37 АС, МИД ППО, 1911, ред 373, ПП Бр. 381, Конзул Милан Ракић – Министарству иностраних дела Краљевине Србије, Приштина 6/19. јун 1911; ДСПКС, књ. IV, св. 3/ II, док. 586, 1082. АС, МИД ППО, 1911, ред 373, ПП Бр. 381, Конзул Милан Ракић – Министарству иностраних дела Краљевине Србије, Приштина 6/19. јун 1911; ДСПКС, књ. IV, св. 3/ II, док. 586, 1082. СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Као и у осталим местима, дочекали су га представници конзуларног кора, виши официри, чиновници, турско свештенство, сви егзархијски и патријаршијски митро- полити из вилајета, међу којима је био и српски Варнава, разне делегације из унутрашњости, представници свих народности и ђаци. Биле су присутне и разне групе Арбанаса из различитих крајева које су дефиловале градом, на чијем се челу налазио Нијази који је предводио кола слободе и окићени топ са којег је објављена уставност 10. јула 1908. године. Као и у Скопљу, султан је даривао поклонима. Међу бројнима који су имали ту привилегију били су грчки и бугарски митрополит, један румунски свештеник и један јеврејски првак. Да ли случајно или намерно, не зна се, али је епископ Варнава био изостављен. Ипак, на исистирање посланика др Димитријевића, добио је златни сат. Султанову посету Битољу, започету 7. јуна, обележила је посеб- на пажња која је посвећена члановима старог младотурског комитета и као Султаново путовање по Старој Србији и Македонији 299 потврда таквог понашања додељења су бројна одликовања или унапређења.38 Срби који су живели и радили у Битољу били су активно укључени око свих припрема за дочек. У томе су се нарочито истакли професори и ученици би- тољске гимназије. Ову, за Србе важну установу, посетио је министар про- свете Абдурахман, бивши директор Галата Сараја у Цариграду, који је и те како био упознат са школским пословима. Приликом обиласка школе био му је показан албум са фотографијама српских и турских школа у Битољу, који је касније у име чланова гимназије уручен султану од стране Ксенофона Шаховића, одличног ученика III разреда.39 Након свечаног дочека и уручених поклона с обе стране, султан се упутио према Солуну. Осим учешћа школа није било већих уличних манифестација и овација, а свечани говори су изостали. Велики напори младотурског коми- тета да посету овом граду учине што топлијом нису уродили плодом. Варош је, по речима вицеконзула Краљевине Србије Јевр. Симића, била окићена у источњачком стилу, с неколико укусних и неукусних капија и својом шарено- ликошћу пружала је слику неког вашара. Полицијске мере су биле појачане, али је младотурски комитет сматрао да су недовољне, па је поред званичне, формирао и своју нарочиту полицију која је упоредо радила на султановој без- бедности.40 У Солуну га је, поред разних делегација, поздравила и српска војна делегација коју су чинили генерали Соларевић и Јовановић. 38 АС, МИД ППО, 1911, ред 373, ПП Бр. 684, Конзул Љуб. М. Михаиловић – Министар- ству иностраних дела Краљевине Србије, Приштина 10/23. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 607, 1117–1118. 39 АС, МИД ППО, 1911, ред 373, Бр. 79, Директор Српске гимнатије А Ст. Јотић – Настојатељу српске гимназије, Битољ 12/25. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 613, 1127–1128. 40 АС, МИД ППО, 1911, ред 373, ПП Бр. 276, Вицеконзул Јевр. Симић – Министарству иностраних дела Краљевине Србије, Солун 12/25. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 616, 1135. 41 ДСПКС, књ. IV, св. 3/II, док. 631, 1158. СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ По доласку у Со- лун су их сачекали султанов ађутант Зија-беј, пуковник Џавер-беј са неколико војника и почасном четом, као и шеф политичког одељења Кара Бибер. Иако се у почетку мислило да српска војна делегација неће наићи на пријатељски пријем, понашање султана и осталих представника османске државе је оповр- гло такво мишљење. Наиме, двојица највећих моћника, министар војни Ма- хмуд Шефкет-паша и најистакнутији члан у младотурском комитету др Назим били су најзаслужнији за такав пријем српске делегације. Они су слање српске војне делегације тумачили као израз пријатељских односа између Србије и Османског царства, што је било и те како битно у том тренутку, када су били заоштрени односи између Турске и Црне Горе.41 По повратку из Старе Србије и Македоније султану је био приређен свечани дочек, исто онако како је и био испраћен. Подаци из различитих 39 АС, МИД ППО, 1911, ред 373, Бр. 79, Директор Српске гимнатије А Ст. Јотић – Настојатељу српске гимназије, Битољ 12/25. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 613, 1127–1128. 40 АС, МИД ППО, 1911, ред 373, ПП Бр. 276, Вицеконзул Јевр. Симић – Министарству иностраних дела Краљевине Србије, Солун 12/25. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 616, 1135. 300 Весна С. Зарковић извора којима је располагао посланик Краљевине Србије у Цариграду Нена- довић нису уливали оптимизам нити су будили наду да ће се Арбанаси при- мирити и да ће завладати мир у тим крајевима. Султанова посета је имала за циљ да код Арбанаса пробуди осећај припадности исламској заједници, па је управо из тог разлога била организована заједничка молитва калифе са верницима, врло ретка у османској историји и великог религиозног значаја. Међутим, прави циљ није ни изблиза био постигнут, што се могло наслутити по разочарењу код владиних кругова и у турској штампи.42 Још се нису стишале приче о султановом одласку, а комисије за процену штете које су Арбанаси претрпели у сукобима са турским властима прет- ходне године, формиране током његове посете, увелико су радиле. Раније донета одлука да најмања надокнада износи пет, а највећа педесет турских лира, због незадовољства виђенијих Арбанаса, пре свега Исе Бољетица, из- мењена је подизањем горње границе којом се није ограничавала надокнада (Перуничић 1988: 459). Ова, као и друге сличне повластице које су турске власти чиниле, нису у потпуности задовољиле Арбанасе. 42 ДСПКС, књ. IV, св. 3/II, док. 620, 1141. 43 АС, МИД ППО, 1911, ред 373, ПП Бр. 401, Конзул Милан Ракић – Министарству иностраних дела Краљевине Србије, Приштина 12/25. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 615, 1131; (Ракић 1985: 251–254). 42 ДСПКС, књ. IV, св. 3/II, док. 620, 1141. СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ Иако су се организатори овог путовања на све начине трудили да при- добију Арбанасе и у неку руку их задовоље новцем, поклонима и амнести- рањем политичких криваца, нису постигли жељени ефекат. Међу њима је и даље постојало незадовољство које је претило да прерасте у отворене су- кобе са турским властима. Арбанаси су били много захтевнији и нису се за- довољавали понуђеним. Према речима Хасан-бега, посланика из Вучитрна, који је имао велики утицај на арбаншко становништво нарочито на Дрени- чане, они су захтевали: постављање за административне, војне и судске чи- новнике у областима насељеним Арбанасима, служење војске у својој земљи, отварање школа у „целој Албанији“ у којима би се учило на албанском језику са латинском азбуком, изградњу путева и железнице кроз Албанију. Иако су ови захтеви представљали зачетак аутономије, Хасан-бег је наглашавао да је из њиховог програма искључена аутономија и тврдио да „су они свесни тога да не могу живети слободни и независни ван граница турске државе“.43 Посета султана Махмуда V Решада није испунила свој циљ. Арбанаси нису били у потпуности задовољени и стално су истицали своје захтеве, а амнестија која им је дата негативно се одразила на српско становништво. Уследила су нова убиства, отмице, пљачке, паљевине и протеривање Срба. Од султановвог доласка, тачније од јула до новембра 1911. године у Старој Србији забележено је 128 крађа, 35 паљевина, 41 разбојништво, 53 отимачине, 30 уцена, 19 застра- шивања, 35 убистава, 37 покушаја убистава, 58 оружаних напада на имовину, Султаново путовање по Старој Србији и Македонији 301 27 примера туче и злостављања, 13, покушаја турчења и 18 примера наношења тешких повреда (Батаковић 1989: 277; Задужбине Косова 1987: 716). Замисао младотурског комитета, захваљујући којој је ово путовање ор- ганизовано, није била само тежња да учврсти своју власт и позицију, него и да скрене пажњу великим силама и одврати их од даљих мешања у зби- вања на овој територији. Иако је султан пре овог путовања обишао Бурсу и Једрене, много већи значај је имала посета Старој Србији и Македонији, које су непрекидно биле предмет интересовања великих сила. Младотуци су мислили да директан контакт са Арбанасима, посебно са оним муслиманске вероисповести, и заједничка молитва могу много утицати на њих, па је зато ова посета углавном имала верски карарктер. НЕОБЈАВЉЕНИ ИЗВОРИ Архив Југославије, Збирка Јован Јовановић Пижон. Архив Југославије, Збирка Јован Јовановић Пижон. Архив Србије, Министарство иностраних дела, Политичко–просветно одељење. ОБЈАВЉЕНИ ИЗВОРИ Алексић Пејковић–Џамбазовски 2009: Љиљана Алексић Пејковић – Климент Џамбазов- ски. Документи о спољној политици Краљевине Србије 1903–1914. Књига IV, свеска 3/I, Београд: Српска академија наука и уметности. ј Алексић Пејковић–Џамбазовски 2009: Љиљана Алексић Пејковић–Климент Џамбазовски. Документи о спољној политици Краљевине Србије 1903–1914. Књига IV, свеска 3/ II, Београд: Српска академија наука и уметности. Перуничић 1988: Бранко Перуничић. Сведочанство о Косову 1901–1913. Београд: Научна књига. Поповић 2007: Јанићије Поповић. Живот Срба на Косову 1812–1912. Грачаница: Никан Ракић 1985: Милан Ракић. Конзулска писма 1905–1911. Београд: Просвета. СУЛТАНОВО ПУТОВАЊЕ ПО СТАРОЈ СРБИЈИ И МАКЕДОНИЈИ ** Уједно, свечани дочек је тре- бало да пред Европом прикаже како се Арбанаси добровољно потчињавају султановој вољи и на тај начин спречи свако друго мешање споља, које је за Османско царство било непотребно и сувишно.44 Упркос таквом мишљењу младотурака, бројни поклони, повластице и уступци учињени од стране сул- тана нису испунили очекивања и жеље Арбанаса који су све више потпадали под страни утицај, пре свега Аустроугарске и Италије, што ће резултирати стварањем Албаније 1912. године. Политика, 1911. 2. 4; 31. 5; 44 АС, МИД ППО, 1911, ред 373, ПП Бр. 276, Вицеконзул Јевр. Симић – Министарству иностраних дела Краљевине Србије, Солун 12/25. јун 1911; ДСПКС, књ. IV, св. 3/II, док. 616, 1133–1134. 302 Весна С. Зарковић SULTAN’S JOURNEY THROUGH OLD SERBIA AND MACEDONIA Summary Рад је предат 1. марта 2019. године, а након мишљења рецензената, одлуком одговорног уредника Баштине, одобрен за штампу. ЛИТЕРАТУРА Батаковић 1989: Душан Т. Батаковић. „Анархија и геноцид над Србима 1897–1912“. Косово и Метохија у српској историји. Београд 1989: Српска књижевна задруга, 249–280. Батаковић 1989: Душан Т. Батаковић. „Анархија и геноцид над Србима 1897–1912“. Косово и Метохија у српској историји. Београд 1989: Српска књижевна задруга, 249–280. Богдановић 1985: Димитрије Богдановић. Књига о Косову. Београд: Српска академија наука и уметности. Задужбине Косова – споменици и знамења српског народа 1989. Призрен: Епархија рашко- призренска; Београд: Богословски факултет у Београду. Задужбине Косова – споменици и знамења српског народа 1989. Призрен: Епархија рашко- призренска; Београд: Богословски факултет у Београду. Задужбине Косова – споменици и знамења српског народа 1989. Призрен: Епархија рашко- призренска; Београд: Богословски факултет у Београду. Зарковић 2018: Весна Зарковић. „Побуне Арбанаса у Косовском вилајету 1911. године“. Баштина, св. 46. Приштина/Лепосавић: Институт за српску културу, 213–229. Зарковић 2018: Весна Зарковић. „Побуне Арбанаса у Косовском вилајету 1911. године“. Баштина, св. 46. Приштина/Лепосавић: Институт за српску културу, 213–229. Зарковић–Савић 2018: Весна Зарковић – Александар Савић. „Арбанашке побуне у Косов- ском вилајету 1910. године“. Баштина, 45. Лепосавић: Институт за српску култу- ру – Приштина, 291–313. Зарковић–Савић 2018: Весна Зарковић – Александар Савић. „Арбанашке побуне у Косов- ском вилајету 1910. године“. Баштина, 45. Лепосавић: Институт за српску култу- ру – Приштина, 291–313. ру р Мантран 2002: Робер Мантран. Историја Османског царства. Београд: Clio. Мантран 2002: Робер Мантран. Историја Османског царства. Београд: Clio. Микић 1988: Ђорђе Микић. Друштвене и економске прилике косовских Срба у XIX и поче- тком XX века (од чифчијства до банкарства). Београд: Српска академија наука и уметности. Микић 1988: Ђорђе Микић. Друштвене и економске прилике косовских Срба у XIX и поче- тком XX века (од чифчијства до банкарства). Београд: Српска академија наука и уметности. 303 Султаново путовање по Старој Србији и Македонији Key words: Ottoman Empire, Old Serbia, Macedonia, sultan, journey, Serbs, Albanians. Vesna S. ZARKOVIĆ SULTAN’S JOURNEY THROUGH OLD SERBIA AND MACEDONIA Vesna S. ZARKOVIĆ Summary Dissatisfaction of the Albanian population in Old Serbia and Macedonia during 1910 and 1911 turned into open conflicts with Turkish authorities. These conflicts resulted in a large number of wounded and killed, and a lot of houses were burnt and destroyed. Turkish authorities, knowing that the Albanians represent a fortress in the preservation of the Ottoman Empire in the Balkans, decided to organize a sultan’s visit to this part of the state. The visit represents the achieve- ment of the idea of Youth Turkish Committee which had the intention to calm down the Albanian population from one side, and to avoid the interference of great powers in internal questions of the state from another side. During his seventy days journey through Old Serbia and Macedonia sultan visited Thessaloniki, Skopje, Pristina and Bitolj, where he was receiving religious and civil representatives of all nationalities including the representatives of foreign countries in those places. Preparation which were done for the arrival of sultan supposed, among other things, the participa- tion of the Serbs. The presence of the Serbian military delegation in Thessaloniki was for Turks the expression of peaceful politics of the Kingdom of Serbia toward Ottoman Empire. Sultan’s visit to Old Serbia and Macedonia and touring of Murat’s turbet, where the decree on amnesty of all political culprits being in the country or outside was read, did not completely satisfy the Albanians. Their dissatisfaction affected Serbs the most. The Albanian fury increased every day and the requests multiplied. They included the appointment of the Albanians for ad- ministrative, military and judicial officers in the areas where they live, doing military service in their country, opening of schools in “the whole Albania” in which the teaching process will be conducted in Albanian language with Latin alphabet, building of roads and railways through Albania. These requests openly led toward the autonomy and creation of new state, which will come true after the First Balkan War in 1912 with dedication of Austria-Hungary and Italy. Key words: Ottoman Empire, Old Serbia, Macedonia, sultan, journey, Serbs, Albanians. Рад је предат 1. марта 2019. године, а након мишљења рецензената, одлуком одговорног уредника Баштине, одобрен за штампу.
https://openalex.org/W3037634746	https://www.nature.com/articles/s41598-020-79964-x.pdf	English	null	The overlap of genetic susceptibility to schizophrenia and cardiometabolic disease can be used to identify metabolically different groups of individuals	medRxiv (Cold Spring Harbor Laboratory)	2,020	cc-by	8,885	The overlap of genetic susceptibility to schizophrenia and cardiometabolic disease can be used to identify metabolically different groups of individuals N Rona J. Strawbridge1,2,3, Keira J. A. Johnston1,4,5, Mark E. S. Bailey5, Damiano Baldassarre6,7, Breda Cullen1, Per Eriksson3, Ulf deFaire8, Amy Ferguson1,9, Bruna Gigante3, Philippe Giral10, Nicholas Graham1, Anders Hamsten3, Steve E. Humphries11, Sudhir Kurl12, Donald M. Lyall1, Laura M. Lyall1, Jill P. Pell1, Matteo Pirro13, Kai Savonen14,15, Andries J. Smit16, Elena Tremoli7, Tomi‑Pekka Tomainen17, Fabrizio Veglia7, Joey Ward1, Bengt Sennblad18 & Daniel J. Smith1 Understanding why individuals with severe mental illness (Schizophrenia, Bipolar Disorder and Major Depressive Disorder) have increased risk of cardiometabolic disease (including obesity, type 2 diabetes and cardiovascular disease), and identifying those at highest risk of cardiometabolic disease are important priority areas for researchers. For individuals with European ancestry we explored whether genetic variation could identify sub-groups with different metabolic profiles. Loci associated with schizophrenia, bipolar disorder and major depressive disorder from previous genome-wide association studies and loci that were also implicated in cardiometabolic processes and diseases were selected. In the IMPROVE study (a high cardiovascular risk sample) and UK Biobank (general population sample) multidimensional scaling was applied to genetic variants implicated in both psychiatric and cardiometabolic disorders. Visual inspection of the resulting plots used to identify distinct clusters. Differences between these clusters were assessed using chi-squared and Kruskall-Wallis tests. In IMPROVE, genetic loci associated with both schizophrenia and cardiometabolic disease (but not bipolar disorder or major depressive disorder) identified three groups of individuals with distinct metabolic profiles. This grouping was replicated within UK Biobank, with somewhat less distinction between metabolic profiles. This work focused on individuals of European ancestry and is unlikely 1Institute of Health and Wellbeing, University of Glasgow, Room 111, Public Health, 1 Lilybank Gardens, Glasgow G12 8RZ, UK. 2Health Data Research, London, UK. 3Cardiovascular Medicine Unit, Department of Medicine Solna, Karolinska Institute, Stockholm, Sweden. 4Deanery of Molecular, Genetic and Population Health Sciences, College of Medicine and Veterinary Medicine, University of Edinburgh, Edinburgh, Scotland, UK. 5School of Life Sciences, College of Medical, Veterinary & Life Sciences, University of Glasgow, Glasgow, Scotland, UK. 6Department of Medical Biotechnology and Translational Medicine, Universit degli Studi di Milano, Milan, Italy. 7Centro Cardiologico Monzino, IRCCS, Milan, Italy. 8Cardiovascular and Nutritional Epidemiology, Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden. 9Usher Institute, University of Edinburgh, Edinburgh, UK. 10Service Endocrinologie‑Metabolisme, Groupe Hôpitalier Pitie‑Salpetriere, Unités de Prévention Cardiovasculaire, Assistance Publique - Hopitaux de Paris, Paris, France. 11Centre for Cardiovascular Genetics, Institute Cardiovascular Science, University College London, London, UK. www.nature.com/scientificreports www.nature.com/scientificreports Results h The IMPROVE and UK Biobank studies. The demographic characteristics of the IMPROVE, UK Biobank subsets 1 (UKB1) and 2 (UKB2) are provided in Table 1. At baseline, individuals in IMPROVE (a Euro- pean high cardiovascular-risk cohort) were older, more overweight and more likely to have T2D, hypertension or medication for hypertension or lipid-lowering medication than the UKB subsets (self-reported white British general population cohort). UKB1 and UKB2 were very similar, with lower frequency of hypertension at follow- up in UKB1 (51.5%) compared to UKB2 (62.0%) but slightly larger carotid Intima-media thickness (cIMT, indicative of vessel wall remodelling) measures in UKB2 to UKB1. Despite different proportions of UKB1 and UKB2 completing the mental health questionnaire, the frequencies of BD, MDD and GAD were similar. Figure 1 provides a schematic overview of the analysis procedure. SCZ‑CM loci can identify metabolically distinct groups of individuals in IMPROVE. When using IMPROVE and single nucleotide polymorphisms (SNPs) with a minor allele frequency (MAF) > 1%, implicated in both SCZ and CMD (SCZ-CMD), plotting the first two multi-dimensional scaling components (C1 and C2) demonstrated 3 groups of individuals (by visual inspection) (Fig. 2a). Separation was predominantly due to C1, and whilst C1 is nominally significantly correlated with latitude (rho = − 0.036, p = 0.0339), the clustering is not being driven by latitude (Supplementary Fig. 1). SNPs with MAF as low as 1% might differ across popula- tions (even within the same ancestry grouping), therefore robustness to MAF threshold also assessed. When using MAF > 5% showed additional groups (Fig. 2b), whereas MAF > 10% showed similar groups to MAF > 1% (Fig. 2c). Assignment to groups was consistent using MAF > 1% and MAF > 10% (Supplementary Table 1). The three groups appear to have modest differences in cardiometabolic profiles (Table 2): Group 3 had a significantly lower frequency of hypertension (group 3: 74% vs groups 1 or 2: 80% or 81% respectively, P = 0.004) and lower fastest progression of cIMT (group 3: 0.156 mm vs groups 1 or 2: 0.176 mm or 0.166 mm, P = 0.002). This is sur- prising given the (non-significant) higher rates of smoking in this group. Group 2 had (non-significantly) lower rates of T2D than the other groups (group 2: 25% vs groups 1 or 3: 28%). to apply to more genetically diverse populations. Overall, this study provides proof of concept that common biology underlying mental and physical illness may help to stratify subsets of individuals with different cardiometabolic profiles. to apply to more genetically diverse populations. Overall, this study provides proof of concept that common biology underlying mental and physical illness may help to stratify subsets of individuals with different cardiometabolic profiles. to apply to more genetically diverse populations. Overall, this study provides proof of concept that common biology underlying mental and physical illness may help to stratify subsets of individuals with different cardiometabolic profiles. Individuals with serious mental illness (such as schizophrenia (SCZ), major depressive disorder (MDD) and bipolar disorder (BD)) have a reduced life expectancy (10–15 years for BD, 15–20 years for SCZ1). This is likely due to the well-established increased prevalence of cardiovascular and metabolic disorders compared to the general population. For example, obesity is up to 3.5-fold higher in those with SCZ2, type 2 diabetes is ~ twofold higher in those with MDD, BD or SCZ2, and cerebrovascular disease is increased by up to 3.3-fold in those with BD2. Understanding this increased risk and identifying individuals at highest risk of metabolic and cardiovascular disease are important priority areas for researchers and healthcare providers. p p y p Historically, the increased risk and prevalence of cardiometabolic disease (CMD) has been attributed to social determinants and lifestyle factors (including poor diet, sedentary behaviour, alcohol and substance use) that co-exist with serious mental illness and effects of psychotropic medication2, however there is growing evidence that there might be common biological mechanisms underlying both mental and psychiatric illness. As genetic data is stable over an individual’s lifetime, and not influenced by disease course, genetic approaches are ideal for investigation of common biology in comorbid conditions. The identification of genetic variants robustly associ- ated with a wide range of psychiatric and cardiometabolic phenotypes by international genetics consortia has enabled the exploration of relationships between psychiatric and cardiometabolic conditions. Genome-wide genetic correlations between psychiatric and cardiometabolic traits provide evidence for underlying common biology. Correlations have been described between depression and obesity (rg = 0.12) or cardiovascular disease (rg = 0.42)3. Evidence of causal relationships between psychiatric and cardiometabolic traits have also been described1,4,5. However, the mechanisms involved have yet to be uncovered and therefore this knowledge has had no clinical impact. g p Here we tested whether a novel approach using multi-dimensional scaling (MDS) of genetic variation asso- ciated with psychiatric and cardiometabolic disorders could aid stratification of individuals into groups with differing cardiometabolic risk profiles. The overlap of genetic susceptibility to schizophrenia and cardiometabolic disease can be used to identify metabolically different groups of individuals N 12Institute of Public Health and Clinical Nutrition, University of Eastern Finland, Kuopio, Finland. 13Internal Medicine, Angiology and Arteriosclerosis Diseases, Department of Clinical and Experimental Medicine, University of Perugia, Perugia, Italy. 14Foundation for Research in Health Exercise and Nutrition, Kuopio Research Institute of Exercise Medicine, Kuopio, Finland. 15Department of Clinical Physiology and Nuclear Medicine, Kuopio University Hospital, Kuopio, Finland. 16Department of Medicine, University Medical Center Groningen and University of Groningen, Groningen, The Netherlands. 17Public Health and Clinical Nutrition, Department of Medicine, University of Eastern Finland, Kupiou, Finland. 18Department of Cell and Molecular Biology, National Bioinformatics Infrastructure Sweden, Science for Life Laboratory, Uppsala University, Uppsala, Sweden. email: rona.strawbridge@glasgow.ac.uk \| https://doi.org/10.1038/s41598-020-79964-x Scientific Reports \| (2021) 11:632 www.nature.com/scientificreports/ Results h Similar groups were observed using T-distributed Stochastic Neighbour Embedding (tSNE) or principal component analyses (PCA, Supplementary Methods), with the majority of individuals being consistently grouped together (Supplementary Figs. 2 and 3, respectively). p y) This result appears specific to SCZ-CMD SNP subset; no separation into groups was observed when using MDD-CMD SNPs, irrespective of the MAF filter used (Fig. 2d–f). For BD-CMD SNPs (Fig. 2g–i), grouping is apparent at MAF > 1%, but not when MAF > 5% or 10% were considered. Validation of method and sensitivity testing of clustering in UKB1. In order to assess whether MDS analysis of SCZ-CMD SNPs could reproducibly identify three groups of individuals, validation of the method was attempted in UKB1. Firstly, to directly replicate the analysis conducted in IMPROVE (Fig. 3a,b), the post-filtering SNPs from IMPROVE were used (Fig. 3c); however the grouping is not convincing as there is little separation between the groups. Secondly, to assess robustness of the method to differences in MAF and LD structure between populations, the SCZ-CMD SNPs were filtered for MAF and LD in UKB1. As noted in Fig. 1, the majority of SNPs included in the two approaches were the same. Unsurprisingly, the SNPs that differed were https://doi.org/10.1038/s41598-020-79964-x Scientific Reports \| (2021) 11:632 \| www.nature.com/scientificreports/ Table 1. Results h Demographic characteristics of IMPROVE and UKB pa estimates of treatment-naïve levels, as per Ehret etl al; na, not avai IMPROVE UKB1 UKB2 Nmax 3300 2202 20,182 Male (%) 1695 (51.4) 1042 (47.3) 9759 (48.4) Baseline Age (years) 64.2 (5.4) 55.7 (7.6) 55.2 (7.5) Weight (kg) 76.7 (15.1) 76.1 (14.5) 76.9 (14.8) Waist (cm) 94 (13) 88 (12) 88 (13) Hip (cm) 102 (10) 102 (8) 102 (8) WHR 0.92 (0.09) 0.86 (0.09) 0.86 (0.09) BMI (kg/m2) 27.3 (4.2) 26.4 (3.9) 26.6 (4.2) SBP (mmHg) 142 (18) 136 (18) 136 (18) DBP (mmHg) 82 (10) 82 (10) 82 (10) SBP* (mmHg) 151 (21) 138 (19) 138 (19) DBP* (mmHg) 88 (11) 83 (11) 83 (11) T2D 880 (26.7) 46 (2.1) 45 (2.2) HTN 2634 (79.8) 974 (45.0) 8765 (45.2) HTN medication 1904 (57.7) 332 (15.2) 2820 (14.0) Lipid-lowering medication 1623 (49.2) 172 (18.4) 1677 (18.9) ISH 0 (0) 21 (1.0) 259 (1.3) IMTmean (mm) 0.891 (0.199) na na IMTmax (mm) 2.037 (0.813) na na Current smoking 498 (15.1) 139 (6.3) 1234 (6.1) Former smoking 1216 (36.9) 746 (33.9) 6656 (33.0) Follow-up Age 66.7 (5.4) 61.8 (7.5) 63.2 (7.5) Weight na 75.6 (14.6) 76.3 (15.0) Waist na 86 (12) 88 (12) Hip na 100 (8) 101 (9) WHR na 0.86 (0.08) 0.87 (0.09) BMI na 26.4 (4.0) 26.5 (4.4) SBP na 138 (20) 137 (18) DBP na 83 (11) 79 (10) SBP* na 140 (20) 141 (20) DBP* na 81 (11) 81 (11) T2D na 79 (3.6) 867 (4.3) HTN na 1121 (51.5) 12,414 (62.0) HTN medication na 466 (21.2) 4568 (22.7) Lipid-lowering medication na 408 (22.3) 4557 (26.4) ISH 119 (3.6) 43 (2.0) 333 (1.7) IMTmean (mm) na 0.672 (0.119) 0.682 (0.125) IMTmax (mm) na 0.888 (0.182) 0.914 (0.297) Progression of IMTmean 0.0186 (0.032) na na Progression of IMTmax 0.0439 (0.163) na na Current smoking na 99 (4.6) 708 (3.5) Former smoker na 742 (34.1) 6800 (33.9) MHQ Nmax (% of group) na 1528 (69.4) 10,079 (49.9) BD na 28 (1.8) 196 (1.4) MDD na 410 (29.9) 3512 (28.6) Table 1. Demographic characteristics of IMPROVE and UKB participants. Where: , adjusted to provide estimates of treatment-naïve levels, as per Ehret etl al; na, not available. mainly those with MAF < 10%. Using SCZ-CMD and conducting MAF and LD filtering in UKB1, nine groups are evident when using SNPs with MAF > 1% (Fig. Figure 1. Schematic of the analysis procedure used to identify clusters. Figure 1. Schematic of the analysis procedure used to identify clusters. were seen (Fig. 4a and Table 3). This is unsurprising, given that it is a smaller cohort with a lower cardiovascular burden. were seen (Fig. 4a and Table 3). This is unsurprising, given that it is a smaller cohort with a lower cardiovascular burden. Validation of metabolic differences between clusters in UKB2. In an attempt to replicate the clus- tering and validate the metabolic differences between groups, the larger UKB2 subset was analysed. As filter- ing with MAF > 10% and 1% gave similar clusters, filtering with MAF > 10% was applied as it is more likely to generalise to other populations. Again, three major groups were identified (Fig. 4b), similar to those identified in IMPROVE and UKB1. Additional clusters between the major three groups were apparent, but they account for ~ 7% of the studied population, and were omitted from the groups.if Consistent with the IMPROVE study, clinically modest (and statistically significant) differences were observed in baseline SBP, SDP adjusted for blood-pressure medication, and frequency of hypertension and T2D (Table3). These effects were not observed at follow-up, potentially due to lifestyle or medications changes in response to baseline observations. It was also noted that the frequency of MDD but not BD differed between the groups. The number of SCZ in UKB2 is too low to provide meaningful statistics. Impact of MDD/BD on clusters. As phenotypes and genetic loci for SCZ overlap with those for MDD and BD, it is perhaps unsurprising to see that the clusters include different proportions of individuals with MDD. To investigate whether these individuals were driving the clustering, the process was repeated in those without BD/ MDD separately from those with these diagnoses (using SNPs with MAF > 10%). In those without mental illness, similar to the overall UKB2, there were there main groups, intermediate clusters accounting for 7.4% of the sam- ple (Fig. 5a). In those with mental illness the three clusters were observed, with better between-group separation and only 1.3% of the sample being ungrouped (Fig. 5b). Small but significant differences between groups were observed for blood pressure measures and rates of hypertension, in both those with and without mental illness (Supplementary Table 2). These results suggest that this method is applicable to the general population, as well as those with increased genetic burden for mental illness. Results h 3d), whereas three groups are observed when using SNPs with MAF > 10% (Fig. 3e). When comparing the metabolic profiles of the 3 groups, no significant differences https://doi.org/10.1038/s41598-020-79964-x Scientific Reports \| (2021) 11:632 \| www.nature.com/scientificreports/ Figure 1. Schematic of the analysis procedure used to identify clusters. Figure 1. Schematic of the analysis procedure used to identify clusters. All genetic loci associated with SCZ do not identify clusters in UKB. To determine whether it is common biology (ie. Overlap in loci for SCZ and CMD) per se, rather than SCZ in general that drives the clus- Scientific Reports \| (2021) 11:632 \| https://doi.org/10.1038/s41598-020-79964-x www.nature.com/scientificreports/ Figure 2. Results of MDS analysis in IMPROVE, using the loci in common between CMD and SCZ with (a) MAF > 1%, (b) MAF > 5% or (c) MAF > 10%; CMD and MDD with (d) MAF > 1%, (e) MAF > 5% or (f) MAF > 10%; CMD and BD with (g) MAF > 1%, (h) MAF > 5% or (i) MAF > 10%. Each data point is an individual therefore the individuals who are closer together are more genetically similar. Figure 2. Results of MDS analysis in IMPROVE, using the loci in common between CMD and SCZ with (a) MAF > 1%, (b) MAF > 5% or (c) MAF > 10%; CMD and MDD with (d) MAF > 1%, (e) MAF > 5% or (f) MAF > 10%; CMD and BD with (g) MAF > 1%, (h) MAF > 5% or (i) MAF > 10%. Each data point is an individual therefore the individuals who are closer together are more genetically similar. tering, the same procedure was followed using all SNPs in loci associated with SCZ in UKB2, with the same MAF and LD filtering being applied prior to MDS analysis. As shown in Supplementary Fig. 4, SNPs in loci associated with SCZ do not separate individuals into groups. A further “negative control” experiment was conducted in UKB2. When repeating the analysis using the genetic loci (Supplementary Table 4) associated with eye colour6,7, there was no evidence of subgroups (Supplementary Fig. 5). These results confirm that it is the overlap of SCZ and CMD loci (rather than a methodological artefact), and therefore probably common biological mechanisms, which are driving the clustering. Discussionh 1 2 3 P* N 1222 (36.2) 1629 (48.2) 526 (15.6) Men (%) 581 (47.6) 842 (51.7) 268 (51.0) 0.954 Age (years) 64.2 (5.4) 64.2 (5.4) 64.4 (5.4) 0.558 BMI (kg/m2) 27.4 (4.3) 27.2 (4.2) 27.0 (4.2) 0.259 Waist (cm) 94.5 (12.8) 93.8 (12.4) 93.9 (12.6) 0.298 Waist_hip 0.92 (0.09) 0.92 (0.09) 0.94 (12.6) 0.859 SBP (mmHg) 142 (18) 142 (19) 140 (18) 0.070 DBP (mmHg) 82 (10) 82 (10) 82 (10) 0.599 HTN 982 (80.4) 1321 (81.1) 389 (74.0) 0.004 HTN medication 694 (56.8) 805 (58.1) 291 (55.3) 0.421 SBP* (mmHg)* 151 (20) 151 (21) 148 (21) 0.080 DBP* (mmHg)* 88 (11) 88 (11) 87 (12) 0.368 NSAIDs 253 (20.7) 260 (18.8) 102 (19.4) 0.622 Current smoking 172 (14.1) 246 (15.1) 87 (16.5) 0.614 Pack years 10.7 (17.2) 10.7 (15.9) 12.3 (19.7) 0.191 T2D 342 (28.0) 414 (25.4) 145 (27.6) 0.252 Lipid-lowering medication 585 (47.9) 828 (50.9) 256 (48.8) 0.159 Framingham risk score 0.27 (0.16) 0.27 (0.16) 0.27 (0.16) 0.743 Cardiac event 74 (6.1) 94 (5.8) 22 (4.2) 0.208 Baseline CC-IMTmean 0.740 (0.130) 0.745 (0.147) 0.749 (0.155) 0.962 IMTmean 0.895 (0.198) 0.888 (0.199) 0.888 (0.205) 0.296 CC-IMTmax 1.184 (0.368) 1.197 (0.400) 1.222 (0.456) 0.705 IMTmax 2.051 (0.812) 2.026 (0.803) 2.040 (0.849) 0.502 IMTmeanmax 1.258 (0.295) 1.252 (0.298) 1.247 (0.306) 0.300 Diameter 7.847 (0.854) 7.834 (0.862) 7.796 (0.839) 0.441 Progression CC-IMTmean 0.009 (0.029) 0.009 (0.027) 0.006 (0.026) 0.193 IMTmean 0.019 (0.032) 0.019 (0.032) 0.017 (0.033) 0.508 CC-IMTmax 0.016 (0.096) 0.017 (0.093) 0.007 (0.098) 0.281 IMTmax 0.049 (0.164) 0.042 (0.156) 0.035 (0.178) 0.209 IMTmeanmax 0.026 (0.053) 0.026 (0.051) 0.023 (0.054) 0.240 IMTfastest 0.176 (0.149) 0.166 (0.139) 0.156 (0.147) 0.002 Diameter 0.003 (0.030) 0.005 (0.033) 0.003 (0.029) 0.749 Table 2. Demographic characteristics of the IMPROVE participants, by cluster (MAF > 10%). Highlighted in bold are the significant (p < 0.05) differences between groups. Where: , adjusted to provide estimates of treatment-naïve levels as per Ehret et al.; Statistical analyses compared levels or frequncies across groups 1, 2 and 3. Ungrouped (.) were omitted from the analyses). P for Pearsons chi square for categorical variables and Kruskal–Wallis for continuous variables; na, not available. It is interesting that the analyses using BD and MDD genetic loci did not enable clustering of individuals in the same way as was observed for SCZ, particularly given that BD and SCZ demonstrate an overlap in genetic loci. Discussionh 1 2 3 P* N 1222 (36.2) 1629 (48.2) 526 (15.6) Men (%) 581 (47.6) 842 (51.7) 268 (51.0) 0.954 Age (years) 64.2 (5.4) 64.2 (5.4) 64.4 (5.4) 0.558 BMI (kg/m2) 27.4 (4.3) 27.2 (4.2) 27.0 (4.2) 0.259 Waist (cm) 94.5 (12.8) 93.8 (12.4) 93.9 (12.6) 0.298 Waist_hip 0.92 (0.09) 0.92 (0.09) 0.94 (12.6) 0.859 SBP (mmHg) 142 (18) 142 (19) 140 (18) 0.070 DBP (mmHg) 82 (10) 82 (10) 82 (10) 0.599 HTN 982 (80.4) 1321 (81.1) 389 (74.0) 0.004 HTN medication 694 (56.8) 805 (58.1) 291 (55.3) 0.421 SBP* (mmHg)* 151 (20) 151 (21) 148 (21) 0.080 DBP* (mmHg)* 88 (11) 88 (11) 87 (12) 0.368 NSAIDs 253 (20.7) 260 (18.8) 102 (19.4) 0.622 Current smoking 172 (14.1) 246 (15.1) 87 (16.5) 0.614 Pack years 10.7 (17.2) 10.7 (15.9) 12.3 (19.7) 0.191 T2D 342 (28.0) 414 (25.4) 145 (27.6) 0.252 Lipid-lowering medication 585 (47.9) 828 (50.9) 256 (48.8) 0.159 Framingham risk score 0.27 (0.16) 0.27 (0.16) 0.27 (0.16) 0.743 Cardiac event 74 (6.1) 94 (5.8) 22 (4.2) 0.208 Baseline CC-IMTmean 0.740 (0.130) 0.745 (0.147) 0.749 (0.155) 0.962 IMTmean 0.895 (0.198) 0.888 (0.199) 0.888 (0.205) 0.296 CC-IMTmax 1.184 (0.368) 1.197 (0.400) 1.222 (0.456) 0.705 IMTmax 2.051 (0.812) 2.026 (0.803) 2.040 (0.849) 0.502 IMTmeanmax 1.258 (0.295) 1.252 (0.298) 1.247 (0.306) 0.300 Diameter 7.847 (0.854) 7.834 (0.862) 7.796 (0.839) 0.441 Progression CC-IMTmean 0.009 (0.029) 0.009 (0.027) 0.006 (0.026) 0.193 IMTmean 0.019 (0.032) 0.019 (0.032) 0.017 (0.033) 0.508 CC-IMTmax 0.016 (0.096) 0.017 (0.093) 0.007 (0.098) 0.281 IMTmax 0.049 (0.164) 0.042 (0.156) 0.035 (0.178) 0.209 IMTmeanmax 0.026 (0.053) 0.026 (0.051) 0.023 (0.054) 0.240 IMTfastest 0.176 (0.149) 0.166 (0.139) 0.156 (0.147) 0.002 Diameter 0.003 (0.030) 0.005 (0.033) 0.003 (0.029) 0.749 Table 2. Demographic characteristics of the IMPROVE participants, by in bold are the significant (p < 0.05) differences between groups. Where treatment-naïve levels as per Ehret et al.; Statistical analyses compared l and 3. Ungrouped (.) were omitted from the analyses). P for Pearsons Kruskal–Wallis for continuous variables; na, not available. Discussionh 1 2 3 P N 1222 (36.2) 1629 (48.2) 526 (15.6) Men (%) 581 (47.6) 842 (51.7) 268 (51.0) 0.954 Age (years) 64.2 (5.4) 64.2 (5.4) 64.4 (5.4) 0.558 BMI (kg/m2) 27.4 (4.3) 27.2 (4.2) 27.0 (4.2) 0.259 Waist (cm) 94.5 (12.8) 93.8 (12.4) 93.9 (12.6) 0.298 Waist_hip 0.92 (0.09) 0.92 (0.09) 0.94 (12.6) 0.859 SBP (mmHg) 142 (18) 142 (19) 140 (18) 0.070 DBP (mmHg) 82 (10) 82 (10) 82 (10) 0.599 HTN 982 (80.4) 1321 (81.1) 389 (74.0) 0.004 HTN medication 694 (56.8) 805 (58.1) 291 (55.3) 0.421 SBP* (mmHg)* 151 (20) 151 (21) 148 (21) 0.080 DBP* (mmHg)* 88 (11) 88 (11) 87 (12) 0.368 NSAIDs 253 (20.7) 260 (18.8) 102 (19.4) 0.622 Current smoking 172 (14.1) 246 (15.1) 87 (16.5) 0.614 Pack years 10.7 (17.2) 10.7 (15.9) 12.3 (19.7) 0.191 T2D 342 (28.0) 414 (25.4) 145 (27.6) 0.252 Lipid-lowering medication 585 (47.9) 828 (50.9) 256 (48.8) 0.159 Framingham risk score 0.27 (0.16) 0.27 (0.16) 0.27 (0.16) 0.743 Cardiac event 74 (6.1) 94 (5.8) 22 (4.2) 0.208 Baseline CC-IMTmean 0.740 (0.130) 0.745 (0.147) 0.749 (0.155) 0.962 IMTmean 0.895 (0.198) 0.888 (0.199) 0.888 (0.205) 0.296 CC-IMTmax 1.184 (0.368) 1.197 (0.400) 1.222 (0.456) 0.705 IMTmax 2.051 (0.812) 2.026 (0.803) 2.040 (0.849) 0.502 IMTmeanmax 1.258 (0.295) 1.252 (0.298) 1.247 (0.306) 0.300 Diameter 7.847 (0.854) 7.834 (0.862) 7.796 (0.839) 0.441 Progression CC-IMTmean 0.009 (0.029) 0.009 (0.027) 0.006 (0.026) 0.193 IMTmean 0.019 (0.032) 0.019 (0.032) 0.017 (0.033) 0.508 CC-IMTmax 0.016 (0.096) 0.017 (0.093) 0.007 (0.098) 0.281 IMTmax 0.049 (0.164) 0.042 (0.156) 0.035 (0.178) 0.209 IMTmeanmax 0.026 (0.053) 0.026 (0.051) 0.023 (0.054) 0.240 IMTfastest 0.176 (0.149) 0.166 (0.139) 0.156 (0.147) 0.002 Diameter 0.003 (0.030) 0.005 (0.033) 0.003 (0.029) 0.749 www.nature.com/scientificreports/ Table 2. Demographic characteristics of the IMPROVE participants, by cluster (MAF > 10%). Highlighted in bold are the significant (p < 0.05) differences between groups. Where: , adjusted to provide estimates of treatment-naïve levels as per Ehret et al.; Statistical analyses compared levels or frequncies across groups 1, 2 and 3. Ungrouped (.) were omitted from the analyses). P for Pearsons chi square for categorical variables and Kruskal–Wallis for continuous variables; na, not available. Discussionh This study provides proof of principle that, using the genetic overlap between SCZ and cardiometabolic disor- ders, subsets of European ancestry individuals with different metabolic profiles can be identified. These findings support the existence of mechanisms common to SCZ and blood pressure regulation.h pp p g The discovery cohort IMPROVE deliberately recruited to identify genes and biomarkers associated with the risk of cardiovascular diseases, at a time when psychiatric disorders were typically excluded from non-psychiatric studies, therefore only a portion of the spectrum of psychiatric genetic burden is represented. In contrast, UKB1 and UKB2 are general population cohorts and therefore have a wider spectrum of both psychiatric and cardiometabolic disorder genetic burden, although it is recognised that the recruitment skews this distribution towards to the healthier segment of the population8. It is therefore both striking that the grouping was present in IMPROVE, and unsurprising that the blood pressure and hypertension differences between groups were more modest in UKB2 than those in IMPROVE.f It is worth noting that similar groups were observed in the IMPROVE cohort, using three different methods and (where applicable) exploring a variety of parameter settings. This suggests that the grouping is robust. The metabolic profiles of the groups did not completely agree between the 3 cohorts, however the repeated observa- tion of between-group differences in T2D and blood pressure/hypertension deserves further attention. If the method can be refined to better identify whether an individual is at increased risk of either hypertension or T2D would be of immense value. Even if the method is only robust in high CMD-risk populations (such as those with family history, multiple risk factors or psychiatric diagnoses), it could be of clinical importance. Scientific Reports \| (2021) 11:632 \| https://doi.org/10.1038/s41598-020-79964-x ficreports/ Table 2. Demographic characteristics of the IMPROVE participants, by cluster (MAF > 10%). Highlighted in bold are the significant (p < 0.05) differences between groups. Where: , adjusted to provide estimates of treatment-naïve levels as per Ehret et al.; Statistical analyses compared levels or frequncies across groups 1, 2 and 3. Ungrouped (.) were omitted from the analyses). *P for Pearsons chi square for categorical variables and Kruskal–Wallis for continuous variables; na, not available. Discussionh There are several possible explanations for this, most notably the ability to identify genetic loci for each mental illness: SCZ is clinically a more severe phenotype with diagnostic criteria that are relatively specific (for exam- ple psychotic episodes). In comparison, MDD spans a wide spectrum severity, with phenotypic heterogeneity potentially diluting or obscuring some true genetic effects. Whilst BD can be considered an intermediate (some symptoms more severe than MDD, most are less severe than for SCZ) diagnostic criteria for MDD and BD over- lap to a large degree as both involve episodes of depression, meaning that there is potential for misdiagnosis and therefore dilution of genetic effects for either trait. Another explanation is that the mechanisms leading to CMD in SCZ differ from those in MDD or BD, with processes that are represented on the CardioMetabo and Immuno chips failing to capture some pathological mechanisms. With this in mind, the finding of different frequencies of MDD in the groups was not anticipated, as the MDD genetics did not achieve any form of grouping, and the overlap of MDD and SCZ genetics is modest. However, MDD is highly heterogeneous, therefore it would be of interest to further explore whether there are any differences between the MDD cases in each group, specifically whether any of the groups corresponds to the recently proposed atypical depression subtype9,10. Genetic correlation analyses have begun to explore the common biology and causal relationships between psychiatric and cardiometabolic diseases1,3,10, however these methods assume that the entire genome influences both sets of traits. The small to moderate correlations could suggest that it is only a portion of the genome that has common effects. In contrast, the current study focuses on only the parts of the genome that have been implicated Scientific Reports \| (2021) 11:632 \| https://doi.org/10.1038/s41598-020-79964-x ww.nature.com/scientificreports/ b h h d C h l h d d b l d d h h Figure 3. Sensitivity testing in UKB1. For comparison, IMPROVE MDS analysis using (a) MAF > 1% and (b) MAF > 10%. MDS analysis in UKB1 using (c) the same post-filtering SNPs as for IMPROVE, (d) the same pre-filtering SNPs with MAF > 1% in UKB1 and (e) the same pre-filtering SNPs with MAF > 10% in UKB1. Each data point is an individual therefore the individuals who are closer together are more genetically similar. www.nature.com/scientificreports/ Figure 3. Sensitivity testing in UKB1. Discussionh For comparison, IMPROVE MDS analysis using (a) MAF > 1% and (b) MAF > 10%. MDS analysis in UKB1 using (c) the same post-filtering SNPs as for IMPROVE, (d) the same pre-filtering SNPs with MAF > 1% in UKB1 and (e) the same pre-filtering SNPs with MAF > 10% in UKB1. Each data point is an individual therefore the individuals who are closer together are more genetically similar. rison, IMPROVE MDS analysis using (a) MAF > 1% and Figure 3. Sensitivity testing in UKB1. For comparison, IMPROVE MDS analysis using (a) MAF > 1% and (b) MAF > 10%. MDS analysis in UKB1 using (c) the same post-filtering SNPs as for IMPROVE, (d) the same pre-filtering SNPs with MAF > 1% in UKB1 and (e) the same pre-filtering SNPs with MAF > 10% in UKB1. Each data point is an individual therefore the individuals who are closer together are more genetically similar. in both psychiatric and CMD. Whilst this study does not bring us any closer to understanding the mechanisms underlying the common pathological mechanisms, it does suggest that exploration of the SCZ-CMD loci could have clinical utility, irrespective of mechanistic understanding. One limitation is that these analyses were conducted in individuals of European ancestry and as SNPs were filtered by MAF and linkage disequilibrium, it is not possible to generalise them to other populations. Indeed, to apply current information from European ancestry individuals to additional ancestry groups has the potential to be misleading and is certainly incomplete. Whilst there is a recognised need11 and growing efforts around the world to explore genetics of disease in non-European ancestry individuals, it will take time to gain full insight into the genetic architecture of diseases in these ancestry groups. Scientific Reports \| (2021) 11:632 \| https://doi.org/10.1038/s41598-020-79964-x www.nature.com/scientificreports/ Figure 4. Comparison of the three clusters identified in (a) UKB1 and (b) UKB2 (lower panel). Each data point is an individual therefore the individuals who are closer together are more genetically similar. Figure 4. Comparison of the three clusters identified in (a) UKB1 and (b) UKB2 (lower panel). Each data point is an individual therefore the individuals who are closer together are more genetically similar. Figure 4. Comparison of the three clusters identified in (a) UKB1 and (b) UKB2 (lower panel). Each data point is an individual therefore the individuals who are closer together are more genetically similar. Figure 4. Discussionh Comparison of the three clusters identified in (a) UKB1 and (b) UKB2 (lower panel). Each data point is an individual therefore the individuals who are closer together are more genetically similar. https://doi.org/10.1038/s41598-020-79964-x Scientific Reports \| (2021) 11:632 \| www.nature.com/scientificreports/ Table 3. Demographic characteristics of the UKB1 and UKB2 participants, by cluster. Highlighted in bold are the significant (p < 0.05) differences between groups. Where: , adjusted to provide estimates of treatment- naïve levels; na, not available; Statistical analyses compared levels or frequencies across groups 1, 2 and 3. Discussionh Ungrouped ( ) were omitted from the analyses) P for Pearson’s chi square for categorical variables and UKB1 (N = 2,202) UKB2 (N = 20,181) Group 1 2 3 P 1 2 3 P** N 443 (20.1) 1054 (47.9) 622 (28.2) 83 (3.8) 5972 (29.6) 9926 (45.7) 3558 (17.6) 1425 (7.1) Male (%) 205 (46.3) 499 (47.3) 300 (48.2) 38 (45.8) 0.788 2837 (47.5) 4459 (48.3) 1745 (49.0) 718 (50.4) 0.323 Baseline Age (years) 55.6 (7.7) 56.0 (7.4) 55.4 (7.7) 54.6 (7.9) 0.121 55.1 (7.4) 55.1 (7.5) 55.4 (7.5) 55.2 (7.5) 0.172 Weight (kg) 76.0 (14.2) 75.7 (14.2) 76.6 (15.2) 76.9 (14.9) 0.805 76.7 (15.0) 77.0 (14.7) 77.0 (14.6) 77.6 (15) 0.352 Waist (cm) 87 (13) 87 (12) 88 (13) 88 (12) 0.883 88 (13) 88 (13) 88 (13) 89 (13) 0.418 Hip (cm) 102 (8) 101 (8) 102 (8) 102 (8) 0.581 102 (8) 102 (8) 102 (8) 102 (8) 0.681 WHR 0.86 (0.09) 0.86 (0.09) 0.86 (0.09) 0.86 (0.09) 0.974 0.86 (0.09) 0.86 (0.09) 0.86 (0.09) 0.86 (0.09) 0.271 BMI (kg/m2) 26.5 (3.9) 26.3 (3.9) 26.4 (4.0) 26.4 (3.7) 0.492 26.6 (4.4) 26.6 (4.2) 26.6 (4.1) 26.8 (4.2) 0.506 SBP (mmHg) 135 (17) 136 (18) 135 (18) 133 (17) 0.822 135 (17) 136 (18) 136 (18) 136 (18) 0.002 DBP (mmHg) 81 (10) 81 (10) 81 (10) 80 (11) 0.822 81 (10) 82 (10) 82 (10) 82 (10) 0.060 SBP* (mmHg) 138 (20) 138 (19) 138 (20) 134 (17) 0.774 137 (19) 138 (20) 138 (19) 138 (20) 0.016 DBP* (mmHg) 83 (11) 83 (11) 83 (11) 81 (11) 0.550 83 (11) 83 (11) 83 (11) 83 (11) 0.164 T2D 14 (3.2) 18 (1.7) 13 (2.1) 1 (1.2) 0.208 154 (2.6) 192 (2.1) 67 (1.9) 32 (2.3) 0.044 HTN 197 (44.9) 457 (44.2) 285 (46.3) 35 (44.3) 0.931 2491 (43.6) 4097 (46.3) 1534 (44.6) 643 (46.6) 0.005 HTN medication 75 (17.1) 158 (15.1) 91 (14.7) 8 (9.8) 0.395 817 (13.7) 1273 (13.9) 515 (14.5) 215 (15.2) 0.530 Lipid-lowering medication 43 (23.5) 80 (17.8) 45 (17.0) 4 (11.1) 0.174 509 (19.7) 741 (18.3) 292 (18.4) 135 (20.1) 0.344 ISH 4 (0.9) 10 (1.0) 6 (1.0) 1 (1.2) 0.361 81 (1.4) 118 (1.3) 47 (1.3) 13 (0.9) 0.920 Current smoking 29 (10.1) 63 (9.3) 45 (9.3) 7 (8.4) 0.900 375 (9.4) 586 (9.5) 193 (8.03) 80 (5.6) 0.092 Former smoking 185 (41.9) 436 (41.4) 252 (36.4) 32 (38.6) 0.072 2341 (39.3) 3622 (39.3) 1340 (37.8) 587 (41.2) 0.221 Follow-up Age (years) 61.7 (7.6) 62.0 (7.3) 61.5 (7.6) 60.9 (7.6) 0.069 61.2 (7.4) 63.2 (7.5) 63.4 (7.5) 63.2 (7.5) 0.190 Weight (kg) 75.8 (14.3) 75.3 (14.3) 75.8 (15.4) 77.4 (14.5) 0.770 76.1 (15.2) 76.4 (14.9) 76.3 (14.8) 76.9 (15.1) 0.225 Waist (cm) 86 (12) 86 (12) 86 (12) 88 (11) 0.943 88 (13) 88 (12) 88 (12) 89 (12) 0.474 Hip (cm) 101 (8) 100 (8) 100 (8) 101 (8) 0.402 101 (9) 101 (9) 101 (8) 101 (9) 0.620 WHR 0.85 (0.08) 0.86 (0.08) 0.86 (0.08) 0.87 (0.08) 0.768 0.87 (0.09) 0.87 (0.09) 0.87 (0.09) 0.87 (0.08) 0.797 BMI (kg/m2) 26.6 (4.0) 26.3 (4.0) 26.3 (4.0) 26.8 (3.7) 0.457 26.5 (4.5) 26.6 (4.4) 26.5 (4.2) 26.7 (4.3) 0.511 SBP (mmHg) 137 (17) 138 (18) 137 (18) 134 (16) 0.332 137 (18) 138 (18) 137 (18) 137 (18) 0.385 DBP (mmHg) 79 (10) 79 (10) 78 (10) 78 (9) 0.381 78 (10) 79 (19) 78 (10) 78 (10) 0.171 SBP* (mmHg) 140 (20) 141 (20) 140 (21) 136 (17) 0.428 141 (20) 141 (20) 141 (20) 141 (20) 0.466 DBP* (mmHg) 81 (11) 81 (11) 81 (12) 80 (10) 0.370 81 (11) 81 (11) 81 (11) 81 (11) 0.316 T2D 22 (5.0) 32 (3.0) 23 (3.7) 2 (2.4) 0.186 267 (4.5) 387 (4.2) 149 (4.2) 64 (4.5) 0.682 HTN 225 (51.6) 556 (53.3) 304 (49.4) 36 (43.4) 0.127 3647 (61.6) 5727 (62.6) 2151 (61.0) 889 (62.8) 0.192 HTN medication 95 (21.5) 222 (21.2) 136 (21.9) 13 (15.7) 0.978 1328 (22.3) 2111 (23.0) 805 (22.7) 324 (22.9) 0.649 Lipid-lowering medication 94 (25.2) 192 (21.9) 107 (20.9) 15 (21.2) 0.341 1374 (26.8) 2056 (26.0) 795 (26.1) 332 (27.2) 0.608 ISH 7 (1.6) 24 (2.3) 8 (1.3) 4 (4.9) 0.491 112 (1.9) 152 (1.7) 53 (1.5) 16 (1.1) 0.334 IMTmean (mm) 0.673 (0.114) 0.674 (0.118) 0.670 (0.122) 0.663 (0.140) 0.475 0.680 (0.123) 0.682 (0.126) 0.686 (0.127) 0.686 (0.127) 0.205 IMTmax (mm) 0.884 (0.172) 0.891 (0.183) 0.886 (0.182) 0.880 (0.217) 0.817 0.911 (0.201) 0.914 (0.208) 0.921 (0.212) 0.916 (0.204) 0.242 Current smoking 17 (6.1) 45 (6.7) 40 (8.2) 2 (2.4) 0.456 221 (5.6) 326 (5.4) 108 (4.6) 53 (3.7) 0.183 Former smoking 177 (40.5) 412 (39.7) 247 (35.8) 26 (31.3) 0.179 2231 (37.7) 3439 (37.6) 1279 (36.4) 559 (39.2) 0.346 MHQ Nmax (% of group) 319 (72.0) 739 (70.1) 462 (74.3) 62 (74.7) 0.289 4285 (71.7) 6636 (66.9) 2549 (71.6) 1008 (70.7) 0.803 bd 3 (0.9) 17 (2.3) 5 (1.1) 3 (4.8) 0.108 55 (1.3) 93 (1.4) 31 (1.2) 17 (1.7) 0.748 mdd 89 (32.4) 178 (28.3) 133 (32.5) 10 (17.9) 0.357 1110 (30.7) 1593 (28.0) 572 (26.6) 237 (28.2) 0.002 Table 3. Methods h Cohorts: phenotyping and genotyping. The IMPROVE study has been described previously12,13. In short, 3700 individuals aged between 54–79 years with high CVD risk profiles (the presence of at least 3 clas- sical CVD risk factors, including family history of CVD, type 2 diabetes, hypertension, hyperlipidaemia and smoking) were recruited from seven centres in Finland, Sweden, the Netherlands, France and Italy. At base- line, individuals completed lifestyle and medical questionnaires and anthropometric measures taken. Blood was sampled for DNA extraction and clinical biochemistry and stored for further biochemical analyses. Detailed ultra-sound examination of the carotid intima-media thickness (cIMT) was conducted at baseline, 15 months and 30 months. Linear regression using all data points was used to calculate progression of cIMT. Mental illness was not assessed; however it is believed that if there is mental illness in this cohort it is likely to be subclinical. All participants provided written informed consent and the study was conducted in accordance with the Hel- sinki Declaration. Ethical approval was granted by the Regional Ethics Review Boards at Karolinska Institutet, Stockholm Sweden, the Groupe Hôpitalier Pitie-Salpetriere, Paris, France, the Comitato Etico delle Aziende Sanitarie della regione Umbria, Perugia, Italy, the Ospedale Niguarda Ca´Granda, Milano, Italy, the University Hospital Groningen, Groningen, the Netherlands, the Hospital District of Northern Savo, Kuopio, Finland and the University of Eastern Finland, Kuopio, Finland.h y p The IMPROVE study was genotyped on the Illumina Cardio-Metabo14 and Immuno chips15, therefore car- diometabolic disorders (including immune and inflammatory components) were well represented. Standard quality control procedures were conducted, namely exclusion of SNPs for low call rate (< 95%) and deviation from Hardy–Weinberg Equilibrium (p < 1 × 10–6) and exclusion of samples for low call rate (< 95%), sex-mismatch, cryptic relatedness. Quality control was conducted on each chip separately, followed by a further round of qual- ity control on the combined chip. y p The UK Biobank (UKB) has been described previously16,17. Approximately 500,000 volunteers aged 39–73 years were recruited from 22 centres across the UK. At baseline, detailed questionnaires on sociodemo- graphic factors, lifestyle factors and medical history were completed by all individuals. Measurements of anthro- pometric variables were recorded and blood samples were taken for DNA extraction. Subsequently (4–8 years after baseline), subsets of participants were invited for follow-up measurements and extensive imaging. All participants provided written informed consent and ethical approval was granted by the NHS national Research Ethics Service. Methods h This work was conducted under projects #6533 (Smith) and #1755 (Pell). h p j Ultrasound measurement of cIMT was conducted in a pilot phase of ~ 2500 individuals (henceforth denoted as UKB1) followed by a subsequent phase including ~ 22,000 individuals (denoted UKB2) using the same recruit- ment and measurement protocol. cIMT measurements were generally consistent with the measurements available in IMPROVE. A mental health/thoughts and feelings questionnaire was also completed by a subset of partici- pants, which enabled estimation of life history of MDD and BD. For both UKB1 and UKB2, 73% of participants completed the mental health questionnaire. p q Genome-wide genotyping was conducted and standard quality control procedures were applied by the UK Biobank team18. Imputation was conducted using the Haplotype reference consortium and 1000 Genomes with standard pre- and post- imputation quality controls being applied by the UK Biobank team (further informa- tion is provided in18). Multi‑dimensional scaling (MDS) to identify clusters. Genome-wide genetic loci reported to be asso- ciated with SCZ19, MDD20 and BD21 were identified. SNPs within these (SCZ, MDD or BD) loci which were present on the CardioMetabo and Immuno chips were selected14,15 (denoted SCZ-CM SNPs, MDD-CM SNPs or BD-CM SNPs, respectively). SNPs with MAF > 1% were included (Supplementary Table 3). A schematic diagram of the analyses steps is provided in Fig. 1. y p p g In IMPROVE, each set of SNPs (SCZ-CM SNPs, MDD-CM SNPs or BD-CM SNPs) were pruned by pairwise LD (parameters 50, 5, 0.1) using PLINK22. Individuals with > 1% missing genetic data were excluded prior to clustering. Clustering was performed using multi-dimensional scaling, implemented in PLINK, using default settings. Multidimensional scaling essentially measures similarity between individuals, in this case using the patterns of genetic variation as the assessment criteria23,24. Individuals with similar genetic sequences are deemed more similar to each other than those with less similar genetic sequences. Clustering was also conducted using tSNE and PCA (Supplementary Methods). pp y Subsequently in UKB1, SCZ-cardiometabolic SNPs only were used and individuals with > 1% missing genetic data were excluded prior to clustering. MDS analyses was conducted using either exactly the same SNPs as were used in IMPROVE (ie SCZ-CM SNPs after filtering for MAF and LD in IMPROVE) or SCZ-CM SNPs with filtering for MAF and LD being done in UKB1. Discussionh Demographic characteristics of the UKB1 and UKB2 participants, by cluster. Highlighted in bold are the significant (p < 0.05) differences between groups. Where: , adjusted to provide estimates of treatment- naïve levels; na, not available; Statistical analyses compared levels or frequencies across groups 1, 2 and 3. Ungrouped (.) were omitted from the analyses). *P for Pearson’s chi square for categorical variables and Kruskal–Wallis for continuous variables. Another limitation is that the CardioMetabo and Immuno chips do not include all loci implicated in cardio- metabolic disorders. Since these chips were described (2012 and 2011 respectively), many more loci involved in many more processes have been identified. However, as more and more samples are available for GWAS analyses, loci are being identified with smaller and smaller effect sizes. Therefore whilst not all possible information is captured by using the CardioMetabo and Immuno chips, the loci with the largest effects are represented. Scientific Reports \| (2021) 11:632 \| https://doi.org/10.1038/s41598-020-79964-x www.nature.com/scientificreports/ www.nature.com/scientificreports/ Figure 5. Comparison of the three clusters identified in UKB2 in individuals (a) without and (b) with mental illness. Each data point is an individual therefore the individuals who are closer together are more genetically similar. Figure 5. Comparison of the three clusters identified in UKB2 in individuals (a) without and (b) with mental illness. Each data point is an individual therefore the individuals who are closer together are more genetically similar. Fi C i f h h l id ifi d i UKB i i di id l ( ) i h d (b) i h l Figure 5. Comparison of the three clusters identified in UKB2 in individuals (a) without and (b) with mental illness. Each data point is an individual therefore the individuals who are closer together are more genetically similar. https://doi.org/10.1038/s41598-020-79964-x Scientific Reports \| (2021) 11:632 \| www.nature.com/scientificreports/ In conclusion, this study provides proof of concept that common biology underlying mental and physical illness is probable and can distinguish subsets of individuals with differing metabolic profiles, even if full under- standing of mechanisms is lacking. Given that large-scale genotyping is not available to healthcare providers and the differences between groups are subtle, there is currently limited potential for translation of this into clinical practice. Further investigation with longitudinal datasets, particularly in high CVD risk populations, would define whether or not there is potential for clinical value in this method. www.nature.com/scientificreports/ www.nature.com/scientificreports/ Choosing a negative control experiment is not straight forward, as current evidence suggests that most genetic variants are highly pleiotropic and that complex traits overlap with each other to a large degree. Despite some overlap with CMD or SCZ-related traits, SNPs in genetic loci associated with eye colour were used as a negative control experiment. The analysis was conducted in UKB2 with MAF > 10% filtering and pruning as described above. Statistical analyses. In IMPROVE, Spearmans rank correlation coefficients were used to assess the rela- tionship between the MDS components and latitude. For IMPROVE, UKB1 and UKB2, Differences between groups were assessed by Pearsons chi squared test for categorical values and Kruskal–Wallis test for continuous variables. All statistical analyses were conducted in Stata (version 11.0). The threshold for significance was set at p < 0.05. No adjustment for multiple testing was applied, because these analyses are exploratory rather than definitive and secondly because most of the cardiometabolic phenotypes tested are interrelated and thus are not independent tests. References e e e ces 1. So, H. C., Chau, K. L., Ao, F. K., Mo, C. H. & Sham, P. C. Exploring shared genetic bases and causal relationships of schizophrenia and bipolar disorder with 28 cardiovascular and metabolic traits. Psychol. Med. 49, 1286–1298. https://doi.org/10.1017/S0033 291718001812 (2019). ( ) 2. De Hert, M. et al. Physical illness in patients with severe mental disorders. I. Prevalence, impact of medications and disparities in health care. World Psychiatry 10, 52–77. https://doi.org/10.1002/j.2051-5545.2011.tb00014.x (2011).fh y y p g j ( ) 3. Amare, A. T., Schubert, K. O., Klingler-Hoffmann, M., Cohen-Woods, S. & Baune, B. T. The genetic overlap between mood dis- orders and cardiometabolic diseases: a systematic review of genome wide and candidate gene studies. Transl. Psychiatry 7, e1007. https://doi.org/10.1038/tp.2016.261 (2017). p g p 4. Tang, B., Yuan, S., Xiong, Y., He, Q. & Larsson, S. C. Major depressive disorder and cardiometabolic diseases: a bidirectiona Mendelian randomisation study. Diabetologia 63, 1305–1311. https://doi.org/10.1007/s00125-020-05131-6 (2020). 5. Mulugeta, A., Zhou, A., King, C. & Hypponen, E. Association between major depressive disorder and multiple disease outcomes: a phenome-wide Mendelian randomisation study in the UK Biobank. Mol. Psychiatry 25, 1469–1476. https://doi.org/10.1038/ s41380-019-0486-1 (2020). 6. Adhikari, K. et al. A GWAS in Latin Americans highlights the convergent evolution of lighter skin pigmentation in Eurasia. Nat Commun. 10, 358. https://doi.org/10.1038/s41467-018-08147-0 (2019). p g ( ) 7. Jonnalagadda, M. et al. A genome-wide association study of skin and iris pigmentation among individuals of South Asian ancestry. Genome Biol. Evol. 11, 1066–1076. https://doi.org/10.1093/gbe/evz057 (2019). p g g 8. Fry, A. et al. Comparison of sociodemographic and health-related characteristics of UK biobank participants with those of the general population. Am. J. Epidemiol. 186, 1026–1034. https://doi.org/10.1093/aje/kwx246 (2017).i 9. Milaneschi, Y., Lamers, F., Bot, M., Drent, M. L. & Penninx, B. W. Leptin dysregulation is specifically associated with major depres- sion with atypical features: evidence for a mechanism connecting obesity and depression. Biol. Psychiatry 81, 807–814. https://doi. org/10.1016/j.biopsych.2015.10.023 (2017). g j p y 0. Milaneschi, Y. et al. Genetic association of major depression with atypical features and obesity-related immunometabolic dysregu- lations. JAMA Psychiatry 74, 1214–1225. https://doi.org/10.1001/jamapsychiatry.2017.3016 (2017). y y g j y y al. Genome-wide association studies in ancestrally diverse popu 11. Peterson, R. E. et al. Genome-wide association studies in ancestrally diverse populations: opportunities, methods, pitfalls, and recommendations. Cell 179, 589–603. https://doi.org/10.1016/j.cell.2019.08.051 (2019). 11. Peterson, R. E. et al. Methods h i g g Finally, in UKB2, Individuals with > 1% missing genetic data were excluded prior to clustering. MDS analysis was conducted on SCZ-CM SNPs with filtering for MAF and LD in UKB2, or on all SCZ SNPs after MAF filter- ing and pruning in UKB2.hi g p g The first two MDS components (C1 and C2) were plotted for visual assessment. Scientific Reports \| (2021) 11:632 \| https://doi.org/10.1038/s41598-020-79964-x References Genome-wide association studies in ancestrally diverse populations: o recommendations. Cell 179, 589–603. https://doi.org/10.1016/j.cell.2019.08.051 (2019). 12. Baldassarre, D. et al. Measurements of carotid intima-media thickness and of interadventitia common carotid diameter improve prediction of cardiovascular events: results of the IMPROVE (Carotid Intima Media Thickness [IMT] and IMT-progression as pre- dictors of vascular events in a high risk European Population) study. J. Am. Coll. Cardiol. 60, 1489–1499. https://doi.org/10.1016/j. jacc.2012.06.034 (2012). j ( ) 13. Baldassarre, D. et al. Cross-sectional analysis of baseline data to identify the major determinants of carotid intima-media thickness in a European population: the IMPROVE study. Eur. Heart J. 31, 614–622. https://doi.org/10.1093/eurheartj/ehp496 (2010).h 14. Voight, B. F. et al. The metabochip, a custom genotyping array for genetic studies of metabolic, cardiovascular, and anthropometric traits. PLoS Genet. 8, e1002793. https://doi.org/10.1371/journal.pgen.1002793 (2012).i . 8, e1002793. https://doi.org/10.1371/journal.pgen.1002793 (201 g j g 15. Trynka, G. et al. Dense genotyping identifies and localizes multiple common and rare variant association signals in celia Nat. Genet. 43, 1193–1201. https://doi.org/10.1038/ng.998 (2011).h , p g g ( ) 16. Matthews, P. M. & Sudlow, C. The UK Biobank. Brain 138, 3463–3465. https://doi.org/10.1093/brain/awv335 (2015). 16. Matthews, P. M. & Sudlow, C. The UK Biobank. Brain 138, 3463–3465. https://doi.org/10.1093/brain/awv335 (2015). 17 S dl C l UK bi b k f id if i h f id f l di f iddl h 17. Sudlow, C. et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of m old age. PLoS Med. 12, e1001779. https://doi.org/10.1371/journal.pmed.1001779 (2015).th g g j 8. Bycroft, C. et al. The UK Biobank resource with deep phenotyping and genomic data. Nature 562, 203–209. https://doi.org/10.1038/ s41586-018-0579-z (2018). 9. Schizophrenia Working Group of the Psychiatric Genomics, C. Biological insights from 108 schizophrenia-associated genetic loci Nature 511, 421–427, https://doi.org/10.1038/nature13595 (2014).i 0. Wray, N. R. et al. Genome-wide association analyses identify 44 risk variants and refine the genetic architecture of major depres- sion. Nat. Genet. 50, 668–681. https://doi.org/10.1038/s41588-018-0090-3 (2018).i p g ( ) 1. Stahl, E. A. et al. Genome-wide association study identifies 30 loci associated with bipolar disorder. Nat. Genet. 51, 793–803. https ://doi.org/10.1038/s41588-019-0397-8 (2019). g 2. Purcell, S. et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am. J. Hum. Genet. 81 559–575. https://doi.org/10.1086/519795 (2007).h p g ( ) 23. Zhang, Z. & Takane, Y. Multidimensional Scaling. International Encyclopedia of Education (Third Edition), 304–311 (2010). Data availabilityh y The datasets generated during and/or analysed during the current study are available from the corresponding author request. Received: 24 June 2020; Accepted: 11 December 2020 Received: 24 June 2020; Accepted: 11 December 2020 References h 24. Jaworska, N. & Chupetlovska-Anastasova, A. A review of multidimensional scaling (MDS) and its utility in various psycholo domains. Tutorials Quantitative Methods Psychol. 5, 1–10 (2009). https://doi.org/10.1038/s41598-020-79964-x Scientific Reports \| (2021) 11:632 \| www.nature.com/scientificreports/ Author contributions d d d Study conception, design, data analysis (R.J.S., B.S.) or acquisition (D.B., U.dF., A.H., S.E.H., E.T., T.P.T.), inter- pretation (R.J.S., K.J.A.J., P.E., B.G., S.E.H., D.M.L., L.M.L., B.S., D.J.S.), manuscript preparation or revision (R.J.S., K.J.A.J., M.E.S.B., D.B., B.C., P.E., U.dF., A.F., B.G., P.G., N.G., A.H., S.E.H., S.K., D.M.L., L.M.L., M.P., J.P.P., K.S., B.S., A.J.S., E.T., T.P.T., F.V., J.W., D.J.S.). Study conception, design, data analysis (R.J.S., B.S.) or acquisition (D.B., U.dF., A.H., S.E.H., E.T., T.P.T.), inter- pretation (R.J.S., K.J.A.J., P.E., B.G., S.E.H., D.M.L., L.M.L., B.S., D.J.S.), manuscript preparation or revision (R.J.S., K.J.A.J., M.E.S.B., D.B., B.C., P.E., U.dF., A.F., B.G., P.G., N.G., A.H., S.E.H., S.K., D.M.L., L.M.L., M.P., J.P.P., K.S., B.S., A.J.S., E.T., T.P.T., F.V., J.W., D.J.S.). Acknowledgementsh g The authors thank all participants and staff of the IMPROVE and UK Biobank studies. A full list of contribu- tors to the IMPROVE Study are listed in the Supplementary Data. This work uses data provided by patients and collected by the NHS as part of their care and support. IMPROVE was supported by the European Commission (Contract number: QLG1-CT-2002-00896), the Swedish Heart-Lung Foundation, the Swedish Research Council (projects 8691 and 09533), the Knut and Alice Wallenberg Foundation, the Foundation for Strategic Research, the Stockholm County Council (project 592229), the Strategic Cardiovascular and Diabetes Programmes of Karolinska Institutet and Stockholm County Council, the European Union Framework Programme 7 (FP7/2007- 2013) for the Innovative Medicine Initiative under grant agreement n° IMI/115006 (the SUMMIT consortium), the Academy of Finland (Grant #110413), the British Heart Foundation (RG2008/08, RG2008/014) and the Italian Ministry of Health (Ricerca Corrente). The UK Biobank was established by the Wellcome Trust, Medical Research Council, Department of Health, Scottish Government and Northwest Regional Development Agency. UK Biobank has also had funding from the Welsh Assembly Government and the British Heart Foundation. Data collection was funded by UK Biobank. This project was completed using UK Biobank applications 6533 (PI. DJS) and 1755 (PI. JPP). RoJS is supported by a UKRI Innovation-HDR-UK Fellowship (MR/S003061/1). LML is supported by the JMAS Sim Fellowship for depression research from the Royal College of Physicians of Edinburgh. AF is supported by an MRC Doctoral Training Programme Studentship at the University of Glasgow (MR/K501335/1). KJAJ is supported by an MRC Doctoral Training Programme Studentship at the Universities of Glasgow and Edinburgh. DJS acknowledges the support of a Lister Prize Fellowship (173096) and MRC Mental Health Data Pathfinder Award (MC_PC_17217). BS is financially supported by the Knut and Alice Wallenberg Foundation as part of the National Bioinformatics Infrastructure Sweden at SciLifeLab. Competing interests Th h d l The authors declare no competing interests. © The Author(s) 2021 Additional informationh Supplementary Information The online version contains supplementary material available at https://doi. org/10.1038/s41598-020-79964-x. Correspondence and requests for materials should be addressed to R.J.S. Correspondence and requests for materials should be addressed to R.J.S. Correspondence and requests for materials should be addressed to R.J.S. Reprints and permissions information is available at www.nature.com/reprints. Reprints and permissions information is available at www.nature.com/reprints. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2021 https://doi.org/10.1038/s41598-020-79964-x Scientific Reports \| (2021) 11:632 \|
W2949563861.txt	https://www.eg-quaternary-sci-j.net/29/179/1979/egqsj-29-179-1979.pdf	de		Tunneltäler in Dänemark	Eiszeitalter und Gegenwart	1,979	cc-by	8,660	179—188 Eiszeitalter u. Gegenwart 29 Hannover 6 Abb. 1979 Tunneltäler in Dänemark STEEN S J Ö R R I N G * ) Upper Pleistocene (Saale a n d Weichsel glaciation), "tunnel v a l l e y " , channel, fluviatile erosion, glacial erosion, geomorphologic sketch map Denmark Kurzfassung: Der Begriff „Tunneltal" umfaßt mehrere verschiedene morphologische Elemente. Neue Ergebnisse weisen darauf hin, daß die klassische Tunneltalhypothese aufgegeben werden muß. Einige der Tunneltäler sind wahrscheinlich als p r ä g l a z i a l e oder proglaziale Wasser läufe angelegt und sind später durch selektive Glazialerosion überprägt worden. [Tunnel V a l l e y s in D e n m a r k ] A b s t r a c t . The term "tunnel v a l l e y " comprises more different geomorphological elements. Based on new investigations it appears that the hypothesis for formations of tunnel v a l l e y s cannot longer be maintained. Some of the tunnel v a l l e y s may have an origin as preglacial or proglacial streems, which later on has been deepend by selective glacial erosion. In e i n e m g r ö ß t e n t e i l s a u s L o c k e r s e d i m e n t e n aufgebauten Flachland wie Dänemark ist d i e A u s f o r m u n g d e r Oberfläche v o r a l l e m a u f d i e W i r k u n g des I n l a n d e i s e s u n d seiner Schmelzwässer zurückzuführen. Auf A b b . 1 sind e i n i g e g e o m o r p h o l o g i s c h e u n d q u a r t ä r g e o l o g i s c h e G r e n z l i n i e n e i n g e zeichnet. D i e westliche L i n i e m u ß i m W e s e n t l i c h e n a l s G r e n z e d e r E i s b e d e c k u n g d e r l e t z t e n G l a z i a l z e i t ( W e i c h s e l / W ü r m ) betrachtet w e r d e n . Es b e s t e h t ein sehr d e u t l i c h e r g e o m o r p h o l o g i s c h e r U n t e r s c h i e d z w i s c h e n d e n Gebieten b e i d e r s e i t s dieser G r e n z l i n i e , d i e die W e i c h s e l - H a u p t s t i l l s t a n d s l i n i e ( o d e r USSINGS E i s r a n d l i n i e ) g e n a n n t w i r d . A u f d e r westlichen S e i t e l i e g e n d i e g r o ß e n w e s t j ü t l ä n d i s c h e n S a n d e r , die n u r v o n den s a a l e - ( r i ß - ) z e i t l i c h e n „ b a k k e 0 e r " , d i e m a n in d i r e k t e r Ü b e r s e t z u n g aus d e m Dänischen a l s „ H ü g e l i n s e l n " b e z e i c h n e n k ö n n t e , u n t e r b r o c h e n w e r d e n . A u f der östlichen S e i t e der G r e n z l i n i e sind d i e G l a z i a l f o r m e n in i h r e m J u g e n d s t a d i u m — mit Moränenflächen, Hügellandschaften und Tunneltälern — gut erhalten. D i e östlichste G r e n z l i n i e , auch a l s d i e H A R D E R ' s c h e E i s r a n d l a g e b e k a n n t , k a n n a l s A u s b r e i t u n g s g r e n z e des l e t z t e n j u n g b a l t i s c h e n E i s s t r o m e s b e t r a c h t e t w e r d e n . Auf d e r A b b . 1 sind a u c h d i e w i c h t i g s t e n T u n n e l t ä l e r in g r o ß e n Z ü g e n e i n g e z e i c h n e t . D i e T u n n e l t ä l e r sind v o n U S S I N G ( 1 9 0 3 , 1 9 0 4 und 1 9 0 7 ) als „f j o r d d a l e " (direkt ü b e r s e t z t : F j o r d - oder F ö r d e n t ä l e r ) b e z e i c h n e t w o r d e n . In s e i n e n A r b e i t e n ü b e r d i e S a n d e r m o r p h o l o g i e W e s t j ü t l a n d s entdeckte U S S I N G , d a ß d i e S a n d e r flache H a l b k e g e l b i l d e n , deren höchster P u n k t d o r t l i e g t , w o d i e F ö r d e n t ä l e r enden. D e s h a l b schien e i n e genetische V e r k n ü p f u n g der S a n d e r m i t den F ö r d e n t ä l e r n a h e z u l i e g e n . U S S I N G l i e ß jedoch d i e F r a g e offen, o b d i e F ö r d e n t ä l e r v o n s u b g l a z i a l e n o d e r s u p r a g l a z i a l e n S c h m e l z w ä s s e r n geschaf fen w o r d e n w a r e n . S p ä t e r a b e r h a t M A D S E N ( 1 9 2 1 ) , ü b e r e i n s t i m m e n d m i t WOLDSTEDT ( 1 9 1 3 ) , e i n e s u b g l a z i a l e E n t s t e h u n g d e r F ö r d e n t ä l e r festgestellt. U m a l l e Z w e i f e l a u s z u r ä u m e n , h a t MAUSEN g l e i c h z e i t i g den N a m e n „Tunneltal" ( d ä n i s c h : „ t u n n e l d a l " ) ein geführt. :: ') Anschrift des Verfassers: Univ.lektor S. S j ö r r i n g , Institut for almen Geologi, Kobenhavns Universitet östervoldgade 10, DK-1350 Köbenhavn K, Dänemark. 12 • Steen Sjörring 180 D e r Begriff „ T u n n e l t a l " u m f a ß t m e h r e r e v e r s c h i e d e n e m o r p h o l o g i s c h e E l e m e n t e , d i e w a h r s c h e i n l i c h u n t e r s c h i e d l i c h e r E n t s t e h u n g sind. D i e d ä n i s c h e n T u n n e l t ä l e r haben L ä n gen v o n 5 bis 7 0 k m . Ä h n l i c h w i e d i e T u n n e l t ä l e r in N o r d d e u t s c h l a n d z e i g e n sie ein u n r e g e l m ä ß i g e s B o d e n r e l i e f . S t e l l e n w e i s e sind sie durch S c h w e l l e n u n t e r b r o c h e n 8 9 10 11 und/oder 12 Abb. 1 : Vereinfachte geomorphologische Karte (frei nach K. MILTHERS 1 9 4 2 ) . Zeichenerklärung: 1. Saale-(Riß-)zeitliche Landschaft, 2 . Sander der Weichsel-(Würm-)eiszeit, 3 . Weichsel-(Würm)zeitliche Landschaft, 4 . Die UssiNc'sche Eisrandlinie, 5 . Die HARDER'sche Eisrandlinie, 6. Tunnel täler, 7 . Das Odder-Elbo-Tal. Tunneltäler in Dänemark 181 m i t S e e n - o d e r B e c k e n a b l a g e r u n g e n g e f ü l l t . I m a l l g e m e i n e n w e i s e n sie h e u t e e i n G e f ä l l e z u m Z e n t r u m d e s e h e m a l i g e n Eises hin a u f . D i e T u n n e l t ä l e r s i n d durchschnittlich V2 bis 3 k m , d i e ostjütischen A u ß e n f ö r d e n jedoch b i s z u 5 — 1 0 k m b r e i t . D i e H a u p t r i c h t u n g d e r g r o ß e n j ü t l ä n d i s c h e n T u n n e l t ä l e r ist i n N o r d j ü t l a n d NE—SW, i m übrigen J u t l a n d i m Wesentlichen E — W . M i t A u s n a h m e e i n i g e r — bezüglich i h r e r G e n e s e unsicherer — T ä l e r ist d i e H a u p t richtung ü b e r a l l p a r a l l e l z u r r e g i o n a l e n B e w e g u n g s r i c h t u n g d e s I n l a n d e i s e s ( u n d w a h r scheinlich a u c h p a r a l l e l z u r R i c h t u n g des E i s a b b a u e s ) . W i e es scheint, sind d i e T u n n e l t ä l e r nördlich D j u r s l a n d ( A b b . l ) s c h m a l e r a l s d i e T ä l e r w e i t e r südlich, w o sie, w i e b e r e i t s v e r m e r k t , h a u p t s ä c h l i c h E — W gerichtet sind. D i e U r s a c h e h i e r z u ist w a h r s c h e i n l i c h i m l e t z t e n E i s v o r s t o ß b i s z u r H A R D E R ' s c h e n E i s r a n d l i n i e z u suchen, d a e i n e Ü b e r e i n s t i m m u n g schen d e r A u s b r e i t u n g dieses Eises u n d d e r V e r t e i l u n g d e r b r e i t e n T u n n e l t ä l e r Die g l a z i a l s t r a t i g r a p h i s c h e n U n t e r s u c h u n g e n in Ü b e r e i n s t i m m u n g zwi besteht. d e r letzten J a h r e in D ä n e m a r k haben m i t H A R D E R ( 1 9 0 8 ) b e w i e s e n , d a ß d i e HARDER'sche E i s r a n d l i n i e (oder O s t j ü t l ä n d i s c h e E i s r a n d l i n i e ) einem s e l b s t ä n d i g e n E i s v o r s t o ß z u z u o r d n e n ist (BER THELSEN 1 9 7 3 ; SJÖRRING 1 9 7 4 , 1 9 7 7 ) . Z u d i e s e n U n t e r s u c h u n g s e r g e b n i s s e n s t e h e n aller d i n g s d i e A u f f a s s u n g e n a n d e r e r Verfasser i m W i d e r s p r u c h ( z . B . K. MILTHERS 1 9 4 2 ; S . HANSEN 1 9 6 5 ; NIELSEN 1967 und MARCUSSEN 1977), die der Meinung sind, daß die H A R D E R ' s c h e L i n i e n u r eine A b s c h m e l z l i n i e r e p r ä s e n t i e r t . A u f g r u n d der erwähnten glazialstratigraphischen Untersuchungen k o n n t e n w i r fest stellen, d a ß d i e T u n n e l t ä l e r östlich d e r H A R D E R ' s c h e n E i s r a n d l i n i e z u m i n d e s t z u m T e i l älter sind a l s der der genannten E i s r a n d l a g e zugeschriebene letzte Eisvorstoß. Diese älte ren T u n n e l t ä l e r , d i e i m W w e i t über d i e H A R D E R ' s c h e L i n i e h i n a u s r e i c h e n , w u r d e n östlich dieser L i n i e v o m e r w ä h n t e n l e t z t e n E i s v o r s t o ß ü b e r p r ä g t . A u c h NORDMANN ( 1 9 5 9 ) h a t auf diese Z u s a m m e n h ä n g e h i n g e w i e s e n . V o n d e n klassischen A u f f a s s u n g e n ü b e r d i e E n t s t e h u n g v o n T u n n e l t ä l e r n abgesehen, w u r d e n i n d e n letzten J a h r e n i n D ä n e m a r k e i n e R e i h e weiterer Gesichtspunkte diskutiert. N O R D M A N N ( 1 9 5 9 ) t e i l t e m i t , d a ß sich d i e s u b g l a z i a l e n S c h m e l z w a s s e r s t r ö m e s e i t l i c h v e r schoben h a b e n (vielleicht a u f g r u n d der d ä m m e n d e n Wirkung herabfallender Eisblöcke). In e i n e r A b h a n d l u n g ü b e r d i e T u n n e l t ä l e r i n D ä n e m a r k u n d Norddeutschland hat K. H A N S E N ( 1 9 7 1 ) f r ü h e r e A u f f a s s u n g e n b e s p r o c h e n . Er macht d a r a u f a u f m e r k s a m , d a ß d i e höchsten P u n k t e d e r S a n d e r k e g e l u n d d i e E n d e n der T u n n e l t ä l e r sich nicht i m m e r i n übereinstimmender Lage zueinander befinden. D i e Ergebnisse v o n W E I S S ( 1 9 5 8 ) u n d JASPERSEN ( 1 9 5 3 ) zeigen, d a ß bei B e r ü c k s i c h t i g u n g der K o r n g r ö ß e n v e r t e i l u n g e n i n den S a n d e r n u n d d e r h y d r o l o g i s c h e n V e r h ä l t n i s s e i m E i s ein genetischer Z u s a m m e n h a n g zwi schen s u b g l a z i a l e n Flüssen u n d S a n d e r n n i c h t m e h r gefolgert w e r d e n k a n n . V o n h y d r o logischen G e s i c h t s p u n k t e n a u s g e h e n d , scheint m a n d i e klassische T u n n e l t a l h y p o t h e s e auf geben z u m ü s s e n . A u f g r u n d seiner S t u d i e n s c h l i e ß t K. HANSEN ( 1 9 7 1 ) , d a ß m e h r e r e d e r Täler bereits saale-(riß-)zeitlich angelegt u n d später durch s e l e k t i v e E r o s i o n übertieft w u r d e n . Ä h n l i c h e A u f f a s s u n g e n bezüglich d e r s e l e k t i v e n Erosion ä u ß e r t e n auch W O L D S T E D T (1961) u n d GRIPP (1964, 1 9 7 5 ) . Eine n e u e H y p o t h e s e ü b e r d i e E n t s t e h u n g e i n i g e r T u n n e l t ä l e r h a t BERTHELSEN ( 1 9 7 2 ) dargelegt. A u f g r u n d der Eisbelastung ( A b b . 2 ) w i r d der U n t e r g r u n d unter d e m Eis her abgedrückt ( A b b 2 B). Der dabei auftretende isostatische D r u c k i n der E r d k r u s t e unter d e m Eis k ö n n t e einen M a t e r i a l t r a n s p o r t a u s d e m Bereich u n t e r d e m Eis z u m E i s r a n d e r z w i n g e n . F a l l s dies geschieht, m ü ß t e dies z u r B i l d u n g eines e i s r a n d p a r a l l e l e n W u l s t e s füh ren ( A b b . 2 C ) . W e n n d a n n d e r w e i t e r e E i s v o r s t o ß rascher v o n s t a t t e n geht a l s d e r b e schriebene isostatisch b e d i n g t e M a t e r i a l t r a n s p o r t , k ö n n t e d a s E i s a u f den v o n i h m geschaf fenen e i s r a n d p a r a l l e l e n W u l s t g e l a n g e n ( A b b . 2 D ) . S c h m e l z w a s s e r , d i e a n d i e s e r P h a s e 182 Steen Sjörring B \\\\\\\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ w \ \ w w w \\u\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ F mwmmmmM T \ i ytTT Abb. 2: Darstellung der hypothetischen Genese der Tunneltäler. Jutland im Querschnitt (140 k m ) . Die Profile sind in fünfzigfacher Überhöhung gezeichnet. Erläuterung: siehe Text. d e n E i s r a n d v e r l a s s e n , f ä n d e n n u n e i n e v o m E i s r a n d w e g gerichtete O b e r f l ä c h e n n e i g u n g v o r u n d h ä t t e n in R i c h t u n g der N e i g u n g subaerisch T ä l e r e r o d i e r e n k ö n n e n . S p ä t e r w a r e n d i e T ä l e r bei Ü b e r s c h r e i t u n g d u r c h d a s Eis durch s e l e k t i v e Erosion ü b e r p r ä g t w o r d e n (Abb. 2 E). Diese s e l e k t i v e E r o s i o n k ö n n t e v i e l l e i c h t m i t P e r m a f r o s t i m Z u s a m m e n h a n g s t e h e n , d a dessen T i e f e u n t e r S c h m e l z w a s s e r f l ü s s e n g e r i n g e r a l s in der U m g e b u n g g e w e s e n s e i n dürfte. B e i m V o r d r i n g e n des Eises k ö n n t e d a s u n t e r d e r P e r m a f r o s t z o n e einem e n t s p r e chend e r h ö h t e n D r u c k a u s g e s e t z t e P o r e n w a s s e r u n t e r U m s t ä n d e n i m Bereich der F l u ß l ä u f e nach oben d r i n g e n u n d d a d u r c h d i e s e l e k t i v e E r o s i o n b e g ü n s t i g e n . W e n n diese H y p o these zutrifft, s o l l t e es m ö g l i c h s e i n , d i e a l t e n F l u ß l ä u f e unter d e n S a n d e r n zu f i n d e n . Bisher h a t n u r H E L L E R ( 1 9 6 1 ) u n t e r S a n d e r n b e g r a b e n e T ä l e r in W e s t j ü t l a n d beschrieben. D i e s e T ä l e r k ö n n t e n w e i c h s e l z e i t l i c h e B i l d u n g e n o d e r a b e r auch ä l t e r s e i n . Ü b e r d a s A l t e r d e r T u n n e l t ä l e r l i e g e n m e h r e r e I n f o r m a t i o n e n v o r . ANDERSEN ( 1 9 7 3 , 1 9 7 4 ) u n d SCHRÖDER ( 1 9 7 4 ) k o n n t e n nach T i e f b o h r u n g e n u n d g e o p h y s i k a l i s c h e n U n t e r suchungen in d e r U m g e b u n g v o n A a r h u s b e w e i s e n , d a ß unter e i n i g e n v o n ihnen ü b e r - Tunneltäler in Dänemark 183 prüften T u n n e l t ä l e r n u n d s p ä t g l a z i a l e n E n t w ä s s e r u n g s t ä l e r n ä l t e r e T ä l e r v o r k o m m e n , d i e s t e l l e n w e i s e Tiefen v o n m e h r a l s 1 0 0 m u n t e r d e m M e e r e s s p i e g e l a u f w e i s e n . In m e h r e r e n dieser b e g r a b e n e n T u n n e l t ä l e r t r e t e n g l a z i a l e u n d i n t e r g l a z i a l e S e d i m e n t e auf. Dies bedeutet, d a ß d i e T ä l e r nicht n u r w e i c h s e l z e i t l i c h e n t s t a n d e n sind, s o n d e r n z u m T e i l n u r w e i c h s e l z e i t l i c h ü b e r p r ä g t w u r d e n . ANDERSEN ( 1 9 7 2 ) h a t e r w ä h n t , d a ß d i e t e r t i ä r e n A b l a g e r u n g e n J ü t l a n d s i m g r o ß e n u n d g a n z e n nach S W e i n f a l l e n . Er d e u t e t a n , d a ß a l t e T ä l e r bereits i m T e r t i ä r g e b i l d e t w u r d e n . MARCUSSEN ( 1 9 7 7 ) v e r m u t e t , d a ß e i n i g e d e r T u n n e l t ä l e r w e i c h s e l z e i t l i c h e n A l t e r s s i n d , d a m a n a n m e h r e r e n S t e l l e n a n e r o d i e r t e E e m a b l a g e r u n g e n in d e n T a l h ä n g e n finden k a n n . D a z u ist z u b e m e r k e n , d a ß sich e i n i g e s e i n e r L o k a l i t ä t e n nicht i n T u n n e l t ä l e r n , s o n d e r n in s p ä t g l a z i a l e n E n t w ä s s e r u n g s t ä l e r n befinden. MARCUSSEN schließt jedoch nicht a u s , d a ß e i n i g e d e r T u n n e l t ä l e r ä l t e r sein k ö n n t e n . D i e a l t e r n a t i v e „ T u n n e l t a l " - h y p o t h e s e ( N i e d e r t a u - t ä l e r n ) v o n MARCUSSEN ( 1 9 7 7 ) scheint ü b r i g e n s d e r H y p o t h e s e v o n H O R M A N N ( 1 9 6 7 ) , d i e v o n GRIPP ( 1 9 7 4 ) besprochen w u r d e , z u m V e r w e c h s e l n ä h n l i c h z u sein. In D ä n e m a r k h a t d a s O d d e r - E l b o - T a l ( A b b . 1 u . 3 ) eine S o n d e r s t e l l u n g a l s „ T u n n e l t a l " g e w o n n e n . Dieses T a l w e i s t eine g a n z a b w e i c h e n d e R i c h t u n g v o n den ü b r i g e n T u n n e l t ä l e r n i m selben R a u m auf. K . MILTHERS ( 1 9 4 2 ) m e i n t e , d a ß d a s O d d e r - E l b o - T a l l ä n g e r e Z e i t m i t T o t e i s gefüllt w a r , d a es einen s p ä t e r e n k r e u z e n d e n E i s v o r s t o ß überleben k o n n t e . Abb. 3: Das Odder-Elbo-Talsystem (frei nach S . HANSEN, NIELSEN 1 9 6 0 und BERTHELSEN 1 9 7 2 ) . Zeichenerklärung: 1 . Sander u n d Entwässerungstäler des jungbaltischen Eisvorstoßes, 2 . Die HARDER'sche Eisrandlinie, 3. Das Odder-Elbo-Tal. 184 Steen Sjörring Abb. 4: Höhenliniekarte im Raum südlich von Aarhus (Ausschnitt nach HARDER 1908). Lokalisie rung s. Abb. 3. Höhenlinieabstand: 30 Fuss = 9,4 Meter. Ausgezogene und gestrichelte Linien sind Hauptwege und Eisenbahnstrecken; Kreuze darstellen Kirchen. Tunneltäler in Dänemark 185 Abb. 5 : Rechts: Höhenliniekarte in der Umgebung von Aabenraa (nach JESSEN 1 9 4 5 ) , Höhenlinie abstand: 1 0 Meter. Links: Höhenliniekarte des p r ä q u a r t ä r e n Untergrundes (nach RASMUSSEN 1 9 6 0 ) in demselben Gebiet. 186 Steen Sjörring BERTHELSEN ( 1 9 7 2 ) n i m m t a n , d a ß es sich u m e i n U r s t r o m t a l h a n d e l n k ö n n t e . E r f o l gert dies a u s d e m U m s t a n d , d a ß es g e n a u p a r a l l e l m i t d e m G u d e n a a - T a l s y s t e m v e r l ä u f t , das als s p ä t g l a z i a l e s E n t w ä s s e r u n g s s y s t e m d e r H A R D E R ' s c h e n E i s r a n d l a g e g e d e u t e t w i r d . Auch die M o r p h o l o g i e w e i s t d a r a u f hin ( A b b . 4 ) , d a ß d a s O d d e r - E l b o - T a l eher e i n e x t r a m a r g i n a l e s T a l ist a l s ein T u n n e l t a l . D i e b r e i t e n T u n n e l t ä l e r in J u t l a n d , i n n e r h a l b d e r HARDER'schen E i s r a n d l a g e , s i n d o h n e Z w e i f e l v o n d e m dieser E i s r a n d l a g e z u g e h ö r i g e n E i s v o r s t o ß g l a z i a l übertieft w o r den ( A b b . 5 ) . W a h r s c h e i n l i c h w u r d e n die T u n n e l t ä l e r a u ß e r h a l b dieses Eisvorstoßes ( A b b . 1 u. 5) e b e n f a l l s d u r c h s e l e k t i v e Erosion (in d i e s e m F a l l durch einen ä l t e r e n E i s v o r s t o ß bis z u r UssiNG'schen E i s r a n d l i n i e ) ü b e r p r ä g t . S i e s i n d a b e r u r s p r ü n g l i c h durch a q u a t i s c h e Erosion p r o g l a z i a l ( o d e r p r ä g l a z i a l ) a n g e l e g t w o r d e n u n d w a h r s c h e i n l i c h p r ä - w e i c h s e l zeitlichen U r s p r u n g s . Die T u n n e l t ä l e r auf N o r d s e e l a n d haben ein a n d e r e s A u s s e h e n ( A b b . 6 ) . S i e k ö n n e n auch über b e g r a b e n e n T ä l e r n l i e g e n , ä h n e l n a b e r m e h r e i n e m F l u ß n e t z im T o t e i s . Oft fin det m a n in i h n e n k l e i n e Oser, v o n denen a n g e n o m m e n w i r d , d a ß sie in S p a l t e n e i n e s a b schmelzenden T o t e i s e s g e b i l d e t w u r d e n . Für eine g e n a u e D e u t u n g dieser verschiedenen T ä l e r b e n ö t i g e n w i r neue Untersuchun gen, die G l a z i a l s t r a t i g r a p h i e u n d isostatische B e w e g u n g e n b e s o n d e r s berücksichtigen. Abb. 6: Kartenausschnitt von Nordseeland (frei nach V. MILTHERS 1935). Gestrichelte Kurven sind Höhenlinien des präquartären Untergrundes (Abstand in Meter). Schraffierte Flächen dar stellen Tunneltäler. Tunneltäler in Dänemark 187 A b s c h l i e ß e n d k a n n gesagt w e r d e n , d a ß w i r in Z u k u n f t den A u s d r u c k n u r m i t g r ö ß t e r V o r s i c h t gebrauchen s o l l t e n . „Tunneltal" Dank. Für k r i t i s c h e D i s k u s s i o n e n u n d für H i l f e bei der Ü b e r s e t z u n g D r . W a l t e r V o r t i s c h ( M a r b u r g ) herzlich d a n k e n . möchte ich Schriftenverzeichnis ANDERSEN, H . LYKKE (1972): Viborgegnens tunneldale. — Museerne i Viborg Amt 2: 10—15; Viborg. — (1973): En begravet dal i prse-Kvartajret ved Aarhus. Dansk geol. Foren., Aarsskrift for 1972: 111—118; Kobenhavn (Reitzel). — (1974): Kortlasgningen af undergrunden i omegnen syd og vest for Aarhus. — In: Laboratoriet for Geofysik ( H r s g . ) : Undersogelser af Aarhusegnens undergrund, 5 7 — 1 0 1 ; Aarhus (Universitetet). BERTHELSEN, A. (1972): Flod-, fjord- og tunneldale. — Dansk geol. Foren., Aarsskrift for 1 9 7 1 : 101—104; Kobenhavn. — (1973): Weichselian Ice Advances and Drift Successions in Denmark. — Bull. geol. Inst. Univ. Uppsala, N . S . , 5: 2 1 — 2 9 ; Stockholm. GRIPP, K. (1964): Erdgeschichte von Schleswig-Holstein. — 411 S.; Neumünster (Wachholz). — (1974): Über die Schwierigkeiten, Vorgänge am R a n d des quartären Inlandeises ohne entspre chende Kenntnis des heutigen Geschehens zu deuten. — Z. Geomorph. N.F., 1 8 , 2: 224—229; Stuttgart. — (1975): 100 J a h r e Untersuchungen über das Geschehen am Rande des nordeuropäischen In landeises. — Eiszeitalter u. Gegenwart, 26: 31—73; Öhringen. HANSEN, K. (1971): Tunnel valleys in Denmark and Northern Germany. — Bull. geol. Soc. Den mark, 20: 295—306; Kopenhagen. HANSEN, S. (1965): The Quaternary of Denmark. — In: RANKAMA, K. (Hrsg.): The Quaternary, 1: 90 S . ; New York, London (Interscience Publishers). — & NIELSEN, A. V. (1960): Glacial Geology of Southern Denmark. — In: SORGENFREI, T. (Hrsg.): International Geological Congress X X I Session in Norden; Guide to Excursions A 44 and C 39: 36 S.; Kopenhagen. HARDER, P. (1908): En ostjysk israndslinie og dens indflydelse paa vandlobene. — Danm. geol. Unders. Rk II, 1 9 : 259 S., und mit A t l a s ; Kopenhagen. [Summary in English: An ice-edge line in East J u t l a n d and its influence on the water-courses]. HELLER, E. (1961): Iagttagelser over tertisere og kvarta^re forhold i Tarm-Brande-Grindstedomraadet. — Meddr. dansk geol. Foren., 14: 374—385; Kopenhagen. [ S u m m a r y in English: Tertiary and Quaternary Observations in the Tarm-Brande-Grindsted A r e a ] . HORMANN, K. (1969): Gibt es Tunneltäler in Schleswig-Holstein? — Sehr. N a t u r w . Ver. Schlesw.Holst., 39: 5 — 1 1 ; Kiel. JASPERSEN, P. (1953): Sanderbildung durch subglaziäre, aufsteigende Schmelzwasserströme? — Eiszeitalter u. Gegenwart, 3: 129—135; Öhringen. JESSEN, A. (1945): Beskrivelse til Geologisk Kort over Danmark, Kortbladet Sonderborg. — Danm. geol. Unders. Rh. I, 20: 91 S.; Kopenhagen. [Resume en francais: Notice explicative de la feuille de Sönderborg]. MADSEN, V. (1921): Terrainformerne pa Skovbjerg Bakkeo. — Danm. geol. Unders. Rk. IV, 1 (12): 24 S.; Kopenhagen. [Resume en francais: Les formes du terrain de la Colline Insulaire de Skovbjerg (en danois: Skovbjerg B a k k e o ) ] . MARCUSSEN, I. (1977): Deglaciation landscapes formed during the wasting of the late Middle Weichselian ice sheet in Denmark. — Danm. geol. Unders. Rk. II, 1 1 0 : 72 S.; Kopenhagen. [Dansk sammendrag: Deglaciations landskaber dannet under smeltningen af isskjoldet i sen mellem weichsei i D a n m a r k ] . MILTHERS, K. (1942): Ledeblokke og landskabsformer i Danmark. — Danm. geol. Unders. Rk. II, 69: 137 S.; Kopenhagen. [Summary in English: Indicator Boulders and Morphology of the Landscape in D e n m a r k ] . iS8 Steen Sjörring MILTHERS, V. (1935): Nordostsjaellands Geologi. — Danm. geol. Unders. Rk. V, 3: 192 S.; Ko penhagen. NIELSEN, A. V. (1967): Landskabets tilblivelse. — In: NORREVANG, A. & MEYER, T. J . : Danmarks Natur, 1: 251—344; Kopenhagen. NORDMANN, V. (1959): Beskrivelse til Geologisk Kort over Danmark, Kortbladet Fredericia. — Danm. geol. Unders. R k . I, 2 2 - A : 125 S.; Kopenhagen. [ S u m m a r y in English: Explanation of sheet Fredericia]. RASMUSSEN, L. B. (1960): Molluscan Faunas and Biostratigraphy of the Marine Younger Miocene Formations in Denmark, Part I: Geology and Biostratigraphy. — Danm. geol. Unders. Rk. II, 88: 358 S.; Kopenhagen. [Dansk sammendrag: De danske marine yngre miocaene formationers molluskfaunaer og biostratigrafi, Del I: Geologi og biostratigrafi]. SCHRÖDER, N . (1974): De geofysisk-geologiske undersogelser af omraadet nord for Aarhus. — In: Laboratoriet for Geofysik (Hrsg.): Undersogelser af Aarhusegnens undergrund, 15—55; Aarhus (Universitetet). SJÖRRING, S. (1974): Über spätpleistozäne Glazialdynamik und -stratigraphie in Ost-Dänemark. — Eiszeitalter u. Gegenwart, 25: 208—209; Öhringen. — (1977): The glacial stratigraphy of the island of Als, southern Denmark. — Z. Geomorph. N. F., Suppl.-Bd., 27: 1—11; Stuttgart. USSING, N. V. (1903): Om J y l l a n d s Hedesletter og Teorierne for deres Dannelse. — Overs. K. danske Vidensk. Selsk. Forh., 1903, 2: 99—165; Kopenhagen. [Resume en francais]. — (1904): Danmarks Geologi i almenfatteligt Omrids. — Danm. geol. Unders. R k . I l l , 2: 358 S.; Kopenhagen. — (1907): Om floddale og randmoraner i J y l l a n d . — Overs. K. danske Vidensk. Selsk. Forh., 1907, 4: 161—213; Kopenhagen. [Resume en francais: Sur les Alluvions Glaciaires et les Moraines Terminales en J u t l a n d ] . WEISS, E. N. (1958): Bau und Entstehung der Sander vor der Grenze der Würmvereisung im Nor den Schleswig-Holstein. — Meyniana, 7: 5—60; Kiel. WOLDSTEDT, P. (1913): Beiträge zur Morphologie von Nordschleswig. — Mitt. Geogr. Ges. u. d. Naturh. Mus. in Lübeck, 2 ( 2 6 ) : 41—109; Lübeck. WOLDSTEDT, P. (1961): Das Eiszeitalter. I: 374 S.; Stuttgart (Enke). Nachtrag Während der Drucklegung erschien die Arbeit KRONBORG, C., BENDER, H. & LARSEN, G. (1978) Tektonik som en mulig medvirkende ärsag til daldannelser i M i d t j y l l a n d . — Danm. Geol. Unders., Ärborg 1977: 63—76; Kopenhagen, in die die Entstehung von Tunneltälern in Zusammenhang mit Tektonik im Untergrund vorgeschlagen worden sind.
https://openalex.org/W3095126155	https://europepmc.org/articles/pmc7843857?pdf=render	English	null	IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG	Netherlands heart journal	2,020	cc-by	5,395	Abstract Background After coronary artery bypass grafting (CABG), healthcare utilisation is high and is partly unplanned. eHealth applications have been proposed to reduce healthcare consumption and to enable pa- tients to get actively involved in their recovery. This way, healthcare expenses can be reduced and the quality of care can be improved. g Conclusion This randomised trial was initiated to test the hypothesis that patients who are partaking in our eHealth programme use less unplanned care and experience a better quality of life, less anxiety and a faster recovery than controls. Objectives We aim to evaluate whether the use of an eHealth programme can reduce unplanned healthcare utilisation and improve mental and physical health in the ﬁrst 6 weeks after discharge in patients who underwent CABG. In addition, patient satisfaction and use of the eHealth programme will be evaluated. Keywords eHealth · Video consultation · Patient education · Healthcare utilisation · Emergency department visits · Coronary artery bypass grafting IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG G. J. van Steenbergen · D. van Veghel · J. ter Woorst · D. van Lieshout · L. Dekker Accepted: 13 October 2020 / Published online: 3 November 2020 © The Author(s) 2020 Results Patient enrolment started in February 2020 and completion of the follow-up period is expected in August 2021. Results Patient enrolment started in February 2020 and completion of the follow-up period is expected in August 2021. Original Article – Study Design Article Neth Heart J (2021) 29:80–87 https://doi.org/10.1007/s12471-020-01508-9 Neth Heart J (2021) 29:80–87 https://doi.org/10.1007/s12471-020-01508-9 Neth Heart J (2021) 29:80–87 https://doi.org/10.1007/s12471-020-01508-9 G. J. van Steenbergen () · D. van Veghel · J. ter Woorst · L. Dekker Catharina Heart Center, Catharina Hospital, Eindhoven, The Netherlands gijs.v.steenbergen@catharinaziekenhuis.nl D. van Lieshout Dutch Heart Foundation, The Hague, The Netherlands Original Article – Study Design Article Original Article – Study Design Article 80 IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG Methods After surgery, patients have an in-hospital physio- therapist consultation. Before discharge, a resident or nurse provides brief information about the permitted level of physical activity and the medication sched- ules and answers remaining questions patients might have. Outpatient postoperative follow-up is sched- uled for 6 weeks after discharge. Structured guidance to improve general condition and strength by a phys- iotherapist is offered (cardiac rehabilitation). Patients have no planned care in the ﬁrst 6 weeks after dis- charge. Study setting This randomised trial is conducted at the Catha- rina Hospital in the Netherlands. The trial will be reported in accordance with relevant sections from the Standard Protocol Items: Recommendations for Interventional Trials [9], and the Consolidated Stan- dards of Reporting Trials of Electronic and Mobile Health Applications and Online Telehealth. The study was approved by the local medical ethics committee (registration number R19.100) and is registered in the Netherlands Trial Registry (www.trialregister.nl, number NL8510). Introduction p g Methods For this single-centre randomised controlled trial, at least 280 patients referred for CABG will be in- cluded at the preoperative outpatient clinic and ran- domised to an intervention or control group. The in- tervention group will have access to an eHealth pro- gramme, which consists of online educational videos developed by the Dutch Heart Foundation and post- operative video consultations with a physician. The control group will receive standard care and will not have access to the eHealth programme. The primary endpoint is healthcare utilisation; other endpoints in- clude anxiety, duration of recovery, quality of life and patient satisfaction. Participants will complete several questionnaires at 6 time points during the study. Coronary artery bypass graft grafting (CABG) is the most prevalent cardiac surgery performed in the Netherlands, with roughly 7000 procedures annually [1]. The care chain of CABG is costly, and several quality improvement initiatives have been success- fully implemented that sought to contain costs and to improve patient outcomes [2, 3]. Despite the pos- itive effects of these initiatives on costs, mortality, postoperative morbidity and process measures such as in-hospital length of stay, healthcare utilisation in the ﬁrst 30 days after CABG remains an issue, placing a signiﬁcant burden on the healthcare system. Read- missions after CABG are commonly reported and the readmission rate can be as high as 34% in the ﬁrst 30 days [4, 5]. Insight into unplanned healthcare utilisation dur- ing this period is scarce (apart from readmissions), but it is reasonable to expect a short hospital stay after CABG is counterbalanced by the use of other healthcare services, especially because planned care is not initiated until 6 weeks after discharge. In this period, patients commonly experience psychological symptoms (e.g. anxiety, depression), have to deal with 80 IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG Original Article – Study Design Article Recruitment and allocation uncertainty and worry about what to expect (e.g. what level of postoperative pain is normal, is physical exer- cise allowed?) [6]. Recall of information provided pe- rioperatively is often incomplete and patients do not always know who to contact in case of complaints. They will then search for (sometimes unreliable) in- formation on surgery or recovery and reach out to dif- ferent healthcare providers, who have a varying degree of expertise in CABG care. Conﬂicting advice on re- covery can further increase fear and insecurity, which will eventually hamper the recovery process and con- tribute to unplanned healthcare utilisation [7, 8]. All patients planned for preoperative outpatient coun- selling for CABG will be contacted by telephone in the week of their scheduled appointment. Patient eligibil- ity for the trial will be assessed according to prespeci- ﬁed inclusion criteria (Tab. 1). Eligible patients will be informed about the study protocol by one of the in- vestigators after their scheduled appointment. If the patient is willing to participate, the informed consent form is signed. At inclusion, patients are randomised to either the intervention or control group in a ratio 1:1 using block randomisation, with a block size of four. We hypothesise that restructuring the postoper- ative period with an eHealth strategy will reduce unplanned healthcare utilisation through improved mental and physical health and faster recovery. In the IMPROV-ED trial, we aim to evaluate whether the use of an eHealth programme that consists of educational videos developed by the Dutch Heart Foundation (Hartstichting) and video consultations, is more effec- tive than standard care in the reduction of unplanned healthcare utilisation and the improvement of patient outcomes in the ﬁrst 6 weeks after CABG. In addition, a process and patient satisfaction evaluation of the newly developed eHealth strategy will be conducted. Intervention group Aside from standard care as described in the previ- ous section, patients in the intervention group will have access to educational videos and will be invited to two video consultations. Access will be granted through a link sent by email. The educational videos will be made available directly after randomisation for the duration of the study, via a secure online portal. The portal provides an orderly index in which patients can navigate by means of preformulated questions stratiﬁed in three categories: treatment, recovery and healthy living (Fig. 1). Each question will be accom- panied by an educational video, which provides infor- mation and practical advice if applicable (Fig. 2). Control group Patients randomised to the control group will receive standard care according to the local protocol. One month prior to surgery, patients are invited to the preoperative outpatient clinic, where they are indi- vidually counselled by a physician and a nurse prac- titioner. They are also handed information brochures after a nurse-led group session during which they re- ceive information on the CABG care process at our hospital. Outcomes and 4); all information is in Dutch. A nurse practi- tioner experienced in care for cardiothoracic surgery patients will conduct the video consultations with pa- tients on their recovery and any complaints. Super- vision will be provided by a cardiothoracic surgeon. The nurse practitioner will be told the study’s aim is to ‘improve the current follow-up procedure’ and that he or she will therefore be blinded for the speciﬁc out- comes. and 4); all information is in Dutch. A nurse practi- tioner experienced in care for cardiothoracic surgery patients will conduct the video consultations with pa- tients on their recovery and any complaints. Super- vision will be provided by a cardiothoracic surgeon. The nurse practitioner will be told the study’s aim is to ‘improve the current follow-up procedure’ and that he or she will therefore be blinded for the speciﬁc out- comes. Original Article – Study Design Article Original Article – Study Design Article Fig. 1 Welcome screen of the eHealth programme with stratification of top- ics on which patients can find information: treatment (Behandeling), recovery (Herstel) and healthy living (Gezond leven) Fig. 2 Overview of the portal for the eHealth pro- gramme Fig. 3 Still from the video ‘What is coronary artery by- pass surgery?’ Fig. 1 Welcome screen of the eHealth programme with stratification of top- ics on which patients can find information: treatment (Behandeling), recovery (Herstel) and healthy living (Gezond leven) Fig. 2 Overview of the portal for the eHealth pro- gramme Fig. 1 Welcome screen of the eHealth programme with stratification of top- ics on which patients can find information: treatment (Behandeling), recovery (Herstel) and healthy living (Gezond leven) Fig. 1 Welcome screen of the eHealth programme with stratification of top- ics on which patients can find information: treatment (Behandeling), recovery (Herstel) and healthy living (Gezond leven) Fig. 3 Still from the video ‘What is coronary artery by- pass surgery?’ Table 1 Eligibility criteria Criterium 1 >18 years of age planned for elective, isolated CABG/OPCAB 2 Sufﬁcient computer knowledge, and internet access. Children can assist, but patients should be able to access their own email and navigate the internet to use the provided eHealth strategy 3 Access to computer with internet connection and webcam or build-in camera 4 Comply to minimal speciﬁcations for use of video consultation: – PC/laptop: Windows 7 or 10 with Chrome or Firefox browser – Android tablet: at least Nougat software installed and use of Chrome browser – Apple iPad: at least iOS 12.3.4 5 Ability to speak, read and interpret the Dutch language 6 Provide informed consent CABG/OPCAB coronary artery bypass grafting/off-pump coronary bypass 1 >18 years of age planned for elective, isolated CABG/OPCAB 2 Sufﬁcient computer knowledge, and internet access. Children can assist, but patients should be able to access their own email and navigate the internet to use the provided eHealth strategy 3 Access to computer with internet connection and webcam or build-in camera 4 Comply to minimal speciﬁcations for use of video consultation: – PC/laptop: Windows 7 or 10 with Chrome or Firefox browser – Android tablet: at least Nougat software installed and use of Chrome browser – Apple iPad: at least iOS 12.3.4 5 Ability to speak, read and interpret the Dutch language 6 Provide informed consent CABG/OPCAB coronary artery bypass grafting/off-pump coronary bypass The aim of the videos is to prepare patients, and their caregivers and family members for the surgery and offer guidance during the recovery process. The videos were developed by the Dutch Heart Foundation and were made available speciﬁcally for this study. Videos contain spoken text with animations (Figs. 3 IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG 81 Fig. 2 Overview of the portal for the eHealth pro- gramme 82 IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG Fig. 1 Welcome screen of the eHealth programme with stratification of top- ics on which patients can find information: treatment (Behandeling), recovery (Herstel) and healthy living (Gezond leven) Secondary outcome measures Secondary outcome measure are deﬁned as the in- dividual care activities of unplanned healthcare util- isation, a composite endpoint of consultations with a general practitioner, allied health professionals and psychologists plus patient-initiated telephone consul- tations with a physician or nurse, and a composite endpoint of in-hospital care (emergency department visits, outpatient clinic visits, rehospitalisation). Primary outcome measure The primary outcome is the volume of unplanned healthcare utilisation (resources used per patient). Healthcare utilisation is deﬁned as a composite end- point of all emergency department visits, outpatient clinic visits, rehospitalisation, patient-initiated tele- phone consultations with a physician or nurse, and visits to a general practitioner, allied health profes- sionals and psychologists. An adaptation of the Institute for Medical Tech- nology Assessment Medical Consumption Question- 82 IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG Original Article – Study Design Article Fig. 4 Still from the video ‘How to prepare for coro- nary artery bypass surgery?’ tive domains on a 5-point scale over the last 7 days. It has been validated in gynaecology patients, but the general nature of the questions suits our patient pop- ulation [19]. naire (iMCQ) will be used to determine the volume of healthcare utilisation and the reasons thereof. This generic questionnaire aims to determine the costs of healthcare based on care consumption and is appli- cable to the Dutch healthcare system [10]. The ques- tionnaire answers will result in absolute frequencies of visits for the relevant care activities. When patients report the use of healthcare, their healthcare provider will be contacted to validate the date and the reason for the healthcare encounter. This information will be crossvalidated with the subjects’ self-reports. naire (iMCQ) will be used to determine the volume of healthcare utilisation and the reasons thereof. This generic questionnaire aims to determine the costs of healthcare based on care consumption and is appli- cable to the Dutch healthcare system [10]. The ques- tionnaire answers will result in absolute frequencies of visits for the relevant care activities. When patients report the use of healthcare, their healthcare provider will be contacted to validate the date and the reason for the healthcare encounter. This information will be crossvalidated with the subjects’ self-reports. Patient and procedural data Patient characteristics (age, sex, comorbidities), pro- cedural characteristics (in-hospital complications, length of stay, duration of surgery), follow-up data (mortality, reoperations, deep sternal wound infec- tions, stroke, recurrent myocardial infarction within 30 days) and sociodemographic information will be collected and analysed to provide insight into our pa- tient population and to adjust endpoints if necessary. Patient, procedural and follow-up data are routinely collected at our facility and are deﬁned by the Nether- lands Heart Registry [22]. Sociodemographic data are part of the iMCQ. Tertiary outcome measures Tertiary outcome measures are the patients’ quality of life and their mental and physical status. To as- sess these domains, the 36-Item Short Form Health Survey (SF-36) [11], the anxiety subscale of the Hos- pital Anxiety and Depression Scale (HADS) [12] and an adaptation of the Recovery Index-10 (RI-10) [13] will be used at different time points during follow-up (Tab. 2). Process measures The performance of the eHealth strategy on several domains will be structurally evaluated using qualita- tive and quantitative measures [20, 21]. The process evaluation will assess participant attitude, eligibility, access, usage and engagement of the eHealth strat- egy (Tab. 3), in order to make recommendations on the development and subsequent implementation of eHealth strategies. Data will originate from the inter- net usage log of the portal and the patient satisfaction questionnaire. Fig. 4 Still from the video ‘How to prepare for coro- nary artery bypass surgery?’ Original Article – Study Design Article Original Article – Study Design Article Table 2 Schedule of enrolment, interventions and assessments Time pointa Enrolment Surgery Follow-up Variable T0 T1 T2 T3 T4 T5 T6 Enrolment – Eligibility screening X – Informed consent X – Allocation X Interventions Intervention group – Standard care X X X X X X X X – Online educational videos X X X X X X X – Video consultation X X Control group – Standard care X X X X X X X X Assessments Primary and secondary outcome measures – Healthcare utilisation (iMCQ) X Tertiary outcome measures – Quality of life (SF-36) X X – Anxiety (HADS subscale) X X X X X – Recovery (RI-10) X X X X Process measures – Patient satisfaction (satisfaction questionnaire) X – Use of intervention (internet usage log) X X X X X X X X Patient and procedural data – Sociodemographic data (iMCQ) X – Patient characteristics (patient ﬁles) X – Follow-up data (patient ﬁles) X iMCQ Institute for Medical Technology Assessment Medical Consumption Questionnaire, SF-36 36-Item Short Form Health Survey, HADS Hospital Anxiety and Depression Scale, RI-10 Recovery Index-10 aT0: 1 month before surgery; T1: 1 week after surgery; T2: 2 weeks after surgery; T3: 3 weeks after surgery; T4: 6 weeks after surgery; T5: 2 months after surgery; T6: 6 months after surgery Table 2 Schedule of enrolment, interventions and assessments pleted the video consultations (‘users only’). A p-value <0.05 will be considered statistically signiﬁcant and all analyses will be performed using SPSS 25 (SPSS Inc., Chicago, IL, USA). questionnaires, they will be contacted and kindly re- quested to ﬁll in and return the next questionnaires. Video consultation will be scheduled for 1 and 3 weeks after surgery. Statistical considerations Studies on eHealth in CABG patients and the effect on healthcare utilisation are scarce and report users per resource [23]. Our primary objective is to reduce healthcare utilisation per patient. To our knowledge, in one study, healthcare utilisation per patient was es- timated using the iMCQ at 0.88± 0.15.[24] Under the assumption of a small or medium effect of our eHealth strategy (d= 0.35), an α of 0.05 and a power of 0.80, a total sample size of 260 patients is required. The total study population is set at 280 patients (140 pa- tients per arm) to account for loss to follow-up and nonadherence to the intervention. 84 IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG Data collection The SF-36 is routinely used at our facility and has been validated in multiple patient populations, in- cluding cardiac surgery patients [14–18]. This ques- tionnaire assesses health-related quality of life in the previous 4 weeks. The HADS questionnaire is widely validated and is most commonly used to assess de- pression or anxiety. The RI-10 is a short, Dutch-lan- guage, 10-item questionnaire measuring postopera- Data will be collected using paper questionnaires at the following time points: 1 month before surgery (T0), 1 week after surgery (T1), 2 weeks after surgery (T2), 3 weeks after surgery (T3), 6 weeks after surgery (T4), 2 months after surgery (T5) and 6 months after surgery (T6) (Tab. 2). If patients do not return two subsequent IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG MPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG 8 Data analysis Descriptive statistics will be used to summarise base- line characteristics of the study population. Health- care utilisation will be expressed as mean± standard deviation (resource use per patient) and absolute and relative frequencies (users per resource). Mann-Whit- ney U test and Fisher exact test will be used to com- pare the intervention and control group. Multivari- ate regression analysis will be performed to adjust the outcomes for confounding factors based on univariate analysis (p< 0.1), literature review and expert opinion. The primary analysis will be an intention-to-treat analysis. In the secondary analysis, we will compare the control group with the intervention group that used the educational videos at least once and com- Original Article – Study Design Article Table 3 Process evaluation of eHealth strategy Domain method of collection Portal Video consultation Eligibility logistic data Number of patients not eligible for inclusion due to technological limitations (e.g. no computer access, digital illiteracy) Access logistic data Number of patients that received access to an account Number of patients that were invited via email Usage internet usage log Number of patients that logged in during the study period Number of patients that completed the planned video consultation Time points at which patients logged in Number of patients that experienced technical errors and the reasons thereof (e.g. magic link not working, bad quality of video connection) Average session length Engagement internet usage log Number of videos started and completed during each visit Average session length Number of times each video was started and completed Participants attitude questionnaire Assessment of the portal/software (e.g. accessibility, interface, navigation) and reasons for not using the portal/software (if applicable) Evaluation of content (comprehensible, use of words, useful) Number of patients that deem video consultation a meaningful addition to standard care Number of patients that would recommend the content as a source of information to family members or other patients Table 3 Process evaluation of eHealth strategy Expected results haviours (e.g. relieve stress on the sternum, gradually increase in physical exercise, follow healthy diet) and to deal with the emotions and worries that go with cardiac surgery through self-management, and, thus, to take responsibility for their own recovery [6]. eHealth has shown to be a useful method for patients to enhance their self-management through better un- derstanding of their disease, increased independence and improved acceptance to adhere to lifestyle advice [37]. The educational videos in our eHealth strategy facilitate self-management. By means of video con- sultation, the physician can guide and supervise the patient’s progress and maintain a good patient-physi- cian relationship, which has been shown to enhance the patient’s self-management skills [37]. The IMPROV-ED trial will be carried out to evaluate whether an eHealth initiative consisting of online ed- ucation and video consultation can reduce healthcare utilisation by improving quality of life, decreasing anx- iety and accelerating recovery within the ﬁrst 6 weeks after discharge for CABG. IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG 85 Limitations The addition of our eHealth strategy to the postop- erative protocol might yield additional costs in com- parison to standard care. We believe that these ad- ditional costs will be balanced by reduced healthcare utilisation and will therefore result in less total costs and better patient outcomes. Another potential limi- tation is that we will only include patients that have sufﬁcient computer and digital literacy skills and have access to a computer or tablet, which might diminish generalisability of our study protocol. Moreover, as in most eHealth research, our trial is not fully blinded, which could lead to bias when patients report health- care utilisation. 6. Lie I, Bunch EH, Smeby NA, Arnesen H, Hamilton G. Patients’experienceswithsymptomsandneedsintheearly rehabilitation phase after coronary artery bypass grafting. EurJCardiovascNurs. 2012;11:14–24. 7. Pedoto A, Perrino AC. Delayed recovery following thoracic surgery: persistentissuesandpotentialinterventions. Curr OpinAnaesthesiol. 2019;32:3–9. 8. Talboom-Kamp EP, Verdijk NA, Harmans LM, Numans ME, Chavannes NH. An eHealth platform to manage chronic disease in primary care: an innovative approach. Interact J MedRes. 2016;5:e5. 9. Chan AW, Tetzlaff JM, Altman DG, et al. SPIRIT 2013 statement: deﬁning standard protocol items for clinical trials. AnnInternMed. 2013;158:200–7. 10. Bouwmans C, Hakkaart-van Roijen L, Koopmanschap M, Krol M, Severens H, Brouwer W. Productivity costs questionnaire. 2013. https://docplayer.nl/60086739- Medical-consumption-questionnaire-productivity-and- health-research-group.html, last access: 15-04-2020, Pro- ductivity and Health Research Group. Inst Med Technol EarsmusUnivRotterdam. Acknowledgements The Dutch Heart Foundation provided the online educational videos and contextMe provided the platform by which video consultation was facilitated. Funding This work was supported by the Dutch Heart Foun- dation and contextMe. Each party was responsible for the quality assurance, maintenance and dispersion of its intel- lectual property. Thus, neither sponsor provided reimburse- ment. 11. Sanderman R, van der Zee KI. Het meten van de algemene gezondheidstoestandmetdeRand-36. 2012. https://www. umcg.nl/SiteCollectionDocuments/research/institutes/ SHARE/assessmenttools/handleiding_rand36_2e_druk. pdf,lastaccess: 15-04-2020,ResInstSHARE. Conﬂict of interest G.J. van Steenbergen, D. van Veghel, J. ter Woorst, D. van Lieshout and L. Dekker declare that they have no competing interests. 12. Spinhoven P, Ormel J, Sloekers PPA, Kempen GIJM, Speck- ensAEM,VanHemertAM.Avalidationstudyofthehospital anxiety and depression scale (HADS) in different groups of Dutchsubjects. PsycholMed. 1997;27:363–70. Discussion According to post-CABG protocols, patients are expected to adopt new be- IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG 85 Original Article – Study Design Article 5. Shah RM, Zhang Q, Chatterjee S, et al. Incidence, cost, and risk factors for readmission after coronary artery bypass grafting. AnnThoracSurg. 2019;107:1782–9. Limitations Open Access This article is licensed under a Creative Com- mons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permis- sion directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. 13. Kluivers KB, Hendriks JCM, Mol BWJ, et al. Clinimet- ric properties of 3 instruments measuring postoperative recovery in a gynecologic surgical population. Surgery. 2008;144:12–21. 14. Jensen BØ, Hughes P, Rasmussen LS, Pedersen PU, Stein- brüchel DA. Health-related quality of life following off- pump versus on-pump coronary artery bypass grafting in elderly moderate to high-risk patients: a randomized trial. EurJCardiothoracSurg. 2006;30:294–9. g 15. Goldsmith IRA, Lip GYH, Patel RL. A prospective study of changes in the quality of life of patients following mitral valve repair and replacement. Eur J Cardiothorac Surg. 2001;20:949–55. 16. Hansen L, Winkel S, Kuhr J, Bader R, Bleese N, Riess FC. Factors inﬂuencing survival and postoperative quality of life after mitral valve reconstruction. Eur J Cardiothorac Surg. 2010;37:635–44. Discussion The IMPROV-ED trial is of clinical signiﬁcance for sev- eral reasons. First, we will evaluate the inﬂuence of an eHealth strategy on healthcare utilisation, anxiety, quality of life and recovery. Positive results will yield a new postoperative protocol that will lead to better patient outcomes and reduced costs [25]. In addi- tion, the process and patient satisfaction evaluation will show the readiness of CABG patients for struc- tured eHealth initiatives and will evaluate the cur- rently used content and mode of administration, given the broad applicability of eHealth in general and the multitude of devices available [26–28]. Second, the control arm of the trial will provide the ﬁrst detailed insight into unplanned, transmural healthcare utili- sation in the early postoperative period after CABG and will thereby show how to further improve post- CABG protocols, aside from eHealth, through multi- disciplinary regional collaboration. p g The message and content of the educational videos were designed in such a way that they provide health information for patients with low/inadequate health literacy (approximately 36.4% of the general popu- lation in the Netherlands [38]), without compromis- ing health communication to patients with adequate health literacy. Meppelink et al. have assessed the fea- tures of health information (written vs spoken text vs animations vs illustrations) and concluded that spo- ken text combined with animation is the most ef- fective way to communicate health information and that it suits both patients with low health literacy and those with adequate health literacy [39]. In addition, to prevent cognitive overload, it is advised to only of- fer information when it is applicable to the patient’s situation instead of presenting all the information at once, especially to not overburden low health liter- ate patients [40]. We therefore decided to divide the information into the three main phases of CABG re- covery (Fig. 2). eHealth strategies in CABG patients have been suc- cessfully applied to guide secondary prevention [29], to improve recovery [30, 31], and to assess physi- cal functioning and quality of life [32–34]. Although evidence on the effect of eHealth on healthcare util- isation in CABG patients is minimal, it is reasonable to expect a positive effect based on reduction of healthcare utilisation by eHealth strategies in other populations [8, 23, 35, 36]. References 1. Nederlandse Hart Registratie. Nederlandse Hart Regis- tratie 2018. 2018;(april2019):249. Available from: https:// nederlandsehartregistratie.nl/wp-content/uploads/2018/ 12/NHR_Publicatie_Registratie_2018.pdf g 17. Santini F, Montalbano G, Messina A, et al. Survival and quality of lifeafter repair of acutetypeA aortic dissection in patients aged 75 years and older justify intervention. Eur J CardiothoracSurg. 2006;29:386–91. 2. Ferguson TB, Peterson ED, Coombs LP, et al. Use of continuousqualityimprovementtoincreaseuseofprocess measures in patients undergoing coronary artery bypass graft surgery: a randomized controlled trial. J Am Med Assoc. 2003;290:49–56. 18. Koertke H, Hoffmann-Koch A, Boethig D, et al. Does the noise of mechanical heart valve prostheses affect quality of life as measured by the SF-36® questionnaire? Eur J CardiothoracSurg. 2003;24:52–8. 19. van der Meij E, Huirne JA, Bouwsma EV, et al. Substitution of usual perioperative care by eHealth to enhance postop- erative recovery in patients undergoing general surgical or gynecological procedures: study protocol of a randomized controlledtrial. JMIRResProtoc. 2016;5:e245. 3. Swaminathan M, Phillips-ButeBG, Patel UD, etal. Increas- ing healthcare resource utilization after coronary artery bypass graft surgery in the United States. Circ Cardiovasc QualOutcomes. 2009;2:305–12. 3. Swaminathan M, Phillips-ButeBG, Patel UD, etal. Increas- ing healthcare resource utilization after coronary artery bypass graft surgery in the United States. Circ Cardiovasc QualOutcomes. 2009;2:305–12. 4. Hannan EL, Zhong Y, Lahey SJ, et al. 30-Day readmissions aftercoronaryarterybypassgraftsurgeryinNewYorkState. JACCCardiovascInterv. 2011;4:569–76. 4. Hannan EL, Zhong Y, Lahey SJ, et al. 30-Day readmissions aftercoronaryarterybypassgraftsurgeryinNewYorkState. JACCCardiovascInterv. 2011;4:569–76. 20. Lilford RJ, Foster J, Pringle M. Evaluating eHealth: how to makeevaluationmoremethodologicallyrobust. PLoSMed. 2009;6:e1000186. 86 IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG Original Article – Study Design Article 31. Zimmerman L, Barnason S, Nieveen J, Schmaderer M. Symptom management intervention in elderly coronary arterybypassgraftpatients. OutcomesManag. 2004;8:5–12. 21. Cissell WB. Process evaluation for public health interven- tionsandresearch. HealthEducRes. 2004;19:739. 22. Nederlandse Hart Registratie. NHR Handboek. 2019. Availablefrom: www.nederlandsehartregistratie.nl. 32. Körtke H, Stromeyer H, Zittermann A, et al. New East- Westfalian postoperative therapy concept: a telemedicine guidefor thestudy of ambulatory rehabilitation of patients aftercardiacsurgery. TelemedJEHealth. 2006;12:475–83. 23. Van der Meij E, Anema JR, Otten RHJ, Huirne JAF, Schaaf- sma FG. The effect of perioperative e-health interventions on the postoperative course: a systematic review of ran- domisedandnon-randomisedcontrolledtrials. PLoS One. 2016;11:1–24. 33. Barnason S, Zimmerman L, Nieveen J, et al. Inﬂuence of a symptom management telehealth intervention on older adults’ early recovery outcomes after coronary artery bypasssurgery. HeartLung. 2009;38:364–76. 24. IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG 87 References Pattamatta M, Smeets BJJ, Evers SMAA, Peters EG, Luyer MDP, Hiligsmann M. Quality of life and costs of patients prior to colorectal surgery. Expert Rev Pharma- coeconOutcomesRes. 2020;20:193–8. 34. Torrance GW. Preferences for health outcomes and cost- utilityanalysis. AmJManagCare. 1997;3:S8–S20. 25. Porter M. What is value in health care?—Supplementary appendix2. NEnglJMed. 2010;363:1–3. 35. Lorig KR, Ritter P, Stewart AL, et al. Chronic disease self- managementprogram: 2-yearhealthstatusandhealthcare utilizationoutcomes. MedCare. 2001;39:1217–23. 26. Piepoli MF, Hoes AW, Agewall S, et al. 2016 European guidelines on cardiovascular disease prevention in clinical practice. The sixth joint task force of the European Society ofCardiologyandothersocietiesoncardiovasculardisease prevention in clinical practice. G Ital Cardiol (Rome). 2017;18:547–612. 36. SteventonA,BardsleyM,BillingsJ,etal. Effectoftelehealth on use of secondary care and mortality: ﬁndings from the WholeSystemDemonstratorclusterrandomisedtrial. BMJ. 2012;344:e3874. 37. Lorig KR, Holman HR. Self-management education: His- tory, deﬁnition, outcomes, and mechanisms. Ann Behav Med. 2003;26:1–7. 27. Frederix I, Caiani EG, Dendale P, et al. ESC e-cardiology working group position paper: overcoming challenges in digital health implementation in cardiovascular medicine. EurJPrevCardiol. 2019;26:1166–77. 38. Sørenson K, Van den Broucke S, Fullam J, et al. (HLS- EU). Consortium Health Literacy Project Europe. Health Literacy and public health: a systematic review and in- tegration of deﬁnitions and models. BMC Public Health. 2012;25:1053–8. 28. Pagliari C, Sloan D, Gregor P, et al. What is eHealth (4): a scoping exercise to map the ﬁeld. J Med Internet Res. 2005;7:e9. 39. Meppelink CS, Van Weert JCM, Haven CJ, Smit EG. The ef- fectivenessofhealthanimationsinaudienceswithdifferent healthliteracylevels: anexperimentalstudy. JMedInternet Res. 2015;17:e11. 29. Brørs G, Pettersen TR, Hansen TB, et al. Modes of e-Health delivery in secondary prevention programmes for patients with coronary artery disease: a systematic review. BMC HealthServRes. 2019;19:364. 30. Miller C, Zimmerman L, Barnason S, Nieveen J. Impact of an early recovery management intervention on function- ing in postoperative coronary artery bypass patients with diabetes. HeartLung. 2007;36:418–30. 40. Heijmans M, Zwikker H, van der Heide I, Rademakers J. NIVEL Kennisvraag 2016: zorg op maat. Hoe kunnen we de zorg beter laten aansluiten bij mensen met lage gezondheidsvaardigheden? Utrecht: NIVEL;2016. IMPROV-ED trial: eHealth programme for faster recovery and reduced healthcare utilisation after CABG 87
https://openalex.org/W4232228294	https://research-information.bris.ac.uk/ws/files/276579221/predicting_prognosis_for_adults_with_depression_using_individual_symptom_data_a_comparison_of_modelling_approaches.pdf	English	null	Predicting prognosis for adults with depression using individual symptom data: a comparison of modelling approaches.	null	2,020	cc-by	10,430	Buckman, J. E. J., Kessler, D., Wiles, N., Pilling, S., & al., E. (2021). Predicting prognosis for adults with depression using individual symptom data: a comparison of modelling approaches. Psychological Medicine, 1-11. Advance online publication. https://doi.org/10.1017/S0033291721001616 Publisher's PDF, also known as Version of record License (if available): CC BY Link to published version (if available): 10.1017/S0033291721001616 Link to publication record on the Bristol Research Porta PDF-document Publisher's PDF, also known as Version of record License (if available): CC BY Link to published version (if available): 10.1017/S0033291721001616 Link to publication record on the Bristol Research Portal PDF-document This is the final published version of the article (version of record). It first appeared online via Cambridge University Press at https://www.cambridge.org/core/journals/psychological-medicine/article/predicting-prognosis- for-adults-with-depression-using-individual-symptom-data-a-comparison-of-modelling- approaches/DB2C74C9B69380288FA6179EB7E2B84F . Please refer to any applicable terms of use of the publisher. Buckman, J. E. J., Kessler, D., Wiles, N., Pilling, S., & al., E. (2021). Predicting prognosis for adults with depression using individual symptom data: a comparison of modelling approaches. Psychological Medicine, 1-11. Advance online publication. https://doi.org/10.1017/S0033291721001616 Buckman, J. E. J., Kessler, D., Wiles, N., Pilling, S., & al., E. (2021). Predicting prognosis for adults with depression using individual symptom data: a comparison of modelling approaches. Psychological Medicine, 1-11. Advance online publication. https://doi.org/10.1017/S0033291721001616 Psychological Medicine cambridge.org/psm Psychological Medicine Introduction Depression affects ∼320 million people worldwide every year (Thornicroft et al., 2017; Vos et al., 2016). Despite the existence of effective treatments, roughly half of depressed patients do not recover with the first treatment they are given. This can lead to disengagement and poor long-term prognoses (Buckman et al., 2018; Judd et al., 1998). Providing accurate predic- tions about the likelihood of treatment response for patients would be of great value, inform- ing clinical management and giving patients and clinicians desired information (Hayden, Windt Van Der, Cartwright, Côté, & Bombardier, 2013; Morgan, Reavley, & Jorm, 2014). However, there are a lack of accurate, validated prognostic models for adults in treatment for depression (Cohen & DeRubeis, 2018). Central to this vacancy in the literature are Abstract Background. This study aimed to develop, validate and compare the performance of models predicting post-treatment outcomes for depressed adults based on pre-treatment data. Methods. Individual patient data from all six eligible randomised controlled trials were used to develop (k = 3, n = 1722) and test (k = 3, n = 918) nine models. Predictors included depres- sive and anxiety symptoms, social support, life events and alcohol use. Weighted sum scores were developed using coefficient weights derived from network centrality statistics (models 1–3) and factor loadings from a confirmatory factor analysis (model 4). Unweighted sum score models were tested using elastic net regularised (ENR) and ordinary least squares (OLS) regression (models 5 and 6). Individual items were then included in ENR and OLS (models 7 and 8). All models were compared to one another and to a null model (mean post-baseline Beck Depression Inventory Second Edition (BDI-II) score in the training data: model 9). Primary outcome: BDI-II scores at 3–4 months. g ) y Results. Models 1–7 all outperformed the null model and model 8. Model performance was very similar across models 1–6, meaning that differential weights applied to the baseline sum scores had little impact. Conclusions. Any of the modelling techniques (models 1–7) could be used to inform prog- nostic predictions for depressed adults with differences in the proportions of patients reaching remission based on the predicted severity of depressive symptoms post-treatment. However, the majority of variance in prognosis remained unexplained. It may be necessary to include a broader range of biopsychosocial variables to better adjudicate between competing models, and to derive models with greater clinical utility for treatment-seeking adults with depression. © The Author(s), 2021. Published by Cambridge University Press. This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0), which permits unrestricted re- use, distribution and reproduction, provided the original article is properly cited. Original Article Cite this article: Buckman JEJ et al (2021). Predicting prognosis for adults with depression using individual symptom data: a comparison of modelling approaches. Psychological Medicine 1–11. https://doi.org/ 10.1017/S0033291721001616 Cite this article: Buckman JEJ et al (2021). Predicting prognosis for adults with depression using individual symptom data: a comparison of modelling approaches. Psychological Medicine 1–11. https://doi.org/ 10.1017/S0033291721001616 1Research Department of Clinical, Educational & Health Psychology, Centre for Outcomes Research and Effectiveness (CORE), University College London, 1-19 Torrington Place, London, UK; 2iCope – Camden & Islington Psychological Therapies Services – Camden & Islington NHS Foundation Trust, St Pancras Hospital, London, UK; 3Department of Psychiatry, University of California, Los Angeles, Los Angeles, CA, USA; 4Department of Clinical Psychology, Leiden University, Leiden, The Netherlands; 5Statistical Science, University College London, 1-19 Torrington Place, London, UK; 6Department of Psychology, School of Arts and Sciences, 425 S. University Avenue, Philadelphia PA, USA; 7Department of Health Sciences, University of York, Seebohm Rowntree Building, Heslington, York, UK; 8Department of Psychology, Vanderbilt University, Nashville, TN, USA; 9Primary Care, Population Sciences and Medical Education, Faculty of Medicine, University of Southampton, Aldermoor Health Centre, Southampton, UK; 10Department of Psychology, University of Exeter, Sir Henry Wellcome Building for Mood Disorders Research, Perry Road, Exeter, UK; 11Social, Genetic and Developmental Psychiatry Centre, Institute of Psychiatry, Psychology & Neuroscience, King’s College London, London, UK; 12Centre for Academic Primary Care, Population Health Sciences, Bristol Medical School, University of Bristol, Canynge Hall, Bristol, UK; 13Centre for Academic Mental Health, Population Health Sciences, Bristol Medical School, University of Bristol, Oakfield House, Bristol, UK; 14Division of Psychiatry, University College London, Maple House, London, UK and 15Camden & Islington NHS Foundation Trust, St Pancras Hospital, London, UK Received: 4 December 2020 Revised: 8 March 2021 Accepted: 12 April 2021 Received: 4 December 2020 Revised: 8 March 2021 Accepted: 12 April 2021 Author for correspondence: Author for correspondence: Joshua E. J. Buckman, E-mail: Joshua.buckman@ucl.ac.uk p Joshua E. J. Buckman, E-mail: Joshua.buckman@ucl.ac.uk cambridge.org/psm J. E. J. Buckman1,2 , Z. D. Cohen3, C. O’Driscoll1, E. I. Fried4, R. Saunders1, G. Ambler5, R. J. DeRubeis6, S. Gilbody7, S. D. Hollon8, T. Kendrick9, E. Watkins10, T.C. Eley11, A. J. Peel11, C. Rayner11, D. Kessler12, N. Wiles13, G. Lewis14 and S. Pilling1,15 J. E. J. Buckman1,2 , Z. D. Cohen3, C. O’Driscoll1, E. I. Fried4, R. Saunders1, G. Ambler5, R. J. DeRubeis6, S. Gilbody7, S. D. Hollon8, T. Kendrick9, E. Watkins10, T.C. Eley11, A. J. Peel11, C. Rayner11, D. Kessler12, N. Wiles13, G. Lewis14 and S. Pilling1,15 Key words: Depressive symptoms; major depression; network analysis; prediction modelling; prognosis University of Bristol – Bristol Research Portal General rights This document is made available in accordance with publisher policies. Please cite only the published version using the reference above. Full terms of use are available: http://www.bristol.ac.uk/red/research-policy/pure/user-guides/brp-terms/ https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291721001616 Downloaded from https://www.cambridge.org/core. IP address: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available at Participants The dataset for this study comes from a larger project investigat- ing prognosis for adults with depression in primary care, the pro- ject involved systematic literature searches to form an individual patient dataset (IPD) from eligible randomised controlled trials (RCTs; Buckman et al., 2020). The final searches were conducted on 1 December 2020 (Buckman et al., 2021b). Studies were included if they were RCTs that recruited adults with depression in primary care, and used the Revised Clinical Interview Schedule (CIS-R) (Lewis, Pelosi, Araya, & Dunn, 1992) to collect depressive and anxiety symptom data and determine diagnoses. This was to ensure uniformity across the studies in the items available for the predictive models. From our previous work we found that the CIS-R is the most commonly used comprehensive measure of this kind in studies of depression in primary care (Buckman et al., 2021a). Studies also had to use the Beck Depression Inventory Second Edition (BDI-II) (Beck, Steer, & Brown, 1996) to collect individual symptoms of depression. Six RCTs met inclusion criteria and were split such that half (k = 3, n = 1722) would form a dataset to develop the predictive models (the ‘training set’) and half (k = 3, n = 1136, of which 918 had outcome data and were used to evaluate the models as detailed below) would form a separate dataset to test the models (the ‘test set’). See online Supplementary Table S1 and Supplementary Fig. S1, for details of each study. It was decided that studies with simi- lar types of treatments would be split across the training and test sets (with all data from one study going into the training set and all data from the other study going into the test set), and where this was the case, those with the larger sample sizes would go into the training data. A recent study used centrality metrics to weight individual items of a depressive symptom questionnaire, which when summed together created a new, or weighted, sum score. A regres- sion model using this weighted sum score was found to outper- form a model containing the original sum score in an exploratory analysis (Boschloo et al., 2016). Participants Other studies have utilised centrality metrics to predict changes in particular symp- toms over time (Boschloo et al., 2016; Koenders et al., 2015; van Borkulo et al., 2015; Wichers & Groot, 2016), or predict post- treatment outcomes (Berlim, Richard-Devantoy, Dos Santos, & Turecki, 2020; Elliott, Jones, & Schmidt, 2020). However, such studies have not tested the developed models against simpler comparative models, nor have they tested the predictive utility of the models in completely external data (Dwyer et al., 2018; Harrell et al., 2004; Webb et al., 2020), or adhered to recent con- ventions for the transparency of conducting such research by fol- lowing pre-registered analysis plans or protocols (Collins, Reitsma, Altman, & Moons, 2015). Therefore, the extent to which the use of centrality metrics can add incremental value in prognostic models remains unclear. The present paper aims to fill this gap and further the consideration of the development of models that can be translated into clinical settings. There are several potentially equally valid ways to estimate item centrality in network models. We will therefore investigate several methods that have been used in the recent network mod- elling literature. One method uses the estimated arrangement of items into communities of highly partially correlated items, we will compare this to a model in which it is assumed that there is a single latent factor. We will use these methods to investigate the benefit of using item centrality scores and factor loadings to create weighted sum scores, and compare these to an unweighted regression model, and to a penalised regression model, as these J. E. J. Buckman et al. J. E. J. Buckman et al. J. E. J. Buckman et al. J. E. J. Buckman et al. are typical methods used to develop predictive models. We will then compare all of these methods against models that use all the individual items rather than sum scores, and to a simple null model (Boschloo et al., 2016). In this way, this study aims to develop, validate and compare the predictive performance of prog- nostic models for depressed adults in primary care, based on pre- treatment data including individual symptoms of depression. methodological inconsistencies, debates about how best to develop predictive models, and what variables to include in such models. Recently, the field has begun to reach consensus on how to best test the utility of predictive models, for example, by evaluating them in datasets that are separate from those used for model development (Adibi, Sadatsafavi, & Ioannidis, 2020; Dwyer, Falkai, & Koutsouleris, 2018; Harrell, Lee, & Mark, 2004; Moons et al., 2015; Steyerberg et al., 2010). ; , ; y g , ) One factor consistently found to be associated with prognosis of depression is the severity of depressive symptoms pre-treatment (Bower et al., 2013; Driessen, Cuijpers, Hollon, & Dekker, 2010; Fournier, Derubeis, Hollon, Shelton, & Fawcett, 2010; Weitz et al., 2015). This is often captured with sum scores on depressive symptom scales. However, depression is heterogeneous (Fried & Nesse, 2015a) so utilising symptom level data might provide more nuanced information on patients experiences of depression, and consequently improve the accuracy of prognostic predictions (Boschloo, van Borkulo, Borsboom, & Schoevers, 2016; Fava, Ruini, & Belaise, 2007; Fried & Nesse, 2014, 2015b). Network the- ory (Borsboom & Cramer, 2013; Fried & Cramer, 2017) has given rise to an approach that can capture the relationships between individual symptoms. These relationships could reflect potential causal pathways, thereby elucidating maintenance mechanisms that could be targeted with treatment, and might therefore inform prognosis (Borsboom, 2017). The arrangement and inter- relationships of symptoms within networks have most often been captured with one or more measures of centrality – i.e. the interconnectedness of each symptom with other symptoms in the network (Bringmann et al., 2019; Fried, Epskamp, Nesse, Tuerlinckx, & Borsboom, 2016). Ethical considerations and trial registrations All included studies were granted ethical approvals and all parti- cipants gave informed consent (online Supplementary Table S5). No additional NHS ethical approval was required for this study: HRA reference 712/86/32/81. Methods The methods for the present study were pre-registered (https://osf. io/vzk65/). We have reported the details in accordance with TRIPOD, brief details are given below, and further information including a TRIPOD checklist is available in the online Supplementary materials. https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291721001616 Downloaded from https://www.cambridge.org/core. IP address: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291721001616 Downloaded from https://www.cambridge.org/core. IP address: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available at 2 Predictors and measures Predictors varied depending on the model used, as detailed below (Table 1). Models either included total scores (with items either weighted or unweighted) or individual items from the BDI-II. All models used total scores for the eight anxiety subscales from CIS-R (generalised anxiety, worry, compulsions, obsessions, phobic anxiety, health anxiety, somatic concerns, and panic; with 3 3 Psychological Medicine Table 1. Description of the modelling approaches for the primary outcome Type of approach Weighting approach Model number Method Predictors included Description Weighted sum scores One-step EI (FGL) 1 OLS CIS-R weighted sum score for anxiety subscales, BDI-II weighted score, SSS score, LE score and AUDIT-PC score Sum of all edges connected to the focal node used to weight items to construct weighted sum scores Two-step EI (FGL) 2 OLS CIS-R weighted sum score for anxiety subscales, BDI-II weighted score, SSS score, LE score and AUDIT-PC score Sum of all edges connected to either the focal node or any other node directly connected to the focal node PC/PR (FGL) 3 OLS CIS-R weighted sum score for anxiety subscales, BDI-II weighted score, SSS score, LE score and AUDIT-PC score the geometric mean between the participation coefficient (PC) and participation ratio (PR) CFA 4 OLS CIS-R weighted sum score for anxiety subscales, BDI-II weighted score, SSS score, LE score and AUDIT-PC score Factor loadings from CFA were used as weights to develop the weighted total scores Unweighted sum scores Shrinkage 5 ENR CIS-R unweighted sum score for anxiety subscales, BDI-II score, SSS score, LE score and AUDIT-PC score ENR built using the unweighted total scores None 6 OLS CIS-R unweighted sum score for anxiety subscales, BDI-II score, SSS score, LE score and AUDIT-PC score OLS model with unweighted total scores on the baseline measures Individual symptoms Shrinkage 7 ENR CIS-R anxiety subscale items, BDI-II individual items, SSS score, LE score and AUDIT-PC score ENR model using all of the individual items of BDI-II, anxiety sub-scores of CIS-R and total scores of other measures None 8 OLS CIS-R anxiety subscale items, BDI-II individual items, SSS score, LE score and AUDIT-PC score OLS regression model with items assessing the same symptoms included in weighted models. Model building Nine models were constructed in the training set (Table 1) for both primary and secondary outcomes, so 18 models were fitted overall. Predictors and measures The sum score therefore represented the best measure of alcohol use. items either weighted or unweighted), and total scores for alcohol use, social support and life events. In previous studies using similar data it has been found that these factors are independently associated with poorer prognoses, and may have utility in predicting treatment outcomes (Buckman et al., 2021a; Buckman et al., 2021b; O’Driscoll et al., 2021). The total scores for the social support, life events and alcohol measures were required instead of the individual items. There was strong topological overlap between the social support items, and all eight items were highly correlated with one another, which would have led to inflated centralityscores were the individual items included in the network models. Further, the level of multi- collinearity went beyond pairs of items, so instead of removing those leading to high multi-collinearity, it was necessary to use the sum score as the best measure of this construct. Modelling binary items into a network is possible but not when using the fused graph- ical least absolute shrinkage and selection operator (LASSO) (FGL) method adopted here to deal with between-study heterogeneity, so the total score from the life events scale was used. As alcohol misuse was an exclusion criterion for some of the eligible RCTs, there was near zero variability for many of the items. The sum score therefore represented the best measure of alcohol use. Data analysis Missing data were imputed in the training set for all variables with <30% missing, using the ‘missForest’ package in R (Stekhoven & Bühlmann, 2012). In the test set, the same approach was used but outcome data were not imputed. The maximum amount of miss- ing data of any of the variables used in the predictive models here, at baseline in any of the six studies was 0.83%. In the test set, 218 participants were missing outcome data and were excluded from the analyses. For one study whose data were included in the train- ing set, ‘COBALT’, BDI-II was not collected at 3–4 months. These scores were imputed using the methods above based on all avail- able variables in that study including baseline BDI-II scores and patient health questionnaire-9 (PHQ-9) scores, 3-month PHQ-9 scores, 6-month BDI-II and PHQ-9 scores, and 12-month BDI-II and PHQ-9 scores (see online Supplementary for additional details). The null models used the BDI-II total score only. See online Supplementary Table S2 for details of the measures. Predictors and measures Null model None 9 OLS Mean BDI-II sum score Mean BDI-II score in training set studies used as prediction for all cases in test set BDI-II, Beck Depression Inventory Second Edition; CFA, confirmatory factor analysis; CIS-R, Revised Clinical Interview Schedule; EI, expected influence; ENR, elastic net regularised regression; FGL, fused graphical LASSO; LE, life events; OLS, ordinary least squares; PC/PR, geometric mean between the participation ratio and participation coefficient; SSS, social support scale. BDI-II, Beck Depression Inventory Second Edition; CFA, confirmatory factor analysis; CIS-R, Revised Clinical Interview Schedule; EI, expected influence; ENR, elastic net regularised regression; FGL, fused graphical LASSO; LE, life events; OLS, ordinary least squares; PC/PR, geometric mean between the participation ratio and participation coefficient; SSS, social support scale. BDI-II, Beck Depression Inventory Second Edition; CFA, confirmatory factor analysis; CIS-R, Revised Clinical Interview Schedule; EI, expected influe FGL, fused graphical LASSO; LE, life events; OLS, ordinary least squares; PC/PR, geometric mean between the participation ratio and participat post-baseline, defined as a score of ⩽10 on the BDI-II. In all but one of the six studies, assessors and analysts were blind to treatment allocation when collecting these data. items either weighted or unweighted), and total scores for alcohol use, social support and life events. In previous studies using similar data it has been found that these factors are independently associated with poorer prognoses, and may have utility in predicting treatment outcomes (Buckman et al., 2021a; Buckman et al., 2021b; O’Driscoll et al., 2021). The total scores for the social support, life events and alcohol measures were required instead of the individual items. There was strong topological overlap between the social support items, and all eight items were highly correlated with one another, which would have led to inflated centralityscores were the individual items included in the network models. Further, the level of multi- collinearity went beyond pairs of items, so instead of removing those leading to high multi-collinearity, it was necessary to use the sum score as the best measure of this construct. Modelling binary items into a network is possible but not when using the fused graph- ical least absolute shrinkage and selection operator (LASSO) (FGL) method adopted here to deal with between-study heterogeneity, so the total score from the life events scale was used. As alcohol misuse was an exclusion criterion for some of the eligible RCTs, there was near zero variability for many of the items. https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291721001616 Downloaded from https://www.cambridge.org/core. IP address: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available at 017/S0033291721001616 ress: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available at Null models A null model was built for each outcome for the purpose of com- parison. For the primary outcome, this used the mean 3–4-month BDI-II score in the training set as the prediction for all patients in the test set, and for the secondary outcome the proportion of par- ticipants in remission in the training set was used as the predic- tion for all patients in the test set. Sensitivity analyses In order to assess the impact of having to impute the 3–4-month BDI-II outcomes for the COBALT study, we conducted two sen- sitivity analyses. All analyses using BDI-II as the outcome were re-done excluding COBALT from the training dataset. Then, a different way of capturing depressive symptoms at 3–4 months was calculated based on a method of converting scores from dif- ferent depressive symptom measures to a single comparable score; the PROMIS T-score (Choi, Schalet, Cook, & Cella, 2014), using a multidimensional item-response theory-based conversion tool (Fischer & Rose, 2016), see online Supplementary for further details. Penalised regression analyses Model 5 was an ENR model built using the unweighted total scores on the same scales that were used for models 1–4. In ENR, variables are selected and model weights are assigned through the use of LASSO and ridge penalisations. Parameter space was searched using 10-fold cross-validation to identify the optimal settings for these parameters before building the final model (Friedman, Hastie, & Tibshirani, 2010; Webb et al., 2020). Model 7 was an ENR using all of the individual items from the BDI-II and the CIS-R anxiety subscales, and total scores for life events, social support and alcohol use. Model evaluation Models were first evaluated in the full test set comprising three studies (TREAD, IPCRESS and MIR), and then separately in each of the three study samples. They were also evaluated in a 10-fold internal cross-validation of the full training set data. For the continuous outcomes, there were three metrics used to evaluate the models: the amount of variance explained (R2), the root mean-squared error (RMSE), and the mean absolute error (MAE). For the binary outcome, there were two metrics used to evaluate the models: the area under the receiver operating charac- teristic curve, and Brier scores. Since the R2 in this study is a com- parison of the predicted BDI-II score values to the mean BDI-II score at 3–4 months in the test set, and the training and test set Non-penalised regression analyses Two simple comparison models were constructed using non- penalised regression (OLS regression for continuous outcomes and logistic regression for binary outcomes). Model 6 used the unweighted total scores on the five baseline measures, and model 8 used the same items as model 7. Outcomes The primary outcome was the BDI-II score at 3–4 months post- baseline. The secondary outcome was remission at 3–4 months For the first four models, we developed separate weighted sum scores for the CIS-R anxiety subscales by summing together 4 J. E. J. Buckman et al. J. E. J. Buckman et al. communities of items via random walks). Factor loadings were rescaled to be between 0 and 1 and summed to develop the weighted total scores. coefficient weights for each of the eight subscales, and for the BDI-II by summing together coefficient weights for each of the 21 BDI-II items. Weighted sum scores for the CIS-R anxiety sub- scales and BDI-II, and coefficient weights for the total scores for social support, life events, and alcohol were used as predictors by entering them into regression models (ordinary least squares (OLS) for the primary outcome and logistic regression for the sec- ondary outcome). This follows a method used by others to develop predictive models from networks (Boschloo et al., 2016). As described below, models 5 and 7 were based on a method that develops model weights internally (elastic net regu- larised regression (ENR)). Models 6 and 8 used the original, unweighted scores as a means of comparison. Model 9 was a null model, detailed further below. Network analyses There are two established ways to estimate a network model across several datasets. First, pool the data and estimate a model. Second, a recent innovation in network methods, the FGL (Costantini & Epskamp, 2017b; Fried et al., 2018), which estimates a model on several datasets and obtains one network. The FGL uses extended Bayesian information criterion, LASSO regularised regression models run separately for each study, and the models are then fused together to get a single network pena- lising differences among corresponding edge weights in the study networks. It is therefore considered better suited to deal with between-study heterogeneity (Costantini et al., 2019), and so was the method used here. For further details on how to estimate and interpret network structures and a comprehensive review of the network literature (see Epskamp & Fried, 2018; Robinaugh, Hoekstra, Toner, & Borsboom, 2020). For models 1–3, the FGL model was estimated using item-level data from CIS-R anxiety subscales and the BDI-II with tuning parameters selected through 10-fold cross validation (Costantini & Epskamp, 2017a; Danaher, Wang, & Witten, 2014). Centrality metrics derived from the FGL were used to construct weights after re-scaling these to be between 0 and 1. The three methods for determining coefficient weights from the estimated networks were: model (1) one-step expected influence (EI: sum of all edges connected to the focal node); model (2) two-step EI (sum of all edges connected to either the focal node or any other node directly connected to the focal node) (Robinaugh, Millner, & McNally, 2016); and model (3) the geometric mean of the participation coefficient (PC) and par- ticipation ratio (PR) (Letina, Blanken, Deserno, & Borsboom, 2019). See online Supplementary materials for details. The EI metrics are widely used and have recently been proposed to be informative for predicting treatment outcomes (Berlim et al., 2020; Elliott et al., 2020). PC/PR is a newer approach, which is thought to be more sensitive to the use of different scale measures within the same network as it takes the community structure (multidimensionality) into account (Letina et al., 2019). This is important here as we used measures of severity beyond depressive symptoms, given their importance for prognosis (Buckman et al., 2021a; Lorenzo-Luaces, Rodriguez-Quintana, & Bailey, 2020). https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291721001616 Downloaded from https://www.cambridge.org/core. IP address: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available at Confirmatory factor analyses Model 4 was a unidimensional confirmatory factor analytic (CFA) model that assumes the data come from a single dimensional latent construct (in contrast to model 3, which is based on a Walktrap algorithm that identifies densely connected 5 Psychological Medicine BDI-II score means at 3–4 months differed, it was expected that some models might have R2 values less than zero. There are limits to the inferences that can be drawn from the above metrics due to the variability in the modelling schemes that were applied (e.g. in which variables were made available; the number of variables made available; whether or not network analysis or factor analysis was used to create weighted sum scores; and whether or not pena- lised regression was applied to the variables that were made avail- able). To make these performance metrics more accessible, we have provided three visualisations that demonstrate the potential clinical relevance of each model. For each of the eight models (excluding the null model) the predicted BDI-II scores at 3–4 months were arrayed from the lowest to the highest, then: (1) we plotted the observed BDI-II score at 3–4 months against the predicted score in groups (‘bins’) of n = 50; (2) predicted scores were split into cat- egories of severity in line with delineations made by the originators of the scale (Beck et al., 1996) (i.e. scores between 0 and 13 were considered minimal, 14 and 19 mild, 20 and 28 moderate, and 29 and 63 severe), and the rate of remission observed in the test set samples was calculated for each category; and (3) to provide a more granular visualisation of remission we plotted the observed percentage of participants in remission against BDI-II predicted scores at 3–4 months, again in bins of n = 50. would get better (remit) and by what magnitude (BDI-II score) at 3–4 months. To illustrate this, the predictions produced for the primary outcome by the models were highly correlated (all correlation coefficients above r = 0.90 for models 1–6 and above r = 0.75 for models 7–8) see online Supplementary Fig. S2. Descriptive statistics In order to evaluate the potential clinical relevance of the mod- els we determined the observed proportion of participants in remission at 3–4 months based on the predicted score made by each model (online Supplementary Fig. S3), and the same based on categories of severity of symptoms taken from the predicted scores (see Fig. 2). From these figures we can see that when the models predicted high BDI-II scores at 3–4 months the chances of being in remission were very low. Models 7 and 8 predicted more participants would have severe depression at 3–4 months than the other models. When the models predicted minimal symptoms (BDI-II scores <10) the observed rate of remission was around 50%. There were few differences between the models overall, although greater variations in the observed rates of remis- sion between the models for patients predicted to have mild to moderate BDI-II scores at 3–4 months. Descriptive statistics and comparisons of the distributions of socio-demographics and markers of severity across the training set and test set samples are provided in Table 2. There were some differences between the training and test datasets: fewer people of non-White ethnicities were in the test set, and more of the training sample were unemployed. On average the test set participants had more comorbid disorders although a higher pro- portion of the training set sample had comorbid panic disorder, specific phobias, or chronic fatigue syndrome. The mean score on the AUDIT-PC was higher in the test set. In addition, the mean BDI-II scores were higher in the test set (by 2.47 points at baseline and 3.53 points at 3–4 months). This corresponded with a large difference in the proportions of each sample reaching remission: 48.83% in the training set and 32.53% in the test set. Sensitivity analyses did not lead to any substantive differences in our findings, see online Supplementary Tables S2 and S3. Discussion There were few differences in the performance of the majority of the predictive models. The first seven models all outperformed the null models on all metrics for primary and secondary outcomes. Those using weighted or unweighted sum scores (the first six models) performed better in the held-out test data than the individual-item models did, particularly model 8 (the OLS regres- sion model using all of the individual BDI-II score items and Characteristics of the included studies Six RCTs met inclusion criteria, three formed the training dataset (n = 1772) and three formed the test dataset (n = 1136, of which n = 918 had outcome data available for analyses), see online Supplementary Fig. S1 for flow of studies and online Supplementary Table S1 for details of each study. Formation of the models The weights given to the individual items for models 1–4 are shown in online Supplementary Table S6. Final model coefficients are presented in online Supplementary Tables S7 and S8. Confirmatory factor analyses ) pp y g For the primary outcome (BDI-II score at 3–4 months post- baseline) in the combined test sets, the RMSE was similar for models 1–6 (the largest difference was between model 2 which had the lowest RMSE and model 4, =0.057) with a slightly higher RMSE for the OLS individual-item model (model 8) (difference between model 2 and model 8 = 0.214). Models 1–8 made similar predictions for those with BDI-II scores at 3–4 months that were <18 or >25, but diverged more in the predictions for those with scores between 18 and 24, see Fig. 1 (for ease of presentation, results are displayed for groups of 50 participants, each point shows the mean predicted and observed score for the 50 partici- pants closest to that point on the graph). All models (1–8) had lower RMSE scores than the null model (ranging from 0.944 for the difference between models 8 and 9 to 1.158 for the difference between models 2 and 9), see Table 3. The amount of variance explained by models 1–7 was again very similar with R2 values between 0.157 and 0.169. Model 8 (R2 = 0.109) explained less variance, but all models had R2 values well above the null model (R2 = −0.01). MAE values were similar for models 1–7 (ranging between 9.089 for model 5 and 9.173 for model 7). MAE was slightly higher in model 8 (=9.279) and higher again in the null model (9.935), see Table 3. For the secondary outcome there was a similar pattern to the results, although the null model (9) had a similar Brier score to models 1–7 and this was slightly lower than that of model 8 (=0.246), see online Supplementary Table S3. There were greater variations between the models in the separate test set studies than in the overall test set and for all models (1–9). Additionally, the RMSE and MAE scores were lower, and R2s were higher, in the internal cross-validation than in the external test set data. https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291721001616 Downloaded from https://www.cambridge.org/core. IP address: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available at Comparison of model performance After the models were developed they were evaluated using the test dataset. Despite slight differences in the formation of some of the models, they made very similar predictions of who J. E. J. Buckman et al. 6 Table 2. Descriptive statistics for training and test set samples, and comparison of the two datasets Table 2. Comparison of model performance BDI-II score at 3–4 months in combined test set data (n = 918) by the eight models (excluding the null model) built in the Training set Fig. 1. Predicted and observed BDI-II score at 3–4 months in combined test set data (n = 918) by the eight models (excluding the null mo d (Buckman et al., 2021a; Buckman et al., 2021b). We also used the most commonly utilised comprehensive measure of depressive and anxiety symptoms and diagnoses from RCTs of depression in primary care, to minimise bias in harmonising data, and ensure a broad range of depressive and anxiety based symptoms could be included in the models we developed. eight CIS-R anxiety subscale scores instead of the sum scores for each). Any of the eight models could be used to predict the sever- ity of depressive symptoms at 3–4 months after starting treatment based on pre-treatment data. The large difference in observed remission rates between those predicted to have high compared to low BDI-II scores at 3–4 months informs the potential clinical relevance of these models. However, there were a number of limitations. Not all import- ant covariates were controlled for: we did not include data on durations of depression or anxiety despite their associations with prognosis for adults with depression (Buckman et al., 2021a; Lorenzo-Luaces et al., 2020). Including such data would have led to problems of multi-collinearity with the symptoms of the individual comorbid anxiety disorders experienced by each participant, and across durations of anxiety disorders and depres- sion, biasing centrality estimates and factor loadings for models 1–4. The intercepts and coefficient weights provided in the online Supplementary materials could be used to derive prognostic pre- dictions for future depressed patients using models developed here. However, there were large amounts of variance in the out- come that could not be explained by any of the models. This is consistent with other studies that developed and validated pre- dictive models for patients with depression (Delgadillo, Huey, Bennett, & McMillan, 2017; Webb et al., 2020). Some of the unex- plained variance is likely due to measurement error and other fac- tors, including those that better capture the biopsychosocial complexity of depression. We speculate that such factors would need to be included before the predictive models could more accurately predict prognosis for any individual patient (Fried & Robinaugh, 2020). Crucially, for this study, such improvements https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291721001616 Downloaded from https://www.cambridge.org/core. IP address: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available at Comparison of model performance Descriptive statistics for training and test set samples, and comparison of the two datasets Train set Test set t-test or χ2 Self-reported baseline characteristics Factor N (%) or mean (S.D.) N (%) or mean (S.D.) p value Sample size 1772 1136 Age in years Mean (S.D.) 42.1 (14.0) 43.2 (14.3) 0.051 Gender Female 1131 (65.7) 769 (67.8) 0.237 Male 59 (34.3) 365 (32.2) Ethnicity White 1613 (93.7) 1085 (95.6) 0.028 Non-White 109 (6.33) 50 (4.41) Employment status Employed 996 (57.8) 643 (56.7) 0.002 Not seeking employment 379 (22.0) 306 (27.0) Unemployed 347 (20.2) 185 (16.3) Marital status Married/cohabiting 819 (47.6) 560 (49.3) 0.608 Single 560 (32.5) 351 (30.9) No longer married 343 (19.9) 225 (19.8) Number of recent life events Mean (S.D.) 1.39 (1.26) 1.28 (1.20) 0.021 Social support total Median (interquartile range) 21 (18 to 24) 22 (18 to 24) 0.752 AUDIT-PC score Mean (S.D.) 2.57 (2.87) 3.13 (3.26) <0.001 Past antidepressant use No 537 (31.2) 371 (32.7) 0.408 Yes 1185 (68.8) 765 (67.3) CIS-R Sum of anxiety Subscales score Mean (S.D.) 13.7 (6.85) 13.9 (6.31) 0.437 CIS-R durations Depression 3.38 (1.44) 3.48 (1.25) 0.056 Average anxiety duration 2.14 (1.00) 2.13 (0.97) 0.780 Comorbid anxiety disorders Mean (S.D.) 2.03 (1.17) 2.19 (1.05) 0.0002 Agoraphobia No 1554 (89.7) 991 (87.3) 0.052 Yes 178 (10.3) 144 (12.7) Chronic fatigue syndrome No 615 (35.7) 348 (30.6) 0.005 Yes 1107 (64.3) 788 (69.4) Generalised anxiety disorder No 701 (40.7) 492 (43.4) 0.162 Yes 1021 (59.3) 643 (56.7) Mixed anxiety and depressive disorder No 1241 (72.1) 798 (70.3) 0.292 Yes 481 (27.9) 338 (29.8) Obsessive compulsive disorder No 1477 (85.8) 962 (84.7) 0.42 Yes 245 (14.2) 174 (15.3) Panic disorder No 1562 (90.7) 1061 (93.4) 0.01 Yes 160 (9.3) 75 (6.60) Specific phobias No 1406 (81.7) 967 (85.1) 0.015 Yes 316 (18.4) 169 (14.9) Baseline BDI-II score Mean (S.D.) 29.5 (11.1) 31.9 (9.45) <0.001 Three–four months BDI-II score Mean (S.D.) 14.4 (11.4) 17.9 (12.4) <0.001 Remission 3–4 months No 742 (51.2) 621 (67.7) <0.001 Yes 708 (48.8) 297 (32.4) Baseline PROMIS score Mean (S.D.) 70.3 (8.38) 73.3 (6.36) <0.001 Three–four months PROMIS score Mean (S.D.) 60.1 (11.5) 60.4 (12.5) 0.499 7 7 Psychological Medicine Fig. 1. Predicted and observed BDI-II score at 3–4 months in combined test set data (n = 918) by the eight models (excluding the null model) built in the Training set data. Strengths and limitations ENRb 11.359 0.157 9.173 11.869 0.138 9.798 11.201 0.178 9.075 11.216 0.123 8.871 9.881 0.233 7.886 8. OLS 11.495 0.137 9.279 12.192 0.090 10.084 11.225 0.174 9.094 11.375 0.098 8.976 9.904 0.230 7.881 Null 9. Null 12.439 −0.010 9.935 12.852 −0.011 10.396 12.544 −0.031 9.993 11.975 0.000 9.521 11.270 −0.001 9.026 p g In this study, predictions of prognosis were made regardless of the type of treatment given, as this may have most utility at the point when patients are seeking treatment, i.e. before a decision on the type of treatment has been made (Buckman et al., 2021a; Marwood, Wise, Perkins, & Cleare, 2018). Although the train and test set studies were split such that where possible, there was a balance of treatment types across the datasets, it may be the case that the models would perform differently between types of treatments. Future studies might address differ- ential model performance by treatment type but adequate data to do so were not available here (Fisher, Carpenter, Morris, Freeman, & Tierney, 2017b). y The present study used prognostic outcomes including depressive symptom severity at 3–4 months and remission, but both of these relied on sum scores from the BDI-II. As the BDI-II items or sum score were used in the development of the predictive models it might have been informative to consider model performance with an entirely separate but clinically mean- ingful outcome such as functioning, quality of life, or mental pain (Fava et al., 2019); data on such outcomes were not available here. In addition, models here used IPD but the networks were esti- mated based on aggregated data, a number of studies have shown the potential utility of using idiographic networks to pre- dict outcomes for individual patients (Fisher & Boswell, 2016; Fisher, Medaglia, & Jeronimus, 2018; Fisher, Reeves, Lawyer, Medaglia, & Rubel, 2017a), this may yet prove the most fruitful avenue for using networks to inform prognostic models which are able to outperform classic regression models of the same factors. dicting BDI-II scores at 3–4 months post-baseline in the test datasets individually and combined Strengths and limitations This study was the first to provide robust tests of the ability of centrality statistics from FGL networks and factor loadings from a factor analytic model to develop weighted total scale scores to inform predictive models of treatment outcomes. This is some- thing that has been proposed as a promising method for using individual symptom data to build informative predictive models (Boschloo et al., 2016). We tested these methods against bone fide predictive models and simple comparison models, and in entirely held-out (test) data, and found there to be little evidence of any advantage to the above approaches. We used a large indi- vidual patient data dataset comprising six RCTs with a variety of widely available treatments for depression, all of the RCTs were situated in primary care, and five were pragmatic trials, increasing the generalisability of these results (Rothwell, 2005). However, the variability in the samples between the studies may have limited the overall performance of the models. We included a range of psychopathology measures at baseline, not just depression symp- toms from a single measure, as there is good evidence that such factors are associated with prognosis for depressed adults J. E. J. Buckman et al. 8 in accuracy may also have been required for us to find substantial differences in the performance of the modelling schemes. Table 3. Performance of the models predicting BDI-II scores at 3–4 months post-baseline in the test datasets individually and combined All studies combined (n = 918) IPCRESS (n = 206) MIR (n = 424) TREAD (n = 288) Internal cross-validation Type of approach Model RMSE R2 MAE RMSE R2 MAE RMSE R2 MAE RMSE R2 MAE RMSE R2 MAE Weighted sum scores 1. EI 1-step 11.285 0.168 9.122 11.642 0.171 9.572 11.216 0.174 8.993 11.127 0.137 8.990 9.995 0.216 7.991 2. EI 2-Step 11.281 0.169 9.119 11.646 0.170 9.575 11.209 0.175 8.987 11.122 0.137 8.985 9.992 0.216 7.989 3. PR/PC 11.326 0.162 9.097 11.626 0.173 9.526 11.226 0.177 9.053 11.253 0.117 8.856 9.941 0.223 7.940 4. CFA 11.338 0.160 9.100 11.655 0.169 9.548 11.219 0.175 9.041 11.284 0.112 8.865 9.953 0.221 7.946 Unweighted sum scores 5. ENRa 11.311 0.165 9.089 11.638 0.171 9.541 11.232 0.173 9.046 11.189 0.127 8.827 9.946 0.223 7.950 6. OLS 11.319 0.163 9.091 11.631 0.172 9.544 11.220 0.175 9.045 11.237 0.119 8.836 9.947 0.222 7.944 Individual symptoms 7. https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291721001616 Downloaded from https://www.cambridge.org/core. IP address: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available at Implications and conclusions Prognoses generated by the models developed here could be informative for depressed patients seeking treatment in primary care. However, there were few differences between the models, with no clear advantage in using individual items over sum scores, or in using network models or factor analytic models to weight individual items, in order to derive prognostic predictions. This may represent a limitation of the available data, or of the model- ling approaches (that e.g. rely on estimating linear relations). In all of the models, the degree of inaccuracy in their predictions might be unacceptable to any individual patient. There were clear differ- ences in the number of people reaching remission when the mod- els predicted patients would have particularly low or high scores, but the models performed less well with BDI-II scores between 18 and 25. It may be informative for future studies to test the utility in giving more intensive treatments or more regular clinical reviews for patients with these mid-range scores, particularly if there is uncertainty about the value of doing so based on clinical severity. It is noteworthy that all of the models utilised both depressive and anxiety symptom data, and all but one included the total score from the life events scale, and six of the eight included the social support scale score. It might therefore be informative for prognosis to assess for these factors routinely in clinic. The individual-item models outperformed the others in the internal cross-validation data suggesting that narrow con- structs (e.g. anhedonia) might be more informative for prognosis than broad constructs (e.g. depression), but issues of measure- ment error arise, particularly with the validity of the single items to measure each narrow construct. The findings presented https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291721001616 Downloaded from https://www.cambridge.org/core. IP address: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available at 9 9 Psychological Medicine Fig. 2. Proportion of participants in remission at 3–4 months post-baseline in the test set studies (n = 918) by predicted category of depressive severity at 3–4 months, for each of the eight models. Fig. 2. Proportion of participants in remission at 3–4 months post-baseline in the test set studies (n = 918) by predicted category of depressive severity at 3–4 months, for each of the eight models. (3) IPCRESS: BUPA Foundation. here also highlight the importance of external validation in accounting for issues of overfitting. Supplementary material. The supplementary material for this article can be found at https://doi.org/10.1017/S0033291721001616. Financial support. This work was supported by the Wellcome Trust through a Clinical Research Fellowship to JEJB (201292/Z/16/Z), MQ Foundation (for ZDC: MQDS16/72), the Higher Education Funding Council for England, the National Institute of Health Research (NIHR), NIHR University College London Hospitals Biomedical Research Centre (CO’D, RS, GL and SP), NIHR Biomedical Research Centre at the University Hospitals Bristol and Weston NHS Foundation Trust and the University of Bristol (NW and DK), University College London (GA, GL), University of Pennsylvania (RJD), Vanderbilt University (SDH), University of Southampton (TK), University of Exeter (EW), and University of York (SG), National Institute for Health Research (NIHR) Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London (TE, AP and CR). The views expressed are those of the authors and not necessarily those of the NIHR or the Department of Health and Social Care. Implications and conclusions (4) MIR: NIHR HTA programme (project 11/129/76) and supported by the NIHR Biomedical Research Centre at University Hospitals Bristol NHS Foundation Trust and the University of Bristol. Authors contributions. JEJB, CO’D, ZDC and EF conceived of the original project, JEJB along with SP, GL, RJD, SDH, SG, TK, EW and GA applied for and received funding to support this work. All ten of the above and RS wrote the initial protocol document and plan for the current study. ZDC, NW, DK, TK, SG and GL provided data and liaison to resolve issues and discrepancies between received datasets and publications about those studies. JEJB, RS, GL and SP were responsible for the screening of studies, data extraction, and add- itional data cleaning. JB, CO’D and ZDC conducted the data analyses with sup- port from EF, GA, RS, GL and SP, and consultation from all other authors. JEJB wrote the original manuscript will support from ZDC, RS, CO’D, EF, GA, GL and SP. All authors contributed to consecutive drafts and approved the final manuscript. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. y (5) PANDA: NIHR Programme Grant for Applied Research (RP-PG-0610-10048). (5) PANDA: NIHR Programme Grant for Applied Research (RP-PG-0610-10048). (6) TREAD: NIHR HTA programme. (6) TREAD: NIHR HTA programme. The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. All authors were fully independent of their respective funders and had responsibility for the decision to submit for this manuscript for publication. https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291721001616 Downloaded from https://www.cambridge.org/core. IP address: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available at Conflict of interest. None. Supplementary material. The supplementary material for this article can be found at https://doi.org/10.1017/S0033291721001616. (2) GENPOD: Medical Research Council and supported by the Mental Health Research Network. References Adibi, A., Sadatsafavi, M., & Ioannidis, J. P. A. (2020). Validation and utility testing of clinical prediction models. JAMA, 324(3), 235. https://doi.org/ 10.1001/jama.2020.1230. Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Manual for the Beck Depression Inventory-II. San Antonio, TX: Psychological Corporation. Berlim, M. T., Richard-Devantoy, S., Dos Santos, N. R., & Turecki, G. (2020). The network structure of core depressive symptom-domains in major depres- sive disorder following antidepressant treatment: a randomized clinical trial. Psychological Medicine, 1–15. https://doi.org/10.1017/S0033291720001002. Borsboom, D. (2017). A network theory of mental disorders. World Psychiatry, 16(1), 5–13. https://doi.org/10.1002/wps.20375. The included studies that make up the Dep-GP IPD database were funded by: Borsboom, D., & Cramer, A. O. J. (2013). Network analysis: an integrative approach to the structure of psychopathology. Annual Review of Clinical Psychology, 9(1), 91–121. https://doi.org/10.1146/annurev-clinpsy-050212- 185608. (1) COBALT: The National Institute for Health Research Health Technology Assessment (NIHR HTA) programme (project number 06/404/02). Boschloo, L., van Borkulo, C. D., Borsboom, D., & Schoevers, R. A. (2016). A prospective study on how symptoms in a network predict the onset of (2) GENPOD: Medical Research Council and supported by the Mental Health Research Network. 10 J. E. J. Buckman et al. depression. Psychotherapy and Psychosomatics, 85(3), 183–184. https://doi. org/10.1159/000442001. depression. Psychotherapy and Psychosomatics, 85(3), 183–184. https://doi. org/10.1159/000442001. Fava, G. A., Ruini, C., & Belaise, C. (2007). The concept of recovery in major depression. Psychological Medicine, 37, 307–317. Retrieved from http:// ovidsp.ovid.com/ovidweb.cgi?T=JS&PAGE=reference&D=psyc5&NEWS= N&AN=2007-03729-001. Bower, P., Kontopantelis, E., Sutton, A., Kendrick, T., Richards, D. A., Gilbody, S., … Liu, E. T.-H. (2013). Influence of initial severity of depression on effect- iveness of low intensity interventions: meta-analysis of individual patient data. BMJ, 346(feb26 2), f540–f540. https://doi.org/10.1136/bmj.f540. Fava, G. A., Tomba, E., Brakemeier, E. L., Carrozzino, D., Cosci, F., Eöry, A., … Guidi, J. (2019). Mental pain as a transdiagnostic patient-reported outcome measure. Psychotherapy and Psychosomatics, 88(6), 341–349. https://doi.org/ 10.1159/000504024. Bringmann, L. F., Elmer, T., Epskamp, S., Krause, R. W., Schoch, D., Wichers, M., … Snippe, E. (2019). What do centrality measures measure in psycho- logical networks? Journal of Abnormal Psychology, 128(8), 892–903. https:// doi.org/10.1037/abn0000446. Fischer, H. F., & Rose, M. (2016). Www.common-metrics.org: a web applica- tion to estimate scores from different patient-reported outcome measures on a common scale. BMC Medical Research Methodology, 16(142), 1–5. https://doi.org/10.1186/s12874-016-0241-0. Buckman, J. E. J., Saunders, R., Cohen, Z. D., Barnett, P., Clarke, K., Ambler, G., … Pilling, S. (2021a). References The contribution of depressive ’disorder character- istics’ to determinations of prognosis for adults with depression: an individ- ual patient data meta-analysis. Psychological Medicine, 1–14. https://doi.org/ 10.1017/S0033291721001367. Fisher, A. J., & Boswell, J. F. (2016). Enhancing the personalization of psycho- therapy with dynamic assessment and modeling. Assessment, 23(4), 496– 506. https://doi.org/10.1177/1073191116638735. Buckman, J. E. J., Saunders, R., Cohen, Z. D., Clarke, K., Ambler, G., DeRubeis, R. J., … Pilling, S. (2020). What factors indicate prognosis for adults with depression in primary care? A protocol for meta-analyses of individual patient data using the Dep-GP database. Wellcome Open Research, 4, 69. https://doi.org/10.12688/wellcomeopenres.15225.3. Fisher, D. J., Carpenter, J. R., Morris, T. P., Freeman, S. C., & Tierney, J. F. (2017b). Meta-analytical methods to identify who benefits most from treat- ments: daft, deluded, or deft approach? Most from particular treatments or other broad approaches used for testing such. BMJ, 356(j573), 1–6. https:// doi.org/10.1136/bmj.j573. Buckman, J. E. J., Saunders, R., O’Driscoll, C., Cohen, Z. D., Stott, J., Ambler, G., … Pilling, S. (2021b). Is social support pre-treatment associated with prognosis for adults with depression in primary care? Acta Psychiatrica Scandinavica, acps.13285. https://doi.org/10.1111/acps.13285. Fisher, A. J., Medaglia, J.D., & Jeronimus, B. F. (2018). Lackofgroup-to-individual generalizability is a threat to human subjects research. Proceedings of the National Academy of Sciences of the United States of America, 115(27), E6106–E6115. https://doi.org/10.1073/pnas.1711978115. Buckman, J. E. J., Underwood, A., Clarke, K., Saunders, R., Hollon, S. D., Fearon, P., & Pilling, S. (2018). Risk factors for relapse and recurrence of depression in adults and how they operate: a four-phase systematic review and meta-synthesis. Clinical Psychology Review, 64(7), 13–38. https://doi. org/10.1016/j.cpr.2018.07.005. Fisher, A. J., Reeves, J. W., Lawyer, G., Medaglia, J. D., & Rubel, J. A. (2017a). Exploring the idiographic dynamics of mood and anxiety via network ana- lysis. Journal of Abnormal Psychology, 126(8), 1044–1056. https://doi.org/10. 1037/abn0000311. Fournier, J. C., Derubeis, R. J., Hollon, S. D., Shelton, R. C., & Fawcett, J. (2010). Antidepressant drug effects and depression severity. Journal of the American Medical Association, 303(1), 47–53. Choi, S. W., Schalet, B. D., Cook, K. F., & Cella, D. (2014). Establishing a com- mon metric for depressive symptoms: linking the BDI-II, CES-D, and PHQ-9 to PROMIS depression. Psychological Assessment, 26(2), 513–527. https://doi.org/10.1037/a0035768. Fried, E. I., & Cramer, A. O. J. (2017). Moving forward: challenges and direc- tions for psychopathological network theory and methodology. Perspectives on Psychological Science, 12(6), 999–1020. https://doi.org/10.1177/ 1745691617705892. Cohen, Z. References D., & DeRubeis, R. J. (2018). Treatment selection in depression. Annual Review of Clinical Psychology, 14(1), 209–236. https://doi.org/10. 1146/annurev-clinpsy-050817-084746. Fried, E. I., Eidhof, M. B., Palic, S., Costantini, G., Huisman-van Dijk, H. M., Bockting, C. L. H., … Karstoft, K.-I. (2018). Replicability and generalizabil- ity of posttraumatic stress disorder (PTSD) networks: a cross-cultural multisite study of PTSD symptoms in four trauma patient samples. Clinical Psychological Science, 6(3), 335–351. https://doi.org/10.1177/ 2167702617745092. Collins, G. S., Reitsma, J. B., Altman, D. G., & Moons, K. G. M. (2015). Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMC Medicine, 13(1), 1–10. https://doi.org/10.1186/s12916-014-0241-z. Costantini, G., & Epskamp, S. (2017a). Package “EstimateGroupNetwork”. pp. 1–10. CRAN. Fried, E. I., Epskamp, S., Nesse, R. M., Tuerlinckx, F., & Borsboom, D. (2016). What are “good” depression symptoms? Comparing the centrality of DSM and non-DSM symptoms of depression in a network analysis. Journal of Affective Disorders, 189, 314–320. https://doi.org/10.1016/j.jad.2015.09.005. Costantini, G., & Epskamp, S. (2017b). Perform the Joint Graphical Lasso and Selects Tuning Parameters. pp. 1–10. CRAN. Costantini, G., Richetin, J., Preti, E., Casini, E., Epskamp, S., & Perugini, M. (2019). Stability and variability of personality networks. A tutorial on recent developments in network psychometrics. Personality and Individual Differences, 136, 68–78. https://doi.org/10.1016/j.paid.2017.06.011. Fried, E. I., & Nesse, R. M. (2014). The impact of individual depressive symp- toms on impairment of psychosocial functioning. PLoS ONE, 9(2), e90311. https://doi.org/10.1371/journal.pone.0090311. Fried, E. I., & Nesse, R. M. (2015a). Depression is not a consistent syndrome: an investigation of unique symptom patterns in the STAR*D study. Journal of Affective Disorders, 172, 96–102. https://doi.org/10.1016/j.jad.2014.10.010. Danaher, P., Wang, P., & Witten, D. M. (2014). for inverse covariance estima- tion across multiple classes. Journal of the Royal Statistical Society, 76(2), 373–397. https://doi.org/10.1111/rssb.12033.The. Delgadillo, J., Huey, D., Bennett, H., & McMillan, D. (2017). Case complexity as a guide for psychological treatment selection. Journal of Consulting and Clinical Psychology, 85(9), 835–853. https://doi.org/10.1037/ccp0000231. Fried, E. I., & Nesse, R. M. (2015b). Depression sum-scores don’t add up: why analyzing specific depression symptoms is essential. BMC Medicine, 13(1), 72. https://doi.org/10.1186/s12916-015-0325-4. Driessen, E., Cuijpers, P., Hollon, S. D., & Dekker, J. J. M. (2010). Does pretreatment severity moderate the efficacy of psychological treatment of adult outpatient depression? A meta-analysis. Journal of Consulting and Clinical Psychology, 78(5), 668–680. https://doi.org/10.1037/a0020570. Fried, E. I., & Robinaugh, D. J. (2020). https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291721001616 Downloaded from https://www.cambridge.org/core. IP address: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available a References Systems all the way down: embracing complexity in mental health research. BMC Medicine, 18(1), 4–7. https:// doi.org/10.1186/s12916-020-01668-w. Friedman, J., Hastie, T., & Tibshirani, R. (2010). Regularization paths for gen- eralized linear models via coordinate descent. Journal of Statistical Software, 33(1), 1–22. https://doi.org/10.18637/jss.v033.i01. Dwyer, D. B., Falkai, P., & Koutsouleris, N. (2018). Machine learning approaches for clinical psychology and psychiatry. Annual Review of Clinical Psychology, 14(1), 91–118. https://doi.org/10.1146/annurev-clinpsy-032816-045037. Harrell, F. E., Lee, K. L., & Mark, D. B. (2004). Prognostic/clinical prediction models: multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Tutorials in Biostatistics: Statistical Methods in Clinical Studies, 1, 223–249. https://doi.org/10.1002/0470023678.ch2b(i). Elliott, H., Jones, P. J., & Schmidt, U. (2020). Central symptoms predict post- treatment outcomes and clinical impairment in anorexia Nervosa: a net- work analysis. Clinical Psychological Science, 8(1), 139–154. https://doi. org/10.1177/2167702619865958. Epskamp, S., & Fried, E. I. (2018). A tutorial on regularized partial correlation networks. Psychological Methods, 23(4), 617–634. https://doi.org/10.1037/ met0000167. Hayden, J. A., Windt Van Der, D. A., Cartwright, J. L., Côté, P., & Bombardier, C. (2013). Assessing bias in studies of prognostic factors. Annals of Internal Medicine, 158, 280–286. https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291721001616 Downloaded from https://www.cambridge.org/core. IP address: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available at Psychological Medicine 11 Judd, L. L., Akiskal, H. S., Maser, J. D., Zeller, P. J., Endicott, J., Coryell, W., … Keller, M. B. (1998). Major depressive disorder: a prospective study of residual subthreshold depressive symptoms as predictor of rapid relapse. Journal of Affective Disorders, 50(2–3), 97–108. https://doi.org/10.1016/ S0165-0327(98)00138-4. Robinaugh, D. J., Hoekstra, R. H. A., Toner, E. R., & Borsboom, D. (2020). The network approach to psychopathology: a review of the literature 2008–2018 and an agenda for future research. Psychological Medicine, 50(3), 353–366. https://doi org/10 1017/S0033291719003404 Robinaugh, D. J., Millner, A. J., & McNally, R. J. (2016). Identifying highly influential nodes in the complicated grief network. Journal of Abnormal Psychology, 125(6), 747–757. https://doi.org/10.1002/cncr.27633.Percutaneous. Koenders, M. A., De Kleijn, R., Giltay, E. J., Elzinga, B. M., Spinhoven, P., & Spijker, A. T. (2015). A network approach to bipolar symptomatology in patients with different course types. PLoS ONE, 10(10), 1–16. https://doi. org/10.1371/journal.pone.0141420. Rothwell, P. M. (2005). Subgroup analysis in randomised controlled trials: Importance, indications, and interpretation. Lancet, 365, 176–186. Letina, S., Blanken, T. F., Deserno, M. K., & Borsboom, D. (2019). References Expanding network analysis tools in psychological networks: minimal spanning trees, participation coefficients, and motif analysis applied to a network of 26 psychological attributes. Complexity, 2019. https://doi.org/10.1155/ 2019/9424605. Stekhoven, D. J. & Bühlmann, P. (2012). Missforest-non-parametric missing value imputation for mixed-type data. Bioinformatics, 28, 112–118. Steyerberg, E. W., Vickers, A. J., Cook, N. R., Gerds, T., Gonen, M., Obuchowski, N., … Kattan, M. W. (2010). Assessing the performance of prediction models. Epidemiology, 21(1), 128–138. https://doi.org/10.1097/EDE.0b013e3181c30fb2. Lewis, G., Pelosi, A. J., Araya, R., & Dunn, G. (1992). Measuring psychiatric disorder in the community: a standardized assessment for use by lay inter- viewers. Psychological Medicine, 22, 465–486. Thornicroft, G., Chatterji, S., Evans-Lacko, S., Gruber, M., Sampson, N., Aguilar-Gaxiola, S., … Kessler, R. C. (2017). Undertreatment of people with major depressive disorder in 21 countries. British Journal of Psychiatry, 210(2), 119–124. https://doi.org/10.1192/bjp.bp.116.188078. Lorenzo-Luaces, L., Rodriguez-Quintana, N., & Bailey, A. J. (2020). Double trouble: do depression severity and duration interact to predicting treatment outcomes in adolescent depression? Behaviour Research and Therapy, 131, 103637. https://doi.org/10.1016/j.brat.2020.103637. van Borkulo, C., Boschloo, L., Borsboom, D., Penninx, B. W. J. H., Waldorp, L. J., & Schoevers, R. A. (2015). Association of symptom network structure with the course of depression. JAMA Psychiatry, 72(12), 1219. https://doi. org/10.1001/jamapsychiatry.2015.2079. Marwood, L., Wise, T., Perkins, A. M., & Cleare, A. J. (2018). Meta-analyses of the neural mechanisms and predictors of response to psycho- therapy in depression and anxiety. Neuroscience and Biobehavioral Reviews, 95(September), 61–72. https://doi.org/10.1016/j.neubiorev.2018. 09.022. Vos, T., Allen, C., Arora, M., Barber, R. M., Bhutta, Z. A., Brown, A., … Coggeshall, M. (2016). Global, regional, and national incidence, prevalence, and years lived with disability for 310 diseases and injuries, 1990–2015: a systematic analysis for the global burden of disease study 2015. Lancet, 388, 1545–1602. https://doi.org/10.1016/S0140-6736(16)31678-6. Moons, K. G. M., Altman, D. G., Reitsma, J. B., Ioannidis, J. P. A., Macaskill, P., Steyerberg, E. W., … Collins, G. S. (2015). Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): explanation and elaboration. Annals of Internal Medicine, 162 (1), W1. https://doi.org/10.7326/M14-0698. Webb, C. A., Cohen, Z. D., Beard, C., Forgeard, M., Peckham, A. D., & Björgvinsson, T. (2020). Personalized prognostic prediction of treatment outcome for depressed patients in a naturalistic psychiatric hospital setting: a comparison of machine learning approaches. Journal of Consulting and Clinical Psychology, 88(1), 25–38. https://doi.org/10.1037/ccp0000451. Morgan, A. J., Reavley, N. J., & Jorm, A. F. (2014). ge.org/core/terms. https://doi.org/10.1017/S0033291721001616 ps://www.cambridge.org/core. IP address: 146.90.13.39, on 20 May 2021 at 10:47:31, subject to the Cambridge Core terms of use, available at References Beliefs about mental disorder treatment and prognosis: comparison of health professionals with the Australian public. Australian and New Zealand Journal of Psychiatry, 48(5), 442–451. https://doi.org/10.1177/0004867413512686. Weitz, E. S., Hollon, S. D., Twisk, J., Van Straten, A., Huibers, M. J. H., David, D., … Cuijpers, P. (2015). Baseline depression severity as moderator of depression outcomes between cognitive behavioral therapy vs pharmaco- therapy: an individual patient data meta-analysis. JAMA Psychiatry, 72 (11), 1102–1109. https://doi.org/10.1001/jamapsychiatry.2015.1516. O’Driscoll, C., Buckman, J. E. J., Fried, E. I., Saunders, R., Cohen, Z. D., Ambler, G., … Pilling, S. (2021). The importance of transdiagnostic symp- tom level assessment to understanding prognosis for depressed adults: ana- lysis of data from six randomized control trials. BMC Medicine, 1–21. https://doi.org/10.1186/s12916-021-01971-0. Wichers, M., & Groot, P. C. (2016). Critical slowing down as a personalized early warning signal for depression. Psychotherapy and Psychosomatics, 85 (2), 114–116. https://doi.org/10.1159/000441458.
https://openalex.org/W1995058339	https://www.cambridge.org/core/services/aop-cambridge-core/content/view/9473F6826A6E57DEAC5FE990F031F812/S0955603600033225a.pdf/div-class-title-service-innovation-psychiatrists-on-call-the-community-at-night-div.pdf	English	null	Service innovation: psychiatrists on call – the community at night	Psychiatric bulletin of the Royal College of Psychiatrists/Psychiatric bulletin	2,007	cc-by	2,721	L AU R E N C E M Y N O R S - WA L L I S A ND DE N I S E C O P E Service innovation: psychiatrists on call - the community at night psychiatry, general adult psychiatry, psychiatry of learning disability, liaison psychiatry, psychiatry of old age, psychotherapy and psychiatry of substance misuse. Knowledge and skills in these areas will need to be maintained and updated. All psychiatrists should be competent to assess and undertake the immediate management of patients for whom they have respon- sibility when on-call over the weekends and in emergencies.’ The document goes on to state that ‘All psychiatrists should have skills in the assessment of psychiatric disorder complicated by or associated with substance misuse and of psychiatric problems in young people, older people and people with learning disability and skills in the immediate (short-term) management of these conditions, together with sufficient knowledge of management strategies and local services to suggest appropriate care for these conditions and knowledge of the differing ethical and legal frameworks to ensure appropriate emergency care’. There have been significant changes in the provision of medical care in hospitals at night. The initial catalyst for this was the New Deal for Junior Doctors but more recently the European Working Time Directive requiring doctors’ hours to be reduced to 56 in 2002 and to 48 by 2009. The reduced availability of junior doctors in hospitals at night has had a range of implications, including the necessity to train other health professionals to do work previously undertaken by doctors and a reduction in the number of specialist doctors available out of hours. The expectation is that staff in the hospital at night will be equipped to deal appropriately and safely with emergency work across specialties, rather than each specialty covering their own patients. In psychiatry reducing the working hours of junior doctors and implementing the Hospital at Night programme (Department of Health, 2005a) have necessitated review of having resident junior doctors on call at night and the training of nurse practitioners to take on some roles of the on-call duty doctor. The extension of prescribing to nurses and other professionals will facilitate and hasten this process; prescribing has been a role which it has not been previously possible to delegate. The trust was also influenced by the emerging proposals from the now published New Ways of Working for Psychiatrists (Department of Health, 2005b) and the White Paper Valuing People (Department of Health, 2001). L AU R E N C E M Y N O R S - WA L L I S A ND DE N I S E C O P E Service innovation: psychiatrists on call - the community at night New Ways of Working for Psychiatrists gives clear guidance about the need for trusts to devise job plans for consultant psychiatrists that will prevent them from becoming burnt out and demoralised through excessive workloads. Valuing People is clear that ‘a person with learning disability who has a mental illness should there- fore expect to be able to access services and be treated in the same way as everyone else’. Although there have been significant and, in some cases, far-reaching changes in out-of-hours hospital work, there has been little discussion of the implications for out-of-hours work in the community and whether the principles of the Hospital at Night programme apply equally to the community at night. This paper sets out the response by a medium-sized specialist mental health and learning disability trust in Dorset to the challenges of providing safe and appro- priate out-of-hours care in the community, balancing the need to have satisfactory working hours, not only for training grade doctors but also for career grade doctors. In meeting this challenge the trust was influenced by the Royal College of Psychiatrists’ Good Psychiatric Practice (Royal College of Psychiatrists, 2004). This states that all psychiatrists should be equipped to deal with emergen- cies across sub-specialties. In the section on competen- cies, Good Psychiatric Practice states that ‘All psychiatrists will have a common basic understanding of the following specialties: child and adolescent psychiatry, forensic Mynors-Wallis & Cope The community at night Mynors-Wallis & Cope The community at night special articles Psychiatric Bulletin (2007), 31, 65^67. doi: 10.1192/pb.bp.106.009811 Psychiatric Bulletin (2007), 31, 65^67. doi: 10.1192/pb.bp.106.009811 Historical on-call arrangements in East Dorset In 1993 there were two separate consultant on-call rotas for general adult and old age services within East Dorset, with consultants on call 1 in 5. In 1996 these rotas merged into one so that there was a single rota covering patients in learning disability, general adult and old age psychiatry. Child and adolescent services were covered by a separate rota. The frequency of on call at this time was 1 in 11 for consultants in general adult, old age and learning disability psychiatry and 1 in 3 for consultants in 65 https://doi.org/10.1192/pb.bp.106.009811 Published online by Cambridge University Press Mynors-Wallis & Cope The community at night child and adolescent psychiatry. The senior doctor on call was supported by a resident junior doctor in the main hospital base and a non-resident junior doctor/staff grade in other peripheral units. both because of the College recommendations about training in the emergency management of other sub- specialties (Royal College of Psychiatrists, 2004) and because specialist registrars were being trained to be consultants, and the on-call rota, although iinnovative, might well be adopted elsewhere. Those specialist regis- trars obtaining consultant posts in Dorset would be helped by the experience of on call across the sub- specialties. In 2001 an out-of-hours nursing service was estab- lished to run alongside and in parallel with the medical on-call rota, to provide additional support for existing patients but not to provide assessment of new patients. In 2003 the out-of-hours nursing service was strength- ened by the appointment of nurse practitioners to undertake first-line assessments from the accident and emergency departments of the two local general hospitals and to take all hospital calls at night in place of junior doctors who became non-resident. Safeguards were incorporated into the system for specialist registrars. . The specialist registrar should contact the consultant on call for discussion of the management of all patients aged15 and under In 2003 the trust also appointed 1.5 additional consultant psychiatrists in child and adolescent mental health services (CAMHS). These new consultants did not wish to participate in a 1 in 4.5 on-call rota or even a 1 in 6 on-call rota, which was the projected development for CAMHS within the trust. A discussion was held with the consultant body as to whether the CAMHS consultants should join the well-established combined learning disability, general adult and old age psychiatry on-call rota. Functioning of the new rota The new rota has now been running successfully for over 2 years. The consultant body continue to be supportive of the rota and feel that they have the necessary skills to manage emergencies as they arise. All new consultants appointed to the rota in all sub-specialties are given appropriate training and induction to ensure that they feel equipped to undertake the role of senior on-call consultant. Feedback from specialist registrars has indicated that they receive appropriate support and supervision in fulfilling their on-call duties. Child and adolescent consultants and general adult consultants, in psychiatry were both anxious about their extended on-call roles. However, the training reassured anxious colleagues in both adult and child and adolescent services. All consultants indicated that they would be happy to be contacted by other colleagues, even if not on call, to provide advice during the trial period. Historical on-call arrangements in East Dorset It was agreed to trial a combined senior on-call rota for a period of 6 months. . The specialistregistrar was encouraged to contact the childand adolescent psychiatrist the next working day for supervisionandadvice about allpatients under the age of16 seen out of hours . Any concerns and difficulties would be discussed in the already established monthly on-call supervision group between specialist registrars and consultant psychiatrists. The system was approved by the Specialist Training Committee in Psychiatry of the Wessex Deanery. Subsequently inspectors from the Royal College of Psychiatrists General Adult and Old Age Higher Specialist Training Scheme expressed concern about such trainees working with children and adolescents and those with learning disability out of hours. They have recommended that this aspect of the rota for these trainees ceases. Discussion The change from traditional ways of working always has the potential to raise anxieties, and in medicine concerns about reduction in standards. We believe that we have developed an on-call system which both maintains high standards of psychiatric practice and provides an appro- priate work^life balance for consultant psychiatrists. The system is in line with all relevant government directives and facilitates achievement of the objectives set out in Good Psychiatric Practice. Feedback from consultant psychiatrists has been uniformly positive. Feedback from specialist registrars has indicated that the on call has been a useful training experience and has been appropri- ately supported by training and supervision. There have been no concerns about clinical practice out of hours. We At the end of a 6-month trial there was unanimous agreement by consultants within the trust that the on-call system had been a success and it was therefore agreed to continue. Preparation for combined rotas Training was given by the child and adolescent consul- tants to their consultant colleagues in the management of psychiatric emergencies in children; in particular, the use of the Children Act 2004, the issues of capacity and consent in children and the links with social services. Arrangements were clarified with the paediatric services that all children under 16 who had taken overdoses would be admitted to a paediatric ward overnight and reviewed by CAMHS the next day. At weekends, the paediatrician would make a decision about whether the child needed to be kept in hospital until the next working day. This decision could be made in conjunction with the on-call psychiatrist. Arrangements were made for emergency assessments by the CAMHS team of all children who had been discharged from the general hospital on the next working day. special articles References hope that this system of on call, which we believe is not replicated elsewhere in the country, provokes discussion about best practice in out-of-hours psychiatric provision. This could be a model as to how high standards can be maintained while ensuring that career grade doctors have satisfactory and enjoyable working lives. Recruitment and retention of consultant psychiatrists remains an ongoing problem and improving the quality of a consultant’s working life both in and out of hours will help address this issue. Enhancing Effective Person Centred ServicesThrough NewWays ofWorking in Multidisciplinary and Multiagency Contexts. Department of Health. http://www.dh.gov.uk/assetRoot/ 04/12/23/43/04122343. pdf DEPARTMENT OF HEALTH (2001) Valuing People: A New Strategy for Learning Disability for the 21st Century. Department of Health. DEPARTMENT OF HEALTH (2005a) The Implementation and Impact of Hospital at Night Pilot Projects: An Evaluation Report. Department of Health. http:// www.dh.gov.uk/assetRoot/04/11/79/ 69/04117969. pdf ROYAL COLLEGE OF PSYCHIATRISTS (2004) Good Psychiatric Practice (2nd edn) (Council Report CR125). Royal College of Psychiatrists. DEPARTMENTOFHEALTH (2005b) New Ways ofWorking for Psychiatrists: LY NDA B R EE N Therapeutic use of soap operas in autistic-spectrum disorders central theme could relate the narratives to their own lives, discuss ‘primordial life values’ and reflect how they might have behaved in similar situations (De Bruin, 2001). Soap opera material has also been used effectively in cognitive skills training with an adolescent with learning disability (Creswell, 2001). A ‘soap therapy’ approach might be useful in children, including those with autistic- spectrum disorders, where descriptions of adapted cognitive therapy are currently relatively uncommon. Since family discussion of favourite television programmes has already been suggested to enhance social learning in autistic-spectrum disorders (Williams & Wright, 2004), therapeutic application of soap opera material might be similarly useful. ‘Soap opera’ is a popular television genre that ‘invites the audience to . . . identify with characters’ (Livingstone, 1990). Storylines tend to be shaped by national and local culture, although they may feature a disproportionate number of unstable relationships and tragedies (Liebes & Livingstone, 1998). Narratives evolve continually, allowing scriptwriters to incite viewer debate on myriad topical social issues, including mental illness (Reveley, 1997). Social change attributable to television drama programmes has already been documented (Singhal & Obregon, 1999). In 1975, the first pro-social soap opera ‘Ven Conmigo’ was credited with a 63% rise in literacy rates in Mexico (Brown et al, 1989). Following the screening of a Tanzanian soap opera on family planning methods, a large field study demonstrated an impressive increase in the uptake of contraception (Rogers et al, 1999). This capacity for community change implies potential for individual change, a concept which might support the therapeutic use of soap opera material. Moreover, the enduring emphasis on inter-character relationships in soap operas might provide a resource for exploring emotions and relationships in a clinical setting. Their rich audio-visual medium and established public popularity might also motivate potential clients (Creswell, 2001). ‘Soap opera’ is a popular television genre that ‘invites the audience to . . . identify with characters’ (Livingstone, 1990). Storylines tend to be shaped by national and local culture, although they may feature a disproportionate number of unstable relationships and tragedies (Liebes & Livingstone, 1998). Narratives evolve continually, allowing scriptwriters to incite viewer debate on myriad topical social issues, including mental illness (Reveley, 1997). Social change attributable to television drama programmes has already been documented (Singhal & Obregon, 1999). Declaration of interest L.M.-W. is the Medical Director of Dorset HealthCare NHS Trust and a general adult psychiatrist. D.C. is Associate Medical Director of Dorset HealthCare NHS Trust and a consultant in old age psychiatry. *Laurence Mynors-Wallis Alderney Community Hospital, Ringwood Road, Parkstone, Poole, Dorset BH124NB, email: laurence.mynorswallis@nhs.net, Denise Cope Alderney Community Hospital, Ringwood Road, Parkstone, Poole Psychiatric Bulletin (2007), 31, 67^69. doi: 10.1192/pb.bp.105.008250 Training grade doctors It was discussed initially whether the specialist registrars in general adult/old age psychiatry and CAMHS should remain on separate rotas. It was decided not to do this, 66 https://doi.org/10.1192/pb.bp.106.009811 Published online by Cambridge University Press Mynors-Wallis & Cope The community at night Enhancing Effective Person Centred ServicesThrough NewWays ofWorking in Multidisciplinary and Multiagency Contexts. Department of Health. http://www.dh.gov.uk/assetRoot/ 04/12/23/43/04122343. pdf LY NDA B R EE N Therapeutic use of soap operas in autistic-spectrum disorders In 1975, the first pro-social soap opera ‘Ven Conmigo’ was credited with a 63% rise in literacy rates in Mexico (Brown et al, 1989). Following the screening of a Tanzanian soap opera on family planning methods, a large field study demonstrated an impressive increase in the uptake of contraception (Rogers et al, 1999). This capacity for community change implies potential for individual change, a concept which might support the therapeutic use of soap opera material. Moreover, the enduring emphasis on inter-character relationships in soap operas might provide a resource for exploring emotions and relationships in a clinical setting. Their rich audio-visual medium and established public popularity might also motivate potential clients (Creswell, 2001). There is an ongoing need for evidence-based methods of teaching emotional recognition and social skills to individuals with autistic-spectrum disorders. A key feature of autistic cognition is delayed development of ’theory of mind’, the concept that these individuals struggle to understand the thoughts, emotions and plans of others (Baron-Cohen et al, 1985). Since they fail to grasp that others think differently, people with autistic- spectrum disorders tend to encounter difficulties in relating to and anticipating the actions of others. Consequently, they may appear to be eccentric or self- centred, which further compounds their potential social isolation. One method used to facilitate social awareness is the ‘social story’, whereby hypothetical scenarios focus discussion on perspectives and cognitions of the self and others (Gray, 1993). Specially commissioned film clips have also been used effectively, but this is expensive and time-consuming, particularly for individualised therapies. More current tools include interactive computer programs Current evidence supporting the therapeutic use of soap operas is limited but they have been used effec- tively to encourage discussion, problem-solving and self- awareness in therapeutic groups (Falk-Kessler & Froschauer, 1978). Qualitative research supports the use of soap opera material in identity work with Asian adolescents (Barker, 1997). Dutch adolescent girls who watched a soap opera in which family conflicts were a 67 https://doi.org/10.1192/pb.bp.106.009811 Published online by Cambridge University Press
https://openalex.org/W4376875790	https://vbn.aau.dk/ws/files/531524097/fpain_04_1191786.pdf	English	null	On determining the mechanical nociceptive threshold in pigs: a reliability study	Frontiers in pain research	2,023	cc-by	6,521	On determining the mechanical nociceptive threshold in pigs li bilit t d Published in: Frontiers in Pain Research Citation for published version (APA): Andreis, F. R., Mørch, C. D., Jensen, W., & Meijs, S. (2023). On determining the mechanical nociceptive threshold in pigs: a reliability study. Frontiers in Pain Research, 4, Article 1191786. https://doi.org/10.3389/fpain.2023.1191786 General rights C i h d General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. - Users may download and print one copy of any publication from the public portal for the purpose of private study or research. - You may not further distribute the material or use it for any profit-making activity or commercial gain - Users may download and print one copy of any publication from the public portal for the purpose of private study Y t f th di t ib t th t i l it f fit ki ti it i l i - Users may download and print one copy of any publication from the public portal for the purpose of private study or research. - You may not further distribute the material or use it for any profit-making activity or commercial gain Users may download and print one copy of any publication from the public portal for the purpose of priv - You may not further distribute the material or use it for any profit-making activity or commercial gain - You may freely distribute the URL identifying the publication in the public portal - y p py y p p p - You may not further distribute the material or use it for any profit-making activity or co Y f l di t ib t th URL id tif i th bli ti i th bli t l Take down policy If you believe that this document breaches copyright please contact us at vbn@aub.aau.dk providing details, and we will remove access to the work immediately and investigate your claim. Take down policy If you believe that this document breaches copyright please contact us at vbn@aub.aau.dk providing details, and we will remove access to the work immediately and investigate your claim. Citation for published version (APA): Andreis, F. R., Mørch, C. D., Jensen, W., & Meijs, S. (2023). On determining the mechanical nociceptive threshold in pigs: a reliability study. Frontiers in Pain Research, 4, Article 1191786. https://doi.org/10.3389/fpain.2023.1191786 Aalborg Universitet On determining the mechanical nociceptive threshold in pigs a reliability study Andreis, Felipe Rettore; Mørch, Carsten Dahl; Jensen, Winnie; Meijs, Suzan Published in: Frontiers in Pain Research DOI (link to publication from Publisher): 10.3389/fpain.2023.1191786 Creative Commons License CC BY 4.0 Publication date: 2023 Document Version Publisher's PDF, also known as Version of record Link to publication from Aalborg University Citation for published version (APA): Andreis, F. R., Mørch, C. D., Jensen, W., & Meijs, S. (2023). On determining the mechanical nociceptive threshold in pigs: a reliability study. Frontiers in Pain Research, 4, Article 1191786. https://doi.org/10.3389/fpain.2023.1191786 Aalborg Universitet Systematic bias was also evaluated. Results: The average ICC was found to be 0.71 and 0.45 for the between-session and within-session, respectively. CV ranged from 17.9% to 20.5%, with a grand average of 19.1%. The grand average SEM was 249.5 kPa (16.6%). No systematic differences were found for the MNT between sessions, which suggests that there was no habituation to the stimulus. Methods: Nine animals were used (23–40 kg), and MNTs were measured at both the right and left limbs at three different sessions, with three repetitions per session. The intraclass correlation coefﬁcient (ICC) was used as a metric for relative reliability. The standard error of measurement (SEM) and coefﬁcient of variation (CV) was used to assess absolute reliability. Systematic bias was also evaluated. © 2023 Rettore Andreis, Mørch, Jensen and Meijs. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Results: The average ICC was found to be 0.71 and 0.45 for the between-session and within-session, respectively. CV ranged from 17.9% to 20.5%, with a grand average of 19.1%. The grand average SEM was 249.5 kPa (16.6%). No systematic differences were found for the MNT between sessions, which suggests that there was no habituation to the stimulus. Conclusion: The reliability indices obtained in this study are comparable to results obtained in other species or anatomical regions and substantiate the use of the pressure algometer as a valuable tool to investigate the nociceptive system in pigs and translation to the human nociceptive withdrawal reﬂex. KEYWORDS pressure algometry, pigs, pain, mechanical nociceptive threshold, reliability EDITED BY Peter Wilhelm Marius Kronen, Veterinary Anaesthesia Services International GmbH, Switzerland REVIEWED BY Daniel Segelcke, University Hospital Münster, Germany Antti Pertovaara, University of Helsinki, Finland CORRESPONDENCE Felipe Rettore Andreis fran@hst.aau.dk RECEIVED 22 March 2023 ACCEPTED 03 May 2023 PUBLISHED 17 May 2023 CITATION Rettore Andreis F, Mørch CD, Jensen W and Meijs S (2023) On determining the mechanical nociceptive threshold in pigs: a reliability study Front. Pain Res. 4:1191786. doi: 10.3389/fpain.2023.1191786 EDITED BY Peter Wilhelm Marius Kronen, Veterinary Anaesthesia Services International GmbH, Switzerland REVIEWED BY Daniel Segelcke, University Hospital Münster, Germany Antti Pertovaara, University of Helsinki, Finland CORRESPONDENCE Felipe Rettore Andreis fran@hst.aau.dk RECEIVED 22 March 2023 ACCEPTED 03 May 2023 PUBLISHED 17 May 2023 CITATION Rettore Andreis F, Mørch CD, Jensen W and Meijs S (2023) On determining the mechanical nociceptive threshold in pigs: a reliability study. Front. Pain Res. 4:1191786. doi: 10.3389/fpain.2023.1191786 EDITED BY Peter Wilhelm Marius Kronen, Veterinary Anaesthesia Services International GmbH, Switzerland REVIEWED BY Daniel Segelcke, University Hospital Münster, Germany Antti Pertovaara, University of Helsinki, Finland CORRESPONDENCE Felipe Rettore Andreis fran@hst.aau.dk RECEIVED 22 March 2023 ACCEPTED 03 May 2023 PUBLISHED 17 May 2023 CITATION Rettore Andreis F, Mørch CD, Jensen W and Meijs S (2023) On determining the mechanical nociceptive threshold in pigs: a reliability study. Front. Pain Res. 4:1191786. doi: 10.3389/fpain.2023.1191786 Felipe Rettore Andreis, Carsten Dahl Mørch, Winnie Jensen and Suzan Meijs Felipe Rettore Andreis, Carsten Dahl Mørch, Winnie Jensen and Suzan Meijs Center for Neuroplasticity and Pain (CNAP), Department of Health Science and Technology, Aalborg University, Aalborg, Denmark Background: A pressure algometer is a valuable tool for assessing the mechanical nociceptive threshold (MNT) in clinical pain studies. Recent research has turned to large animal models of pain because of the closer anatomy and physiology to humans. Although the reliability and usefulness of the MNT have been extensively validated in humans, similar data from large animals is still sparse. Objective: Therefore, the aim of the current study was to evaluate the reliability (within- and between-session) of MNT in the forelimb of pigs using a pressure algometer. Methods: Nine animals were used (23–40 kg), and MNTs were measured at both the right and left limbs at three different sessions, with three repetitions per session. The intraclass correlation coefﬁcient (ICC) was used as a metric for relative reliability. The standard error of measurement (SEM) and coefﬁcient of variation (CV) was used to assess absolute reliability. KEYWORDS pressure algometry, pigs, pain, mechanical nociceptive threshold, reliability pressure algometry, pigs, pain, mechanical nociceptive threshold, reliability Downloaded from vbn.aau.dk on: October 24, 2024 Downloaded from vbn.aau.dk on: October 24, 2024 TYPE Original Research PUBLISHED 17 May 2023 DOI 10.3389/fpain.2023.1191786 2.3. Instrumentation and procedure The pressure algometer system consisted of an active actuator mounted in a placement boot on the animal’s forelimb and connected to a strap and buckle kit to ﬁxate the probe (ProdPro, Topcat Metrology Ltd, United Kingdom). A blunt-ended pin (2 mm diameter) protruded from the placement boot to allow the experimenter to apply pressure on the animal’s limb. The blunt-ended pin was positioned approximately two cm below the middle carpal joint, roughly 45° from the sagittal plane. The pressure was manually induced via air injection from a syringe. The injection rate was controlled with the assistance of green and red lights that indicated if the pressure rate should be increased or decreased to maintain a constant slew rate of 2 N/s. The pressure algometer kit ensured a perpendicular angle between the pressure point and the skin surface. Finally, both limbs were tested, and while the mounting boot and actuator were placed on one limb, a dummy actuator and mounting boot were placed on the contralateral limb (see Figure 1). NTT testing has, over the last decade, been extended to larger animal species (e.g., calves, horses, sheep, and dogs) (10–14), and the reliability of these measures have varied signiﬁcantly between different species and body sites (15–17). In pigs, prior studies have, however, focused on suitability and factors inﬂuencing mechanical nociceptive threshold (MNT). Giminiani et al. demonstrated the feasibility of using a pressure algometer for measuring MNT in pigs’ tails. Janczak et al. and Nalon et al. evaluated confounding factors using hand-held and limb- mounted algometers in assessing MNT in piglets and sows, respectively (16–18). To the best of our knowledge, no studies are currently available focusing on estimating the reliability of mechanical sensory testing in the pig. As the pig is gaining interest as a translational model, the aim of the present study was to quantify the reliability (within-session and between- session) of MNT using a pressure algometer in the forelimb of pigs. The pressure was increased gradually until a foot lift was visually observed at which the respective force level was annotated. Stimulation was also immediately stopped when the cut off force of 25 N was reached. Three measurements were obtained on each limb at each experimental session, with a minimum rest interval of approximately 15 s. Finally, the animals were measured for three days, with a one-day interval between each measurement day. 2.3. Instrumentation and procedure In total, 162 measurements were obtained, representing nine animals measured three times per session, on three sessions, at both limbs. The experimental procedure is described in Figure 2. 1. Introduction Pain is a multifaceted and subjective experience resulting from the intricate interplay between psychological, biological, and social elements (1). Because of its subjective nature, studies in humans rely on the subject’s ability to express their pain experience through standardised questionnaires and quantitative scales. In animals, however, pain cannot be directly measured, and researchers can only infer the animal’s pain state through surrogate behaviours (2). Nociceptive threshold testing (NTT) is a well-validated method to investigate experimentally painful conditions in animals, such as allodynia (i.e., pain due to a stimulus that does not normally provoke pain), hyperalgesia (i.e., increased pain response to a painful stimulus), and to test the efﬁciency of analgesic compounds (3). NTT is stimulus-dependent and entails the application of a quantiﬁable stimulus to a particular body location until a behavioural or physiological response is noticed (e.g., withdrawal, vocalisation) (4). Frontiers in Pain Research frontiersin.org 01 Rettore Andreis et al. Rettore Andreis et al. 10.3389/fpain.2023.1191786 2.2. Habituation and training There are mainly four types of stimuli used in NTT: mechanical, thermal, electrical, and chemical (5). Thermal and mechanical stimuli are the most adopted sensory modalities because they provide natural stimuli that are easy to control and can be applied on a continuous scale, while chemical stimuli need to be dosed and cause sustained stimulation (5). Mechanical stimulation can be further subdivided into static (triggered by pressure), dynamic (triggered by brushing), and punctate (triggered by touch) (6). The pigs were habituated and trained daily at roughly the same time (08:00 to 10:00 AM) to decrease stress levels and increase the method’s reliability. The pigs were habituated to the stable, the caretakers, the researchers, the equipment and separation from the mate for one week after arrival at the facility. The pigs then underwent clicker training individually for one week to train them to stand still and accept the mounting of the boots. When both the active and dummy boot were mounted, the animal received a food bowl with their regular commercial food to allow them to relax and stand still. At the conclusion of the training period, no retraction of the limbs was observed during the mounting boot attachment to the leg. In the third week, measurements of the MNT were conducted while pigs were eating calmly by their food bowl with both boots mounted. Most pain preclinical studies have been conducted in rodents (3) and therefore, numerous techniques have been developed to assess “pain-like” behaviour in this species [for a comprehensive review, please read (6)]. The almost sole dependence on rodents as preclinical models might be an important factor explaining the poor translational record of the pain ﬁeld, and researchers suggested using larger animal models to bridge the translational gap between rodents and humans (7). Pigs, in particular, are promising models because they share many physiological and anatomical characteristics with humans (e.g., skin structure, sequence homology, metabolism, and nerve ﬁbre classes) (8). The interest in pigs was highlighted in a recent systematic review that revealed a substantial increase in the number of studies looking at pain in pigs using various model types (i.e., evoked pain models, production procedures, naturally occurring pain and disease models) (9). 2.1. Animals Nine adolescent female Danish Landrace pigs acquired from a commercial farm were included (23–40 kg). The animals were housed in pairs in iron enclosures with a 13:11 h light-dark cycle. Commercial food was provided twice daily, and nipple drinkers allowed the animals unlimited access to water. The room was maintained at ≈24°C. The study was approved by the Danish Veterinary and Food Administration under the Ministry of Environment and Food of Denmark (protocol number: 2020- 15-0201-00514). 3. Results There were no signiﬁcant differences between the average MNTs of the left and right limbs (session 1: p = 0.95, session 2: p = 0.54, session 3: p = 0.60); therefore, the following analysis was performed on a pooled dataset for the left and right limb. The average MNT for all sessions and trials are shown in Figure 3. The MNT was not signiﬁcantly different between trials in sessions 1 and 3; however, there was a signiﬁcant difference in MNT between trials in session 2 (p = 0.02). The post-hoc analysis revealed a lower MNT in trial 1 than in trials 2 and 3. There were no signiﬁcant differences in average MNT for the between-session analysis (i.e., session 1 vs. session 2, session 2 vs. session 3, and session 1 vs. session 3). FIGURE 1 Mounting boots are attached to both limbs; one contains a dummy actuator and the other the active probe, which is ﬁxated through a strap and buckle kit. The actuator contains a blunt-ended pin that works by pressing against the skin via air injection from a syringe. The results from the MNT reliability analysis for the within- and between-session reliability analysis are shown in Table 1. Six relative reliability measures were obtained with ICC values ranging from 0.30 to 0.81. The average ICC for the within- session analysis was 0.71, while the average ICC for the between- session analysis was 0.45. Interestingly, all values of ICC from the between-session analysis are lower than the ones from the within-session analysis, demonstrating a higher day-to-day variability compared to the variability within the same day. reliability (19). Within and between-session systematic errors were tested with the one-way repeated measures ANOVA. The relative reliability was determined by calculating the intraclass correlation coefﬁcient 2-way mixed-effects model type absolute agreement (ICC2,k) where k indicates the average of three repetitions. The average form, rather than the single measurement of ICC, was selected because a few repetitions are often performed in MNT. ICC values were interpreted based on a previously proposed category, according to which an ICC between than 0.81 and 1.00 is considered almost perfect, from 0.61 to 0.80 it is considered substantial, values between 0.41 and 0.60 are considered moderate, and below 0.40 it is considered unacceptable (20). The obtained measures of absolute reliability were CV and SEM. The CV, expressed as a percentage, ranged from 17.9% to 20.5%. 2.4. Statistical analysis The force measurements were converted to pressure by dividing the force by the probe area. The statistical analysis was split into two parts: the ﬁrst to assess internal consistency (within-session) reliability and the second to assess stability (between-session) Frontiers in Pain Research 02 frontiersin.org 10.3389/fpain.2023.1191786 10.3389/fpain.2023.1191786 10.3389/fpain.2023.1191786 Rettore Andreis et al. The absolute reliability was evaluated by the coefﬁcient variation (CV) and standard error of measurements (SEM). CV was computed by the within-subject standard deviation as a proportion of the within-subject mean, indicating the stability of a measure across repeated trials (21). The SEM was deﬁned as SEM ¼ SD ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ 1 ICC p and reﬂects the precision of individual scores on a test, meaning that it can be considered an estimation of expected random variation when no real change has occurred. FIGURE 1 Mounting boots are attached to both limbs; one contains a dummy actuator and the other the active probe, which is ﬁxated through a strap and buckle kit. The actuator contains a blunt-ended pin that works by pressing against the skin via air injection from a syringe. Finally, results are presented as mean and SD unless otherwise speciﬁed. The adopted signiﬁcance level was 0.05, and the assumptions of normality and homoscedasticity were veriﬁed through residual analysis (Q–Q plots and histograms). Statistical analysis was performed with R software (22). 3. Results The average within-session CV was 18.6%, while the mean between-session CV was 19.7%. The SEM ranged from FIGURE 2 Experimental procedure for the reliability measurements. At each session, the animal is measured three times, at both limbs. FIGURE 2 Experimental procedure for the reliability measurements. At each session, the animal is measured three times, at both limbs. Frontiers in Pain Research 03 frontiersin.org frontiersin.org Rettore Andreis et al. 10.3389/fpain.2023.1191786 Within-session analysis Between-session analysis Session 1 Session 2 Session 3 Sessions 1–2 Sessions 1–3 Sessions 2–3 Mean ± SD 1426 ± 376 1537 ± 368 1522 ± 447 1481 ± 374 1474 ± 414 1530 ± 407 RM-ANOVA p = 0.73 p = 0.02 p = 0.41 p = 0.06 p = 0.13 p = 0.82 CV% 18.1 17.9 19.8 18.9 19.9 20.5 ICC2,k [95% CI] 0.68 [0.40–0.85] 0.66 [0.37–0.84] 0.81 [0.64–0.91] 0.55 [0.30–0.71] 0.51 [0.23–0.69] 0.30 [0–0.56] SEM [absolute vs. %] 215.8 [15.1%] 201.8 [13.1] 195.6 [12.9] 249.4 [16.8] 284.9 [19.6] 340.8 [22.3] Indicates p < 0.05. algometer in three distinct tail regions for animals with different ages and obtained on average, ICCs ranging from 0.33 to 0.46, depending on the tail region. A study investigating mechanical thresholds at the back of the metacarpus in piglets’ legs found across different days ICC values in the range of 0.29 to 0.65 (18). The results obtained in this study indicate that in terms of relative reliability, our method is consistent with previous reports and substantiates the use of MNT in large animals. It must be stated that human studies tend to ﬁnd higher reliability coefﬁcients; for instance, in a study investigating interrater reliability of pressure pain threshold, the authors found high ICCs (>0.92) for several body sites such as wrist, leg, neck and back (27). It is expected that human studies display higher reliability since instructions can be given such that the participants speciﬁcally respond to a painful stimulus. In animals, however, one cannot be sure whether the animal is responding to a noxious sensation or other sensations. Another likely reason for higher ICCs in humans is that humans offer a more heterogeneous sample and therefore larger individual differences. As relative reliability is dependent on the between- subjects variability, this, in turn, can result in a higher ICC. 195.6 kPA (12.9%) to 340.8 kPA (22.3%), with an average of 204 kPA (13.7%) for the within-session and 291.7 kPA (19.5%) for the between-session reliability. The measures of absolute reliability indicate a lower reliability of the data from day to day compared to the variability of multiple repetitions within the same day. Finally, the overall reliability between the legs was computed, resulting in an ICC of 0.61 [95% CI: 0.43–0.73]. 4.2. Absolute reliability Regarding absolute reliability, we obtained a grand-average CV of 19.1%, lower than previous studies that found an MNT CV of 4. Discussion The assessment of pain behaviour in pigs has been mostly concerned with pig production procedures such as tail amputation (23) and castration (24). Biomedical research has also seen an increased interest in pigs as subjects for translational pain models (25, 26), where the assessment often relies on evoked responses of mechanical and thermal nociceptive stimulation. In order for a behavioural model for nociception to be useful, the measurement must meet ﬁve different requirements: speciﬁcity, sensitivity, validity, reliability, and reproducibility (5). The literature on the reliability of quantitative-sensory testing in these animal models is still sparse. Therefore, this study focused on estimating the reliability of MNT longitudinally in the forelimb of pigs. Frontiers in Pain Research 4.1. Relative reliability FIGURE 3 Mean mechanical nociceptive thresholds (MNTs) across every trial and session. The error bars represent standard deviations. p < 0.05. Six relative reliability measures were obtained with one almost perfect ICC (>0.81), two substantial ICCs (0.61–0.80), two moderate ICCs (0.41–0.60), and one unacceptable ICC (<0.40). The average ICC for the within-session analysis was substantial (i.e., 0.71), while the average ICC for the between-session analysis was moderate (i.e., 0.45), conﬁrming that a pressure algometer is a valuable tool in assessing the nociception in the forelimb of pigs. We have also found that the within-session reliability was higher than the between-session reliability, indicating that the most considerable variability occurs between different days, even though measurements were obtained at roughly the same time every day to control for circadian patterns. This result could be explained by subtle changes in the position of the blunt-ended pin. Contrary to the hand-held algometer, where the exact position can be seen, the placement boot masks a clear view of the pin location. Earlier studies were carried out to quantify the reliability of these methods in other species or at different body locations of the pig, such as the tail (16), where the authors used a hand-held pressure FIGURE 3 Mean mechanical nociceptive thresholds (MNTs) across every trial and session. The error bars represent standard deviations. p < 0.05. Mean mechanical nociceptive thresholds (MNTs) across every trial and session. The error bars represent standard deviations. p < 0.05. Frontiers in Pain Research 04 frontiersin.org Rettore Andreis et al. 10.3389/fpain.2023.1191786 10.3389/fpain.2023.1191786 25.5% (17) and 35% (14) for the limbs of pigs and dogs, respectively. A CV of 19.1% can be considered low and is comparable to values obtained in human studies (14.6% for the leg and 17.7% for the arm) (28). Only a few studies reported the SEM, making the comparison with existing literature challenging. Still, the SEM is an important parameter to be compared with future studies as it indicates the precision of individual scores on the test (29). The grand-average SEM of 249.5 kPA obtained in this study is considerably higher than the 93 kPA obtained in the leg of humans (27), which can be explained by the fact that SEM tends to increase at higher scale values (20), and pigs have a higher MNT than humans. 4.1. Relative reliability Therefore, to allow for comparison across different species, we also calculated the SEM in terms of percentages of the mean, which resulted in an average value of 13.7% and 19.5% for the within- and between-session, respectively. These values are comparable to human studies (20). make the subjects stay still long enough to obtain accurate measures, especially in large animals. For instance, rodents can be immobilised with the hand to obtain the measurements (34). A “forced” immobilisation in large animals would be practically impossible and can generate stress-induced analgesia, affecting the measurements by increasing thresholds (35). Still, recent studies have shown the feasibility of assessing the nociceptive system in the limbs of pigs using von Frey ﬁlaments (36), laser stimulation (37), and mechanical stimulation (38). The latter used a perforated test platform to which the animals were acclimatised. In our study, the animals were tested in their pen, and a food bowl was sufﬁcient to keep them standing still for the duration of the task. It must be noted that several factors can inﬂuence the MNT; therefore, caution must be taken when translating the results from this experiment to other studies. Previous studies demonstrated that mechanical threshold increases with larger tip diameters (39) and time of the day, where thresholds are higher in the morning than in the afternoon (17). The range of MNT values obtained in this study is similar to other studies in pigs at the same weight range (38), but it is far smaller than MNTs in the limbs of heavier animals. In pigs weighing an average of 267 kg, thresholds of 16,500 kPA were observed (31). Therefore, the direct comparison of MNT between studies should also consider the effect of animal weight, as mechanical thresholds are positively correlated with body weight (18). No systematic differences in MNT were observed between the sessions, suggesting no effect of habituation (i.e., increased thresholds) or sensitisation (i.e., decreased thresholds). A systematic difference was observed only within session 2, where MNT increased in trials 2 and 3, suggesting some adaptation has occurred. Still, the lack of systematic changes between sessions highlights the importance of adequate training of the animals for the task prior to the experiment so that measurements are not obtained in a period where the animal’s familiarisation curve is changing. 4.1. Relative reliability A prior study investigating MNT in dairy cows reported that pre-test habituation decreased the variability and increased the reliability of MNT (30). In sows, a study assessing anatomical and methodological factors inﬂuencing MNT reported increased MNTs over measurement days up to a stabilisation in the fourth and ﬁfth day, indicating habituation to the stimulus (17). We did not observe a difference in the average MNT between the right and left limbs; however, contradictory evidence exists in the literature, with studies also reporting no left-to-right differences in MNT (31) and others reporting different values on the left vs. right side of the body, which might be a result of left- or right-side dominance (32). This study was conducted only on female pigs for two reasons. First, female subjects are underrepresented in preclinical pain research (40), despite the fact that the majority of chronic pain sufferers are female (41). This trend resulted in a male-based literature (42). The second reason relates to swine housing. Due to the fact that the animals in this experiment were housed in pairs, it is known that entire male pigs tend to exhibit more aggressive behaviour and ﬁghting activity, particularly during puberty (43), which could impede the continuation of the study. Additionally, housing mixed-sex groups also result in more aggressive behaviour than housing only females (44). Sex differences in MNT have been investigated in dogs (39) and piglets (18), and none of the studies reported signiﬁcant differences between males and females. 4.3. Methodological considerations The effect of different examiners on reliability was not investigated in this study and should be considered in future experiments, as different examiners can display signiﬁcantly different reliability levels that may be related to the examiner’s experience or timing in detecting avoidance reactions (15). An advantage of the “remotely-controlled” actuator used in this study is that the animal has no visual cue when the stimulus is given, which can generate anticipation of the stimulus (17). Another factor concerning the device that may inﬂuence the measurements’ reliability is the pressure application rate; the device uses a light system (green and red diodes) to indicate if the pressure is at the selected level. We observed that prior training with the device was enough for the researcher to keep the pressure rate stable during the experiment. Still, further improvements could include computer-controlled algometers with ﬁxed pressure rates. Frontiers in Pain Research Acknowledgments The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. The authors would like to thank the animal caretakers at the laboratory animal facility at Aalborg University Hospital, especially Pernille Mikkelsen, for their valuable assistance during the experiments. References 1. Cohen SP, Vase L, Hooten WM. Chronic pain: an update on burden, best practices, and new advances. Lancet (2021) 397(10289):2082–97. doi: 10.1016/ S0140-6736(21)00393-7 11. Schambourg M, Taylor PM. Mechanical nociceptive thresholds in endurance horses. Vet Rec. (2020) 186(4):124. doi: 10.1136/vr.105499 12. Love EJ, Taylor PM, Murrell J, Whay HR. Effects of acepromazine, butorphanol and buprenorphine on thermal and mechanical nociceptive thresholds in horses. Equine Vet J. (2012) 44(2):221–5. doi: 10.1111/j.2042- 3306.2011.00412.x 2. Du Sert NP, Rice ASC. Improving the translation of analgesic drugs to the clinic: animal models of neuropathic pain. Br J Pharmacol. (2014) 171(12):2951–63. doi: 10. 1111/bph.12645 2. Du Sert NP, Rice ASC. Improving the translation of analgesic drugs to the clinic: animal models of neuropathic pain. Br J Pharmacol. (2014) 171(12):2951–63. doi: 10. 1111/bph.12645 13. Rohrbach H, Andersen OK, Zeiter S, Wieling R, Spadavecchia C. Repeated electrical stimulations as a tool to evoke temporal summation of nociceptive inputs in healthy, non-medicated experimental sheep. Physiol Behav. (2015) 142:85–9. doi: 10.1016/j.physbeh.2015.02.008 3. Mogil JS. Animal models of pain: progress and challenges. Nat Rev Neurosci. (2009) 10(4):283–94. doi: 10.1038/nrn2606 3. Mogil JS. Animal models of pain: progress and challenges. Nat Rev Neurosci. (2009) 10(4):283–94. doi: 10.1038/nrn2606 4. Love EJ, Murrell J, Whay HR. Thermal and mechanical nociceptive threshold testing in horses: a review. Vet Anaesth Analg. (2011) 38(1):3–14. doi: 10.1111/j. 1467-2995.2010.00580.x 14. Briley JD, Williams MD, Freire M, Grifﬁth EH, Lascelles BDX. Feasibility and repeatability of cold and mechanical quantitative sensory testing in normal dogs. Vet J. (2014) 199(2):245–50. doi: 10.1016/j.tvjl.2013.10.025 5. Le Bars D, Gozariu M, Cadden SW. Animal models of nociception. Pharmacol Rev. (2001) 53(4):597–652. 15. Menke ES, Blom G, van Loon JPAM, Back W. Pressure algometry in Icelandic horses: interexaminer and intraexaminer reliability. J Equine Vet Sci. (2016) 36:26–31. doi: 10.1016/j.jevs.2015.10.007 6. Deuis JR, Dvorakova LS, Vetter I. Methods used to evaluate pain behaviors in rodents. Front Mol Neurosci. (2017) 10(September):1–17. doi: 10.3389/fnmol.2017. 00284 16. Di Giminiani P, Sandercock DA, Malcolm EM, Leach MC, Herskin MS, Edwards SA. Application of a handheld pressure application measurement device for the characterisation of mechanical nociceptive thresholds in intact pig tails. Physiol Behav. (2016) 165:119–26. doi: 10.1016/j.physbeh.2016.07.006 7. Castel D, Sabbag I, Meilin S. The effect of local/topical analgesics on incisional pain in a pig model. J Pain Res. (2017) 10:2169–75. doi: 10.2147/JPR.S144949 8. Gigliuto C, De Gregori M, Malafoglia V, Raffaeli W, Compagnone C, Visai L, et al. Publisher’s note The study was designed by SM and WJ. FRA and SM collected the data. Data processing was performed by FRA with the supervision of CDM. FRA wrote the ﬁrst draft and subsequent versions of the manuscript. SM, WJ, and CDM critically revised the ﬁrst draft and subsequent versions of the article. All authors contributed to the article and approved the submitted version. All claims expressed in this article are solely those of the authors and do not necessarily represent those of their afﬁliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. Funding and comparable to reliability studies performed in humans. Lastly, the absence of systematic differences between sessions corroborates the need for proper training of the animals prior to obtaining measurements. This work was funded by the Center for Neuroplasticity and Pain (CNAP). CNAP is supported by the Danish National Research Foundation (DNRF121). 5. Conclusion The aim of the present work was to quantify the reliability (within-session and between-session) of mechanical nociceptive threshold (MNT) using a pressure algometer. This study indicates that mechanical nociceptive testing through a pressure algometer is a reliable research tool for investigating nociceptive thresholds in the limbs of pigs. Measures of absolute and relative reliability were superior to other animal studies The limbs are particularly important in neuropathic pain models, where the disease is induced by some form of peripheral nerve injury (33). In animals, it is challenging to Frontiers in Pain Research 05 frontiersin.org 10.3389/fpain.2023.1191786 10.3389/fpain.2023.1191786 Rettore Andreis et al. Conﬂict of interest The animal study was reviewed and approved by Danish Veterinary and Food Administration. The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. References Pain assessment in animal models: do we need further studies? J Pain Res. (2014) 7:227–36. doi: 10.2147/JPR.S59161 17. Nalon E, Maes D, Piepers S, Taylor P, van Riet MMJ, Janssens GPJ, et al. Factors affecting mechanical nociceptive thresholds in healthy sows. Vet Anaesth Analg. (2016) 43(3):343–55. doi: 10.1111/vaa.12313 9. Meijs S, Schmelz M, Meilin S, Jensen W. A systematic review of porcine models in translational pain research. Lab Anim (NY). (2021) 50(11):313–26. doi: 10.1038/ s41684-021-00862-4 18. Janczak AM, Ranheim B, Fosse TK, Hild S, Nordgreen J, Moe RO, et al. Factors affecting mechanical (nociceptive) thresholds in piglets. Vet Anaesth Analg. (2012) 39 (6):628–35. doi: 10.1111/j.1467-2995.2012.00737.x 10. Williams HJ, Duncan JS, Grove-White DH, Mahen PJ, Gillespie AV. Repeatability and feasibility of pressure algometry for quantifying mechanical nociceptive threshold in the thoracic region of calves. Front Vet Sci. (2020) 7 (July):1–8. doi: 10.3389/fvets.2020.00442 19. Atkinson G, Nevill AM. Statistical methods for assessing measurement error (reliability) in variables relevant to sports medicine. Sport Med. (1998) 26 (4):217–38. doi: 10.2165/00007256-199826040-00002 Frontiers in Pain Research frontiersin.org 06 Rettore Andreis et al. 10.3389/fpain.2023.1191786 32. Haussler KK, Hill AE, Frisbie DD, McIlwraith CW. Determination and use of mechanical nociceptive thresholds of the thoracic limb to assess pain associated with induced osteoarthritis of the middle carpal joint in horses. Am J Vet Res. (2007) 68(11):1167–76. doi: 10.2460/ajvr.68.11.1167 20. Walton D, Macdermid J, Nielson W, Teasell R, Chiasson M, Brown L. Reliability, standard error, and minimum detectable change of clinical pressure pain threshold testing in people with and without acute neck pain. J Orthop Sports Phys Ther. (2011) 41(9):644–50. doi: 10.2519/jospt.2011.3666 33. Castel D, Sabbag I, Brenner O, Meilin S. Peripheral neuritis trauma in pigs: a neuropathic pain Model. J Pain. (2016) 17(1):36–49. doi: 10.1016/j.jpain.2015.09.011 21. Doi SAR, Williams GM, editors. Methods of clinical epidemiology. Berlin, Heidelberg: Springer Berlin Heidelberg (2013). Springer Series on Epidemiology and Public Health. Available at: http://link.springer.com/10.1007/978-3-642-37131-8 34. Santos-Nogueira E, Redondo Castro E, Mancuso R, Navarro X. Randall-Selitto test: a new approach for the detection of neuropathic pain after spinal cord injury. J Neurotrauma. (2012) 29(5):898–904. doi: 10.1089/neu.2010.1700 22. R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing (2020). Available at: https://www.r-project-org/ 35. Di Giminiani P, Stausholm JS, Viitasaari E, Petersen LJ, Herskin MS. The effect of social isolation, gender and familiarity with the experimental procedure on tests of porcine nociceptive thresholds. Vet Anaesth Analg. References (2015) 42(6):648–56. doi: 10.1111/ vaa.12254 23. Di Giminiani P, Edwards SA, Malcolm EM, Leach MC, Herskin MS, Sandercock DA. Characterization of short- and long-term mechanical sensitisation following surgical tail amputation in pigs. Sci Rep. (2017) 7(1):1–9. doi: 10.1038/s41598-017- 05404-y 36. Herskin MS, Rasmussen JS. Pigs in pain—porcine behavioural responses towards mechanical nociceptive stimulation directed at the hind legs. Scand J Pain. (2010) 1 (3):175–6. doi: 10.1016/j.sjpain.2010.05.026 24. Viscardi AV, Hunniford M, Lawlis P, Leach M, Turner PV. Development of a piglet grimace scale to evaluate piglet pain using facial expressions following castration and tail docking: a pilot study. Front Vet Sci. (2017) 4(APR):1–9. doi: 10. 3389/fvets.2017.00051 37. Herskin MS, Ladewig J, Arendt-Nielsen L. Measuring cutaneous thermal nociception in group-housed pigs using laser technique-effects of laser power output. Appl Anim Behav Sci. (2009) 118(3–4):144–51. doi: 10.1016/j.applanim. 2009.02.016 25. Castel D, Sabbag I, Nasaev E, Peng S, Meilin S. Open ﬁeld and a behavior score in PNT model for neuropathic pain in pigs. J Pain Res. (2018) 11:2279–93. doi: 10. 2147/JPR.S172300 38. Sandercock DA, Gibson IF, Brash HM, Rutherford KMD, Scott EM, Nolan AM. Development of a mechanical stimulator and force measurement system for the assessment of nociceptive thresholds in pigs. J Neurosci Methods. (2009) 182 (1):64–70. doi: 10.1016/j.jneumeth.2009.05.019 26. Janjua TA, dos Santos Nielsen TG, Andreis FR, Meijs S, Jensen W. The effect of peripheral high-frequency electrical stimulation on the primary somatosensory cortex in pigs. IBRO Neurosci Rep. (2021) 11(June):112–8. doi: 10.1016/j.ibneur.2021.08.004 39. Harris LK, Murrell JC, van Klink EGM, Whay HR. Inﬂuence of experimental protocol on response rate and repeatability of mechanical threshold testing in dogs. Vet J. (2015) 204(1):82–7. doi: 10.1016/j.tvjl.2015.02.008 27. Waller R, Straker L, O’Sullivan P, Sterling M, Smith A, Arendt-Nielsen L. Reliability of pressure pain threshold testing in healthy pain free young adults. Scand J Pain. (2015) 9:28–9. doi: 10.1016/j.sjpain.2015.06.002 40. Mogil JS, Chanda ML. The case for the inclusion of female subjects in basic science studies of pain. Pain. (2005) 117(1–2):1–5. doi: 10.1016/j.pain.2005.06.020 28. Graven-Nielsen T, Vaegter HB, Finocchietti S, Handberg G, Arendt-Nielsen L. Assessment of musculoskeletal pain sensitivity and temporal summation by cuff pressure algometry: a reliability study. Pain. (2015) 156(11):2193–202. doi: 10.1097/ j.pain.0000000000000294 41. Ruau D, Liu LY, Clark JD, Angst MS, Butte AJ. Sex differences in reported pain across 11,000 patients captured in electronic medical records. J Pain. (2012) 13 (3):228–34. doi: 10.1016/j.jpain.2011.11.002 29. Weir JP. Rettore Andreis et al. References Quantifying test-retest reliability using the intraclass correlation coefﬁcient and the SEM. J Strength Cond Res. (2005) 19(1):231–40. doi: 10.1519/ 15184.1 42. Mogil JS. Qualitative sex differences in pain processing: emerging evidence of a biased literature. Nat Rev Neurosci. (2020) 21(7):353–65. doi: 10.1038/s41583-020- 0310-6 30. Raundal PM, Andersen PH, Toft N, Herskin MS, Forkman B, Munksgaard L, et al. Pre-test habituation improves the reliability of a handheld test of mechanical nociceptive threshold in dairy cows. Res Vet Sci. (2015) 102:189–95. doi: 10.1016/j. rvsc.2015.07.020 30. Raundal PM, Andersen PH, Toft N, Herskin MS, Forkman B, Munksgaard L, et al. Pre-test habituation improves the reliability of a handheld test of mechanical nociceptive threshold in dairy cows. Res Vet Sci. (2015) 102:189–95. doi: 10.1016/j. rvsc.2015.07.020 43. von Borell E, Bonneau M, Holinger M, Prunier A, Stefanski V, Zöls S, et al. Welfare aspects of raising entire male pigs and immunocastrates. Animals (Basel). (2020) 10(11):1–12. doi: 10.3390/ani10112140 44. Hintze S, Scott D, Turner S, Meddle SL, D’Eath RB. Mounting behaviour in ﬁnishing pigs: stable individual differences are not due to dominance or stage of sexual development. Appl Anim Behav Sci. (2013) 147(1–2):69–80. doi: 10.1016/j. applanim.2013.04.023 31. Nalon E, Maes D, Piepers S, van Riet MMJ, Janssens GPJ, Millet S, et al. Mechanical nociception thresholds in lame sows: evidence of hyperalgesia as measured by two different methods. Vet J. (2013) 198(2):386–90. doi: 10.1016/j.tvjl. 2013.08.016 31. Nalon E, Maes D, Piepers S, van Riet MMJ, Janssens GPJ, Millet S, et al. Mechanical nociception thresholds in lame sows: evidence of hyperalgesia as measured by two different methods. Vet J. (2013) 198(2):386–90. doi: 10.1016/j.tvjl. 2013.08.016 07 Frontiers in Pain Research 07 frontiersin.org
https://openalex.org/W4205618419	https://figshare.com/articles/thesis/Development_and_demonstration_of_an_energy_feedback_research_platform_in_a_field_study_with_real-time_social_comparisons/14646003/1/files/28126083.pdf	English	null	Development and demonstration of an energy feedback research platform in a field study with real-time social comparisons	null	2,021	cc-by	40,038	DEVELOPMENT AND DEMONSTRATION OF AN ENERGY FEEDBACK RESEARCH PLATFORM IN A FIELD STUDY WITH REAL-TIME SOCIAL COMPARISONS Author’s Declaration I hereby declare that I am the sole author of this thesis. This is a true copy of the thesis, including any required final versions, as accepted by my examiners. I authorize Ryerson University to lend this thesis to other institutions or individuals for the purpose of scholarly research. Kevin Trinh Kevin Trinh I further authorize Ryerson University to reproduce this thesis by photocopying or by other means, in total or in part, at the request of other institutions or individuals for the purpose of scholarly research. I understand that my thesis may be made electronically available to the public. Dated: _ Dated: _ Kevin Trinh ii DEVELOPMENT AND DEMONSTRATION OF AN ENERGY FEEDBACK RESEARCH PLATFORM IN A FIELD STUDY WITH REAL-TIME SOCIAL COMPARISONS Kevin Trinh Master of Applied Science Program of Building Science Kevin Trinh Kevin Trinh Abstract Providing residential tenants with feedback on their energy use can be an effective intervention, promoting savings ranging from 4-12%. However, advancements in feedback design have been hindered by methodological limitations, the lack of specification of visual feedback designs, and a poor understanding of the behaviour changes that are induced by feedback. This thesis presents the design and demonstration of an Internet-of-Things-based feedback research platform, which was intended to help address these issues, and which will be made freely available for re-use and reconfiguration. Providing residential tenants with feedback on their energy use can be an effective intervention, promoting savings ranging from 4-12%. However, advancements in feedback design have been hindered by methodological limitations, the lack of specification of visual feedback designs, and a poor understanding of the behaviour changes that are induced by feedback. This thesis presents the design and demonstration of an Internet-of-Things-based feedback research platform, which was intended to help address these issues, and which will be made freely available for re-use and reconfiguration. Configured for a rental apartment building in Toronto, Canada, the platform was a central component of a conservation program and field study examining the efficacy of real-time social comparisons. Results showed a statistically significant effect of the conservation program with a relative year-over-year, weather-normalized savings of approximately 11%. An encouraging, but non-significant, finding of a 3.5% relative improvement with real-time social comparisons warrants future large scale studies. Configured for a rental apartment building in Toronto, Canada, the platform was a central component of a conservation program and field study examining the efficacy of real-time social comparisons. Results showed a statistically significant effect of the conservation program with a relative year-over-year, weather-normalized savings of approximately 11%. An encouraging, but non-significant, finding of a 3.5% relative improvement with real-time social comparisons warrants future large scale studies. iii iii This thesis would not have been possible without the help of many people. First I would like to thank my supervisors Prof. Alan Fung and Prof. Vera Straka for providing guidance and a tremendous and timely opportunity. I would also like to thank the members of my project team: Dr. Sara Alsaadani, Samira Zare Mohazabieh. To Danilo Yu, Gabriel Leong, and Edward Vuong, thank you for your friendship and technical assistance in setting up this study. To the ever hospitable staff at Phoenix Place, Predrag Milenkovic and Glenda Moore: I only hope to have demonstrated your cooperation and trust in me was wise. A special thank you to Dr. Winnie Chen for more than refreshing me on statistical analyses. I would like to acknowledge our project sponsors: Canada Mortgage and Housing Corporation (CMHC), Ontario Ministry of Municipal Affairs and Housing (MAH), City of Toronto, Enbridge Gas Distribution Inc., and MITACS. To Wendy: as I’ve written before, I cannot say thank you enough. Your unfailing support made this journey possible. Finally, to Amber: I hope this work may one day demonstrate and instill in you the values of courage and perseverance. iv iv Contents Contents AUTHOR’S DECLARATION ...................................................................................................................................... II ABSTRACT ............................................................................................................................................................. III ACKNOWLEDGEMENTS ......................................................................................................................................... IV CONTENTS ............................................................................................................................................................. V LIST OF FIGURES .................................................................................................................................................. VIII LIST OF TABLES ................................................................................................................................................... VIII LIST OF ABBREVIATIONS ....................................................................................................................................... IX 1. INTRODUCTION ............................................................................................................................................. 1 2. LITERATURE REVIEW ..................................................................................................................................... 5 2.1 ENERGY BEHAVIOURS......................................................................................................................................... 5 2.1.1 Behaviour Taxonomies ........................................................................................................................ 5 2.1.2 Behaviour Change Challenges ............................................................................................................. 7 2.1.3 Behavioural Models ............................................................................................................................. 8 2.2 ENERGY FEEDBACK .......................................................................................................................................... 11 2.2.1 How feedback works ......................................................................................................................... 11 2.2.2 The different dimensions of energy feedback .................................................................................... 12 2.2.3 Feedback in context: Frameworks and Complementary Engagement Strategies ............................. 14 2.2.4 Summary of literature review on good feedback .............................................................................. 19 2.3 WHERE DO WE GO FROM HERE? ........................................................................................................................ 20 3. FEEDBACK RESEARCH PLATFORM DESIGN ................................................................................................... 22 3.1 PLATFORM TECHNICAL REQUIREMENTS ............................................................................................................... 22 3.2 OPEN SOURCE PROJECTS AND FREE UTILITIES ....................................................................................................... 26 3.2.1 The Open Energy Monitor Project ..................................................................................................... 26 3.2.2 SafePlugs ........................................................................................................................................... 30 3.2.3 Open Data Kit for Surveys .................................................................................................................. 31 3.2.4 Android App ....................................................................................................................................... 32 3.2.5 Web / Android Analytics .................................................................................................................... 33 3.2.6 Weather Data .................................................................................................................................... 33 3.3 SYSTEM ARCHITECTURE .................................................................................................................................... 34 3.4. IMPLEMENTATION AND MAINTENANCE SKILLS REQUIREMENTS ................................................................................ 34 v 4. PROJECT IMPLEMENTATION AT PHOENIX PLACE . vi 4. PROJECT IMPLEMENTATION AT PHOENIX PLACE ......................................................................................... 36 4.1 MURBS CONTEXT .......................................................................................................................................... 36 4.2 PHOENIX PLACE – HISTORY AND FACTS ............................................................................................................... 38 4.3 RESULTS FROM A POST OCCUPANCY EVALUATION AT PHOENIX PLACE ....................................................................... 40 4.4 THE CBSM FRAMEWORK IN APPLICATION ........................................................................................................... 41 4.5 SUMMARIZING THE FEEDBACK DESIGN CONTEXT, OBJECTIVES, AND CONSTRAINTS ...................................................... 43 4.6 MEASURING THE IMPACT OF THERMAL COMFORT ................................................................................................. 44 4.7 FEEDBACK DESIGN PROCESS .............................................................................................................................. 45 4.7.1 Iteration 1 – Heuristic Design ............................................................................................................ 46 4.7.2 Iteration 2 – Usability Test Prototype based on Initial Feedback ...................................................... 52 4.7.3 Iteration 3 – Final Prototype for Field Study ...................................................................................... 54 5. FIELD STUDY DESIGN AT A MURB IN TORONTO ........................................................................................... 58 5.1 HYPOTHESES .................................................................................................................................................. 58 5.2 EXPERIMENT DESIGN ....................................................................................................................................... 59 5.3 RECRUITMENT AND PARTICIPANTS ...................................................................................................................... 59 5.4 PROCEDURE ................................................................................................................................................... 60 5.5 EQUIPMENT AND INSTALLATION ......................................................................................................................... 62 5.6 MEASURES .................................................................................................................................................... 65 5.7 THERMAL COMFORT MEASURES ........................................................................................................................ 65 6. RESULTS AND DISCUSSION .......................................................................................................................... 67 6.1 SYSTEM PERFORMANCE .................................................................................................................................... 67 6.2 PARTICIPANT NOISE AND VARIABILITY .................................................................................................................. 67 6.3 CONSERVATION PROGRAM: ENERGY SAVINGS ...................................................................................................... 68 6.4 TEST OF HYPOTHESIS 1 – THE EFFECT OF THE CONSERVATION PROGRAM ................................................................... 69 6.5 TEST OF HYPOTHESIS 2 – THE EFFECT OF SOCIAL COMPARISONS .............................................................................. 72 6.5 EXPLORATORY ANALYSIS OF ENGAGEMENT ........................................................................................................... 74 6.6 EXIT SURVEY RESULTS ...................................................................................................................................... List of Figures Figure 1-1: Examples of different feedback approaches .............................................................................. 1 Figure 2-1: Heuristic model of environmentally relevant behavior .............................................................. 8 Figure 2-2: Simplified default construal-time relationship based on temporal construal theory .............. 10 Figure 2-3: Household savings broken down by feedback type ................................................................. 13 Figure 3-1: Open Energy Monitor system components and connectivity .................................................. 27 Figure 3-2: emonTx V3 ruggedized power and temperature sensing now ................................................ 29 Figure 3-3: emonTH temperature and humidity, battery-powered sensing node ..................................... 30 Figure 3-4: SafePlug Home Energy Manager Kit ......................................................................................... 31 Figure 3-5: System architecture for the feedback research platform ........................................................ 34 Figure 4-1: Typical floor plan at Phoenix Place ........................................................................................... 39 Figure 4-2: Basic Feedback + Social Comparisons – Suite Dashboard – Iteration 1 ................................... 46 Figure 4-3: Basic Feedback + Social Comparisons – FCU Dashboard – Iteration 1 ..................................... 47 Figure 4-4: Basic Feedback + Social Comparisons – Suite Dashboard – Iteration 2 ................................... 52 Figure 4-5: Basic Feedback + Social Comparisons – FCU Dashboard – Iteration 2 ..................................... 53 Figure 4-6: Basic Feedback + Social Comparisons – Suite Dashboard – Iteration 3 ................................... 55 Figure 4-7: Basic Feedback + Social Comparisons – FCU Dashboard – Iteration 3 ..................................... 55 Figure 4-8: Basic Feedback Display 1 of 2 – Total Suite Energy Use ........................................................... 57 Figure 4-9: Basic Feedback Display 2 of 2 – Fan Coil Unit - Energy Use ...................................................... 57 Figure 5-1: Architecture for the system installed at Phoenix Place ............................................................ 63 Figure 5-2: Rich-picture diagram of feedback hardware ............................................................................ 63 Figure 6-1: Aggregated year-over-year savings by feedback condition ..................................................... 69 Figure 6-2: Error bar graph of experimental condition on savings percentage .......................................... 70 Figure 6-3: Box plot of experimental conditions on savings percentage ................................................... 71 Figure 6-4: Error bar graph of feedback condition on savings % ................................................................ 72 Figure 6-5: Savings and Engagement by Quarter ........................................................................................ 74 List of Tables Table 2-1. Energy Behaviors as a Function of Frequency and Cost............................................................... 6 Table 2-2. Distinguishing Low-Level and High-Level Construals. ................................................................ 10 Table 2-3. Selecting Tools Based on Barriers and Benefits. ........................................................................ 15 Table 4-1. Johnson Controls Enviro-Tec Model VFE Size 20. 115 Volts. ..................................................... 45 Table 5-1. Participant breakdown by Experimental Group, Gender, Place of Birth, and Age ................... 60 Figure 1-1: Examples of different feedback approaches .............................................................................. 1 Figure 2-1: Heuristic model of environmentally relevant behavior .............................................................. Contents 75 7. CONCLUSIONS ............................................................................................................................................. 77 7.1 CONTRIBUTIONS ............................................................................................................................................. 78 7.2 FUTURE WORK ............................................................................................................................................... 79 APPENDIX A – RECRUITMENT POSTER ................................................................................................................. 83 APPENDIX B – CONSERVATION PROGRAM CONSENT FORM ................................................................................ 85 vi APPENDIX C – FIELD STUDY CONSENT FORM ....................................................................................................... 88 APPENDIX D – CONSERVATION PROGRAM PRESENTATION SLIDES ...................................................................... 93 APPENDIX E – ENERGY TRACKING PRESENTATION SLIDES.................................................................................... 98 APPENDIX F – NEP AND DEMOGRAPHICS SURVEY ............................................................................................. 101 APPENDIX G – PLEDGE FORM............................................................................................................................. 105 APPENDIX H – THERMAL COMFORT SURVEY ..................................................................................................... 107 APPENDIX I – ENERGY AUDIT SAMPLE RESULTS ................................................................................................. 112 APPENDIX J – USABILITY TEST SCRIPT ................................................................................................................ 114 APPENDIX K – EXIT SURVEYS .............................................................................................................................. 120 APPENDIX L – RESEARCH ETHICS BOARD APPROVAL LETTERS ............................................................................ 123 REFERENCES ....................................................................................................................................................... 126 APPENDIX C – FIELD STUDY CONSENT FORM ....................................................................................................... 88 APPENDIX D – CONSERVATION PROGRAM PRESENTATION SLIDES ...................................................................... 93 APPENDIX E – ENERGY TRACKING PRESENTATION SLIDES.................................................................................... 98 APPENDIX F – NEP AND DEMOGRAPHICS SURVEY ............................................................................................. 101 APPENDIX G – PLEDGE FORM............................................................................................................................. 105 APPENDIX H – THERMAL COMFORT SURVEY ..................................................................................................... 107 APPENDIX I – ENERGY AUDIT SAMPLE RESULTS ................................................................................................. 112 APPENDIX J – USABILITY TEST SCRIPT ................................................................................................................ 114 APPENDIX K – EXIT SURVEYS .............................................................................................................................. 120 APPENDIX L – RESEARCH ETHICS BOARD APPROVAL LETTERS ............................................................................ 123 REFERENCES ....................................................................................................................................................... 126 vii List of Figures List of Figures 8 Figure 2-2: Simplified default construal-time relationship based on temporal construal theory .............. 10 Figure 2-3: Household savings broken down by feedback type ................................................................. 13 Figure 3-1: Open Energy Monitor system components and connectivity .................................................. 27 Figure 3-2: emonTx V3 ruggedized power and temperature sensing now ................................................ 29 Figure 3-3: emonTH temperature and humidity, battery-powered sensing node ..................................... 30 Figure 3-4: SafePlug Home Energy Manager Kit ......................................................................................... 31 Figure 3-5: System architecture for the feedback research platform ........................................................ 34 Figure 4-1: Typical floor plan at Phoenix Place ........................................................................................... 39 Figure 4-2: Basic Feedback + Social Comparisons – Suite Dashboard – Iteration 1 ................................... 46 Figure 4-3: Basic Feedback + Social Comparisons – FCU Dashboard – Iteration 1 ..................................... 47 Figure 4-4: Basic Feedback + Social Comparisons – Suite Dashboard – Iteration 2 ................................... 52 Figure 4-5: Basic Feedback + Social Comparisons – FCU Dashboard – Iteration 2 ..................................... 53 Figure 4-6: Basic Feedback + Social Comparisons – Suite Dashboard – Iteration 3 ................................... 55 Figure 4-7: Basic Feedback + Social Comparisons – FCU Dashboard – Iteration 3 ..................................... 55 Figure 4-8: Basic Feedback Display 1 of 2 – Total Suite Energy Use ........................................................... 57 Figure 4-9: Basic Feedback Display 2 of 2 – Fan Coil Unit - Energy Use ...................................................... 57 Figure 5-1: Architecture for the system installed at Phoenix Place ............................................................ 63 Figure 5-2: Rich-picture diagram of feedback hardware ............................................................................ 63 Figure 6-1: Aggregated year-over-year savings by feedback condition ..................................................... 69 Figure 6-2: Error bar graph of experimental condition on savings percentage .......................................... 70 Figure 6-3: Box plot of experimental conditions on savings percentage ................................................... 71 Figure 6-4: Error bar graph of feedback condition on savings % ................................................................ 72 Figure 6-5: Savings and Engagement by Quarter ........................................................................................ 74 List of Tables Table 2-1. Energy Behaviors as a Function of Frequency and Cost............................................................... 6 Table 2-2. Distinguishing Low-Level and High-Level Construals. ................................................................ 10 Table 2-3. Selecting Tools Based on Barriers and Benefits. ........................................................................ 15 Table 4-1. Johnson Controls Enviro-Tec Model VFE Size 20. 115 Volts. ..................................................... 45 Table 5-1. Participant breakdown by Experimental Group, Gender, Place of Birth, and Age ................... 60 List of Tables List of Tables Table 2-1. Energy Behaviors as a Function of Frequency and Cost............................................................... 6 Table 2-2. Distinguishing Low-Level and High-Level Construals. ................................................................ 10 Table 2-3. Selecting Tools Based on Barriers and Benefits. ........................................................................ 15 Table 4-1. Johnson Controls Enviro-Tec Model VFE Size 20. 115 Volts. ..................................................... 45 Table 5-1. Participant breakdown by Experimental Group, Gender, Place of Birth, and Age ................... 60 viii List of Abbreviations ANCOVA Analysis of Covariance API Application Programming Interface CBSM Community-Based Social Marketing CMS Content Management System COP Coefficient of Performance CT Current Transformer EE Energy Efficient FBRP Feedback Research Platform FCU Fan Coil Unit FOSS Free and Open Source Software GSHP Ground Source Heat Pump HEMS Home Energy Management System HCI Human-Computer Interaction HF Human Factors HLC High-Level Construals HVAC Heating, Ventilation, and Air Conditioning IDE Integrated Development Environment IoT Internet of Things LLC Low-Level Construals MECHanisms Make Energy Change Happen MURB Multi-Unit Residential Building NEP New Environmental Paradigm Survey POE Post Occupancy Evaluation TCT Temporal Construal Theory List of Abbreviations ANCOVA Analysis of Covariance API Application Programming Interface CBSM Community-Based Social Marketing CMS Content Management System COP Coefficient of Performance CT Current Transformer EE Energy Efficient FBRP Feedback Research Platform FCU Fan Coil Unit FOSS Free and Open Source Software GSHP Ground Source Heat Pump HEMS Home Energy Management System HCI Human-Computer Interaction HF Human Factors HLC High-Level Construals HVAC Heating, Ventilation, and Air Conditioning IDE Integrated Development Environment IoT Internet of Things LLC Low-Level Construals MECHanisms Make Energy Change Happen MURB Multi-Unit Residential Building NEP New Environmental Paradigm Survey POE Post Occupancy Evaluation TCT Temporal Construal Theory ix ix 1. Introduction 1. Introduction 1. Introduction Research in energy conservation behaviours for building inhabitants burgeoned during the 1970’s energy crisis to reduce dependence on foreign oil. As climate change has emerged on the political agenda in recent years, energy conservation has also regained traction; and is now acknowledged as perhaps the most cost-effective way of reducing greenhouse gas emissions (IEA, 2010). In the field of residential energy conservation, providing tenants with feedback on their energy use has been demonstrated as an effective intervention with savings ranging from 4-12% (Ehrhardt-Martinez, Donnelly, & Laitner, 2010). When considering that Canadian residential sector consumes 410 TWh of energy per year (Government of Canada, 2012), a 4-12% savings amounts to approximately 16-49 TWh. In Toronto, Canada, the current flat-rate, post-tax, marginal price for delivered electricity is approximately 0.14 $/kWh. This means with feedback there is a potential to save residential consumers $2.3B to $7.0B. In addition to the political and social influences on energy conservation research, technological advances have also enabled new ways to promote conservation, with feedback as a key strategy. In the past several years, on the strength of the smart grid technology and advanced metering infrastructure, industry has produced many feedback instruments on the market. These have ranged from smart bills, in-home displays, to web-based dashboards. Figure 1-1 shows an example in each of these categories. Figure 1-1: Examples of different feedback approaches (From left to right). OPOWER paper bill shows neighborhood comparisons. Aztech In-Home Display shows aggregate home energy consumption. LucidDesign Building Dashboard is a web-based portal offering interactive views of energy use. Figure 1-1: Examples of different feedback approaches t). OPOWER paper bill shows neighborhood comparisons. Aztech In-Home Display shows aggregate home mption. LucidDesign Building Dashboard is a web-based portal offering interactive views of energy use. 1 1 However, despite the many commercial implementations of feedback and a plethora of studies on the efficacy of feedback approaches, researchers (Ehrhardt-Martinez et al., 2010; Fischer, 2008; Flemming, Hilliard, & Jamieson, 2008) have pointed to two key challenges that have limited our understanding of how best to design feedback. First, methodological problems have hindered consolidation of the literature. In her review of 26 original feedback projects Fischer (2008, p. 87) writes: “[Feedback projects] differ markedly with respect to study design, sample, and method of data gathering, differences occurring both in substance and in scientific elaborateness. 1. Introduction What is more, results are not always reported quantitatively or in sufficient detail to make a comparison. And if they are reported, studies use very diverse reporting schemes. They vary in baseline, in time and duration of measurement, and in the unit for which savings are reported.” Second, there is no consensus on how best to visually design feedback. Feedback designs range from traditional quantitative representations (e.g., charts and graphs) to artistic, data-driven renderings of energy. Furthermore, design decisions on graph choice, measurement units, or wording may also impact user satisfaction, and overall adoption of the feedback. Unfortunately, very few studies have focussed on evaluating such design decisions. While it could be argued that this second challenge is being addressed through the success or failure of commercial products, market capitalization can often mask the underlying reasons for these outcomes – with the design of a product being only one such reason. Business models, marketing campaigns, regulatory landscape, and strategic partnerships are often major factors that can impact the wide adoption of a company’s product. For example, while OPOWER’s paper bills commercial success can be attributable to their science-driven design, it has also been written how they have been able to navigate the regulatory landscape in the US requiring utilities to cost-effectively induce energy conservation (e.g., St. John (2014), Tweed(2015)) A third challenge, identified by Ehrhardt-Martinez (2012), reflects the lack of details known about behaviours induced by feedback. As will be detailed later, a popular dichotomization of behaviours distinguishes between efficiency and curtailment behaviours. However, this dichotomy does not describe the variety of ways in which technology can be used, maintained, or interchanged (ibid). Without a clear understanding of how feedback can be designed to shape behaviour, it is likely difficult 2 to make optimal design choices. As utilized in Ehrhardt-Martinez’s study, surveys are an effective method to understand these nuanced behaviour changes. Any one of these challenges described above is worthy of further exploration. Interestingly, a common theme in all three is that they fundamentally point to methodological limitations of past feedback research. What appears problematic is that these limitations have also made it difficult for feedback researchers to advance the state of the art and science on feedback design and on maximizing the potential of feedback strategies. 1. Introduction While there is no straightforward solution to any of these challenges, it is clear that a common platform on which feedback research was conducted could help advance the field more rapidly and with coherence. The goal of the feedback research platform would be to afford a systematic approach to evaluating feedback designs given the wide variety of contexts possible. Having the platform freely available should also help it grow a community of researchers to further support this agenda. The development of such a feedback research platform is the purpose of this thesis. While on the surface this may appear overly ambitious (and perhaps it is!) and the platform unwise to freely share (and perhaps that is the case too!) there are two key socio-technical developments that have helped enable and inform this objective: The Internet of Things (IoT), and free and open source software (FOSS). The IoT refers to the interconnection of electronic devices (e.g., energy or indoor environment sensors) through the Internet. The movement towards an IoT has increased access to sensors and wireless communication technologies enabling the cost-effective collection of real time and disaggregated energy data amongst other applications. In conjunction with web-enabled mobile devices (i.e., smart phones, tablets), energy feedback can be as easily delivered as a consumer phone app is to download. FOSS (e.g., Linux, GIMP) is software available to use, copy, study, and modify for free. This is in contrast to proprietary software (e.g., Windows, Photoshop), which is restricted under copyright and has source code hidden from users. FOSS communities have developed with the belief that their approach fosters learning, collaboration, community, and innovation. As will be discussed in Chapter 3, the feedback research platform leverages several FOSS projects with an IoT focus. So it is with the same spirit, that the feedback research platform will be shared back to the FOSS community and to the energy feedback research community. In addition to helping address the three challenges discussed above, the FOSS-IoT-based platform developed in this thesis will be tailored to provide near real-time social comparisons. Real-time social 3 comparisons, to the author’s knowledge, has not been evaluated in conjunction with feedback. As will be described later, this focus was enabled due to the homogenous nature of the suites at the target building. 2. Literature Review The purpose of this chapter is to review the literature surrounding energy feedback to provide a better understanding of its history, target behaviour changes, the different characteristics of feedback, as well as how it can be supplemented by or used to supplement other motivators for conservation. This bulk of literature review draws from several key reviews on feedback (Abrahamse, Steg, Vlek, & Rothengatter, 2005; Darby, 2006; Ehrhardt-Martinez et al., 2010; Fischer, 2008) amongst others. A brief overview of the various topics and a summary of key takeaways are discussed in the forthcoming sub-chapters. 2.1 Energy Behaviours 1. Introduction For the purposes of a larger project initiative the feedback research platform was customized to measure the impact of thermal comfort and relevant feedback on energy use. Thermal comfort is an important topic because it is both visceral and energy intensive; heating and cooling related energy use accounts for 65% of energy use in Canadian homes (Government of Canada, 2012). To accommodate this objective, the platform incorporated in-situ surveys to help understand thermal comfort related behaviours as well as a complementary dashboard on fan coil unit (FCU) usage as the FCU is the primary way participants maintain comfort in their suites. Chapters 4 and 5 include descriptions of the design aspects of the platform related to thermal comfort as was implemented in the field study; however, the analysis of thermal comfort data was outside this thesis’s scope. The rest of this thesis is outlined as follows. Chapter 2 summarizes a literature review of feedback research to help provide guidance on the requirements and also inform best practices for feedback design and delivery. Chapter 3, based on the identified requirements, describes an architecture and design for the platform. This is followed by a matching of open source projects. Chapter 4 then describes an implementation of the platform as part of an energy conservation program in a multi-unit residential building (MURB). Chapter 5 details the field study methodology used in the study. Chapter 6 reports the results of the field study. Finally, Chapter 7 reviews the contributions from this work and outlines a way forward. 4 2. Literature Review 2. Literature Review 2.1 Energy Behaviours This chapter reviews the predominant views on behaviour including taxonomies, challenges to eliciting behaviour changes, and models that can guide feedback design. 2.1.1 Behaviour Taxonomies To encourage residential energy conservation, it is useful to first identify and understand the different types of conservation behaviours. Gardner and Stern (1996) divide these into two categories: efficiency and curtailment behaviours. Efficiency behaviours are one-time behaviours associated with initial capital investment in energy saving technologies such as home insulation or energy efficient appliances. Curtailment behaviours involve repeated or frequent efforts to reduce energy consumption, such as turning off lights or unplugging appliances when not in use. Curtailment behaviours involve repeated or frequent efforts to reduce energy consumption, such as turning off lights or unplugging appliances when not in use. While the bulk of past studies target curtailment behaviours it is worth noting that efficiency behaviours are considered to have greater energy-savings potential (Ehrhardt-Martinez et al., 2010). For example, installing compact fluorescent light bulbs will likely save more electricity than promptly turning incandescent bulbs off when they are not in use. However, efficiency behaviours do not necessarily result in net energy savings if, for example, energy efficient appliances are used more frequently than less-efficient models. This is an example of the rebound effect (Moezzi & Diamond, 2005): the often counter-productive behavioural response to the introduction of new, more efficient technologies. 5 Ehrhardt-Martinez et al. (2010) provide a more detailed breakdown by frequency and cost as shown in Table 2-1. They define habitual behaviours as frequent and low-cost. Energy stock-taking behaviours are infrequent but are still low-cost. Finally, consumer behaviours such as upgrading windows and appliances happen infrequently but are higher in cost. Table 2-1. Energy Behaviors as a Function of Frequency and Cost. Adapted from (Ehrhardt-Martinez et al., 2010) Infrequent Actions Frequent Actions Low-Cost / No Cost Energy Stocktaking Behavior Install CFLs, Pull fridge away from wall, Inflate tires adequately, Install weather stripping Habitual Behaviors and Lifestyles Slower highway driving Slower acceleration Air dry laundry Turn off devices Higher Cost / Investment Consumer Behavior New energy efficient (EE) windows New EE appliances Additional insulation New EE AC or furnace Table 2-1. Energy Behaviors as a Function of Frequency and Cost. 2.1 Energy Behaviours This makes it difficult to consciously save energy. The operating hypothesis of feedback is that it helps to make energy use visible. However, it does not help that feedback on energy use has been typically infrequent, delayed from the time of consumption, and only reaches those who pay the energy bills. 2. Energy is cheap. In Toronto, Canada, the current flat rate, marginal price for delivered electricity is about $0.14/kWh. This puts the cost of watching an hour of television on a modern 42” LCD HDTV at about three cents. It can be argued that unless prices rise substantially, it will be tough to motivate people to conserve. 2. Energy is cheap. In Toronto, Canada, the current flat rate, marginal price for delivered electricity is about $0.14/kWh. This puts the cost of watching an hour of television on a modern 42” LCD HDTV at about three cents. It can be argued that unless prices rise substantially, it will be tough to motivate people to conserve. 3. Split incentives do not foster conservation behaviours. In rental housing situations, there is no direct financial motivation for tenants to conserve since they pay only flat rental rate regardless of their energy usage. Landlords may be reluctant to invest in energy efficient appliances as those have higher upfront costs. In many cases if there is nothing broken, it will not be fixed or upgraded. 4. The invisibility of energy leads to poor mental models and habits. The invisibility of energy use can lead to poor energy use habits (Verplanken & Wood, 2006) because the environment was not at the forethought at the time those habits were developed. This is especially problematic because most residential energy is consumed through routine and habitual behaviour (Lutzenhiser, 1993; Sauer, Wiese, & Ruettinger, 2003). 4. The invisibility of energy leads to poor mental models and habits. The invisibility of energy use can lead to poor energy use habits (Verplanken & Wood, 2006) because the environment was not at the forethought at the time those habits were developed. This is especially problematic because most residential energy is consumed through routine and habitual behaviour (Lutzenhiser, 1993; Sauer, Wiese, & Ruettinger, 2003). 4. The invisibility of energy leads to poor mental models and habits. 2.1 Energy Behaviours Adapted from (Ehrhardt-Martinez et al., 2010) Infrequent Actions Frequent Actions Low-Cost / No Cost Energy Stocktaking Behavior Install CFLs, Pull fridge away from wall, Inflate tires adequately, Install weather stripping Habitual Behaviors and Lifestyles Slower highway driving Slower acceleration Air dry laundry Turn off devices Higher Cost / Investment Consumer Behavior New energy efficient (EE) windows New EE appliances Additional insulation New EE AC or furnace Higher Cost / Investment In a more recent study, Ehrhardt-Martinez (2012) argues for a need for further classification based on the multitude of ways in which technology can be used, maintained, or interchanged. In so doing, this classification provides a more nuanced perspective by which householders can achieve feedback- induced energy savings. The nine classes of actions include: The nine classes of actions include: 1. alternative technology c 2. conservation behaviour 3. conservation settings 4. enhanced control 5. investment 6. low cost investment 7. turn off 8. unplug 1. alternative technology choices 2. conservation behaviour 3. conservation settings 4. enhanced control 6. low cost investment 6 1.2 Behaviour Change Challenges 2.1.2 While having a grasp of behaviours that feedback can influence is essential, it is also necessary to review why behaviour change can be difficult. With this context, designers can begin thinking about how such obstacles may be overcome when designing feedback. In this chapter, five key behavioural challenges are reviewed. While having a grasp of behaviours that feedback can influence is essential, it is also necessary to review why behaviour change can be difficult. With this context, designers can begin thinking about how such obstacles may be overcome when designing feedback. In this chapter, five key behavioural challenges are reviewed. 1. Energy is invisible. Energy use is embedded in our buildings, our food and our transportation systems. Yet it is largely invisible; most people do not think or talk about the energy they use. This makes it difficult to consciously save energy. The operating hypothesis of feedback is that it helps to make energy use visible. However, it does not help that feedback on energy use has been typically infrequent, delayed from the time of consumption, and only reaches those who pay the energy bills. 1. Energy is invisible. Energy use is embedded in our buildings, our food and our transportation systems. Yet it is largely invisible; most people do not think or talk about the energy they use. 2.1 Energy Behaviours The invisibility of energy use can lead to poor energy use habits (Verplanken & Wood, 2006) because the environment was not at the forethought at the time those habits were developed. This is especially problematic because most residential energy is consumed through routine and habitual behaviour (Lutzenhiser, 1993; Sauer, Wiese, & Ruettinger, 2003). Ordinary people may also develop faulty mental models of how energy is consumed; relying instead on folk theories that often lead to sub-optimal energy use (Karjalainen & Vastamaeki, 2007; Kempton, 1986). Kempton (1986) found that between 25-50% of Americans believe that a thermostat works like a valve, in that a higher temperature setting will deliver heat at a faster rate than a lower setting. However, many conventional residential heating and air conditioning systems produce or remove heat at a constant rate, and can only be turned on or off by the thermostat. 7 7 5. Visceral Influences compete against conservation goals. Even if residents can be made conscious of their energy consumption, visceral influences can compete with conservation behaviours at the point of consumption (Trinh & Jamieson, 2014). Visceral influences (Loewenstein, 1996) such as inconvenience, fatigue, or physical discomfort focus attention on the immediate and direct hedonic impact of behaviours rather than long term objectives. At sufficient intensity, visceral influences can cause people to act contrary to their pro- environmental attitudes in favour of impulsive behaviours. This effect is compounded as many energy consuming technologies are intentionally designed to be viscerally attractive to use (Norman, 2004). 2.1.3 Behavioural Models 2.1.3 Behavioural Models If designers are to address the behavioural challenges identified above, it is helpful to have a theoretical basis on how decisions and behaviours are formed. Toward this end, environmental psychology researchers have developed models to understand how environmentally relevant behaviour change can be obtained. One such heuristic model, as shown in Figure 2-1, is discussed by Fischer (2008). Figure 2-1: Heuristic model of environmentally relevant behavior Adapted from Fischer (2008) Figure 2-1: Heuristic model of environmentally relevant behavior Adapted from Fischer (2008) This heuristic model distinguishes between habits and conscious decisions and recognizes that habits, while not reflected upon consciously, influence the decision making process. What is particularly useful 8 8 in this model, is the acknowledgement that for new norms to be activated, they must be consciously reflected on. This reflection process has three parts. First, the person must realize there is a problem. Second, the person must realize that his/her behaviour is relevant to the problem. Third, the person must have a sense of control, acknowledging the possibility to have influence. Once this norm activation process is completed, the person enters an evaluation process where he/she must weigh various motives that may be in conflict with one another. Such motives may include personal norms, social norms, or other motives such as comfort or convenience. What is not explicit in the model is that considerable amount of information is necessary to perform the decision process. However, this is why feedback can have such an influential role. Feedback can interject the process with information to raise awareness to inform new norms and break old habits. Some caution should be identified with this model, though, as it assumes a rational decision-making process. From the field of behavioural economics, we learn that decisions are not always consistent as the rational model would have us believe. Rather, cognitive biases and decision-making heuristics are known to lead to sub-rational decisions. As Trinh and Jamieson (2014) identify, the outcomes of the decision- making process vary, also in part, as a function of temporal distance. For example, one might make deliberate plans to take the stairs for health or environmental reasons; but, at the moment of decision for convenience, succumb to using the elevator. Trinh (2010) adopted temporal construal theory (TCT) (Liberman & Trope, 1998) to help characterize the effect of visceral influences. 2.1.3 Behavioural Models TCT describes how temporal distance systematically changes people’s mental representations (i.e., “construals”) and associated valuations of future events. TCT posits that an increased temporal separation from an event or activity shifts preferences to more abstract goals. Conversely, more temporally immediate events are associated with contextualized features that are more concrete. These features are examples of high level construals (HLCs) and low level construals (LLCs), respectively (see Figure 2-2). HLCs are relatively simple, decontextualized representations that consist of general, superordinate (i.e., goal relevant, “why” features), and essential features of events (Trope & Liberman, 2003). By contrast, LLCs are akin to visceral influences in that they are more concrete and include subordinate, contextual, and incidental features of events. For example, composting may bring about HLCs such as environmental preservation or financial benefits (reasons why Trinh (2010) adopted temporal construal theory (TCT) (Liberman & Trope, 1998) to help characterize the effect of visceral influences. TCT describes how temporal distance systematically changes people’s mental representations (i.e., “construals”) and associated valuations of future events. TCT posits that an increased temporal separation from an event or activity shifts preferences to more abstract goals. Conversely, more temporally immediate events are associated with contextualized features that are more concrete. These features are examples of high level construals (HLCs) and low level construals (LLCs), respectively (see Figure 2-2). HLCs are relatively simple, decontextualized representations that consist of general, superordinate (i.e., goal relevant, “why” features), and essential features of events (Trope & Liberman, 2003). By contrast, LLCs are akin to visceral influences in that they are more concrete and include subordinate, contextual, and incidental features of events. For example, composting may bring about HLCs such as environmental preservation or financial benefits (reasons why 9 one would want to compost) but may also evoke LLCs such as negative thoughts of dirt and odors (contextual factors associated with the act of composting). Table 2-2. Distinguishing Low-Level and High-Level Construals. Adapted from Trope and Liberman (2003) Low-level Construals (LLCs) High-level Construals (HLCs) Concrete Complex Unstructured, Incoherent Contextualized Secondary, Surface Subordinate (“how”) Goal Irrelevant Abstract Simple Structure, Coherent Decontextualized Primary, Core Superordinate (“why”) Goal Relevant Figure 2-2 shows a highly simplified time-construal function, depicting the conflicting impacts of HLCs and LLCs on decisions over time. In the near future, LLCs spurred by visceral influences have more impact on decisions than HLCs. 2.1.3 Behavioural Models In the distant future, HLCs representing one’s attitudes towards conservation have more influence than LLCs. For stubborn or habitual consumption behaviours Trinh and Jamieson (2014) posit that the default time perspective is typically near term, as represented in Figure 2-2 by the dotted line in the near future. Figure 2-2 shows a highly simplified time-construal function, depicting the conflicting impacts of HLCs and LLCs on decisions over time. In the near future, LLCs spurred by visceral influences have more impact on decisions than HLCs. In the distant future, HLCs representing one’s attitudes towards conservation have more influence than LLCs. For stubborn or habitual consumption behaviours Trinh and Jamieson (2014) posit that the default time perspective is typically near term, as represented in Figure 2-2 by the dotted line in the near future. Figure 2-2: Simplified default construal-time relationship based on temporal construal theory Adapted from Trinh and Jamieson (2014) Time from Decision Influence on Decision (near future) (distant future) Time perspective held when making decisions for near-future behaviors Goal-conflicting LLCs influence decisions of near- future behaviors more than HLCs Goal-aligned HLCs influence decisions of distant-future behaviors more than LLCs Time perspective held when making decisions for near-future behaviors Figure 2-2: Simplified default construal-time relationship based on temporal construal theory Adapted from Trinh and Jamieson (2014) The visual representation of TCT in Figure 2-2 leads to some distinct insights. First, it suggests an explanation for a related finding in energy conservation research; that attitudes do not necessarily predict behaviours (Gatersleben, Steg, & Vlek, 2002; McKenzie-Mohr, 2011). Second, the representation 10 10 forms the basis of a conceptual framework to help characterize four strategies for behavioural interventions in energy conservation (Trinh, 2010) When considering both Fischer’s and Trinh’s models together, one can begin envisioning how feedback might operate. From Fischer’s model she points that feedback can direct attention towards a problem and increase the consciousness of the relevance of one’s behaviour. It may also motivate, for example, a sense of competition or incent behaviours through the use of comparison or goal settings, respectively. Trinh’s model points to the importance of frequent feedback. It also points to how feedback can be framed to target conservation motives. The following chapter was dedicated to providing a more full characterization of energy feedback. 2.1.3 Behavioural Models In the meanwhile and in agreement with Fischer (2008), from these considerations, one can deduce the hypotheses that feedback is most effective if:  It successfully captures the user’s attention  Draws a close link between specific behaviours and their effects  Activates various HLCs or motives that may appeal to different user groups, such as cost savings, resource conservation, emissions reduction, competition and others.  Activates various HLCs or motives that may appeal to different user groups, such as cost savings, resource conservation, emissions reduction, competition and others. 2.2 Energy Feedback Chapter 2.1.3, through a review of decision-making models, identified the potential for how conservation behaviour challenges can be addressed with feedback. This chapter deals with how feedback works, its different dimensions, and how it has been used in context. 2.2.1 How feedback works 2.2.1 How feedback works As mentioned earlier energy feedback has been demonstrated as an effective intervention with savings ranging from 4-12% (Ehrhardt-Martinez et al., 2010). To justify the approach and explain findings, much of the early feedback research sought to describe the psychological mechanisms that feedback As mentioned earlier energy feedback has been demonstrated as an effective intervention with savings ranging from 4-12% (Ehrhardt-Martinez et al., 2010). To justify the approach and explain findings, much of the early feedback research sought to describe the psychological mechanisms that feedback supported. The most predominant of these is that feedback facilitates learning by making the invisible visible (Abrahamse et al., 2005; Benders, Kok, Moll, Wiersma, & Noorman, 2006; Darby, 2006; Holmes, 2007; Katzev & Johnson, 1987). While some aspects of energy use in the home are highly visible, other aspects are largely hidden from view. For example, the energy consumed from a television set or from room lighting are much more salient than energy lost due to poor insulation in the attic or from water heating. By making such consumption visible, it is argued that one is then able to understand and address the challenge of conservation (Trinh, 2010). 11 A subset of the studies following the learning perspective aimed to specifically improve conservation competence. In one study looking at the control of a simulated central heating system, participants were asked to maintain thermal comfort in the home while minimizing energy waste using a feedback system (Sauer, Schmeink, & Wastell, 2007). Results showed participants improved conservation competence when provided with additional energy use information. Van Raaij and Verhallen (1983) extend the learning perspective by suggesting that feedback works through a three-step process: learning, habit formation, and internalization of behaviour. In the learning phase, households observe or become aware of the specifics of their consumption patterns and learn about how their specific actions affect their consumption levels. They respond by making small changes in their behaviour, initially to view the effects on the feedback they received and over time as a way to maintain a lower consumption level. These changes that persist become habit that may work even with the withdrawal of feedback. The third phase is the internalization of behaviour. As energy-conserving behaviour becomes habit, an individual’s attitude will also change to reflect the adjustment in behaviour. Ehrhardt-Martinez, Donnelly and Laitner (2010) found evidence showing the effect of feedback to be persistent. 2.2.1 How feedback works Taking a different approach, Seligman et al. (1981) argue that feedback works by providing goal-relevant information. Given the important precondition that one is motivated to conserve, increased effort can be triggered by showing when actual conservation is below the level the person wants to achieve. They identify that setting a performance goal and providing feedback relevant to that goal are basic elements in self-control. From the explanations above, the central rationale of feedback is that people are hampered by an information deficient world. Furthermore, if this information became available in a timely and comprehensible way then motivated individuals would be enabled to make more competent decisions. 2.2.2 The different dimensions of energy feedback 2.2.2 The different dimensions of energy feedback Darby (2001) distinguishes between direct and indirect feedback. Direct feedback refers to feedback that is available on demand in the form of a real-time meter or electronic display. Indirect feedback, by contrast, is typically processed by the electrical utility and sent out in the form of a bill. Currently, direct feedback strategies are receiving increased attention as a result of continuing proliferation of new information and communications technologies that facilitate the effective delivery of feedback. Over the 12 past decade, researchers have been exploring feedback delivered over the internet via personal computers (Benders et al., 2006; Petersen, Shunturov, Janda, Platt, & Weinberger, 2007) and on mobile devices (e.g.,Froehlich et al., 2009). Based on these recent advances in feedback delivery mechanisms, the Electric Power Research Institute (EPRI) expanded Darby’s feedback spectrum (see Figure 2-3) to offer greater resolution of the type and frequency of information provided (Ehrhardt-Martinez et al., 2010). The trade-off with the additional information availability and resolution offered by direct feedback strategies is the associated costs of implementation and maintenance of the feedback systems. Figure 2-3: Household savings broken down by feedback type Adapted from Ehrhardt-Martinez et al. (2010) Figure 2-3: Household savings broken down by feedback type Adapted from Ehrhardt-Martinez et al. (2010) In Trinh’s (2010) assessment, indirect feedback may be more appropriate for efficiency than curtailment behaviours because they are not time sensitive and implementation costs are also kept minimal. On the other hand, direct feedback may be more appropriate for curtailment than efficiency behaviours because information needs to be provided more frequently to support learning and performance tracking specific energy-related activities. As shown in Figure 2-3, Real-Time Plus Feedback has been 13 shown to promote the most savings of feedback types. Chapter 3 details the development of such a platform. By aiming for this level of feedback, other forms can be easily derived through subtraction of features (e.g., lowering the frequency, or making print-out “bills” rather than showing feedback online. Feedback strategies seldom work alone in a conservation program. That is because without goals, baselines for comparison, or clear objectives for example, feedback is just information. Consideration must be given to the sociotechnical context for which conservation programs with feedback are designed to help ensure optimal program implementation. One line of thought is that for behaviours to become habitual, they require community support and reinforcement. 2.2.2 The different dimensions of energy feedback To guide the development of conservation programs there are two prominent behaviour change frameworks: Community-based Social Marketing (CBSM) and the Make Energy Change Happen (MECHanisms) Toolkit. This chapter reviews each in turn while comparing their broad similarities and differences. It then dives into engagement strategies that are often used to complement an effective feedback approach. CBSM (McKenzie-Mohr, 2011) is a sustainability program design process that draws from research in social psychology, which indicates that behaviour change is most effective when they are implemented interactively at the community level. It leverages from the observation that raising awareness of sustainability issues is alone insufficient to evoke behaviour changes. In addition, CBSM emphasizes direct and personal contact with community members using many of the occupant engagement strategies reviewed later in this chapter. McKenzie-Mohr (2011) reviews these strategies with examples in agriculture & conservation, energy, transportation, waste & pollution, and water. CBSM also pragmatically emphasizes the identification and removal of barriers that may impede change. It is only by understanding these barriers, that a program designer can effectively implement change. Barriers include: lack of motivation, social pressure, or knowledge; forgetfulness or structural barriers. To help address these barriers, CBSM offers guidance on pairing tools with barriers as summarized in Table 2-3. 14 Table 2-3. Selecting Tools Based on Barriers and Benefits. Adapted from McKenzie-Mohr (2011) BARRIERS TOOLS Lack of Motivation Commitment Norms Incentives Forget to Act Prompts Lack of Social Pressure Norms Lack of Knowledge Communication Social Diffusion Structural Barriers Convenience Table 2-3. Selecting Tools Based on Barriers and Benefits. Adapted from McKenzie-Mohr (2011) BARRIERS TOOLS Lack of Motivation Commitment Norms Incentives Forget to Act Prompts Lack of Social Pressure Norms Lack of Knowledge Communication Social Diffusion Structural Barriers Convenience To ensure program success, CBSM recommends small-scale piloting and refinement before broad mplementation to the target community. This follows from well-known cyclical design patterns for any uccessful product or program development. In summary, the CBSM process prescribes five steps: 1) Select behaviours Table 2-3. Selecting Tools Based on Barriers and Benefits. Adapted from McKenzie-Mohr (2011) BARRIERS TOOLS Lack of Motivation Commitment Norms Incentives Forget to Act Prompts Lack of Social Pressure Norms Lack of Knowledge Communication Social Diffusion Structural Barriers Convenience Table 2-3. Selecting Tools Based on Barriers and Benefits. Adapted from McKenzie-Mohr (2011) TOOLS Table 2-3. Selecting Tools Based on Barriers and Benefits. 2.2.2 The different dimensions of energy feedback Adapted from McKenzie-Mohr (2011) TOOLS To ensure program success, CBSM recommends small-scale piloting and refinement before broad implementation to the target community. This follows from well-known cyclical design patterns for any successful product or program development. In summary, the CBSM process prescribes five steps: 1) Select behaviours 2) Identify barriers and benefits 3) Develop strategies 4) Pilot 5) Broad-scale implementation 2.2.3.2 Make Energy Change Happen (MECHanisms) Toolkit The MECHanisms Toolkit (Changing Behaviour, 2013) is designed for project managers who are looking to promote energy conservation with small energy end-users such as households, housing managers, small businesses and local communities. It is based on both practice and research and was established by the European-led Changing Behaviour project. Its objective is to support the development of programs for enduring energy conservation. Similar to CBSM, MECHanisms focuses on change at the community level rather than with specific individuals, acknowledging the importance of the socio-technical context. As such, it also encourages an interactive program design approach working with the target group. It offers guidance to develop a deep understanding of target groups and their socio-technical context. Within the MECHanisms Toolkit are detailed descriptions for individual tools (e.g., competitions, using fun activities, feedback, etc.) intended 15 for the practitioner. Furthermore, guidance is provided for its best use along with caveats of which to be mindful. In addition, detailed, printable checklists are provided to facilitate the implementation of the tools. Compared to CBSM, MECHanisms emphasises a broader approach to change and energy conservation and this difference is highlighted in the breakdown of Steps in Stage A (see below). Inherently, the MECHanisms process appears to cater more to project managers who are open to either a top-down (driven by the program’s agenda) vs a bottom-up (driven by community needs through participation in design) approach. Whereas CBSM presumes a target community in mind and targets specific behaviours, MECHanisms takes a more abstract initial stance to challenge the program designer to consider the higher project objectives. Doing so, it encourages thinking about the context, the timing, and relevant stakeholders, and possible barriers before specific conservation behaviours. Furthermore, it suggests using broad interventions such as energy audits and information campaigns and only after some initial testing, supporting them with engagement strategies, many of which are described later in this chapter. By contrast, in CBSM these engagement strategies are the core of the proposed intervention. Stage A: Understand 1) Pinpoint your problem 2) Get to know your target group 3) Understand your context 4) Determine if the time is right 5) Identify relevant stakeholders Stage B: Plan and Do 2.2.2 The different dimensions of energy feedback The MECHanisms process is executed by following these fourteen steps divided into three stages: The MECHanisms process is executed by following these fourteen steps divided into three stages: Stage A: Understand Stage A: Understand Stage B: Plan and Do Stage B: Plan and Do 6) Define goals 7) Plan with your target group 8) Select and adapt your instruments 10) Engage your target group 11) Motivate through feedback Stage C: Evaluate and Learn 16 14) Develop a learning culture 2.2.3.3 Information and Prompts Curtailment campaigns provide consumers with information in an attempt to change their attitudes or highlight economic benefits. McKenzie-Mohr (2011) posits that for a message to be effective and influential, it should (to name a few): capture the reader’s attention; be vivid and captivating; be tailored to the attitudes and beliefs of the intended audience, and their perceived barriers and benefits to taking action; cite a credible source; frame the message to highlight a potential loss; provide actionable solutions when highlighting something that may threaten the reader; keep instructions clear, specific, and easy to remember; and be combined with other approaches. While providing information may change attitudes, it does not necessarily change related behaviours (McKenzie-Mohr, 2011; Verplanken & Wood, 2006) or lower energy consumption (Abrahamse et al., 2005). For example, Sauer, Wiese, and Ruettinger (2003) found that knowledge of environmental impacts did not predict environmental performance in the use of consumer appliances. However, if the information is delivered at the point of consumption, it can prompt specific behaviours. The purpose of such prompts is to spur people to do something they are already predisposed to do but may have forgotten (McKenzie-Mohr, 2011). Effective prompts are noticeable, specific, and actionable. Prompts placed around taps and showers displaying the environmental impacts of water use decreased water consumption by 23% (Kurz, Donaghue, & Walker, 2005). Prompts placed over waste bins produced a 50% reduction in litter(Kort, McCalley, & Midden, 2008). 2.2.3.4 Goal Setting and Commitments Setting a performance goal and providing feedback relevant to that goal are basic elements in self- control (Seligman et al., 1981). However, goals should be achievable and challenging to have an impact on energy conservation (Becker, 1978). Goals are effective when they are clear, agreed upon, and measureable and when frequent feedback is available (Changing Behaviour, 2009). Commitment strategies can be used to promote a variety of sustainable behaviours. Becker (1978) showed that both feedback and goal setting were responsible for motivating individuals to reduce electricity use. In particular, the more difficult the goal, the more effort the individuals put into meeting that goal. In his study, Becker found that subjects who chose to reduce their energy consumption by 17 20% conserved significantly more than those who chose to reduce by only 2%, even if their goals were not reached. However, for commitments to be most effective, they should be made publicly and written rather than non-public and verbal (Shippee & Gregory, 1982). 2.2.3.5 Comparisons Comparisons can also motivate conservation behaviours. There are two types of comparisons: historic comparisons (e.g., with one’s past consumption); and normative comparisons (i.e., social comparisons; e.g., with one’s neighbor). Using energy consumption from a previous billing period is an effective historic comparison (S. Darby, 2006). However, weather and occupancy fluctuations may make this form of comparison less meaningful unless normalizing factors are modeled. Social comparisons often happen in the form of competitions. Competitions can be effective but it is unclear whether their effects persist once they end (Ehrhardt-Martinez et al., 2010). Social comparisons also suffer from the perception of unfair comparison groups (Darby, 2006). However using “injunctive” norms, which describe how one should behave, rather than “descriptive” norms, which describe how others have behaved, can prolong the effect of social comparisons (McKenzie-Mohr, 2011; Schultz, Nolan, Cialdini, Goldstein, & Griskevicius, 2007). Perhaps the champion of social comparisons has been Opower (https://opower.com/), who have developed a commercially viable business around their proprietary home energy reports. Their home energy reports (See Figure 1-1) combined with customer data mining has led to documented energy savings from 1.4-3.3% (Allcott, 2011) (on average 2%) and persistence (Allcott & Rogers, 2012). However, in agreement with Froehlich (2009), more research is needed to understand how social comparisons can be effectively integrated with feedback information. This is especially challenging in real-time applications where providing an individual’s feedback in near real-time has been costly (i.e., relative to monthly home energy reports). 2.2.3.6 Rewards and Incentives Incentives, rewards, and disincentives provide extrinsic motivation to perform existing, or learn new, behaviours that consumers would otherwise be indifferent or resistant to (Abrahamse et al., 2005; McKenzie-Mohr, 2011). Implemented correctly, incentives foster sustainable behaviours. For example, introducing bottle deposits in Oregon, Vermont, and Michigan, saw decreases in litter of 68%, 76%, and 82%, respectively (Syrek & Legislature, 1980). A program in California that charged residents for the amount of waste they put out on the curb, saw a 46% reduction in landfill-bound waste and a 158% increase in recycling (Federation of Canadian Municipalities, 1996). However, Abrahamse et al. (2005) 18 found that while rewards produce large effects, these effects quickly diminish once the reward is discontinued. found that while rewards produce large effects, these effects quickly diminish once the reward is discontinued. 2.2.4 Summary of literature review on good feedback 6) Make the feedback information task relevant (Sauer et al., 2007) or related to behaviour in an intelligible way (Winett, Neale, & Grier, 1979). When possible, feedback should be related to specific behaviours of interest. 6) Make the feedback information task relevant (Sauer et al., 2007) or related to behaviour in an intelligible way (Winett, Neale, & Grier, 1979). When possible, feedback should be related to specific behaviours of interest. 7) Use concrete consequences by framing consumption data using tangible equivalents (Pierce, Odom, & Blevis, 2008). It is useful to explain measurements in alternative equivalents to which users can relate. For example, while energy is reported in kilowatt-hours (kWh), an average homeowner is more likely to understand that amount in terms of light-bulb equivalents. Trees are recommended as equivalent units of carbon offsets (Katzev & Johnson, 1987) since people have positive feelings towards trees and they are also public symbols of carbon sinks. Another alternative is to allow users to select their own frames as Schott et al. (2012) proposed with options ranging from the empathic to data driven. While the above two lists represent the best practices from literature, there is no guarantee that adhering to these will be best for any given community of users. The review of CBSM and MECHanisms, indicated that an interactive approach can help feedback designers identify barriers and develop sound strategies (even if not with feedback) to address them. In fields such as Human Computer Interaction (HCI) or Human Factors (HF), this approach is analogous to a user-centered design (UCD). A UCD approach recommends an iterative design philosophy beginning with perhaps surveys or interviews to understand the context. This is often followed by design iterations with prototypes of increasing fidelity. An example of this approach is Stragier et al’s (2013) work developing a Home Energy Management System (HEMS) by involving input from end users in the design phase. 2.2.4 Summary of literature review on good feedback 2.2.4 Summary of literature review on good feedback 2.2.4 Summary of literature review on good feedback Fischer (2008) concludes in her review that successful feedback that stimulates conservation and is satisfying to users are likely: Fischer (2008) concludes in her review that successful feedback that stimulates conservation and is satisfying to users are likely: 1) Based on actual consumption 2) Given frequently (ideally, daily or more) 3) Involves interaction and choice for households 4) Involves appliance-specific breakdown 5) Is given over a longer period 6) May involve historical or normative comparisons 1) Based on actual consumption 2) Given frequently (ideally, daily or more) 3) Involves interaction and choice for households 4) Involves appliance-specific breakdown 5) Is given over a longer period 6) May involve historical or normative comparisons 7) Is presented in an understandable and appealing way. Trinh (2010) summarizes feedback visual design heuristics and best practices to encourage conservation behaviours. They are listed as follows. 1) Make visible important but normally imperceptible information (the basic premise of providing feedback). 2) Design the message carefully to filter out unimportant information (Gardner & Stern, 2002). This is related to the data-ink ratio concept by Tufte (1983) who proposed that a high proportion of a graphic’s ink should be devoted to the non-redundant display of data information. 3) Consider the audience; be specific and personalized (Benders et al., 2006; Brandon & Lewis, 1999; Gardner & Stern, 2002). The information needs to be tailored to the environment for which it is intended, the task that it supports, and the users who will need to act on the information. 4) Benchmark in a meaningful and fair way (Abrahamse et al., 2005; Egan, 1999; Seligman et al., 1981). If feedback is to be comparative, comparisons should be perceived as equitable. Comparisons of consumption of one house to the average house in one’s neighborhood may seem unfair if the home has more occupants than the average. Similarly, comparisons of total home energy consumption by month may not show actual conservation improvements if warmer weather required more air conditioning usage. 19 5) Average feedback over meaningful intervals (Seligman et al., 1981). There is little purpose, even if possible, to report energy consumption by the second if target activities take place at a slower time scale. On the other hand, monthly averages may not be specific enough to promote learning. 3. Feedback Research Platform Design 3. Feedback Research Platform Design This chapter describe the design and integration of a near real-time feedback platform for research applications. This platform is informed by the best practices identified in Chapter 2, and is motivated by challenges that have plagued the feedback design community as identified in Chapter 1. This chapter may be especially useful for research project managers and designers. Where do we go from here? 2.3 This literature review has shown the various dimensions of feedback design and the variety of feedback interventions programs. Given this space, it is not surprising that there have been a breadth of studies in this field. However, as introduced earlier, the advancement has been hindered without a standard way of designing and delivering feedback. Thus, the following seven functional requirements were identified for the feedback research platform. 20 1. It should allow for the implementation of feedback on a multitude of design dimensions such as visual design, frequency, and delivery format. 1. It should allow for the implementation of feedback on a multitude of design dimensions such as visual design, frequency, and delivery format. 2. It should allow for aggregated and disaggregated feedback data. 3. It should allow for historical and social comparisons to be integrated with feedback. 4. It should support researchers by not only delivering feedback but also standardizing how data is collected and managed. 5. The data collected should not be limited to simply energy measurements, but should be widened to include survey data, thermal comfort data and data related to energy use. 6. It should allow for data to be collected with a common structure and data format, to afford cross-experiment data analysis. 7. Finally, because the platform is built on open-source technology it should be freely available for others to use and customize. In Chapter 3, this thesis explores the detailed design of the feedback research platform. 21 3.1 Platform Technical Requirements At a minimum, the platform should integrate sensors for power, temperature, and humidity measurement. Power measurements should be able to be captured using both clamp-on style (non-invasive) current transducers (CTs) or from electrical plugs. CT sensors must be able to sample every 10 seconds to support precise power measurements of high wattage appliances with short duty cycles (e.g. electric kettles, microwaves). Temperature and humidity sensors must be able to sample every minute to capture changes in the ambient space due to power and FCU use. The sensors should have the option to be battery-powered, support a low-power wireless transmission (e.g., Zigbee, Z-wave), and have at least a year of data sensing and transmission function if battery powered. Each sensor must be uniquely identifiable. 2. Fieldable sensors. At a minimum, the platform should integrate sensors for power, temperature, and humidity measurement. Power measurements should be able to be captured using both clamp-on style (non-invasive) current transducers (CTs) or from electrical plugs. CT sensors must be able to sample every 10 seconds to support precise power measurements of high wattage appliances with short duty cycles (e.g. electric kettles, microwaves). Temperature and humidity sensors must be able to sample every minute to capture changes in the ambient space due to power and FCU use. The sensors should have the option to be battery-powered, support a low-power wireless transmission (e.g., Zigbee, Z-wave), and have at least a year of data sensing and transmission function if battery powered. Each sensor must be uniquely identifiable. 3. Data processing. The key differentiating requirement for the CMS is its ability to process time series data as this is what separates the requirements of this CMS from those that handle news feeds or blogs. The CMS should be able to perform basic data manipulations such as addition, subtraction multiplication, and division of data feeds. This should allow for averaging of data feeds, which is important for historical and social comparison applications. They should also support conversion of time stamped power measurements in watts (W) to energy units (e.g., kWh) and to kWh/day to afford visualizations on demand. 4. Efficient data storage and transfer. It is important for any software, especially one that is intended for real-time data monitoring, to have a smooth user interface and interaction. Ensuring that data is stored and transferred to and from the CMS is an important objective in this regard. 3.1 Platform Technical Requirements When reviewing the list of seven functional requirements for the platform identified in Chapter 2.3, it can be taken for granted that a web-based solution is critical for content management, experimental configuration, and robust deployability. The promise of a connected world is premised on the internet as a common communications platform. Web-based technologies have been in rapid development since the early 90’s. Today, many of the world’s key communications services (e.g., email, telephone, video conferencing, news, television) are delivered over the internet. Online data storage and web-hosting for content management is also become more cost-efficient and reliable. Moving towards a web-based solution for content management is also practical from an administration standpoint as it allows remote and shared access amongst team members. A web-based platform also allows deployment over a range of internet-connected devices such as smart phones, tablet, laptops, or desktop computers. In addition to being web-based, the author determined 12 technical requirements that guided the design of the platform and selection of components to meet the platform’s functional requirements: 1. Manages content. A content management system (CMS) is a necessary and arguably the core component for the platform. According to Wikipedia, a CMS is a computer application that allows publishing, editing and modifying content, organizing, deleting as well as maintenance from a central interface. Such systems of content management provide procedures to manage workflow in a collaborative environment. In this context, the content is referring to the energy and energy-related data and all its attributes that will be collected from sensors in the field. As 1. Manages content. A content management system (CMS) is a necessary and arguably the core component for the platform. According to Wikipedia, a CMS is a computer application that allows publishing, editing and modifying content, organizing, deleting as well as maintenance from a central interface. Such systems of content management provide procedures to manage workflow in a collaborative environment. In this context, the content is referring to the energy and energy-related data and all its attributes that will be collected from sensors in the field. As 22 such, it will need to effectively handle data feeds at fixed (i.e., time series data (e.g., from temperature sensors)) and variable (e.g., survey data) intervals. such, it will need to effectively handle data feeds at fixed (i.e., time series data (e.g., from temperature sensors)) and variable (e.g., survey data) intervals. 2. Fieldable sensors. 3.1 Platform Technical Requirements Amongst other things, this means storing time series data using time series database technology that leverages the fixed interval nature of time series to minimize data transfer. Data must also be available to be downloaded on demand for analysis. 5. Supports up to 50-100 users. While there is no specific upper limit for the maximum number of homes that this platform should accommodate, it should be noted that utility-scale implementations (i.e., hundreds to thousands of users) are not the intended use case. Rather, this platform is intended for the testing of feedback designs that have passed the stage of Wizard-of-Oz prototypes and usability studies and are ready for a high-fidelity field implementation and piloting. 50-100 users were deemed appropriate for this scale. 23 23 6. Allows visual design customization. As discussed earlier, the need for this platform extends from methodological issues that have surfaced in the literature. A key part of the problem has been the lack of detail when describing technical feedback implementations. Quintessentially, this includes the specification of visual feedback design and testing of design variations. Such a platform should not only support the customization of feedback, but also sharing of the design specification for ease of replicability. 7. Clearly specifies accuracy and precision of collected data. Ideally sensors will produce accurate and precise data. Due to manufacturing tolerances, wear and tear, this cannot always be guaranteed. This is especially the case for low-cost sensors that are not certified, but are likely to be used in anticipated implementations due to budget considerations. In some cases, ensuring a high precision (with a wider tolerance for accuracy) is acceptable if social comparisons are the focus for example. Manual calibration efforts may also be required to bring accuracy to within an acceptable range. 8. Has an open Application Programming Interface (API). At present the IoT world is still in its infancy. As such, there are few dominant standards for communication. However, wifi and the internet via Hypertext Transfer Protocol (HTTP) are quickly becoming the lowest common denominator with which many IoT devices can interoperate. The platform should have an API to accept data feeds (e.g., from various IoT sensors and surveys), manipulate existing values, and to export data for analytical purposes or visualization. 9. Allow for in-situ surveys to be designed, deployed and filled. 3.1 Platform Technical Requirements Not all data can be captured with sensors and often times it is useful to ask participants what they feel, think, and why they held those thoughts or acted a certain way. Such a tool would be useful to immediately gauge information about issues like thermal comfort, which are very subjective. A good survey tool would allow for the design of surveys with multiple question types and response types. In comparison to printed and manually entered surveys, electronic equivalents are quick and convenient to disseminate and complete. Completed surveys should be time stamped and stored securely on the CMS. It should allow individual surveys to be delivered through an app on the Android platform. 10. Web analytics. Since the days of web page hit counters, web analytics have given administrators insightful information to the usage of their webpage. Nowadays, web analytics has exploded in capability and can now track information on users’ devices, location, visit frequency and duration, and conversion rates if the web site is for commercial use. To ensure adequate 10. Web analytics. Since the days of web page hit counters, web analytics have given administrators insightful information to the usage of their webpage. Nowadays, web analytics has exploded in capability and can now track information on users’ devices, location, visit frequency and duration, and conversion rates if the web site is for commercial use. To ensure adequate 24 interaction, and to potential track confounds on the efficacy of feedback, the platform should support usage analytics. At a minimum, it should be able to track how frequently and duration with which the feedback tool is being utilized. The same analytics may also be used to track any usability issues to help improve future iterations of feedback tools. interaction, and to potential track confounds on the efficacy of feedback, the platform should support usage analytics. At a minimum, it should be able to track how frequently and duration with which the feedback tool is being utilized. The same analytics may also be used to track any usability issues to help improve future iterations of feedback tools. 11. Tablet and software. While, the platform is designed to be robust and capable to deliver energy feedback information on multiple devices, for experimental purposes it is helpful to provide a common device to help ensure the same user experience. 3.1 Platform Technical Requirements A tablet was considered the best choice as it could be used equivalently as a typical home energy monitoring display would (e.g., Aztech In-Home Display). Furthermore, a tablet can also be used as part of a reward for participation since they are also able to connect to the internet for browsing purposes or to play games, etc. This should not preclude other forms of feedback delivery such as printed statements, or traditional website portals. Rather, by developing a tablet app, the platform is robust to the most demanding of feedback implementations. For cost and availability, the Android platform was selected for development of an app. However, to minimize platform- specific development, the concept for the app could be to simply display web content from the CMS. 12. Data security and privacy considerations. Data security and privacy are important considerations when running any web-based service and when collecting personal data from participants. As a first line of defense, researchers should aim to keep personally identifying information in a separate database, preferably kept only on local storage. This will help ensure privacy if data is illicitly retrieved from the CMS database or during transmission. To enable this, the platform should be able to accommodate arbitrary participant identifiers. The CMS should offer standard login security for administrators and participants. To allow write access to the CMS database from third party devices (e.g., SafePlugs, and online weather data feeds), the CMS should support specially coded read/write API keys. To allow tablets and other web-enabled devices to display data from the CMS, the CMS should support read-only API keys. These keys should be kept private at all times in the same way passwords would. Given the time and technical resource constraints for this thesis, commercial-off-the-shelf solutions were considered for building sensing infrastructure. Several options were considered from manufacturers/vendors including National Instruments (http://canada.ni.com/), BlueLine (http://www.bluelineinnovations.com/), LaCrosse (http://www.lacrossetechnology.com/), and SensorSuite (www.sensorsuite.com). With the exception of National Instruments, no solution fully met 25 the sensing requirements. National Instruments solutions, while technically feasible were dismissed due to cost prohibits. Additionally, commercial solutions were dismissed for two broad reasons. First, they can be difficult or impossible to customize in an agile fashion needed for scientific exploration and rigor. Second, commercial solutions are often based on business models that keeps ownership of the data within the manufacturer; this is not acceptable for research purposes. 3.1 Platform Technical Requirements As will be explained later, SafePlugs were the one exception to this policy and there were no FOSS equivalents to its capabilities. In general however, with the broad requirements identified above, an IoT-FOSS approach was decided upon to allow for flexibility in design and full ownership of the data collected. 3.2 Open Source Projects and Free Utilities Open source hardware such as Arduinos and Raspberry Pis, were an appealing starting point for this feedback research platform because of the flexibility of the hardware to build ‘Internet of Things’ solutions. The Arduino platform in particular is well known to undergraduate engineering and computer science students as a rapid prototyping platform for many Do-It-Yourself (DIY) style of projects such as graphing thermostats, power meters, or home automation applications (e.g., http://playground.arduino.cc/projects/ideas). Raspberry Pi projects are often geared towards low-cost computer applications to serve speciality purposes. For example, there are a multitude of Raspberry Pi projects to build media centres or computer network servers. (http://www.raspberrypi.org/forums/viewforum.php?f=15). While Arduino and Raspberry Pi boards themselves serve as a platform for rapid prototyping and DIY projects, there is considerable amount of work required to scale the platform for larger data-centric applications as is required for the current project. In the following sub-chapters, two open source projects used as a basis for the feedback research platform are discussed. The Open Energy Monitor project serves as the core technology behind the platform. The Open Data Kit project was leveraged for survey deployments. Google Analytics and Piwik were used to provide usage data, while Weather Underground was leveraged for real-time weather data. 3.2.1 The Open Energy Monitor Project 3.2.1 The Open Energy Monitor Project Through a search of open source projects based on Arduino or Raspberry Pis, the Open Energy Monitor project surfaced quickly as a robust platform for energy monitoring applications. The Open Energy Monitor system comprises of wireless sensor nodes that send data at periodic intervals to a web- 26 connected base-station. From there data can be visualized locally using the base station as a server, or the data can be sent to an online content management software (CMS). Figure 3-1 illustrates how these components are connected. connected base-station. From there data can be visualized locally using the base station as a server, or the data can be sent to an online content management software (CMS). Figure 3-1 illustrates how these components are connected. 3.1 Platform Technical Requirements Figure 3-1: Open Energy Monitor system components and connections Adapted from www.openenergymonitor.com Figure 3-1: Open Energy Monitor system components and connections Adapted from www.openenergymonitor.com Given that the OEM platform is free and highly configurable it was an obvious choice on which to base the feedback research platform. At the time of writing, with the OEM platform it was possible to sense: AC electricity (apparent power, current, voltage, real power, power factor), temperature, humidity, pulses (from pulse output utility meters), Elster IrDA (direct utility meter interface) and solar PV power diversion; thus, the sensing requirements for the feedback research platform were met. Furthermore, there is ongoing work to extend this list to include CO2 and other air quality measurements. Below, detailed components of the platform are reviewed as they relate to the requirements specified in Chapter 3.1. Emoncms content management software (CMS) Emoncms is an open-source web-app for processing, logging and visualising energy, temperature and other environmental data. It has an open API to accept inputs from any data source, and out-of-the-box it can accept inputs from sensing devices offered from OEM. By leveraging the TimeStore database 27 technology (Sterling, 2014), it meets the requirements for efficient data processing and storage from Chapter 3.1. Data feeds can also be visualized through a dashboard creation tool. technology (Sterling, 2014), it meets the requirements for efficient data processing and storage from Chapter 3.1. Data feeds can also be visualized through a dashboard creation tool. A full and public installation of Emoncms (can be accessed at www.emoncms.org; or, the source code may be freely downloaded and installed on a separate server. For research purposes it is important to have ownership and complete control of the data set so Emoncms was installed on a separate server. As will be detailed later, there were several modifications made to this platform for it to be more amenable for research. Raspberry Pi base station The purpose of a gateway is to bridge two networks: the low-powered radio-based network for local data sensors and, wifi for relaying data to the internet. There are two options for base station gateway in the platform: NanodeRF or Raspberry Pi. The Raspberry Pi (currently using Model B, the latest model as of this writing) was selected because it offered the potential to have wireless internet connectivity via wifi USB dongle and because it provides options for local back-up, and the flexibility of control that a Linux-based machine provides. Having wifi access to the internet also makes for a less-intrusive installation. A ready-to-go software configuration of the Raspberry Pi can be found on OEM’s website (http://emoncms.org/site/docs/raspberrypigateway) to forward the data to the CMS. This configuration file has been modified for research and reliability purposes. As listed above from the OEM site, there are a multitude of relevant sensors that have been configured to work with the platform. In particular electric current, voltage, temperature, and humidity were identified as the part of the core requirement. However, other sensors to aid in measuring air quality, information from utility meters, and occupancy would be helpful for a more comprehensive research or home automation application. Conveniently, the emonTXv3 and emonTH sensor nodes were available to meet the core requirement. emonTx V3 (http://openenergymonitor.org/emon/modules/emonTxV3) As of this writing, the emonTx V3 was the latest generation of the emonTx low power wireless energy monitoring node. See Figure 3.2. It was designed for monitoring AC electrical power on up to four separate household electrical circuits using non-invasive clip on current transformer (CT) sensors and an AC-AC Voltage adaptor to provide a voltage signal for full real power calculations. One of the unique advantages of this device is that it can be powered by a standard 5V Universal Serial Bus (USB) cable or 28 with 3 AA batteries for simplicity of installation. With standard alkaline batteries, the device is rated to last for approximately one year. Figure 3-2: emonTx V3 ruggedized power and temperature sensing now Imaged adapted from http://openenergymonitor.org/emon/modules/emonTxV3 Figure 3-2: emonTx V3 ruggedized power and temperature sensing now Imaged adapted from http://openenergymonitor.org/emon/modules/emonTxV3 Using the ATmega328 microprocessor the emonTx V3 runs standard Arduino programs (i.e., sketches) and is fairly easy to customise and upload code using the standard Arduino integrated development environment (IDE) and a USB to Universal Asynchronous Receiver/Transmitter (UART) cable. The data from the emonTx V3 is transmitted via a 433 MHz radio to an the Raspberry Pi Gateway with similar radio, which then posts the data onto an Emoncms server for logging, processing and graphing. emonTH (http://openenergymonitor.org/emon/modules/emonTH) The emonTH is an open-source, battery powered (2xAA), temperature and humidity monitoring wireless node and was designed to be an easy to deploy tool. See Figure 3-3. Like the emonTx V3, the emonTH uses an ATmega328 chip, runs standard Arduino sketches, and is easy to customise and upload code using the Arduino IDE and a USB to UART cable. The data from the emonTH is transmitted via 433 MHz radio signals to the Rasbperry Pi gateway. 29 Figure 3-3: emonTH temperature and humidity, battery-powered sensing node Imaged adapted from http://openenergymonitor.org/emon/modules/emonTH Figure 3-3: emonTH temperature and humidity, battery-powered sensing node Imaged adapted from http://openenergymonitor.org/emon/modules/emonTH 3.2.2 SafePlugs One limitation with the OEM platform is the lack of a convenient sensor for capturing plug loads. Disaggregated feedback gives task-relevant feedback and if delivered in real-time can quickly allow users to draw connections between their behaviour and environmental impact. The SafePlug (www.safeplug.com) is a commercial-off-the-shelf product that meets this requirement. The SafePlug is a power receptacle placed overtop standard wall receptacles. Out of the box, it provides fire protection, shock protection, and power protection aimed to help keep family members and property safe. When configured as part of the Home Energy Manager Kit (see Figure 3-4), it also can be used for energy monitoring and automation applications. Each kit comes with a gateway device, two SafePlugs, and a collection of RFID tags. The SafePlugs themselves incorporate safety and energy metering circuitry. They also have actuators to control power flow to plugs and Zigbee radios to relay data and controls to and from the gateway. The gateway device serves a bridge between Zigbee and Ethernet networks allowing connected SafePlugs to be tracked and controlled wirelessly over the internet. It also has internal storage for energy use data. The RFID tags are used to help the system uniquely identify appliances. 30 Figure 3-4: SafePlug Home Energy Manager Kit Imaged adapted from www.safeplug.com Figure 3-4: SafePlug Home Energy Manager Kit Imaged adapted from www.safeplug.com The key enabling feature of the SafePlug Energy Manager Kit is its open API, allows full access to the SafePlugs. Using this API, software hooks were developed to allow the SafePlugs to integrate into the Emoncms platform. Technically there are a couple of limitations of the SafePlug. First, as with many other power meters, it cannot reliability detect power draws below 20W. This limitation may be critical if power draw from phantom loads is important. Second, due to the nature of the Zigbee network, data cannot be collected more frequently than every 20 seconds. High frequency sampling rate is an important consideration for some load disaggregation calculations. Also, if high draw, short cycle appliances are used (e.g., microwave ovens), this could be problematic. In a search of free or open source tools for this purpose, Open Data Kit surfaced as the de facto candidate. As mentioned a key criteria for such a tool is its deployability in the field with a tablet. 3.2.2 SafePlugs Open Data Kit allows researchers to design custom surveys with a range of question types and response types using templates based in Microsoft Excel. Using a tool called XLSForm and ODK Collect, survey designs are then transformed into deployable format. ODK Aggregate is both a server and data repository for completed surveys. Administrators can download data for analysis or see basic descriptive visualizations of results within ODK Aggregate. Ideally, the ODK Aggregate server would be fully integrated with the Emoncms. However, this is not presently the case due to different underlying server and database technologies. This is not surprising 31 given the separate nature of both projects. Integrating the platforms would require a porting of one platform to the other’s technology base. With a common platform from which researchers could design tools and collect data, new feedback opportunities would arise. For example, one could imagine an adaptive form of feedback that provides tailored recommendations based on a survey response. However, the integration of functionality between services was not determined to be a priority at this time. Despite the separate back-end technologies between Emoncms and OpenDataKit, it was determined that having a coherent front-end software application was the minimum level of integration necessary. As will be discussed later, this was achieved through the development of an Android app to integrate data from both sources. 3.2.4 Android App 3.2.4 Android App To help control for the effectiveness of feedback, it was important to ensure a common user experience. Laptops have different sized screens, different operating systems, and different web browsers. Smartphones suffer similar issues but are further constrained by screen size. It was determined that it was important to not only provide a common user experience, but to also provide it on a common hardware device. The OEM platform incorporates a desktop LCD display (called emonGLCD – see Figure 3-1) and this was one such option. It was determined early on in the development of the research platform, however, that this would not be sufficient to test a variety of feedback designs due mainly to its screen limitations. It was black and white, had a low resolution, and was smaller than most smartphones. Instead, a tablet- based solution was decided upon. For cost and hardware selection considerations, the Android platform was chosen. In particular, the feedback app was designed for the ASUS MemoPad 7 HD, an Android 4.1 (Jellybean) based tablet. It should be noted that, with additional effort, the same app may be scoped for a wider range of mobile devices. To maintain central control over the feedback design and delivery, it was determined early on that the Android app would not contain code for visualization but rather leverage from the OEM dashboard tool. In this vein then, the app would essentially serve as a window into the web. However, much of the usability design around the app would function to give the user the feel of an app. For example, 404 error pages would be replaced with tablet-styled pop-ups. Also, the app would simplify navigation between different sub-dashboards as required. Furthermore, the app would handle login credentials. 32 Web / Android Analytics 3.2.5 There were two freely available options from which to choose from this purpose: Piwik and Google Analytics. Google Analytics is the de facto standard in the web analytics field and is a free solution for small-scale applications. It comes with many standard visualization widgets and a tool for designing custom dashboards. Its primary advantage for the FBRP is that it can also be used to track analytics from Android devices. This allows the tracking of app-specific interactions like user login and interface clicks. However, any analytics data collected resides on Google servers and incoming data often takes several minutes to days to appear on dashboards. 3.2.4 Android App There were two freely available options from which to choose from this purpose: Piwik and Google Analytics. Google Analytics is the de facto standard in the web analytics field and is a free solution for small-scale applications. It comes with many standard visualization widgets and a tool for designing custom dashboards. Its primary advantage for the FBRP is that it can also be used to track analytics from Android devices. This allows the tracking of app-specific interactions like user login and interface clicks. However, any analytics data collected resides on Google servers and incoming data often takes several minutes to days to appear on dashboards. Piwik is an open source web analytics solution that can be freely installed on any server. At the time of writing it was not as fully featured ad Google Analytics, however it appears to have an active and growing user community. With its current test installation on the FBRP server, there are no limits to the number of feeds or database size. Furthermore, because of the small-scale installation, it can provide real-time analytics. Both Google and Piwik analytics solutions were installed with concurrent data collection since they were both free to use, their co-existence would not introduce any conflicts, and each had their unique benefits. Besides minor bandwidth concerns, there was very little disadvantage to this approach. 3.2.6 Weather Data Obtaining live weather data is important to help homeowners manage their HVAC-related energy use. Weather Underground is an online, commercial weather data source with more than 34,000 weather stations around the world (www.wunderground.com). Its API can be used to gather weather data such as temperature, humidity, wind speed and direction, pressure, etc. It is free to use for small-scale applications; API calls are limited to a frequency of approximately five minutes for the free usage tier. This was deemed acceptable for the intended application within the feedback research platform as it was determined that only hourly API calls were required for the purposes of this platform. This allows up to 12 cities (i.e., at a frequency of one hour) to be monitored at a time. 33 System Architecture 3.3 Overall, the FOSS and commercial products described in Chapter 3.2 were sufficient in meeting the requirements for the feedback research platform specified in Chapter 3.1. With some integration and configuration work, a system architecture was developed as depicted in Figure 3-5. The architecture emphasizes the flow of data between the major components. It also distinguishes between physical devices and components that are virtual and lie within the internet cloud. However, it should be noted that as depicted, its scope is not meant to be rigid and exhaustive, but rather flexible and configurable to the scope of a given project. For example, display devices can be limited to simply tablets, and more sensors can be integrated to communicate through the base station. In this sense, the architecture proposed in Figure 3-5 can be viewed as a framework. Figure 3-5: System architecture for the feedback research platform Figure 3-5: System architecture for the feedback research platform 3.4. Implementation and Maintenance Skills Requirements As expected, when leveraging open source projects, considerable work is still required to integrate and customize the feature-set to your requirements. There are many details to the integration of the components; a description for which is better suited for a technical report and is outside the scope of 34 34 this thesis. However, to summarize the corresponding skillset and tools required to customize the platform, the following is a list of key software technologies that are leveraged: this thesis. However, to summarize the corresponding skillset and tools required to customize the platform, the following is a list of key software technologies that are leveraged: - HTML and CSS for general web development and front-end design; - Advanced Javascript including JQuery, AJAX for core web development with Emoncms; - PHP for Emoncms server-side scripting and customizing the Raspberry Pi gateway behaviour; - PHP for Emoncms server-side scripting and customizing the Raspberry Pi gateway - JSON for data interchange with Weather Underground, Emoncms, SafePlugs; - XML for data interchange format for OpenDataKit; - XML for data interchange format for OpenDataKit; - MySQL for Emoncms administration; - WAMP for configuring a local instantiation of Emoncms for testing - Java for Android development in the Eclipse-based Android Development Tools integrated development environment (IDE); - C++ for Arduino sketch development; - C++ for Arduino sketch development; - C++ for Arduino sketch development; - Linux for working with Raspberry Pi and web servers; and - Python for big data processing and analysis - Python for big data processing and analysis Further implementation details and source code for the feedback research platform will be made available here: https://github.com/kevinci29/fbrp/. available here: https://github.com/kevinci29/fbrp/. 35 4. Project Implementation at Phoenix Place This chapter describes an implementation of a real-time feedback solution as part of a tenant engagement program in a MURB, leveraging the aforementioned feedback research platform presented in Chapter 3. This implementation process was not linear, but rather cyclical. This was necessary in order to find a solution that met technical challenges, accommodated and leveraged existing building infrastructure and involved the target user community of tenants. As would be expected, much coordination was needed between various stakeholders. Fortunately, this thesis work extended from an existing and on-going relationship between Ryerson University and Phoenix Place, the target site for this research. Furthermore, this work was also part of a larger tenant engagement program with a team consisting of the author, Dr. Sara Alsaadani, Samira Zare Mohazabieh, Professor Alan Fung, and Professor Vera Straka, all from Ryerson University. This chapter begins with a review of the general MURBs context and the efforts previously conducted at the target site for the current study. This background is leveraged as part of a Community-Based Social Marketing approach for the design of a tenant engagement program intended to promote energy conservation. The chapter concludes with a discussion on the iterative design process for the visual feedback design – the focus of this research. 4.1 Approximately 30% of the Canadian households reside in MURBs (Government of Canada, 2012). With an overall aging of the stock of MURBs there has been a growing effort on the part of industry and government to develop measures to improve their efficiency. The City of Toronto's Tower Renewal project is an example of one such initiative. However, while there are many conventional approaches to improving MURB energy, water and indoor environmental performance, most are directed at improving the building itself. It can be argued that reducing energy consumption in buildings and enhancing their 36 36 performance is equally a social problem and technical one. Proponents of this vantage point argue that “buildings don’t use energy: people do” (Janda, 2011). Rental MURB dwellers tend to be of a lower or working class relative to their peers in single family homes. This likely results in energy use per tenant to be lower and it can be argued that there is less savings to promote. Neilsen (1993, from Fischer, 2008) found that savings were harder to tease out. However, it can be argued that low income households have most to gain since, low-income households spend about twice the percentage of their income on energy as compared to middle- or upper-class homes (Tweed, 2013). This sentiment appears to corroborate the view that feedback is not as effective for affluent homes where the cost of energy is low relative to income. (Geller et al., 1982, via Froehlich, 2009) Likely less contentious is the negative effect that split incentives have in motivating energy conservation. Split incentives take rise in scenarios where the building occupant, who consumes utilities, does not pay (or directly pay) the utility bill. This is often the case in rental MURBs or in condominiums that have utilities built into flat monthly fees. The result is that there is very little external incentive reward (cost savings, or fee decrease) for inhabitants to conserve. There has also been very little investment by the HEMS industry to MURB renter demographic. Split incentives are one reason, but these tenants are less likely to have the same level of discretionary funds to allow for the purchase of the latest energy efficiency gadget – even if the anticipated savings would more than recover the upfront costs. 4.1 For similar reasons, or perhaps as a consequence of the above, very little is still known about the efficacy of energy interventions on this demographic. Nonetheless, there is a strong case for why MURBs can be an invaluable backdrop for field study research. First, rental units in such buildings are relatively homogenous in size and, naturally, in vintage and construction. In addition, major white appliances (i.e., stove, fridge) are often provided and they too are of the same vintage. Second, such buildings often attract a relatively homogenous tenant-base. Taken together, such conditions lend themselves well to controlled studies close to what may be simulated in a laboratory, but in the field. 37 Phoenix Place – History and Facts 4.2 Given a pre-existing and longstanding relationship, this research was set at Phoenix Place, a mid-rise MURB in Toronto’s Parkdale community. Phoenix Place is an affordable housing project built by the Parkdale United Church Foundation in 1976. According to the Green Phoenix website (www.greenphoenix.ca): Given a pre-existing and longstanding relationship, this research was set at Phoenix Place, a mid-rise MURB in Toronto’s Parkdale community. Phoenix Place is an affordable housing project built by the Parkdale United Church Foundation in 1976. According to the Green Phoenix website (www.greenphoenix.ca): “These apartments are home to many who would otherwise find housing too expensive or difficult to obtain, including persons who are new to Canada, who have been living in shelters or sub-standard housing, or who lack the resources to find decent shelter elsewhere.” Directed by the Parkdale United Church Foundation (PUCF) and its congregation, Phoenix Place underwent retrofits based on principles of sustainability and green construction. 4.1 Completed in the summer of 2010, the retrofits included: - An upgrade to double-glazed, argon filled, low-e coated, fibreglass framed window - An upgrade of wall assemblies using exterior insulation and finish system (EIFS); - An upgrade from electric-baseboard heaters in each suite to hydronic fan-coil units - An upgrade of makeup air-handling unit with enthalpy recovery; An upgrade to high efficiency gas boilers that replaced the original atmospheric boiler - The installation of flat plate solar thermal collectors with capacity to fully meet domestic hot demands during the summer, reducing the need for natural gas to run the existing boilers; an - The installation of ground source heat pumps as the source of heating and sole system of cooling; however, with a resultant air supply of 17-19oC during the cooling season there have been complaints that this is insufficient (Prada, 2013). The tower itself contains 136 suites; 134 of which are nearly identical bachelors each with approximately 20.5m2 of space. Figure 4-1 shows a typical floor plan in the 11 storey tower. The near-identical units are intended for single occupancy and also contain the same standard fridges, stoves, range hoods, and light fixtures – all of the same vintage as well. 38 Figure 4-1: Typical floor plan at Phoenix Place Figure 4-1: Typical floor plan at Phoenix Place Additionally, the electrical wiring in each suite was isolated from others allowing for energy sub- metering as is currently being conducted by the property manager. The sub-metering system afforded two key enablers for the field study. First, it allowed for a validation of measurements captured with the feedback research platform since both systems were running concurrently during the study. Second, it provided approximately three years of historical data with which to establish a baseline of energy use. As will be discussed in Chapters 5 and 6, the analysis of savings in the field study was done using the sub-metering data as it was deemed more consistent and reliable than the data captured with the feedback research platform. The limitation of the sub-metering data, however, was that it could not be used for the real-time feedback – so the sensors from the feedback research platform were still required for that functionality. Additionally, the electrical wiring in each suite was isolated from others allowing for energy sub- metering as is currently being conducted by the property manager. 4.1 The sub-metering system afforded two key enablers for the field study. First, it allowed for a validation of measurements captured with the feedback research platform since both systems were running concurrently during the study. Second, it provided approximately three years of historical data with which to establish a baseline of energy use. As will be discussed in Chapters 5 and 6, the analysis of savings in the field study was done using the sub-metering data as it was deemed more consistent and reliable than the data captured with the feedback research platform. The limitation of the sub-metering data, however, was that it could not be used for the real-time feedback – so the sensors from the feedback research platform were still required for that functionality. Overall, given the homogeneity of suites and the electrical isolation and sub-metering of each suite, Phoenix Place was an excellent test-bed for social comparisons strategies since many normalization estimates (e.g. home size, appliances, occupants) were not required to ensure fair comparisons. 39 Results from a Post Occupancy Evaluation at Phoenix Place 4.3 In their survey of energy use by tenants at Phoenix Place, Roque, Straka and Fung (2012) sought to understand relationships between household energy use and demographic information amongst other variables. The survey was comprised of questions on ownership, usage of various consumer appliances, and the frequency in which occupants turned appliances off when not in use. The questions were grouped by usage categories like heating/cooling, cooking, and lighting. Additionally, it posed questions on satisfaction with the indoor environment focussing on thermal comfort. The survey results helped provide insight to the demographics at Phoenix Place and where potential energy savings may lie. Of the 48 tenants who completed the survey the following demographics information were reported: Of the 48 tenants who completed the survey the following demographics information were reported: - 80% of respondents were male, - 56% of respondents were over the age of 46, - 56% of respondents were over the age of 46, - 45% of respondents had lived in this MURB for over 7 years, - 49% of respondents reported spending between 9-13 hours a day at home (including p p p g y ( g p), - 45% of respondents reported growing up in Africa, and - 45% of respondents reported growing up in Africa, and - 66% of respondents have a total annual household income below $29,999. 4.1 In a meta-analysis of the survey results, Dr. Sara Alsaadani found several statistically significant correlations between suite-metered energy use and specific energy use behaviours as defined in (Roque et al., 2012). By prioritizing these behaviours, a set of 27 energy conservation tips were developed to inform the energy conservation program’s information campaign. These tips were reinforced within the feedback dashboard: In a meta-analysis of the survey results, Dr. Sara Alsaadani found several statistically significant correlations between suite-metered energy use and specific energy use behaviours as defined in (Roque et al., 2012). By prioritizing these behaviours, a set of 27 energy conservation tips were developed to inform the energy conservation program’s information campaign. These tips were reinforced within the feedback dashboard: 1. "Switch off your TV and cable boxes when you are not watching TV.", 2. "Make sure the brightness of your TV is just how you need it for your room. Factory settings brightness is usually brighter than necessary.", 3. "If you own both a desktop and a laptop, try to use your laptop more often as a laptop is generally more energy-efficient.", 4. "Switch off your computer when you are not using it.", 5. "When you are cooking put your lid on the pot or pan.", 6. "Use your microwave, rather than your stove, especially to heat already cooked food 7. "When boiling foods on the stove (e.g., pasta, potatoes, eggs, etc.) switch off the stove burner a few minutes early.", 40 8. "Stove – use the correct sized burner for the pot or pan.", 9. "Turn the heat down to the minimum setting required to cook your food.", 10. "Use the minimum amount of water when boiling your food.", 11. "Increase the amount of food you cook to that you can refrigerate or freeze it, and re-heat it later.", 12. "Cutting food into smaller pieces reduces cooking time.", 13. "Thaw your frozen food in the refrigerator rather than the microwave or the oven.", 14. "Use an electric kettle to boil water for coffee or tea (or even cooking) instead of a stove-top kettle or pan.", 15. "Turn off your fan-coil unit when you are not at home.", 16. "Use window shades or blinds to reduce or completely block sun and heat during the summer, especially if you receive direct sunlight.", 17. "Use your hairdryer sparingly and don’t use the maximum heat setting to save energy.", 18. 4.1 "Remember to switch off the lights when you are not in the room.", 19. "Install Compact Fluorescent Lights (CFL) rather than incandescent bulbs.", 20. "Dust your bulbs and light fixtures with the power off.", 21. "Only do your laundry when you are ready to load your washing machine to full capacity.", 22. "Use lower temperature settings on washing machines – use warm or cold water for the wash cycle rather than hot water, and only use cold water for rinses.", 23. "When drying, separate your clothes and dry similar types of clothes together.", 24. "Don’t over-dry your clothes. Take your clothes out of the dryer while they are still slightly dam if you intend to iron them immediately, to reduce energy.", 25. "When possible, dry full loads.", 26. "Consider hang-drying clothes when/if possible.", 27 "If you live on the lower floors consider taking the stairs rather than the elevator " 22. "Use lower temperature settings on washing machines – use warm or cold water for the wash cycle rather than hot water, and only use cold water for rinses.", 4.4 The CBSM Framework in Application Given that this research was being conducted with a known target community (i.e., tenants at Phoenix Place), the team had arrived at a set of conservation behaviours, and considered CBSM as an appropriate framework to follow. As outlined in Chapter 2.2.3, the CBSM process consists of five steps. This sub-chapter details each in application. 41 1. Select behaviors. Target behaviours include simple electricity savings tips in and around the home. As listed above these behaviours were identified to have a significant correlation to measured energy use. Additionally, this work focussed on thermal comfort related behaviours for a couple reasons. First, heating and cooling are a large component of overall MURB energy use. Second, as was identified by Prada (2013), cooling was deemed a concern at Phoenix Place due to the circulation of cold water from the ground loop without using the heat pump. 2. Identify barriers and benefits. Given the MURBs context, lack of motivation was determined to be a key barrier against conservation. This is partly due to split incentives since tenants do not directly pay for their electricity use. The lack of knowledge on their energy use and how best to conserve is another barrier. While tenants’ energy use has been sub-metered since 2010, they have not been shown this use, nor do they have a point of reference to know whether their use is above or below average neighbors or if it has gone up or down from past use. 3. Develop strategies. Several strategies were applied in this project. First, to provide motivation, an information campaign was launched to raise awareness and make a case for the need for energy conservation at Phoenix Place. This was combined with community goal of 10% in overall savings and written individual commitments to help reach that goal. This was deemed a reasonable initial target given similar approaches to feedback. Since this value was essentially set through software it was possible to change it mid-way through the study if needed. To help improve energy conservation knowledge, specific tips were provided as part of campaign materials. These were reinforced through their inclusion in a feedback implementation – the subject of this thesis. The feedback implementation would also focus on providing historical and social comparisons to provide energy use in context, further improving user knowledge. To help improve energy conservation knowledge, specific tips were provided as p campaign materials. 4.4 The CBSM Framework in Application These were reinforced through their inclusion in a feedback p g g implementation – the subject of this thesis. The feedback implementation would also focus on providing historical and social comparisons to provide energy use in context, further improving user knowledge. As noted in Chapter 2.2.3.5, the efficacy of motivating through real-time social comparisons has been under-explored. Furthermore, given the homogenous nature of suites within Phoenix Place, this field study offered an excellent opportunity to evaluate a social comparisons strategy. Naturally, the evaluation of social comparisons approach to feedback was of key interest in this field study. 4. Pilot. Design is not a linear, but a cyclical process with an evaluation stage at the end of each cycle. As will be discussed in the following chapters, the visual feedback design underwent several iterations. Similarly, the information campaign materials underwent considerable internal review with the project team and with the property manager. While the efforts in this 4. Pilot. Design is not a linear, but a cyclical process with an evaluation stage at the end of each cycle. As will be discussed in the following chapters, the visual feedback design underwent several iterations. Similarly, the information campaign materials underwent considerable internal review with the project team and with the property manager. While the efforts in this 42 program were of considerable scale, they may also be considered a pilot in the larger context of MURBs in general. program were of considerable scale, they may also be considered a pilot in the larger context of MURBs in general. program were of considerable scale, they may also be considered a pilot in the larger context of MURBs in general. 5. Broad scale implementation. The larger purpose of this project is to provide knowledge on how such interventions may be replicated in other MURBs. Thus, the focus of this thesis was to ensure that success could first be demonstrated within the current MURB. Towards this goal, the next chapter clarifies the detailed objectives and constraints for this project. Summarizing the Feedback Design Context, Objectives, and Constraints 4.5 Summarizing the Feedback Design Context, Objectives, and Constraints Thus far, this chapter has reviewed some recent history and facts at Phoenix Place as well as POE efforts to understand the demographic and personas within the community. Those POE efforts have been used to grasp how electricity is being used and subsequently leveraged to identify a set of tips that are relevant to the community as a whole. Chapter 3 also outlined technical details of the feedback research platform. Building from that, this sub-chapter reviews the thought process behind the tailoring of the feedback displays specifically. The implications of the visual feedback design cascaded naturally to requirements for the customization of the rest of the feedback research platform. The objectives of the feedback at Phoenix Place is to engage users to learn about their energy use and motivate them to conserve and reach individual target savings of 10%. That is, if all participants were to share in the same individual goal, the collective 10% savings would be achieved. From the literature review and analysis of the Phoenix Place context, the key strategies to achieving this objective include the following goal setting, written commitments, and historical and social comparisons. Thus, the design will: Thus, the design will: - Focus on benchmarking. Providing appropriate comparisons and showing them visually can be an effective way to inform, frame, and motivate energy conservation. - Aim to keep a simple message. This was decided early on to accommodate understandability and to be approachable given the sample demographic. - Provide reasons for users to explore their data on a regular basis, but at a minimum on a weekly basis. It is important to keep users engaged and benefiting from the feedback information. - Compare the effectiveness of design details. The purpose of this study is not only to design an effective feedback display for wide consumption. Rather, given the gaps identified in the literature, 43 43 it is to test which detailed design choices are most appropriate. In particular, this thesis explores the efficacy of delivering real-time social comparison data. it is to test which detailed design choices are most appropriate. In particular, this thesis explores the efficacy of delivering real-time social comparison data. There were two key design decisions that were made early in the process. The first was to omit the SafePlugs from the current study. There were several reasons for this. Summarizing the Feedback Design Context, Objectives, and Constraints The main reason was that plug load disaggregation was not prioritized for this study. Additionally, it was important for the installation to be minimally invasive. Given the size of apartment suites at Phoenix Place, the SafePlugs would also have required more hardware and considerable installation effort. Second, to limit the scope of development of the feedback dashboards, only out-of-box dashboard tools available from Emoncms (current version was v8.0.3). Theoretically though, given that this is an open source product, a multitude of dashboard widgets and visualizations are possible. Data-Driven Documents (http://d3js.org/) for example, offers many visualization examples. However, this decision was made given time and resource constraints and also because developing a polished product was not a priority at this time. Where appropriate, minor customizations were made to the existing visualization source files. 4.6 Measuring the Impact of Thermal Comfort Measuring the Impact of Thermal Comfort Measuring the Impact of Thermal Comfort 4.6 For the purposes of a larger project initiative the feedback research platform was customized to measure the impact of thermal comfort and relevant feedback on energy use. This thesis details the platform customization of meeting this objective for demonstration purposes. However, detailed analysis on thermal comfort data is outside the scope. Towards providing relevant feedback on thermal comfort, the feedback dashboard was configured with separate, but navigable, displays for the total suite and FCU energy use. FCUs, while not solely responsible for, are a key contributor in delivering thermal comfort. The decision to disaggregate the FCU energy use was to draw special attention to heating and cooling energy use and raise awareness for how electrical energy use was tied to an occupant’s thermal comfort. It was important to build in the flexibility to calculate or estimate the true energy draw from the FCU at a later time. As is, the sub-metering at Phoenix Place for each suite measures the plug loads, lighting, and oven and FCU electricity use. This does not include the energy required to heat or cool the liquid in the radiators inside the FCU. Currently, in the cooling season, the liquid circulated through the ground 44 loop. In the heating season, it is heated by the GSHP along with the gas-fired boilers. To account for this energy use, estimates can be made measuring the hot/cold output from the FCU and multiplying by a simple COP factor. For example, GSHPs have an approximate Coefficient of Performance (COP) of 5 in the cooling season and 3 for the heating season (Safa, Fung, & Kumar, 2015). While the flexibility has been retained, detailed modelling was not prioritized in the dashboard because a precise absolute FCU energy use was not deemed as important as the relative energy use when compared to neighbors. Furthermore, it would mean considerable effort obtaining measures from building wide equipment and more overhead cost. Instead, while less ideal, it was decided simply to provide feedback on what could be directly and precisely measure, and this was the electrical power draw from the FCU and its heating/cooling output as measured using a temperature probe. From the manufacturer’s specification the fan has the following rated properties: Table 4-1. Johnson Controls Enviro-Tec Model VFE Size 20. 115 Volts. Measuring the Impact of Thermal Comfort Fan Speed Rated CFM Rated Power Draw Measured Draw* Low 250 33 33 Medium 550** 39 42 High 1200 57 89 * Average measurements from 3 different suites ** Estimated through linear interpolation using rated CFM and rated power draws Table 4-1. Johnson Controls Enviro-Tec Model VFE Size 20. 115 Volts. Table 4-1. Johnson Controls Enviro-Tec Model VFE Size 20. 115 Volts. Fan Speed * Average measurements from 3 different suites ** Estimated through linear interpolation using rated CFM and rated power draws Section 4.7 details how both the total suite and FCU energy dashboards evolved through three design iterations as part of the CBSM process. Feedback Design Process With the context, objectives, and constraints in mind an iterative design process was used to hone in on an appropriate visual feedback design for the field study. The field study used two types of feedback: real-time feedback with historical comparisons (herein called basic feedback); and the same with additional social comparisons (herein called basic feedback + social comparisons). This chapter focusses on the feedback + social comparison designs since the basic feedback versions were simply derived by subtracting the social comparison features. While considerable description on process could be provided for any given iteration, the intent with each was simply to generate improvements in relative quick 45 succession. As will be detailed in Chapter 5, a detailed evaluation was planned for the field study and this is where a more rigorous scientific approach was followed. succession. As will be detailed in Chapter 5, a detailed evaluation was planned for the field study and this is where a more rigorous scientific approach was followed. 4.7.1 Iteration 1 – Heuristic Design 4.7.1 Iteration 1 – Heuristic Design The first design iteration was guided by the project objectives and strategies earlier and largely informed from the design heuristics identified in Chapter 2.2.4. The result of this analysis led to the prototypes shown in Figures 4-2 and 4-3. This chapter first explains the layout and functioning of each widget in the dashboards. Then it reviews how the feedback design heuristics were applied, or not. Figure 4-2: Basic Feedback + Social Comparisons – Suite Dashboard – Iteration 1 46 Figure 4-3: Basic Feedback + Social Comparisons – FCU Dashboard – Iteration 1 igure 4-3: Basic Feedback + Social Comparisons – FCU Data to populate this dashboards in this iteration were simulated using a PHP script attached to a Linux cron job that would trigger as frequent as every 10 seconds on the server. The large trends in the simulated data were determined by the time of day using trigonometric functions. Random noise was built in to the simulated data to provide some realism. A limitation to the approach was that there was no memory in the model to capture transient effects. For example, as the FCU was turned on in heating season, the temperature in the room did not gradually rise over time. Rather, it rose only during the late afternoon and fell at night. Thus, the data shown in the following prototypes should not be taken literally. 4.7.1.1 Android App and Design Template As shown, the dashboards were designed to fit within an Android application. The title and navigation bar on top shows the two dashboard tabs. Users can click on these two tabs to navigate between the dashboards, or they may simply use a swipe gesture between them. On the right of the title bar there are buttons to complete a comfort survey, to refresh the data, and to enter some administrative and preference settings. 47 The comfort survey can be completed at any time. However, the app was designed to also raise a notification every eight days to prompt users to complete this. The reasoning behind the 8-day interval was to capture the user’s thermal comfort on different days of the week to remove possible scheduling confounds. The notification appeared similar to what a voice mail notification would be like. Details of the comfort survey will be shown later. 4.7.1 Iteration 1 – Heuristic Design Users could always click to refresh the entire dashboard to ensure the latest data is retrieved from the server. In case of any crash, the entire dashboard was also set to reload every five minutes. However, the dashboard widgets were also designed to refresh as soon as data was received on the server in asynchronous fashion (i.e., without reloading the entire dashboard). This helped ensure an overall smooth user experience. 4.7.1.2 Feedback Design Walkthrough 4.7.1.2 Feedback Design Walkthrough Within the dashboard itself (i.e the centre panel of the app with a white background) there were 5 distinct regions. the top-left moving clockwise, there are “Last 7 Days”, “Last 24 Hours”, “Right Now”, a comparison widget, and finally the dashboard title. Energy savings tips were not prepared for this iteration. This layout reflects a chronological ordering of information from left to right. Power draw (and current indoor temperature from FCU dashboard) dials emphasize the real-time nature of the dashboard. The analogy of dials is most similar to what be found in cars. This data is converted to energy use and shown as part of the daily cumulative energy graph. The cumulative energy graph was chosen for its ability to summarize the total energy use for the day (height of curve), while also showing when energy was most or least consumed throughout the day (slope of curve). The color coding of dials and curves were meant to show linkages of data between the graphs. For example, the red temperature dial corresponds to the red temperature curve in the FCU dashboard. Each day at midnight, the cumulative energy graph was set to reset, at which point, concretizing that day’s bar in the Last 7 Days bar graph. Both the Last 24 Hours and 7 Days graphs showed moving windows of data as their name implies and these were in fact updated every 10 seconds, asychronously along with the Right Now dials. The comparator at the bottom of each dashboard was intended to portray the key takeway from the dashboard. For the Total Suite dashboard, this was whether or not the user had surpassed the target limit for the day and by how much. For the bulk of any day, this comparator would return a green smiley. Otherwise, a peach colored sad face would sppear. Similarly on the FCU dashboard, the 48 comparator returned a smiley or sad face. However, in this case, it depended on whether or not they were above or below the cumulative usage when compared against their neighbors average. Note how the FCU graph did not have an absolute goal line. This was because a historical baseline from which to draw was not available in contrast to the baseline for the Total Suite dashboard. Chapter 2 introduced seven feedback design heuristics. 4.7.1.2 Feedback Design Walkthrough While it would be ideal to address all seven heuristics, for practical reasons only the first five were implemented. The following points highlight the considerations for each. - Design the message to filter out unimportant information. This heuristic was especially applicable to utility bills which often contained non-relevant clutter and even advertising materials. Here, just about every graphical widget and label was tailored to providing useful data or clarification of them. - Design the message to filter out unimportant information. This heuristic was especially applicable to utility bills which often contained non-relevant clutter and even advertising materials. Here, just about every graphical widget and label was tailored to providing useful data or clarification of them. - Consider the audience; be specific and personalized. A lot of legwork to tailoring this display took place before even this first design was conceived. The feedback information provided on both displays are by definition personalized to the user. - Benchmark in a fair and meaningful way. The focus of this feedback is to provide not only real-time data for users to learn from, but also to provide clear and motivational points of references. The goal of the conservation program was to encourage 10% savings building-wide; a goal that is designed to cascade down to individual tenants. In the Total Suite display, the suite baseline curves (red lines) were determined based on the average monthly data from over three years’ worth of data from the sub-metering system. The Suite Goal curve (green lines) reflect a 90% value from that goal (here it showed an 80% goal because 20% was the target initially). - Average feedback over meaningful intervals. When real-time feedback is first introduced it was anticipated that the “Right Now” power use dial would be most useful. However, as users begin learning about the energy impact from specific behaviours and appliances, they will likely want to see this data averaged over a longer period to more effectively track savings. In anticipation of this trend, the Daily Cumulative Energy Use chart for the “Past 24 Hours” was also provided. Similarly this was the rationale for the Last 7 Day graphs. It could be argued that weekly or monthly levels of aggregation would be useful as well; however, they were not included for a couple reasons. 4.7.1.2 Feedback Design Walkthrough First, they would have required more display real-estate or more interactivity, possibly overcomplicating the dashboard given the nature of the study demographic. Second, having such displays might condone less frequent checks into the dashboard, providing less reason to check in at least weekly. - Average feedback over meaningful intervals. When real-time feedback is first introduced it was anticipated that the “Right Now” power use dial would be most useful. However, as users begin learning about the energy impact from specific behaviours and appliances, they will likely want to see this data averaged over a longer period to more effectively track savings. In anticipation of this trend, the Daily Cumulative Energy Use chart for the “Past 24 Hours” was also provided. Similarly this was the rationale for the Last 7 Day graphs. It could be argued that weekly or monthly levels of aggregation would be useful as well; however, they were not included for a couple reasons. First, they would have required more display real-estate or more interactivity, possibly overcomplicating the dashboard given the nature of the study demographic. Second, having such displays might condone less frequent checks into the dashboard, providing less reason to check in at least weekly. 49 If users did not check in weekly, they would miss data. The third reason extends to the next heuristic with the intent of keeping users focused on their task of saving energy today and not necessarily dwelling on the distant past. - Make the feedback information task relevant. The overall task for the user is to keep their total suite energy use within their target upper limit of 90% of their baseline energy use (i.e., achieving a 10% overall reduction in their energy use). We also wanted to see how social norming may motivate them to conserve. As mentioned, specific comparisons were made to summarize these in the form of the happy or sad faces. As reflected in the list of tips, there were multiple target behaviours. While it could be useful to disaggregate the feedback to individual appliances, as stated earlier, such details would not necessarily significantly contribute to overall savings goals. However, to demonstrate and explore the efficacy of the disaggregated approach FCU energy use data were included as it related to the task of achieving thermal comfort. Frame feedback data using concrete, tangible equivalents. Currently, energy and power use are communicated in kW and kWh. 4.7.1.2 Feedback Design Walkthrough Using better units of measure is an area that could be useful to help users understand the data. However, it was not clear what equivalent would resonate with users best. An easy solution would be to show the equivalents in terms of its cost in dollars. However, with tiered pricing, and time of use factors, this calculation was not trivial. Estimations could have been completed, but introduces sources of confusion and inaccuracy that may complicate matters. As mentioned earlier, deriving a polished dashboard was not deemed a requirement for this study so this was left out. However, to help overcome the obstacles of understanding standard energy units, the measures were explained as part of the information campaign. In future, evaluating different units would be a worthwhile endeavor. In the meanwhile, though, for the purposes of historical and social comparisons and goal setting, the unit of measure is of less importance. This is especially the case with visual comparisons of data as is used heavily in these dashboards. In summary, this heuristic was not as critical to meet. Use feedback to support a distant future retrospect. As would be expected, this form of feedback requires a considerable amount of modelling to show the future impact of savings. While such feedback offers an additional layer of motivation potential, it was not critical for this study since historical and social comparisons were already utilized as well as goal setting. However, follow-up studies to compare sources of motivation would be worthwhile. 50 4.7.1.4 Internal and Informal Design Review The first prototype was shopped around to the project members and to the author’s peers for internal review and informal critiquing. As necessary, details of the project background and usage context (see Chapter 4.5) were explained in advance. The prototype was demonstrated on the target Android tablet (Asus MemoPad 7 HD) with live, albeit simulated, data. The following is a summary of the three main criticisms from this review. First, the Last 24 Hours graphs were perceived to be too data dense and it was unclear how the curves were intended to be related to one another. Thus this display required considerable explanation. Part of the problem is that such a cumulative energy graph, while very information rich, is not typically used. Given that the target user community was not expected to perform a detailed interrogation of the graphs, a simpler solution was recommended. 4.7.1.2 Feedback Design Walkthrough Second, it was not clear to the reviewers how useful it would be to correlated temperature with FCU energy use. The original intent was to draw that linkage visually to help users better rationalize their comfort-related energy use. However, it was reasoned that users could already sense their thermal comfort and such an explicit display would be redundant at best and most likely insufficient given that thermal comfort is comprised of several additional factors such as relative humidity and air speed. Since it was not the goal of this feedback display to dive deep into thermal comfort, further simplification was recommended. Third, the overlapping bars in the Last 7 Days graph were difficult to distinguish. This was due to additional colors that were produced due to transparency effects. Thus, some respondents initially perceived the bars as stacked and wondered why such colors were not shown on the legend. This particular overlapping design was chosen, initially because it was the default format using the Emoncms multigraph tool. Alternatively, line graphs were considered. However, line graphs tend to suffer from visual peculiarities as viewers attempt to interpret and compare line slopes. For that reason, line graphs are also better tuned for presenting time series data (e.g., the cumulative energy use graph); here, the data points were discrete and aggregated. Overlapping bars is certainly atypical, but rather than dismissing it for that purpose alone, the following factors were considered. The main benefit of overlapping bars is that they draw a direct comparison while taking less visual real-estate in comparison to side-by-side bars, which are the standard. Also, it was believed that the live data being shown would clarify this relationship over time as users experience the dynamics of the graph – something that was not possible during the present design review. Thus, this design was retained. 51 4.7.2 Iteration 2 – Usability Test Prototype based on Initial Feedback 4.7.2 With the results from the first design review, a second prototype was created. This second iteration was prepared for a more formal usability test with two volunteer participants from the target user population. The two volunteers were identified by the property manager at Phoenix Place. The purpose of this usability test was not to collect quantitative data on readability, or users satisfaction or preferences as is typically done (Cialdini & Goldstein, 2004). 4.7.1.2 Feedback Design Walkthrough Rather, it was intended as a sanity check to ensure that representative users from the target community could understand and appreciate the data presented within the displays. The prototypes shown in Figure 4-4 and 4-5 were presented to the users. The test followed the script shown in Appendix J. With the results from the first design review, a second prototype was created. This second iteration was prepared for a more formal usability test with two volunteer participants from the target user population. The two volunteers were identified by the property manager at Phoenix Place. The purpose of this usability test was not to collect quantitative data on readability, or users satisfaction or preferences as is typically done (Cialdini & Goldstein, 2004). Rather, it was intended as a sanity check to ensure that representative users from the target community could understand and appreciate the data presented within the displays. The prototypes shown in Figure 4-4 and 4-5 were presented to the users. The test followed the script shown in Appendix J. Figure 4-4: Basic Feedback + Social Comparisons – Suite Dashboard – Iteration 2 52 52 Figure 4-5: Basic Feedback + Social Comparisons – FCU Dashboard – Iteration 2 4.7.2.1 Design Walk Through Based on comments received from the first review, the Last 24 Hours graph was removed in favor of a simpler, though less informative, comparative bar graph display. These bars now only showed the cumulative energy use for current day without any further information on how the bars arrived there. There is no longer any historical temperature data shown here. The benefit of the side-by-side bars is that they more intuitively link with the bars shown in the last 7 days. In fact, in this example, the Today bars are intentionally redundant with the most recent bars in the Last 7 Days. This visual relationship benefits from the Gestalt principle of common fate (Todorovic, 2008), which states that objects that move in a similar direction are perceived to be related. To provide further task relevant information, a daily quota gauge was provided in the Right Now section. This was color coded green to match the green curve in the Last 7 Days and Today graphs. This gauge essentially served as a fuel gauge in a car. The comparator at the bottom of this dashboard in this version showed comparisons against your neighbours’ average. On the FCU dashboard, temperature was represented as a numerical figure with a thermostat to denote that it is the measured indoor temperature. Live weather data was shown just below it with a numerical indication out outdoor temperature and an icon corresponding to the sky condition (i.e., here shown as 53 sunny). By showing both indoor and outdoor temperatures, the idea was to promote smarter FCU behaviours such as opening windows to cool the apartment when cooler outside. A placeholder for daily tips was also created to serve as prompts and reinforcement of target behaviours from the information campaign. 4.7.2.2 Results from Usability Test Overall, both participants were able to read and comprehend the information shown in the dashboards. As intended, the bars served as useful means in comprehending relative performance. For example, the participants were able to see how some days consumption was higher or that their consumption was higher (i.e., in Total Suite dashboard) or lower (i.e., in FCU dashboard) than their neighbours’ average. Interestingly, it appears that goal-related figures were more important to the users than comparisons against neighbours. One participant commented that he probably uses less than his average neighbor because his work schedule has him away for large portions of the day. There were three friction points however. First, there were issues in understanding the units of measure (i.e., kWh and kW) and explanations of these were required. Some explanation was required. Second, the two temperature figures in the FCU dashboard were not understood from just the graphic icons. It was clear that more labelling was required. Third, the users did not immediately understand the difference between the Total Suite and FCU dashboards. This was likely due to the similar color coding between the displays. The two participants in this usability test confirmed that they did not user their FCU very much. 4.7.3 Iteration 3 – Final Prototype for Field Study Using the results of the usability test, the prototypes were refined resulting in the iterations shown in Figures 4-6 and 4-7. These prototypes were used as part of the field study detailed in Chapter 5. 54 Figure 4-6: Basic Feedback + Social Comparisons – Suite Dashboard – Iteration 3 Figure 4-6: Basic Feedback + Social Comparisons – Suite Dashboard – Iteration 3 Figure 4-7: Basic Feedback + Social Comparisons – FCU Dashboard – Iteration 3 Figure 4-7: Basic Feedback + Social Comparisons – FCU Dashboard – Iteration 3 55 4.7.3.1 Design Walkthrough 4.7.3.1 Design Walkthrough As per the results from the usability test, textual labels were included to clearly distinguish weather data from indoor readings. As per the results from the usability test, textual labels were included to clearly distinguish weather data from indoor readings. To improve clarity, a color coding was applied to visually distinguish Total Suite (now grey colored) from FCU energy and power use. This color coding should be pre-attentively processed faster than reading a title label. With this coding, the title was deemed no longer necessary so it was removed. 4.7.2.2 Results from Usability Test Removing the title should not create confusion as there are legends or labels on all charts and widgets and on top for each dashboard. To further reduce clutter, a standard refresh icon was used instead of the textual Refresh label. Space was dedicated for goal-related comparators under the heading of “How You’re Doing”. The wording for this title was chosen to align with the purpose of this section, which is to summarize the key task-related results in a friendly way. In this section for the Total Suite dashboard, both social and goal- related comparisons are shown. As will be discussed in the next chapter, this was done deliberately to examine the impact of real-time social norming. Finally, the day of week labels were added to the last 7 days to help improve recall of energy related behaviours for that date. 5. Field Study Design at a MURB in Toronto 5. Field Study Design at a MURB in Toronto This chapter describes the methodology for a yearlong field study (run from September 2014 through August 2015) intended to both demonstrate the feedback research platform (FBRP) and evaluate the visual feedback design described in chapters 3 and 4, respectively. As introduced in Chapter 4, the field study was conducted at Phoenix Place, a MURB in the Parkdale community in Toronto, Ontario, Canada. 4.7.3.2 Basic Feedback Design From the basic feedback + social comparison dashboard designs the following basic feedback displays, as shown in Figures 4-8 and 4-9, were derived simply by subtracting the social comparison features from Figures 4-6 and 4-8, respectively. 56 56 Figure 4-8: Basic Feedback Display 1 of 2 – Total Suite Energy Use Figure 4-9: Basic Feedback Display 2 of 2 – Fan Coil Unit - Energy Use Figure 4-8: Basic Feedback Display 1 of 2 – Total Suite Energy Use Figure 4-9: Basic Feedback Display 2 of 2 – Fan Coil Unit - Energy Use Figure 4-8: Basic Feedback Display 1 of 2 – Total Suite Energy Use 57 5.1 Hypotheses The purpose of the development of the feedback research platform was to afford the testing of a multitude of visual feedback designs. A limitation of past feedback studies has been the lack of testing of real-time social norming strategies. Given, the homogenous layouts of suites within the target building was very amenable to testing social norming strategies. Thus, the primary research question was: Can combining real-time feedback with real-time social comparisons help communities of users reach individual and collective energy conservation goals? In the context of a broader conservation program, described later in this chapter, this led to the following two hypotheses for the study: Hypothesis 1: The conservation program comprised of an information campaign, participant commitment, an energy audit, and real-time feedback promotes energy conservation. Hypothesis 1: The conservation program comprised of an information campaign, participant commitment, an energy audit, and real-time feedback promotes energy conservation. Hypothesis 2: Real-time feedback with social comparisons promotes more energy conservation than with just real-time feedback alone for total home energy use. Hypothesis 2: Real-time feedback with social comparisons promotes more energy conservation than with just real-time feedback alone for total home energy use. As mentioned, another objective of the larger project was to understand the impact of thermal comfort and relevant feedback on energy usage. However, the analysis of this data was outside the scope of this thesis. 58 Experiment Design 5.2 The conservation program’s objective was to help participants collectively reach 10% in electricity savings from the previous year’s energy use. To achieve this, the program was comprised of four interventions: The conservation program’s objective was to help participants collectively reach 10% in electricity savings from the previous year’s energy use. To achieve this, the program was comprised of four interventions: 1) An information campaign which outlined reasons for saving energy and providing energy saving tips. 2) A personal pledge to save 10% of their own energy use from the year prior. 3) An energy audit of electrical appliances within the suite 3) An energy audit of electrical appliances within the suite 4) Real-time feedback for a full year. 4) Real-time feedback for a full year. The first three interventions were common across all actively recruited participants. However, feedback was treated as a sole between-subjects variable in the univariate study design. 5.1 Hypotheses There were two levels of feedback: basic feedback (which contained real-time feedback with historical comparisons), and basic feedback + social comparisons (the same with additional social comparisons). The rationale for these designs was described in depth in Chapter 4.6 and in particular Chapter 4.6.3. Participants were randomly assigned to a feedback condition. 5.3 Recruitment and Participants The results of this survey were used first, as part of a separate study to advance a predictive model of energy conservation, and second, as a potential covariate for this study. For their attendance and completion of the survey, they were remunerated $20. In total, there were 50 participants in this portion of the study. outlines the breakdown of the participants in each experimental group by gender, place of birth, and age. Table 5-1. Participant breakdown by Experimental Group, Gender, Place of Birth, and Age Experimental Group Basic Feedback Basic + Social Feedback Gender Male 10 7 Female 2 5 Place of Birth Canada 3 4 Europe 1 -- Africa 5 6 Asia 1 1 Central or South America 2 1 Age Range 18-30 2 1 31-45 3 3 46-60 4 5 61+ 3 3 5.3 Recruitment and Participants With permission and collaboration of the board and property management at Phoenix Place, tenants were first recruited to take part in an information session to kick-start the program. The conservation program recruitment poster is shown in Appendix A. Prior to attending, each participant signed an informed consent form (see Appendix B). As part of the recruitment for this study, a short presentation was given twice on separate weeknights – see presentation slides in Appendix D. After the initial recruitment phase, additional canvassing took place in the building lobby for a week to reach a wider audience. Interested tenants were given the same information, but on a one-on-one basis. Eligible participants fulfilled the screening criteria of having lived at Phoenix Place for at least one year prior, being 18 or older, and having working knowledge of the English language. Of the 134 tenants at Phoenix place, 28 participants were recruited. The remaining 106 tenants in the building were not actively participating. However, with permission, their energy usage data from the building’s sub-metering system was used for comparative purposes, effectively as part of a control group. As detailed in Section 6.2, only 24 participants met the final eligibility requirements. Table 5-1 59 outlines the breakdown of the participants in each experimental group by gender, place of birth, and age. Table 5-1. Participant breakdown by Experimental Group, Gender, Place of Birth, and Age Experimental Group Basic Feedback Basic + Social Feedback Gender Male 10 7 Female 2 5 Place of Birth Canada 3 4 Europe 1 -- Africa 5 6 Asia 1 1 Central or South America 2 1 Age Range 18-30 2 1 31-45 3 3 46-60 4 5 61+ 3 3 5.4 Procedure Following the initial canvassing and recruitment phase, those interested in participation took part in an information session early in August 2014. In this session they were provided with energy saving tips, presented by Dr. Sara Alsaadani, and introduced to the conservation program goal of saving 10% of energy use throughout the building (see Appendix D). Additionally, led by Samira Zare Mohazabieh, they were asked to complete the New Environmental Paradigm (NEP) questionnaire and demographics survey (see Appendix F), which can be used to provide further insights into user personas through their attitudes toward the environment. 5.4 Procedure 5.4 Procedure Following the initial canvassing and recruitment phase, those interested in participation took part in an information session early in August 2014. In this session they were provided with energy saving tips, presented by Dr. Sara Alsaadani, and introduced to the conservation program goal of saving 10% of energy use throughout the building (see Appendix D). Additionally, led by Samira Zare Mohazabieh, they were asked to complete the New Environmental Paradigm (NEP) questionnaire and demographics survey (see Appendix F), which can be used to provide further insights into user personas through their attitudes toward the environment. The results of this survey were used first, as part of a separate study to advance a predictive model of energy conservation, and second, as a potential covariate for this study. For their attendance and completion of the survey, they were remunerated $20. In total, there were 50 participants in this portion of the study. 60 60 The same participants in attendance were then introduced to the energy conservation study involving feedback. This presentation (see Appendix E) was delivered by the author and served as part of the recruitment. They were informed about the level of commitment required, should they be interested in participating in this portion of the study, and the hardware that would be installed in their suites. Additionally, they were also asked to commit, in writing, to saving the 10% (see Appendix G). For their participation in this portion of the study, they were informed that they would receive an Android tablet with a high-speed internet connection for the duration of the study as remuneration. They were also informed that the tablet would be used to deliver the feedback information but could also be used for personal purposes (e.g., games, internet surfing). However, caution was given not to transmit sensitive information. We intended to allow participants to keep the tablets at the completion of the yearlong study; however, to avoid deliberate drop-outs, they were not informed of this at the on-set of the study. Collectively, they were walked through the basic feedback dashboard (i.e., the version without the real- time social norming) and the thermal comfort survey (see Appendix H). Following the information session, hardware installations were scheduled with participants in the following two weeks. 5.4 Procedure This was conducted alongside a basic energy audit of electrical appliances (see Appendix I) to help further understand energy use and potential areas for savings within the suites. In total, the hardware and installation and energy audit took on average 40 minutes to complete. Before proceeding with installation of any hardware, tenants signed an informed consent form – see Appendix C. C. Prior to the distribution of tablets, two days of data had been collected, to ensure proper installation and to provide data for the first contact with the feedback. Participants were randomly grouped into one of the two experimental feedback conditions. Regardless of their experimental condition, participants were given a one-on-one walkthrough and tutorial. As part of the walkthrough, their FCU and a readily available appliance (e.g., floor lamp, or oven) were power cycled to show the impact of its power use on the display. This was followed by a basic hands-on quiz intended to ensure they understood the information being displayed and how to navigate through the app. Participants were reminded of the overall 10% savings goal for the program and how the feedback dashboards showed how much kWh per month that goal meant for them given their own historical energy use from the year prior to the study. Prior to the distribution of tablets, two days of data had been collected, to ensure proper installation and to provide data for the first contact with the feedback. Participants were randomly grouped into one of the two experimental feedback conditions. Regardless of their experimental condition, participants were given a one-on-one walkthrough and tutorial. As part of the walkthrough, their FCU participants were given a one-on-one walkthrough and tutorial. As part of the walkthrough, their FCU and a readily available appliance (e.g., floor lamp, or oven) were power cycled to show the impact of its power use on the display. This was followed by a basic hands-on quiz intended to ensure they understood the information being displayed and how to navigate through the app. Participants were reminded of the overall 10% savings goal for the program and how the feedback dashboards showed how much kWh per month that goal meant for them given their own historical energy use from the year prior to the study. Participants in the feedback + social norming condition had the additional comparative information explained. 5.4 Procedure However, to avoid adding external motivation, these participants were informed that such 61 information and comparisons were for their knowledge only, and that being better than average was not a program objective. At the onset of the study, all participants were informed their level of engagement with the app would be tracked by the research team. It was also recommended to them that they check their dashboards daily and that their participation required them to check at least weekly. Similarly, they were asked to fill the thermal comfort survey on a weekly basis. A software reminder on the tablet would notify them when the survey should be filled. If they had any questions, they could always call or email the author or another member of the research team. Also, any specific concerns could be conveyed through the open- ended question at the end of their thermal comfort survey. They were also informed that check-ins would happen approximately every two months by the author to ensure proper functioning of equipment and to answer any questions or concerns about the study. Participants were reminded that their participation was voluntary and that they may withdraw from the study at any time. The feedback portion of the study ran from September 2, 2014 through August 31, 2015. At the conclusion of the study, participants were asked to complete an exit survey (see Appendix K) to get their opinions on the usefulness of the dashboards and their experience in the study. 5.5 Equipment and Installation 5.5 Equipment and Installation The system architecture implemented for this study (see Figure 5-1) was a modification from the general FBRP architecture shown in Figure 3-5. Each suite was fitted with the following components: The system architecture implemented for this study (see Figure 5-1) was a modification from the general FBRP architecture shown in Figure 3-5. Each suite was fitted with the following components: - A battery-powered emonTXv3 installed inside the FCU to measure its fan power draw and output temperature measured at the top diffusing grate of the unit. Data were sampled every 10 seconds. - A battery-powered emonTH installed in the “neck” of the apartment where the corridor opens up to the main living space. - A Raspberry Pi gateway to relay the data collected for the suite to the content management system on-line. Additionally, Android Tablets (ASUS MemoPad 7 HD with Jellybean 4.2) were given to each participant for three purposes: to view their own energy feedback dashboards, to complete in-situ thermal comfort surveys on a weekly basis, and for general internet browsing as part of the compensation for their participation. Figure 5-2 offers a rich-picture illustration of these devices in the context of an empty suite. Additionally, Android Tablets (ASUS MemoPad 7 HD with Jellybean 4.2) were given to each participant for three purposes: to view their own energy feedback dashboards, to complete in-situ thermal comfort surveys on a weekly basis, and for general internet browsing as part of the compensation for their participation. Figure 5-2 offers a rich-picture illustration of these devices in the context of an empty suite. 62 62 Figure 5-1: Architecture for the system installed at Phoenix Place Figure 5-1: Architecture for the system installed at Phoenix Place Figure 5-2: Rich-picture diagram of feedback hardware Figure 5-2: Rich-picture diagram of feedback hardware 63 The Emoncms CMS was installed on a private virtual server on the internet to manage data for the study. Data from all sensors in the study were stored on the same account and database to afford centralized data management. Weather data was pulled in from the Weather Underground service to the same database. Finally, tailored feedback dashboards were created for each user using the CMS. In addition to the hardware installed in suites, there were components installed in the main hallway corridors and in electrical cabinets. emonTHs were installed in the hallway corridor of each floor. 5.5 Equipment and Installation Inside the electrical cabinets, emonTXv3s were installed to measure up to two suites’ total energy use (i.e., with each suite requiring two 120V lines). Raspberry Pi gateways were installed in the same electrical cabinets to relay all data collected from sensors in these spaces. A building-wide internet connection was provided to allow all sensor data to be forwarded to the online CMS and for data to be downloaded to tablets. To enable this building-wide internet connection, a series of wired and wireless Wi-Fi repeaters from OpenMesh (2014) were utilized. Access to this network was restricted via MAC address to only the Raspberry Pis and tablets associated with this study. Transfer speeds were throttled to meet minimum data transfer requirements for the purposes of this study while ensuring equal and maximum benefit of the shared internet to all participants. With any Internet of Things application, data privacy is a major concern. In this study, data privacy was handled through data confidentiality at all times and data security when possible. While data being transmitted over radio was not encrypted it was specially coded without linking the data to a specific participant; thus, keeping the data anonymous to those outside the research team. Due to the nature of the data collection scheme, it was necessary for all data to be collected on a single user account (an account only the researchers have full access to through password protection). Similarly data collected from Google Analytics and OpenDataKit are password protected. Through the Android application, participants only had access to their own dashboard (through a special 10-digit code assigned to them by the researchers). Even if they became aware of another dashboard code through trial-and-error or other means, the dashboard itself would not identify the participant to which it belongs; thus, maintaining anonymity. While it is impossible to provide full data security, the use of passwords and coding schemes were deemed adequate to protect the participants’ privacy. maintaining anonymity. While it is impossible to provide full data security, the use of passwords and coding schemes were deemed adequate to protect the participants’ privacy. 64 Measures 5.6 The primary dependent measure of interest is the total suite percentage energy use difference between the study period and the year prior. To ensure consistency between historical and study measures, the building’s sub-metering system, which was installed by Intellimeter (http://intellimeter.on.ca/) and Measurement Canada certified, was used for this purpose. 5.5 Equipment and Installation To form a stronger basis for determining energy use savings or increases, all energy use data was weather-normalized using climate data obtained for Toronto since 1978 from the Government of Canada (http://climate.weather.gc.ca). For statistical analyses, savings would be measured at the individual level. However, for overall program performance, aggregate savings percentage would be calculated for each experimental condition and from the entire study population. There are several potential covariates that will be examined to understand how they modify the dependent variable: There are several potential covariates that will be examined to understand how they modify the dependent variable: - Environmental attitudes using NEP scores, - Engagement (via dashboard page view statistics), and - Engagement (via dashboard page view statistics), and - Pre-study average daily energy use (Examining pre-use as a covariate acknowledges its potential impact on absolute savings.) 5.7 Thermal Comfort Measures Fan coil energy usage and related thermal comfort data (from surveys, and temperature readings), while important for platform demonstration purposes, were not the primary focus for this experiment. Thus, the analysis of this data is outside the scope of this thesis. However, this sub-chapter offers how that data might be leveraged. To capture thermal comfort related behaviours and strategies, the following could be used: To capture thermal comfort related behaviours and strategies, the following could be used: - FCU average daily energy use, - FCU thermostat set points (see Appendix H, Thermal Comfort Survey page 3), - Clothing levels (see Appendix H, Thermal Comfort Survey page 4), and - Alternative behaviours (see Appendix H, Thermal Comfort Survey page 5). The ASHRAE Predicted Mean Vote (PMV) model for thermal comfort utilizes six parameters that were directly measured or estimated in this study: 65 - Self-reported comfort level (see Appendix H, Thermal Comfort Survey page 2), - Ambient temperature from emonTH device, - Ambient relative humidity from emonTH device, - Clothing levels (see Appendix H, Thermal Comfort Survey page 4), Assumed MET value of seated position (see Appendix H, Thermal Comfort Survey page 1) - Assumed MET value of seated position (see Appendix H, Thermal Comfort Survey page 1), - Estimated average air speeds from FCU usage from the manufacturer, and p ( pp y p g ) - Estimated average air speeds from FCU usage from the manufacturer, and - Estimated average air speeds from FCU usage from the manufacturer, and - Assumed radiant temperature from indoor temperature readings. These parameters could help provide a comprehensive picture of thermal comfort and its impact on FCU and total suite energy use. 66 6. Results and Discussion 6.1 System performance Outside of a server issue from October 20 – November 14, 2014, the system was up approximately 90% of the time. The 10% downtime was attributable to a combination of internet outages, building-wide power outages, sensor battery outages, and wifi and wireless connection drops between the sensors, gateways and router. The sensors were pre-calibrated to within +/- 5% of a ‘Watts Up? Pro’ power meter. However, in the field monthly aggregated energy measurements were within -8 to +18% of the Intellimeter readings with an average measurement of approximately +6%. This discrepancy was likely in large part due to the limitation of the platform’s power readings which were deduced using a fixed voltage of 120V, not taking into account voltage drops in the building. While these error figures are not ideal, as they hamper trust and confidence in the system, they were reasonable for the purposes of the pilot. It is worth noting that this discrepancy impacted just the feedback delivered to the participants and not the data that was used for analysis later in this chapter, which was gathered from the Intellimeter readings for year-over-year consistency. 6.2 Participant noise and variability At the onset of the study, it was discovered that two of the 28 tenants recruited had not been tenants at Phoenix place for a full year prior to the study. While historical comparisons were provided from data gathered for previous tenants living in these suites, it was determined that their data was ineligible for use in the analyses. Nonetheless, their feedback on the platform was taken. Data from another two participants were deemed unacceptable and were effectively removed from the analysis due to extended periods of abnormal energy usage; one participant had a life partner co- occupying the suite at the study’s onset and another was using an unsafe personal space heater, which the property manager had disallowed halfway through the study. In general, however, noise factors 67 including vacations, and temporary changes in living arrangements with family members, and significant others were not specially treated despite the temptation to omit periods of known vacation, for example. The reasoning is that such life factors are bound to take place in any field study of such duration. Furthermore, while these events were noted for the study year, they may have as likely happened the year prior. Of the 28 participants recruited, data from only 24 were considered as part of the quantitative analysis. This included data from two other participants who had moved out of the building 9 months into the study; their data set was truncated at that point. 6.3 Conservation Program: Energy Savings As mentioned earlier, one objective of the conservation program was to achieve an overall 10% in energy savings year-over-year. Naturally, this program-wide objective cascaded to individual tenants, who were asked to save 10% of their own year-over-year energy use. Figure 6-1 illustrates the findings looking at percentage savings of actual group-aggregated kWh use and normalized group-aggregated kWh use across the three groups of participants. The average actual savings percentage between the two feedback groups was 10.8% compared to an increased use of 4.4% for those outside the study. This led to a net delta of 15.2% in relative savings. Similarly for normalized savings percentage, the average for those with feedback was 8.4% compared to an increase of 5.1% for those outside the study for a net delta of 13.5% in relative savings. It appears that the program was successful in surpassing the 10% savings target. 68 Feedback Condition Basic Basic + Social No Feedback Year-Over-Year Savings % -6 -4 -2 0 2 4 6 8 10 12 14 Actual Normalized Figure 6-1: Aggregated year-over-year savings by feedback condition Figure 6-1: Aggregated year-over-year savings by feedback condition In addition to the conservation program objectives, this thesis also sought to test whether the savings would be statistically significant and thus reliable; and whether providing real-time social comparisons would achieve improved savings in a similarly reliable fashion. The results shown in Figure 6-1 would suggest that there was a small normalized savings improvement between the feedback conditions of approximately 9.4% and 7.3% in favor of having real-time social comparisons. The next chapter describes the results of the hypothesis testing. 6.4 Test of Hypothesis 1 – The Effect of the Conservation Program As introduced in Chapter 5.1: 6.4 Test of Hypothesis 1 – The Effect of the Conservation Program As introduced in Chapter 5.1: 6.4 Test of Hypothesis 1 – The Effect of the Conservation Program As introduced in Chapter 5.1: Hypothesis 1: The conservation program comprised of an information campaign, participant commitment, and real-time feedback promotes energy conservation. Hypothesis 1: The conservation program comprised of an information campaign, participant commitment, and real-time feedback promotes energy conservation. To test Hypothesis 1, a 2-level (participation type: feedback, no feedback) between subjects ANCOVA was run for the weather-normalized, annual savings percentage dependent variable. Individual participant’s energy use (in kWh) for the year prior to the study was entered as a covariate. To test Hypothesis 1, a 2-level (participation type: feedback, no feedback) between subjects ANCOVA was run for the weather-normalized, annual savings percentage dependent variable. Individual participant’s energy use (in kWh) for the year prior to the study was entered as a covariate. 69 There was a significant difference in savings between participation type (F(1,128)=3.938, p=.049). There was a significant difference in savings between participation type (F(1,128)=3.938, p=.049). There was a significant difference in savings between participation type (F(1,128)=3.938, p=.049). Figures 6-2 and 6-3 illustrate the effect; those who participated and received feedback saved 8.4% on average, whereas those without feedback used 5.1% more for a 13.5% difference between the groups. Note that the group averages in these figures are calculated by averaging each participant’s savings percentages (kWhs reduced compared to kWh used the previous year), whereas the group aggregated figures from Figure 6-1 represent savings percentages calculated based on the entire participant group’s combined kWh savings. There was also a significant effect for the Baseline energy use covariate (F(1,128)=5.085, p=.026). This indicated that the higher the baseline energy use, the more savings potential there was concurring with Allcott’s (2011) finding with Opower home energy reports. Experimental Condition No Feedback Feedback Savings % -15 -10 -5 0 5 10 15 20 Figure 6-2: Error bar graph of experimental condition on savings percentage Note: Error bar graphs represent 95% confidence intervals Figure 6-2: Error bar graph of experimental condition on savings percentage Note: Error bar graphs represent 95% confidence intervals 70 Experimental Condition No Feedback Feedback Savings % -150 -100 -50 0 50 100 Figure 6-3: Box plot of experimental conditions on savings percentage Figure 6-3: Box plot of experimental conditions on savings percentage This result is encouraging and provides further evidence that the conservation program surpassed its 10% savings target. 6.4 Test of Hypothesis 1 – The Effect of the Conservation Program As introduced in Chapter 5.1: Furthermore, when enlarging the perspective of the field study, this is a very encouraging result for a couple reasons. First, the participants were not financially motivated to save since their monthly rent would be flat regardless of their performance. Second, many of the participants could be considered low power users with a baseline from which there was very little excess to trim. In a study with participants from a broader sample of home owners in townhomes, semi- or detached homes, it could be reasonably expected that such participants would save more. However, this test does not allow us to make any conclusive statements on the efficacy of feedback since the experimental condition was additionally comprised of an information campaign, a personal commitment to save 10%, and an energy audit. Additionally, an overall limitation of the field study was that the self-selection bias makes it difficult to discern whether energy savings are attributable to the participant’s characteristics or due to their reaction towards the feedback and conservation program. Unfortunately, a randomized control trial, where we might have recruited twice as many participants and randomly denied half, was not possible given the small overall study population. 71 Test of Hypothesis 2 – The Effect of Social Comparisons 6.5 Compared to Hypothesis 1, the experimental design was more deliberately intended for the testing of Hypothesis 2 with the type of feedback provided as the sole controlled difference between the two groups. All participants who received feedback, observed the information campaign, pledged to save 10%, and had an energy audit. To recap: Compared to Hypothesis 1, the experimental design was more deliberately intended for the testing of Hypothesis 2 with the type of feedback provided as the sole controlled difference between the two groups. All participants who received feedback, observed the information campaign, pledged to save 10%, and had an energy audit. To recap: Hypothesis 2: Real-time feedback with social comparisons promotes more energy conservation than with just real-time feedback alone for total home energy use. To test Hypothesis 2, a 2-level (feedback type: basic, basic+social comparisons) between subjects ANCOVA was run for the weather-normalized annual savings percentage dependent variable. NEP scores (a proxy for environmental attitudes), page views (a proxy for engagement), and pre-study energy use were entered as subject level covariates. 6.4 Test of Hypothesis 1 – The Effect of the Conservation Program As introduced in Chapter 5.1: There were no significant findings on normalized savings percentage for NEP scores (F(1,19)=.485, p=.50, n.s.) , page views (F(1,19)=.568, p=.46, n.s.), pre-study energy use (F(1,19)=.094, p=.76, n.s.) or feedback (F(1,21)=.114, p=.74, n.s.). Thus, hypothesis 2 was rejected. Figure 6-4 illustrates this non-significant effect. Feedback Condition Basic Feedback Basic + Social Savings % -10 -5 0 5 10 15 20 25 30 Figure 6-4: Error bar graph of feedback condition on savings % Note: Non-significant effect. Error bars represent 95% confidence intervals Feedback Condition Basic Feedback Basic + Social Savings % -10 -5 0 5 10 15 20 25 30 Figure 6-4: Error bar graph of feedback condition on savings % Note: Non-significant effect. Error bars represent 95% confidence intervals 72 This non-significant effect is not unexpected given the wide variability of energy savings observed year- over-year, the relatively small difference in savings between the two feedback conditions of 3.5% (i.e., 6.6% vs 10.1%) and the relatively small sample size for each feedback condition. Given the effect size between the two conditions is rcontrast = .072, a power analysis (using an alpha = .05, beta = .8) suggests that a study sample size of 1,516 participants would have been required to obtain significant results. While a sample of this size may not be feasible for a pilot study, at a utility scale, this may be realistic. The trends shown in these results may warrant future consideration on that front. This non-significant effect is not unexpected given the wide variability of energy savings observed year- over-year, the relatively small difference in savings between the two feedback conditions of 3.5% (i.e., 6.6% vs 10.1%) and the relatively small sample size for each feedback condition. Given the effect size between the two conditions is rcontrast = .072, a power analysis (using an alpha = .05, beta = .8) suggests that a study sample size of 1,516 participants would have been required to obtain significant results. While a sample of this size may not be feasible for a pilot study, at a utility scale, this may be realistic. The trends shown in these results may warrant future consideration on that front. Interestingly, this improvement in savings was in line with findings from several Opower studies (Allcott, 2011) leveraging their home energy reports as part of large scale utility projects. 6.4 Test of Hypothesis 1 – The Effect of the Conservation Program As introduced in Chapter 5.1: However, there are two key differences to consider between those utility projects and the current field study. The first difference is in the intervention design. Homeowners either received Opower’s home energy reports (with social comparisons data and energy saving tips) or nothing; whereas the current field study compared feedback dashboards differing only in the availability of social comparisons data. Because the difference in treatment conditions in the current field study and analysis were smaller (basic feedback data vs basic feedback with social comparisons), it could be expected that the social comparisons data design had more net impact than Opower’s solution. The second difference deals with the feedback delivery mechanism and frequency. Opower’s study was essentially a paper-based report delivered either monthly, bi-monthly, or quarterly. In the current study, the feedback was delivered electronically and in near real-time. However, one would anticipate that real-time feedback should be more effective than less frequent home energy reports (Darby, 2006). As a more fair basis for comparison between the current field study and Allcott’s findings, we might compare savings between the basic+social feedback treatment group (n=12) and the control group (n=104). From Figure 6-1, the difference between average treatment savings was 14.5% (i.e., 9.4% vs - 5.1% in normalized savings). Compared to Allcott’s findings of savings averaging 2%, the large improvement here is most likely attributable to the delivery of social comparisons feedback in real-time. As a reminder, the treatment group in the current field study was, however, also exposed to the information campaign, commitment, and energy audit interventions. As also mentioned, there may have been self-selection bias in play. Finally, it should be noted that the Opower studies were conducted at the utility scale with tens of thousands of participants, whereas the current field study had a sample of just 12 with social comparison data. Nonetheless, this finding warrants further work exploring the 73 73 efficacy of real-time social comparisons. As suggested above, one way forward would be to pursue the solution at a larger scale. efficacy of real-time social comparisons. As suggested above, one way forward would be to pursue the solution at a larger scale. 6.5 Exploratory Analysis of Engagement At the onset of the hypothesis testing, it was important to assess whether year-over-year savings were sufficient to capture trends. In addition to year-over-year analyses, all results were binned by season – Fall, Winter, Spring, and Summer. Analyses of covariance (ANCOVAs) were run with seasons as within- subject variables; however, it was determined that the results were not necessarily more insightful than year-over-year analyses – so only year-over-year results were reported earlier in Chapter 6. The one exception was the noticeable drop in dashboard views beyond the Fall months, which applied to both feedback groups as shown in Figure 6-5. Interestingly, despite this trend, there was not a significant change in seasonal savings. This suggests that while feedback was of more interest for the first few months, the benefit may persist despite lower engagement. Figure 6-5: Savings and Engagement by Quarter 0 10 20 30 40 50 60 70 0 2 4 6 8 10 12 14 16 18 20 Fall Winter Spring Summer Actual Energy Savings % Basic Feedback Savings % Engagement 0 10 20 30 40 50 60 70 0 2 4 6 8 10 12 14 16 18 20 Fall Winter Spring Summer Dashboard views / Month Basic + Social Feedback Savings % Engagement Figure 6-5: Savings and Engagement by Quarter Interestingly, the levels of engagement appeared to be higher for the Basic + Social feedback group. However this difference was not statistically significant due to the wide variability in page views across all users and the small sample sizes in each group. However, the trend in higher engagement levels 74 among basic + social feedback participants may warrant further investigation on whether having social comparison data improves overall user interest and usage experience. among basic + social feedback participants may warrant further investigation on whether having social comparison data improves overall user interest and usage experience. 6.6 Exit Survey Results 6.6 Exit Survey Results Of the 24 eligible participants, only eight returned their exit surveys (See Appendix K). Thus, rather that attempting inferential statistics on the dataset, it was deemed reasonable to simply provide qualitative insight as to what worked and what did not. Not surprisingly, many of the respondents were also of the most engaged in the study judging by their number of dashboard views. Overall (Questions 1 and 2), respondents commented that the dashboards raised their awareness of their energy use. 6.5 Exploratory Analysis of Engagement Additionally, one respondent stated that “[It] was a good experience and I enjoyed the competition with my neighbors”. Another user liked “[being] able to compare my usage to what I thought I was using”. This suggested that the comparative elements were engaging. However, other users seemed to find it difficult to reach their targets as suggested by the following comments: “It’s not easy to change lifestyle to save energy”, and “The target limit was not appropriate”. On the negative front (Questions 3 and 6), users commented about being frustrated at times with slow or inconsistent internet connection. One user also commented on “a constant feeling the readings were incorrect”. Taken together, these comments would suggest that there is work to be done to improve the reliability of the system. For the dashboard widgets (Question 4), users with basic feedback found the “LAST 7 DAYS” charts and the historical and daily target lines to be the most useful. Interestingly, those with social comparisons valued the “TODAY” widget and seeing their neighbors’ usage most. The “RIGHT NOW” and “HOW YOU’RE DOING” widgets were found least useful overall. The daily tips appeared to be most polarizing amongst respondents – they appeared to be either liked the most or least. As for top savings strategies (Question 9), comments appear to fall into either cooking related strategies (e.g., cooking for multiple days, or reheating with the microwave) or being more diligent about turning appliances off when not in use. Reassuringly, all respondents were able to estimate their savings to within one neighboring 10% bin on the survey (Question 8). This included a participant, who had used 33.6% more than the previous year, who had estimated he had used 10-20% more. 75 Overall, the results of the exit survey suggest that the feedback dashboard was useful and achieved the intended effect of raising awareness, and motivating users to save energy. This observation corroborated the statistical results presented earlier in the chapter, which demonstrated the efficacy of the feedback intervention. However, it should be noted that there is bias in these survey results due to only having eight responses from some of the most engaged users in the study. Overall, the results of the exit survey suggest that the feedback dashboard was useful and achieved the intended effect of raising awareness, and motivating users to save energy. 6.5 Exploratory Analysis of Engagement This observation Overall, the results of the exit survey suggest that the feedback dashboard was useful and achieved the intended effect of raising awareness, and motivating users to save energy. This observation corroborated the statistical results presented earlier in the chapter, which demonstrated the efficacy of the feedback intervention. However, it should be noted that there is bias in these survey results due to only having eight responses from some of the most engaged users in the study. 76 7. Conclusions 7. Conclusions The purpose of this thesis was to design, develop, and demonstrate a feedback research platform (FBRP) to afford a systematic approach to evaluating feedback designs. The implementation of this platform leveraged heavily on the advancement of Internet of Things (IoT) and free and open source software (FOSS). The FOSS-IoT-based platform developed in this thesis was tailored to demonstrate three key features: disaggregated feedback, real-time social comparisons, and in-situ surveys to help understand user behaviours. Evaluating the efficacy of real-time social comparisons – something that, to the author’s knowledge, has not been evaluated in conjunction with feedback – was the analytical focus of this thesis. Feedback interventions should not and do not exist in a vacuum. For this reason, the energy conservation program presented in this thesis employed several interventions in addition to feedback, including an information campaign and participant pledges to save 10% towards a collective 10% savings for the program. The program was framed in a Community-Based Social Marketing (CBSM) program implemented at Phoenix Place, an affordable housing project in Toronto, Canada comprised of 136 near- identical bachelor suites. The conservation program was also designed as a field study to examine the efficacy of two feedback designs. In total, 28 participants were recruited to receive a feedback condition; however, only 24 of those were deemed eligible for the statistical analysis. The results showed a statistical significant effect of the conservation program with a relative year-over-year, weather-normalized savings of approximately 11%, surpassing the goal of 10%. While there was a 3.5% difference in savings favoring an enhanced feedback with social comparisons (vs basic feedback), this was not statistically significant. The non-significant findings were not unexpected as the sample sizes in the study were of a pilot scale rather than a utility-wide implementation across hundreds or thousands of customers. By evaluating the efficacy of social comparisons feedback from the current field in a similar fashion as Opower’s home energy reports were evaluated in Allcott’s (2011) work, there was a 12.5% discrepancy 77 (i.e., 14.5% savings improvement vs 2%). The improvement in the current field study may be largely attributable to the delivery of the social comparisons data in real-time. However, this finding is unclear due to confounding factors from the study design and limited sample size. 7. Conclusions While the benefit of real-time social comparisons is unclear, it would be prudent to still ask the question: Would such a feedback strategy be worth the cost? The cost side of this questions is less of an unknown and, on a superficial inspection at least, may not be too large for utilities who already have existing smart meter infrastructure in place. Through the government organized GreenButton initiative (http://www.greenbuttondata.org/), much of this data is accessible. The main component missing is software comparison algorithms. Ensuring a fair social comparison would be perhaps the most difficult, but far from impossible, challenge. While atypical, the benefit of conducting the present study at Phoenix Place was that such complexities were circumvented by virtue of the homogenous nature of the suites and tenant population. Other than requiring the design and processing of software algorithms and visual feedback design, there would be very little added technical cost to such a system. However, the timeliness and frequency of the feedback is still an open question worth answering. To the author’s knowledge, GreenButton data is still laggy by a full day. 7.1 Contributions 7.1 Contributions There are three main contributions from this work; they are detailed in order of significance. There are three main contributions from this work; they are detailed in order of significan First, a feedback research platform was developed with intentions to release the specifications and source to the open source community. It will be made available here: First, a feedback research platform was developed with intentions to release the specifications and source to the open source community. It will be made available here: source to the open source community. It will be made available here: https://github.com/kevinci29/fbrp/. To the author’s knowledge, this is the first time such a platform would be openly released in this fashion for the wide public benefit. By releasing the specifications and source for the platform, the author hopes to build a community of researchers who can more easily build off each other’s work to discover more effective feedback designs. The platform can help form a common methodological approach to delivering feedback and especially real-time feedback. Being widely and freely available, it should also help clarify the specifications of feedback designs. This will help results be more comparable and reproducible. Through an Internet of Things approach this platform enables some key benefits that have been difficult to produce in the past. For example, it supports disaggregated feedback (i.e., multiple appliances around 78 78 the home) as well as real-time social comparisons as was demonstrated in this study. Furthermore, in- situ surveys allows researchers to better understand energy related behaviours in the home. The second contribution of this work naturally extends the first contribution by demonstrating the platform in action. The implementation of the platform at Phoenix Place in Toronto as part of a field study also allowed the research team to develop a program of research to easily examine the efficacy of a variety of feedback designs and techniques. Additionally, it allowed for the exploration of thermal comfort on energy use. Furthermore, through the literature review, it was identified that there has been a dearth of studies focussing on MURBs. Through this research the author hopes to have demonstrated why they can be quite beneficial for progressing research from lab studies to field studies. Last but not least, this research demonstrated a novel approach to feedback design that leveraged the advantages of the research platform and the MURBs context at Phoenix Place. 7.1 Contributions By implementing real- time social comparisons, this thesis demonstrated how such an approach can help further motivate energy conservation beyond levels shown in past studies. The author has argued that, while the measurable benefit of real-time social comparisons may still be unclear, the cost of implementing it at a wide scale is probably reasonable given existing metering infrastructure. The design of the dashboards in this thesis followed an iterative design cycle. Designs were first informed by theory (i.e., using design heuristics proposed by Trinh and Jamieson (2014), then by usability testing, and now field testing, the author hopes to raise the standard to which feedback designs are rigorously specified and thusly advanced. Such an approach is not unique in the Human Factors or Human Computer Interaction communities, but historically has been lacking in the energy feedback realm. Conversely, perhaps those communities may also benefit from post occupancy evaluations that are often conducted by building engineering firms; and one of which was leveraged for this study. Future Work 7.2 While the formal study at Phoenix Place has concluded, the sub-metering system is still in place. With permission, the research team will be pursuing follow-up analyses with that data to measure the persistence of savings from both the conservation program and feedback implementation. This will provide insight into the necessity for continual conservation interventions. 79 While the feedback research platform was configured to capture data to help understand the impact of thermal comfort on energy use, it was out scope in this thesis to fully explore that area. Analyses should be conducted to assess perceived thermal comfort from the in-situ surveys as well as measured energy draw from appliances and, in particular, the FCUs. Furthermore, engagement data was collected on the FCU dashboards which can provide insight on the efficacy of disaggregated feedback information. To date, the feedback research platform was configured with a subset of features but it may be reconfigured to meet different applications and research questions. Another benefit of the platform is that it can be scaled for use in a single family home, or for multiple homes in a MURB. Larger implementations are possible with enough server processing and bandwidth. However, the feature set of the platform is rather limited when considering the foreseeable growth areas possible as will be discussed in this chapter. There are several key features worth pursuing in future versions of the platform. 7.1 Contributions Firstly, in the short term, it is important to ensure a higher rate of feedback up-time and feedback accuracy. In the study, there was an estimated a 90% uptime. However, it is not unusual to see industry strive for 99.99% uptime (approximately 1 hour of downtime per year). Improved uptime might be achieved through land-powered sensors vs relying on battery power. Providing a tighter mesh of wifi-repeaters may also help ensure sensor readings are not lost. Also the impact of downtime may be thwarted if sensors could locally store data when there are server connection issues and post them when reconnected. Upgrading power meters to account for voltage measurements would help reduce the discrepancy between official sub-metering systems. More effort in calibrating sensors in the field is also warranted. Such changes would help promote trust and confidence in the system. Second, there may be benefit in fully integrating the Open Data Kit survey tool. Having survey responses in the same database as sensor data affords a more adaptive feedback approach. For example, in the context of the current field study with the thermal comfort survey data collected, it is possible show, in the form of a recommendation, the most popular strategies others had used to achieve their thermal comfort. Or to show empathy, one could show how neighbors were experiencing similar levels of thermal discomfort. To help users understand their energy consumption better, feedback can be framed in terms of how FCU energy use correlates negatively with their thermal comfort. This list is not intended to be exhaustive as there are likely many more possibilities when such data becomes available for feedback. 80 Third, push notifications may be used to alert or inform tenants when key thresholds have been crossed. For example, to warn a tenant when 90% of a daily energy use quota has been reached, or to provide acknowledgement when a monthly savings goal has been achieved. Notifications may also be used to prompt users to perform specific tasks. For example, if the weather forecast indicates a cool day is ahead, a notification can be used to recommend users open the window and turn off their fans or air conditioners. Fourth, another feature would be to incorporate time of use (TOU) pricing. This can be an important feature because lower energy use does not necessarily equate to lower energy costs. 7.1 Contributions Furthermore, it is well known that managing peak demands by shifting energy use to lower peak times can save utilities and the general public billions of dollars in infrastructure costs. TOU optimization algorithms can help tenants save energy and money by recommending or even automating the shifting of high intensity appliances. TOU Services from companies like Bidgely (www.bidgely.com) may enable such a feature. Fifth, another popular trend in the HEMS industry are the integration of controls and automation functionality. At present, the feedback research platform serves mainly a monitoring function, relying on the tenant to manually actuate changes within the environment to reach conservation goals. By implementing controls in the system users can, with a simple click of a button on any internet-enabled device, turn the lights off or program the dishwasher to start when energy costs are lower. At present, the SafePlugs have built into their API the ability to control the power flow to such appliances. Setting a schedule to un-power devices, especially those known to have large phantom loads, when not in use can save more than half of their total daily consumption (Fung, Aulenback, Ferguson, & Ugursal, 2003). In the same vein, programmable thermostats have aimed to save energy by allowing tenants to keep HVAC systems use to a minimum when they are away. Rather than keeping the human as an essential part of the feedback-control loop, automation approaches relieve users from their manual tasks and place them instead in a supervisory role. Such strategies may be built and tested on this platform. Sixth and finally, another emerging trend in the HEMS industry is the shift towards managing micro- generation from solar and wind power generation. The Open Energy Monitor platform was premised on enabling the monitoring of such generation capabilities. Conceptually, given the ideas presented above (i.e., for push notifications, time of use pricing, controls and automation) that one could shape the platform to one which ensures that a home operates on net-zero energy; that is, a home that uses only as much energy as it generates. Such a system could enable homes to be sustained off the grid. Many other technologies would obviously be required to turn this vision into a reality – for example, large 81 batteries like the Tesla Power Wall (http://www.teslamotors.com/powerwall) and passive housing designs come to mind. Perhaps with small steps such a vision can be realized. 7.1 Contributions In combination with demonstrations like the one presented in this thesis, the author hopes to have shown how a sustainable future is closer than we might have imagined. 82 Appendix A – Recruitment Poster 83 84 4. Risks or discomfort: There is very little risk involved in participating in this study. You may be concerned that someone else may find out your responses to the questions. Please note that there is no right or wrong answer to the questions, we are seeking your individual opinions to each of the statements in the questionnaire. Please note that we will not be collecting any names and will not publish information in any reports that will identify you by unit number or by any other kind of personal information. When we publish reports from this research project, we will be using only general information, not individual information and your confidentiality will be protected. 2. Purpose of the Study: The purpose of this study is to promote energy literacy and conservation, and to gain an understanding of whether there is a relationship between energy-conscious attitudes and energy consumption. 3. Description of the study: Participation in the study entails completing the attached survey. This survey consists of a fifteen questions documenting your opinions and perceptions about energy use and the environment, as well as a few short questions about yourself. Completion of this survey should take no more than 10 minutes. 84 84 Appendix B – Conservation Program Consent Form 85 6. Benefits of the study: The following are potential benefits of the research:  To engage and educate tenants about environmental issues.  To promote a community and teamwork spirit.  To gain access to valuable information about energy-conscious attitudes, and whether they have an impact on energy consumption.  To gain access to valuable information about energy-conscious attitudes, and whether they have an impact on energy consumption. While this project promises benefit for social good, individual benefit to any of the tenants hile this project promises benefit for social good, individual benefit to any of the tenants can 1. Investigators:  Prof. Alan Fung, Associate Professor, Department of Mechanical and Industrial Engineering.  Prof. Alan Fung, Associate Professor, Department of Mechanical and Industria  Prof. Vera Straka, Associate Professor, Department of Architectural Science.  Dr. Sara Alsaadani, Post-Doctoral Fellow, Department of Architectural Science.  Kevin Trinh, Graduate student supervised by Prof. Vera Straka and Prof. Alan Fung.  Samira Zare Mohazabieh, Graduate student supervised by Prof. Vera Straka and Pro Research Study: Tenant Engagement and Energy Conservation, Toronto You are being asked to participate in a research study. Before you give your consent to participate, it is important that you read the following information and ask as many questions as necessary to be sure you understand what you will be asked to do and the degree of your involvement. 9. Compensation: You will be compensated with $20 for your time and participation. 7. Confidentiality: All data collected will be handled confidentially. We will not be collecting any names. Unit numbers will be collected to enable us to link energy consumption data, data from the thermal comfort survey and data from the attitude survey, and to enable 86 us to provide you with feedback about your individual energy consumption. We will not publish unit numbers or specific information about individuals in any publication or report. us to provide you with feedback about your individual energy consumption. We will not publish unit numbers or specific information about individuals in any publication or report. 8. Voluntary nature of participation: Participation in this study is voluntary. Participation in the study is completely voluntary, will not be coerced by any undue influence from any party and will not influence your present or future relations with Ryerson University or the Property Manager. If you decide to participate, you are free to withdraw your consent and to stop your participation at any time. If you choose to withdraw your participation, any data gathered to that point, provided by you, would be destroyed. At any particular point in this study, you may refuse to answer any particular question or stop participation altogether. 10. Questions about the study: If you have questions about the research, you may contact Prof. Vera Straka by email: vst 416-979-5000 extension 6495. If you have questions about the research, you may contact Prof. Vera Straka by email: vstraka@ryerson.ca or by phone at 416-979-5000 extension 6495. If you have any questions regarding your rights as a participant in this study, you may contact the Ryerson University Research Ethics Board for information: Research Ethics Board, c/o Office of the Vice President, Research and Innovation, Ryerson University, 350 Victoria Street, Toronto, ON M5B 2K3, 416-979-5042. Research Study: Energy Conservation and Feedback Study, Toronto You are being asked to participate in a research study. Before you give your consent to participate, it is important that you read the following information and ask as many questions as necessary to be sure you understand what you will be asked to do and the degree of your involvement. 1. Investigators:  Prof. Alan Fung, Associate Professor, Department of Mechanical and Industrial Engineering.  Prof. Alan Fung, Associate Professor, Department of Mechanical and Industrial Engineerin  Prof. Vera Straka, Associate Professor, Department of Architectural Science.  Dr. Sara Alsaadani, Post-Doctoral Fellow, Department of Architectural Science. p  Kevin Trinh, Graduate student supervised by Prof. Vera Straka and Prof. Alan Fung. p y g  Samira Zare Mohazabieh, Graduate student supervised by Prof. Vera Straka and Prof. Alan Fung.  Samira Zare Mohazabieh, Graduate student supervised by Prof. Vera Straka and Pro 2. Purpose of the Study: The purpose of this study is to investigate the effects of real-time feedback on energy conservation. The purpose of this study is to investigate the effects of real-time feedback on energy conservation. 11. Agreement: By signing the following agreement and returning it to us, you are indicating that: By signing the following agreement and returning it to us, you are indicating that: 1. You have read the information in this agreement 2. You have had a chance to ask any questions you have about the study 3. You understand that you can change your mind and withdraw your consent to participate. 4. You are providing your consent to take part and have your information used in our st I, ____________________________________ consent to participate in the study conducted by Dr. Sara Alsaadani, Kevin Trinh and Samira Zare Mohazabieh, and supervised by Prof. Vera Straka and Prof. Alan Fung, Ryerson University. 87 Appendix C – Field Study Consent Form 88 1 Note that the provided internet connection should be treated as a public network like you would find at a coffee shop or hotel. We strongly discourage its use for sensitive information such as online banking. If your participation terminates prior to the completion of the study, we will be collecting the tablet for redistribution. Will encourage you to make any back- up of your data. In your presence, we will clear all data from its local memory. b) Completing a short thermal comfort survey approximately once per week: b) This thermal comfort survey is also to be completed through your Android tablet. This survey consists of seven short questions about the means undertaken to maintain your thermal comfort during different weather conditions. You will receive a prompt on your tablet once every eight days asking you to complete your thermal comfort survey. You will be receiving this prompt for a total period of two months. Completion of this survey should take less than 2 minutes each time. If you have any questions about this consent or require any technical support or training on how to use you Android tablet, members of the research team will be available in the Lobby at the times posted on the Notice Board to answer your questions and assist you. 4. Eligibility criteria: Before giving your consent to participate in this study, you must satisfy the following eligibility criteria: g your consent to participate in this study, you must satisfy the following eligibility criteria: Before giving your consent to participate in this study, you must satisfy the following eli  Be aged 18 or above. e aged 8 o abo e  That you have a basic level of English literacy to allow you to understand the research material and to complete the surveys required as part of your participation in the study (see section 3 above).  That you have a basic level of English literacy to allow you to understand the research material and to complete the surveys required as part of your participation in the study (see section 3 above).  That you intend to reside in your unit in Green Phoenix for at least the twelve-month period between June 2014- May 2015.  That you intend to reside in your unit in Green Phoenix for at least the twelve-month period between June 2014- May 2015. 3. Description of the study: You do not need to be physically present in your unit during the time of installation. Installation of this equipment will enable us to provide with you with direct feedback about your energy consumption for a two-month period. This feedback will include information about your energy consumption over the last 24 hours and over the last 7 days. In addition to information about your own consumption, you will also receive feedback about your neighbours’ average energy consumption over the last 24 hours. You will receive feedback through an application designed specifically for this purpose and installed on an Android Tablet, which will be given to you once you sign this agreement. 3. Description of the study: Participation in the study entails the following: Participation in the study entails the following: a) Agreeing that equipment is installed in your unit to capture your energy consumption data over a twelve month period, and agreeing to long-term monitoring of the energy consumption from your unit for a twelve-month period (June 2014-May 2015). Equipment installed in your unit will consist of: Equipment installed in your unit will consist of:  A battery-powered sensor installed in the fan coil unit to measure its power draw and air temperature output. A b d i ll d i h i bi d l i h idi  A battery-powered sensor installed in the suite to measure ambient temperature and relative humidity.  A free internet connection to allow energy data collected via the afore-described sensors to be transmitted back to the research group. You will also be able to use this internet connection for casual web-browsing1.  A battery-powered sensor installed in the suite to measure ambient temperature and relative humidity.  A free internet connection to allow energy data collected via the afore-described sensors to be transmitted back to the research group. You will also be able to use this internet connection for casual web-browsing1.  A free internet connection to allow energy data collected via the afore-described sensors to be transmitted back to the research group. You will also be able to use this internet connection for casual web-browsing1. 1 Note that the provided internet connection should be treated as a public network like you would find at a coffee shop or hotel. We strongly discourage its use for sensitive information such as online banking. If your participation terminates prior to the completion of the study, we will be collecting the tablet for redistribution. Will encourage you to make any back- up of your data. In your presence, we will clear all data from its local memory. 89 The time taken to complete the installation in you unit will be approximately 20 minutes. We will complete the installation at an agreed time that is convenient for you. You do not need to be physically present in your unit during the time of installation. The time taken to complete the installation in you unit will be approximately 20 minutes. We will complete the installation at an agreed time that is convenient for you. 8. Voluntary nature of participation: Participation in this study is voluntary. Participation in the study is completely voluntary, will not be coerced by any undue influence from any party and will not influence your present or future relations with Ryerson University or the Property Manager. If you decide to participate, you are free to withdraw your consent and to stop your participation at any time. If you choose to withdraw your participation, any data gathered to that point, provided by you, would be destroyed. At any particular point in this study, you may refuse to answer any particular question or stop participation altogether. Please note that, in the event of withdrawing your participation from the study, we would require you to return the Android tablet that was provided to you as a research instrument. We would also need access to your unit to remove the installed equipment. The time taken to remove the installed equipment would be approximately 5 minutes. Should you choose to withdraw, we will remove the equipment at an agreed time that is convenient for you. Again, you do not need to be physically present in your unit during this time. 5. Risks or discomfort: There is little risk in participating in this study. You may also be concerned that someone else may find out your responses to the thermal comfort survey. Please note that the responses are individual to you; there is no ‘right’ or ‘wrong’ answer. In addition, we will not be collecting any names and will not publish information in any reports that will identify you by unit number or by any other kind of personal information. When we publish reports from this research project, we will be using only general information, not individual information and your confidentiality will be protected. You may have concerns about data privacy and security; particularly as the data is going to be transmitted over the internet. We assure you that the data will be completely confidential, and is going to be password-protected and specially coded 90 before it is transmitted over radio. Only members of the research team will be able to link coded data to the participating individual. before it is transmitted over radio. Only members of the research team will be able to link coded data to the participating individual. A final concern you may have is that the internet we have installed for the purpose of our study may affect your own private internet subscription and connection. It will not. 9. Compensation: You will be allowed full use of the Android tablet assigned to you as a research instrument for the purpose of this study. You will also be given access to an internet connection for a full twelve-month period. 6. Benefits of the study: The following are potential benefits of the research:  To design and aid a system that will help reduce energy consumption g y p gy p  To potentially reduce energy consumption and thus help to save money.  To promote a community and teamwork spirit.  To promote a community and teamwork spirit. While this project promises benefit for social good, individual benefit to any of the tenants 7. Confidentiality: All data collected will be handled confidentially. We will not be collecting any names. Unit numbers will be collected to enable us to link energy consumption data, data from the thermal comfort survey and data from the attitude survey, and to enable us to provide you with feedback about your individual energy consumption. We will not publish unit numbers or specific information about individuals in any publication or report. 10. Questions about the study: If you have questions about the research, you may contact Prof. Vera Straka by email: vstraka@ryerson.ca or by phone at 416-979-5000 extension 6495. 91 If you have any questions regarding your rights as a participant in this study, you may contact the Ryerson University Research Ethics Board for information: Research Ethics Board, c/o Office of the Vice President, Research and Innovation, Ryerson University, 350 Victoria Street, Toronto, ON M5B 2K3, 416-979-5042. If you have any questions regarding your rights as a participant in this study, you may contact the Ryerson University Research Ethics Board for information: Research Ethics Board, c/o Office of the Vice President, Research and Innovation, Ryerson University, 350 Victoria Street, Toronto, ON M5B 2K3, 416-979-5042. 11. Agreement: By signing the following agreement and returning it to us, you are indicating that: 5. You have read the information in this agreement 6. You have had a chance to ask any questions you have about the study 7. You understand that you can change your mind and withdraw your consent to participate at any time 8. You are providing your consent to take part and have your information used in our study. 5. You have read the information in this agreement Please feel free to keep a copy of the following agreement for your own records before submitting it. Please feel free to keep a copy of the following agreement for your own records before submitting it. I, ____________________________________ consent to participate in the study conducted by Dr. Sara Alsaadani, Kevin Trinh and Samira Zare Mohazabieh, and supervised by Prof. Vera Straka and Prof. Alan Fung, Ryerson University. ____ consent to participate in the study conducted by Dr. Sara Alsaadani, Kevin upervised by Prof. Vera Straka and Prof. Alan Fung, Ryerson University. 92 92 Appendix D – Conservation Program Presentation Slides 93 94 95 96 97 97 Appendix E – Energy Tracking Presentation Slides 98 99 RYERSON UNIVERSITY Energy Conservation and the Green Phoenix 100 Appendix F – NEP and Demographics Survey 101 RYERSON UNIVERSITY Energy Conservation and the Green Phoenix General Information Pl l ll i b l CHECK OFF  h i i General Information Please complete all questions below. CHECK OFF  the appropriate option. 3. What part of the world did you grow up in? Canada USA Europe South or Central America or Caribbean South Asia (e.g. India, Pakistan, Sri Lanka) East Asia (e.g. China, Japan, Korea) Southeast Asia (e.g. Vietnam, Philippines, Malaysia) West Asia & Middle East (e.g. Lebanon, Iran) Africa (e.g. Ethiopia) Australia, New Zealand or the South Pacific Other, please specify._______________ Prefer not to answer. 4. How many years have you been living in Phoenix Place? 4. How many years have you been living in Phoenix Place?  0 to 1 year 2 to 4 years 5 to 7 years More than 7 years 5. How many people live in your household? y p p 1 person 2 persons 3 or more persons 102 Environmental Attitudes: To the best of your understanding, please answer whether you agree or disagree with the following statements by checkin the box on the following scale: To the best of your understanding, please answer whether you agree or disagree with the following statements by checking the box on the following scale: Strongly Agree Agree Neither Agree nor Disagree Disagree Strongly Disagree 1. We are approaching the limit of the number of people the earth can support. 2. Humans have the right to modify the natural environment. 3. When humans interfere with nature it often produces disastrous consequences. 4. Human ingenuity will ensure that we do NOT make the earth unlivable. 5. Humans are severely abusing the environment. 6. The earth has plenty of natural resources if we just learn how to develop them. 7. Plants and animals have as much right as humans to exist. 8. The balance of nature is strong enough to cope with the impacts of modern industrial nations. 9. Despite our special abilities humans are still subject to the laws of nature. 10. The so-called ’ecological crisis’ facing humankind has been greatly exaggerated. 103 11. The earth is like a spaceship with very limited room and resources. 12. Humans were meant to rule over the rest of nature. 13. The balance of nature is very delicate and easily upset. 14. Humans will eventually learn enough about how nature works to be able to control it. 15. If things continue on their present course, we will soon experience a major ecological catastrophe. General Information 104 Appendix G – Pledge Form Appendix G – Pledge Form 105 ENERGY CONSERVATION PLEDGE I (first name, last name)_________________________________________, understand that energy consumption affects our natural environment, human health and overall well-being. Therefore, to show my support for the tenant engagement program at Green Phoenix, I pledge to make every effort to reduce my energy consumption at home as much as possible, and to contribute toward the building- wide energy reduction goal of 10%. ENERGY CONSERVATION PLEDGE I (first name, last name)_________________________________________, understand that energy consumption affects our natural environment, human health and overall well-being. Therefore, to show my support for the tenant engagement program at Green Phoenix, I pledge to make every effort to reduce my energy consumption at home as much as possible, and to contribute toward the building- wide energy reduction goal of 10%. ENERGY CONSERVATION PLEDGE I (first name, last name)_________________________________________, understand that energy consumption affects our natural environment, human health and overall well-being. I (first name, last name)_________________________________________, u consumption affects our natural environment, human health and overall well-being. Therefore, to show my support for the tenant engagement program at Green Phoenix, I pledge to make every effort to reduce my energy consumption at home as much as possible, and to contribute toward the building- wide energy reduction goal of 10%. 106 Appendix H – Thermal Comfort Survey 107 Figure 1: Thermal Comfort Survey Page 1 of 8. Figure 2: Thermal Comfort Survey Page 2 of 8. Figure 1: Thermal Comfort Survey Page 1 of 8. Figure 2: Thermal Comfort Survey Page 2 of 8. Figure 1: Thermal Comfort Survey Page 1 of 8. Figure 2: Thermal Comfort Survey Page 2 of 8. 108 Figure 3: Thermal Comfort Survey Page 3 of 8. Figure 3: Thermal Comfort Survey Page 3 of 8. Figure 3: Thermal Comfort Survey Page 3 of 8. Figure 4: Thermal Comfort Survey Page 4 of 8. Figure 3: Thermal Comfort Survey Page 3 of 8. Figure 3: Thermal Comfort Survey Page 3 of 8. Figure 4: Thermal Comfort Survey Page 4 of 8. Figure 3: Thermal Comfort Survey Page 3 of 8. 109 Figure 4: Thermal Comfort Survey Page 4 of 8. Figure 4: Thermal Comfort Survey Page 4 of 8. Figure 4: Thermal Comfort Survey Page 4 of 8. 109 Figure 5: Thermal Comfort Survey Page 5 of 8. Figure 6: Thermal Comfort Survey Page 6 of 8. Figure 5: Thermal Comfort Survey Page 5 of 8. Figure 6: Thermal Comfort Survey Page 6 of 8. Figure 6: Thermal Comfort Survey Page 6 of 8. Figure 6: Thermal Comfort Survey Page 6 of 8. 110 Figure 7: Thermal Comfort Survey Page 7 of 8. 111 Figure 7: Thermal Comfort Survey Page 7 of 8. Figure 8: Thermal Comfort Survey Page 8 of 8. Figure 7: Thermal Comfort Survey Page 7 of 8. Figure 7: Thermal Comfort Survey Page 7 of 8. Figure 7: Thermal Comfort Survey Page 7 of 8. Figure 8: Thermal Comfort Survey Page 8 of 8. Figure 7: Thermal Comfort Survey Page 7 of 8. Figure 7: Thermal Comfort Survey Page 7 of 8. Figure 7: Thermal Comfort Survey Page 7 of 8. Figure 8: Thermal Comfort Survey Page 8 of 8. Figure 8: Thermal Comfort Survey Page 8 of 8. 111 Appendix I – Energy Audit Sample Results 112 Suite No. Discuss Privacy I am not collecting any information that can be used to personally identify you, such as name, age, gender, etc. I will only collect your verbal answers, and we will not share these outside of my research team. Introduction Thank you for helping me with your participation. I am exploring ideas to improve my energy dashboard for use by tenants like yourself at Green Phoenix, and I am very interested in your feedback. Today I will be exploring some concepts visualizing the energy usage of home heating and cooling systems. Appliance Make Model Year Rated Pwr(W) On Pwr (W) Stdy Pwr (W) #Hrs on .Wk A Refregerator LG GR-292R 16.4 1.5 A Electric Kettle Rival RV-KE5754o 1000 1070 0 0.35-0.583 A TV Insignia NS-39D310NA15 120v - 1.5A 49.6 0 28-35 A Laptop Toshiba PSCFWC-005002 19v - 2.37A 62.7 0 Jul-14 A Desk Lamp 30 15 0 B Microwave B Toaster B Electric Kettle B Rice Cooker Black and Decker RC426C 500 490 52 1.5 B TV Samsung T23A360 2011 50 32 0 35 B BluRay Samsung BDH5100 9 4.5 2.2 2.5 B Cable box Cisco 4642HD 15 14.4 13.7 42 B Speaker Logitech LSH-00035 5 1.8 1.5 168 B Laptop Toshiba PSCDW-005002 12.5 0 168 B Printer Canon MX452 875 5.2 0 0.02083 B Heater Intertek PTC-700 1500 1800 0 0 C Microwave Sharp R-2428C 1030 1057 1.4 C Refrigerator Magic Chef 168 C Toaster Danby 0.25 C TV Sony KDL-32DX420 115 49.8 0 C DVD RCA 8 5.4 C Laptop Acer Aspire 3680 29.7 0 1.5 C Fan 1 month old 42 2 C Curling Iron 167 0.5 C Straightening Iron 267 0.5 D Toaster Proctor Silex 900 937 0 0.583 D Laptop Acer Aspire 35735- 4401 45.98 31.8 3.8 96 on standby D Clock/radio Sylvannia 12 1.8 1.5 21 D Humidifier Air o Swiss 2.8 2.7 0 E TV Samsung UH32EH5300F 69 41.3 0 56 E Cable box Motorola (from bell) 17.2 15 56 E Modem Bell 8.6 168 E Land phone Panasonic KXT94771C 2.75 0 0 E Desk Lamp Sylvannia (CFL) 13 13 21 21 13 13 113 Appendix J – Usability Test Script 114 114 Session Procedure This is how the testing session will work: Using my computer, I will show you some prototypes and ask you to tell me how you interpret an screen elements, or an entire screen. Please say whatever comes into your mind. There are no wrong answers. We are evaluating the concepts, not you. You can end the test at any time you choose, for whatever reason. You can end the test at any time you choose, for whatever reason. 115 Questions Introduction: You’re a tenant who is participating in our energy conservation study and are given a tablet to help you track your energy use. You just opened the dashboard application to take a look at the energy usage data. (Show Screen 1). gy j p pp gy g ( ) 1. Please look at this screen closely and tell me what you think of the information you see. Possible probe (asked if users don’t mention anything specific):  What is your initial reaction upon seeing this screen?  What does the information under “Last 7 Days” mean to you?  What does the information under “Today” mean to you?  What does the information under “Right Now” mean to you?  What does the “Tip” mean to you?  What information makes sense to you?  What information doesn’t make sense to you?  What does the “Tip” mean to you?  What information makes sense to you?  What information doesn’t make sense to you? 116 116 2. Your Fan Coil Unit (FCU) can account for a large portion of your suite’s total energy use. Please look at this screen closely and tell me what you think of the information you see. Possible probes (asked if users don’t mention anything specific):  What does the information under “Right Now” mean to you?  What does the “Tip” mean to you?  What information makes sense to you?  What information doesn’t make sense to you? 117 117 3. Here is a variation of the first screen “Total Suite Energy Use” I showed you. 3. Here is a variation of the first screen “Total Suite Energy Use” I showed you. Please look at this screen closely and tell me what you think of the information you see. Possible probes (asked if users don’t mention anything specific):  How is this screen different from the first version?  What is your initial reaction upon seeing this screen? 4. Here is a variation of the second screen “FCU Energy Use” I showed you. 4. Here is a variation of the second screen “FCU Energy Use” I showed you. Please look at this screen closely and tell me what you think of the information you see. Possible probes (asked if users don’t mention anything specific):  How is this screen different from the first version? 118 Overall 5. Introduction: Please rank the usefulness of the following features on the “Suite Dashboard”: (1 = most useful, 6 least useful) ___ “LAST 7 DAYS” Energy Use Chart ___ Historical Use and Daily Targets (Red and Green lines) ___ “TODAY” Energy Use Chart ___ “RIGHT NOW” Power Meter ___ “HOW YOU’RE DOING” Faces ___ “DAILY TIP” 5. Please rank the usefulness of the following features on the “FCU Dashboard”: (1 = most useful, 6 least useful) ___ “LAST 7 DAYS” FCU Energy Use Chart ___ “TODAY” FCU Energy Use Chart ___ “RIGHT NOW” Power Meter ___ “Weather and Indoor Temperature” ___ “DAILY TIP” 6. How would you change the presentation of the information, if at all? 7. Do you have any other comments about the feedback system, the tablet, the internet connection? Introduction: Overall, what do you think of what you’ve seen today? 6. What information do you find valuable? Why? 7. What information do you find useless? Why? 8. Would you use this information on a regular basis? If yes, with what frequency? Would you come back and look at this information again? 7. What information do you find useless? Why? 8. Would you use this information on a regular basis? If yes, with what frequency? Would you come back and look at this information again? 8. Would you use this information on a regular basis? If yes, with what frequency? Would you come back and look at this information again? 9. Does the information motivate you to make changes to your energy use? 119 Appendix K – Exit Surveys 120 Research Study: Tenant Engagement and Energy Conservation, Toronto EXIT SURVEY Name: ______________________ Suite #: _____ OVERALL EXPERIENCE 1. What was your overall impression of the experience using the energy dashboard app on the tablet? 2. What did you like most about using the dashboard? 3. What did you like least about using the dashboard? 4. Please rank the usefulness of the following features on the “Suite Dashboard”: (1 = most useful, 6 least useful) ___ “LAST 7 DAYS” Energy Use Chart ___ Historical Use and Daily Targets (Red and Green lines) ___ “TODAY” Energy Use Chart ___ “RIGHT NOW” Power Meter ___ “HOW YOU’RE DOING” Faces ___ “DAILY TIP” 5. Please rank the usefulness of the following features on the “FCU Dashboard”: (1 = most useful, 6 least useful) ___ “LAST 7 DAYS” FCU Energy Use Chart ___ “TODAY” FCU Energy Use Chart ___ “RIGHT NOW” Power Meter ___ “Weather and Indoor Temperature” ___ “DAILY TIP” 6. How would you change the presentation of the information, if at all? Research Study: Tenant Engagement and Energy Conservation, Toronto EXIT SURVEY Name: ______________________ Suite #: _____ OVERALL EXPERIENCE 1. What was your overall impression of the experience using the energy dashboard app on the tablet? 2. What did you like most about using the dashboard? 3. What did you like least about using the dashboard? 4. Research Study: Tenant Engagement and Energy Conservation, Toronto EXIT SURVEY Research Study: Tenant Engagement and Energy Conservation, Toronto EXIT SURVEY 2. What did you like most about using the dashboard? 3. What did you like least about using the dashboard? 4. Please rank the usefulness of the following features on the “Suite Dashboard”: (1 = most useful, 6 least useful) ___ “LAST 7 DAYS” Energy Use Chart ___ Historical Use and Daily Targets (Red and Green lines) ___ “TODAY” Energy Use Chart ___ “RIGHT NOW” Power Meter ___ “HOW YOU’RE DOING” Faces ___ “DAILY TIP” 5. Please rank the usefulness of the following features on the “FCU Dashboard”: (1 = most useful, 6 least useful) ___ “LAST 7 DAYS” FCU Energy Use Chart ___ “TODAY” FCU Energy Use Chart ___ “RIGHT NOW” Power Meter ___ “Weather and Indoor Temperature” ___ “DAILY TIP” 6. How would you change the presentation of the information, if at all? 7. Do you have any other comments about the feedback system, the tablet, the internet connection? SAVINGS STRATEGIES 8. Compared to the year before, how much energy do you think you used during this study? Circle the option you feel is most accurate. I used 20%+ more I used 10-20% more I used 0-10% more I saved 0-10% I saved 10-20% I saved 20%+ 9. Please list the top 2-3 changes you made to save energy this past year. - - 4. Please rank the usefulness of the following features on the “Suite Dashboard”: (1 = most useful, 6 least useful) ___ “LAST 7 DAYS” Energy Use Chart ___ Historical Use and Daily Targets (Red and Green lines) ___ “TODAY” Energy Use Chart ___ “RIGHT NOW” Power Meter ___ “HOW YOU’RE DOING” Faces ___ “DAILY TIP” 5. Please rank the usefulness of the following features on the “FCU Dashboard”: (1 = most useful, 6 least useful) ___ “LAST 7 DAYS” FCU Energy Use Chart ___ “TODAY” FCU Energy Use Chart ___ “RIGHT NOW” Power Meter ___ “Weather and Indoor Temperature” ___ “DAILY TIP” 6. How would you change the presentation of the information, if at all? 7. Do you have any other comments about the feedback system, the tablet, the internet connection? SAVINGS STRATEGIES SAVINGS STRATEGIES AVINGS STRATEGIES 8. Compared to the year before, how much energy do you think you used during this study? Circle the option you feel is most accurate. I used 20%+ more I used 10-20% more I used 0-10% more I saved 0-10% I saved 10-20% I saved 20%+ 9. Please list the top 2-3 changes you made to save energy this past year. 8. Compared to the year before, how much energy do you think you used during this study? Circle the option you feel is most accurate. I used 20%+ more I used 10-20% more I used 0-10% more I saved 0-10% I saved 10-20% I saved 20%+ 9. Please list the top 2-3 changes you made to save energy this past year. 9. Please list the top 2-3 changes you made to save energy this past year. 121 OVERALL EXPERIENCE 2. What did you like most about using the dashboard? 3. What did you like least about using the dashboard? 4. Please rank the usefulness of the following features on the “Suite Dashboard”: (1 = most useful, 7 least useful) ___ Historical Use + Daily Target (Red & Green lines) ___ Neighbour Comparisons (Yellow bars) ___ “LAST 7 DAYS” Energy Use Chart ___ “TODAY” Energy Use Chart ___ “RIGHT NOW” Power Meter ___ “HOW YOU’RE DOING” Faces ___ “DAILY TIP” 5. Please rank the usefulness of the following features on the “FCU Dashboard”: (1 = most useful, 7 least useful) ___ Neighbour Comparisons (yellow bars) ___ “LAST 7 DAYS” FCU Energy Use Chart ___ “TODAY” FCU Energy Use Chart ___ “RIGHT NOW” Power Meter ___ “Weather and Indoor Temperature” ___ “HOW YOU’RE DOING” Faces ___ “DAILY TIP” 6. How would you change the presentation of the information, if at all? 7. Do you have any other comments about the feedback system, the tablet, the internet connection? SAVINGS STRATEGIES 8. Compared to the year before, how much energy do you think you used during this study? Circle the option you feel is most accurate. I used 20%+ more I used 10-20% more I used 0-10% more I saved 0-10% I saved 10-20% I saved 20%+ 9. Please list the top 2-3 changes you made to save energy this past year. - 4. Please rank the usefulness of the following features on the “Suite Dashboard”: (1 = most useful, 7 least useful) ___ Historical Use + Daily Target (Red & Green lines) ___ Neighbour Comparisons (Yellow bars) ___ “LAST 7 DAYS” Energy Use Chart ___ “TODAY” Energy Use Chart ___ “RIGHT NOW” Power Meter ___ “HOW YOU’RE DOING” Faces ___ “DAILY TIP” 5. Please rank the usefulness of the following features on the “FCU Dashboard”: (1 = most useful, 7 least useful) ___ Neighbour Comparisons (yellow bars) ___ “LAST 7 DAYS” FCU Energy Use Chart ___ “TODAY” FCU Energy Use Chart ___ “RIGHT NOW” Power Meter ___ “Weather and Indoor Temperature” ___ “HOW YOU’RE DOING” Faces ___ “DAILY TIP” 6. How would you change the presentation of the information, if at all? SAVINGS STRATEGIES SAVINGS STRATEGIES AVINGS STRATEGIES 8. Compared to the year before, how much energy do you think you used during this study? Circle the option you feel is most accurate. I used 20%+ more I used 10-20% more I used 0-10% more I saved 0-10% I saved 10-20% I saved 20%+ 9. Please list the top 2-3 changes you made to save energy this past year. 8. Compared to the year before, how much energy do you think you used during this study? Circle the option you feel is most accurate. I used 20%+ more I used 10-20% more I used 0-10% more I saved 0-10% I saved 10-20% I saved 20%+ 9. Please list the top 2-3 changes you made to save energy this past year. 122 Appendix L – Research Ethics Board Approval Letters 123 124 References References Abrahamse, W., Steg, L., Vlek, C., & Rothengatter, T. (2005). A review of intervention studies aimed at household energy conservation. Journal of Environmental Psychology, 25(3), 273–291. http://doi.org/10.1016/j.jenvp.2005.08.002 Abrahamse, W., Steg, L., Vlek, C., & Rothengatter, T. (2005). A review of intervention studies aimed at household energy conservation. Journal of Environmental Psychology, 25(3), 273–291. http://doi.org/10.1016/j.jenvp.2005.08.002 Allcott, H. (2011). Social norms and energy conservation. Journal of Public Economics, 95(9–10), 1082–1095. http://doi.org/10.1016/j.jpubeco.2011.03.003 Allcott, H., & Rogers, T. (2012). The Short-Run and Long-Run Effects of Behavioral Interventions: Experimental Evidence from Energy Conservation (Working Paper No. 18492). National Bureau of Economic Research. Retrieved from http://www.nber.org/papers/w18492 Becker, L. J. (1978). Joint effect of feedback and goal setting on performance: A field study of residential energy conservation. Journal of Applied Psychology, 63(4), 428–433. http://doi.org/10.1037/0021-9010.63.4.428 Benders, R. M., Kok, R., Moll, H. C., Wiersma, G., & Noorman, K. J. (2006). New approaches for household energy conservation–In search of personal household energy budgets and energy reduction options. Energy Policy, 34(18), 3612–3622. Brandon, G., & Lewis, A. (1999). Reducing household energy consumption: A qualitative and quantitative field study. Journal of Environmental Psychology, 19(1), 75–85. Changing Behaviour. (2009). Interaction Schemes For Successful Energy Demand Side Management (No. Deliverable 5). Europe. Retrieved from http://www.energychange.info/component/content/article/67-deliverables/191-d5- interaction-schemes-for-successful-energy-demand-side-management Changing Behaviour. (2013). MECHanisms \| Make Energy Change Happen Toolkit. Retrieved August 22, 2013, from http://mechanisms.energychange.info/ Cialdini, R. B., & Goldstein, N. J. (2004). Social Influence: Compliance and Conformity. Annual Review of Psychology, 55(1), 591–621. Cialdini, R. B., & Goldstein, N. J. (2004). Social Influence: Compliance and Conformity. Annual Review of Psychology, 55(1), 591–621. Darby, S. (2001). Making it Obvious: Designing Feedback into Energy Consumption. In D. P. Bertoldi, D. A. Ricci, & P. A. de Almeida (Eds.), Energy Efficiency in Household Appliances and Lighting (pp. 685–696). Springer Berlin Heidelberg. Retrieved from http://link.springer.com/chapter/10.1007/978-3-642-56531-1_73 Darby, S. (2001). Making it Obvious: Designing Feedback into Energy Consumption. In D. P. Bertoldi, D. A. Ricci, & P. A. de Almeida (Eds.), Energy Efficiency in Household Appliances and Lighting (pp. 685–696). Springer Berlin Heidelberg. Retrieved from http://link.springer.com/chapter/10.1007/978-3-642-56531-1_73 Darby, S. (2006). The effectiveness of Feedback on energy Consumption: A Review for DEFRA of the Literature on Metering, Billing and direct Displays. Environmental Change Institute, University of Oxford. Darby, S. (2006). The effectiveness of Feedback on energy Consumption: A Review for DEFRA of the Literature on Metering, Billing and direct Displays. Environmental Change Institute, University of Oxford. Egan, C. (1999). 125 References Graphical displays and comparative energy information: What do people understand and prefer? In Proceedings of the Summer Study of the European Council for an Energy Efficient Economy (pp. 2–13). Egan, C. (1999). Graphical displays and comparative energy information: What do people understand and prefer? In Proceedings of the Summer Study of the European Council for an Energy Efficient Economy (pp. 2–13). 126 Ehrhardt-Martinez, Donnelly, K. A., & Laitner, J. A. (2010). Advanced Metering Initiatives and Residential Feedback Programs: A Meta-Review for Household Electricity-Saving Opportunities (Research Report No. E105). Washington, DC: American Council for an Energy-Efficient Economy. Retrieved from http://aceee.org/research-report/e105 Ehrhardt-Martinez, K. (2012). A Comparison of Feedback -Induced Behaviors from Monthly Energy Reports, Online Feedback, and In-home Displays. In Proceedings of 2012 ACEEE Summer Study on Energy Efficiency in Buildings. Retrieved from http://www.aceee.org/files/proceedings/2012/data/papers/0193-000244.pdf#page=1 Federation of Canadian Municipalities. (1996). A Municipal Guide on Economic Instruments to Support Municipal Waste Management Programs. Toronto, ON: Resource Integration Systems Ltd. (RIS). Fischer, C. (2008). Feedback on household electricity consumption: a tool for saving energy? Energy Efficiency, 1(1), 79–104. http://doi.org/10.1007/s12053-008-9009-7 Flemming, S. A., Hilliard, A., & Jamieson, G. A. (2008). The Need for Human Factors in the Sustainability Domain. Human Factors and Ergonomics Society Annual Meeting Proceedings, 52, 748–752. Fischer, C. (2008). Feedback on household electricity consumption: a tool for saving energy? Energy Efficiency, 1(1), 79–104. http://doi.org/10.1007/s12053-008-9009-7 Flemming, S. A., Hilliard, A., & Jamieson, G. A. (2008). The Need for Human Factors in the Sustainability Domain. Human Factors and Ergonomics Society Annual Meeting Proceedings, 52, 748–752. Froehlich, J. (2009). Promoting energy efficient behaviors in the home through feedback: The role of human-computer interaction. In HCIC 2009 Winter Workshop. Froehlich, J., Dillahunt, T., Klasnja, P., Mankoff, J., Consolvo, S., Harrison, B., & Landay, J. A. (2009). UbiGreen: Investigating a Mobile Tool for Tracking and Supporting Green Transportation Habits. In Proceedings of CHI2009 (pp. 4–9). Boston, MA, USA: ACM. Fung, A. S., Aulenback, A., Ferguson, A., & Ugursal, V. I. (2003). Standby power requirements of household appliances in Canada. Energy and Buildings, 35(2), 217–228. http://doi.org/10.1016/S0378-7788(02)00086-5 Gardner, G., & Stern, P. C. (2002). Environmental problems and human behavior (2nd ed. Boston MA: Pearson Custom Pub. Gardner, G. T., & Stern, P. C. (1996). Environmental Problems and Human Behavior (1st ed. Allyn & Bacon. Gatersleben, B., Steg, L., & Vlek, C. (2002). Measurement and determinants of environmentally significant consumer behavior. Environment and Behavior, 34(3), 335–362. Government of Canada, N. R. References C. (2012). Report to Parliament Under the Energy Efficiency Act 2010-2011. Retrieved August 2, 2014, from http://oee.nrcan.gc.ca/publications/statistics/parliament10-11/chapter1.cfm?attr=0 Holmes, T. G. (2007). Eco-visualization: combining art and technology to reduce energy consumption. In Proceedings of the 6th ACM SIGCHI conference on Creativity & Cognition (pp. 153–162). Washington DC. Holmes, T. G. (2007). Eco-visualization: combining art and technology to reduce energy consumption. In Proceedings of the 6th ACM SIGCHI conference on Creativity & Cognition (pp. 153–162). Washington DC. IEA. (2010). Energy Technology Perspectives 2010. Retrieved from http://www.iea.org/publications/freepublications/publication/name,26100,en.html IEA. (2010). Energy Technology Perspectives 2010. Retrieved from http://www.iea.org/publications/freepublications/publication/name,26100,en.html 127 Janda, K. B. (2011). Buildings don’t use energy: people do. Architectural Science Review, 54(1), 15–22. http://doi.org/10.3763/asre.2009.0050 Karjalainen, S., & Vastamaeki, R. (2007). Occupants Have a False Idea of Comfortable Summer Season Temperatures. In Proceedings of Clima 2007 WellBeing Indoors. Helsinki (Finland): FINVAC. Retrieved from http://www.irb.fraunhofer.de/CIBlibrary/search- quick-result-list.jsp?A&idSuche=CIB+DC7482 Katzev, R. D., & Johnson, T. R. (1987). Promoting energy conservation: An analysis of behavioral research. Kempton, W. (1986). Two theories of home heat control. Cognitive Science, 10(1), Kort, Y. A. W. de, McCalley, L. T., & Midden, C. J. H. (2008). Persuasive Trash Cans: Activation of Littering Norms by Design. Environment and Behavior, 40(6), 870–891. http://doi.org/10.1177/0013916507311035 Kurz, T., Donaghue, N., & Walker, I. (2005). Utilizing a Social-Ecological Framework to Promote Water and Energy Conservation: A Field Experiment. Journal of Applied Social Psychology, 35(6), 1281–1300. http://doi.org/10.1111/j.1559-1816.2005.tb02171.x Liberman, N., & Trope, Y. (1998). The role of feasibility and desirability considerations in near and distant future decisions: A test of temporal construal theory. Journal of Personality and Social Psychology, 75(1), 5–18. Liberman, N., & Trope, Y. (1998). The role of feasibility and desirability considerations in near and distant future decisions: A test of temporal construal theory. Journal of Personality and Social Psychology, 75(1), 5–18. Loewenstein, G. (1996). Out of Control: Visceral Influences on Behavior. Organizational Behavior and Human Decision Processes, 65(3), 272–292. Loewenstein, G. (1996). Out of Control: Visceral Influences on Behavior. Organizational Behavior and Human Decision Processes, 65(3), 272–292. Lutzenhiser, L. (1993). Social and Behavioral Aspects of Energy use. Annual Review of Energy and the Environment, 18(1), 247–289. http://doi.org/10.1146/annurev.eg.18.110193.001335 Lutzenhiser, L. (1993). Social and Behavioral Aspects of Energy use. Annual Review of Energy and the Environment, 18(1), 247–289. http://doi.org/10.1146/annurev.eg.18.110193.001335 McKenzie-Mohr, D. (2011). Fostering Sustainable Behavior: An Introduction to Community- Based Social Marketing (Third Edition). Gabriola, BC, Canada: New Society Publishers. McKenzie-Mohr, D. (2011). References Fostering Sustainable Behavior: An Introduction to Community- Based Social Marketing (Third Edition). Gabriola, BC, Canada: New Society Publishers. Moezzi, M., & Diamond, R. (2005). Is Efficiency Enough? Towards a New Framework for Carbon Savings in the California Residential Sector. Lawrence Berkeley National Laboratory: Lawrence Berkeley National Laboratory. Moezzi, M., & Diamond, R. (2005). Is Efficiency Enough? Towards a New Framework for Carbon Savings in the California Residential Sector. Lawrence Berkeley National Laboratory: Lawrence Berkeley National Laboratory. Norman, D. A. (2004). Emotional design: Why we love (or hate) everyday things. New York, NY, USA: Basic Civitas Books. Norman, D. A. (2004). Emotional design: Why we love (or hate) everyday things. New York, NY, USA: Basic Civitas Books. Petersen, J. E., Shunturov, V., Janda, K., Platt, G., & Weinberger, K. (2007). Dormitory residents reduce electricity consumption when exposed to real-time visual feedback and incentives. International Journal of Sustainability in Higher Education, 8(1), 16–33. Petersen, J. E., Shunturov, V., Janda, K., Platt, G., & Weinberger, K. (2007). Dormitory residents reduce electricity consumption when exposed to real-time visual feedback and incentives. International Journal of Sustainability in Higher Education, 8(1), 16–33. Pierce, J., Odom, W., & Blevis, E. (2008). Energy aware dwelling: a critical survey of interaction design for eco-visualizations. In Proceedings of the 20th Australasian Conference on Computer-Human Interaction: Designing for Habitus and Habitat (pp. 1–8). New York, NY, USA: ACM. 128 Prada, J. (2013). Comparing Occupant Self-assessed Behaviour to Actual Metered Consumption (Unpublished Masters thesis). Ryerson University, Toronto, Canada. Raaij, V., F, W., & Verhallen, T. M. M. (1983). A behavioral model of residential energy use. J l f E i P h l J l f E i P h l 3(1) 39 63 Raaij, V., F, W., & Verhallen, T. M. M. (1983). A behavioral model of residential energy use. Journal of Economic Psychology, Journal of Economic Psychology, 3(1), 39–63. Roque, M., Straka, V., & Fung, A. (2012). Survey of Household Energy Use in a Toronto Rental High-rise Multi-unit Residential Building (MURB). In In Proceedings of the 2nd World Sustain. Forum (Vol. 2). Safa, A. A., Fung, A. S., & Kumar, R. (2015). Comparative thermal performances of a ground source heat pump and a variable capacity air source heat pump systems for sustainable houses. Applied Thermal Engineering, 81, 279–287. http://doi.org/10.1016/j.applthermaleng.2015.02.039 Sauer, J., Schmeink, C., & Wastell, D. G. (2007). Feedback quality and environmentally friendly use of domestic central heating systems. References Ergonomics, 50(6), 795–813. Sauer, J., Wiese, B. S., & Ruettinger, B. (2003). Designing low-complexity electrical consumer products for ecological use. Applied Ergonomics, 34(6), 521–531. http://doi.org/10.1016/j.apergo.2003.07.001 Schott, M., Long, N., Scheib, J., Fleming, K. B., & Brackney, L. (2012). Progress on Enabling an Interactive Conversation between Commercial Building Occupants and Their Building to Improve Comfort and Energy Efficiency. In ACEEE Summer Study on Energy Efficiency in Buildings (Vol. 7, pp. 250–265). Retrieved from http://aceee.org/files/proceedings/2012/data/papers/0193-000234.pdf Schultz, P. W., Nolan, J. M., Cialdini, R. B., Goldstein, N. J., & Griskevicius, V. (2007). The constructive, destructive, and reconstructive power of social norms. Psychological Science, 18(5), 429–434. http://doi.org/10.1111/j.1467-9280.2007.01917.x Seligman, C., Becker, L. J., & Darley, J. M. (1981). Encouraging residential energy conservation through feedback. In A. Baum & J. Singer (Eds.), Advances in Environmental Psychology Energy Conservation: Pyschological Perspectives (Vol. 3). Hillsdale, NJ: Erlbaum Associates. Shippee, G., & Gregory, W. L. (1982). Public commitment and energy conservation. American Journal of Community Psychology, 10(1), 81–93. http://doi.org/10.1007/BF00903306 Shippee, G., & Gregory, W. L. (1982). Public commitment and energy conservation. American Journal of Community Psychology, 10(1), 81–93. http://doi.org/10.1007/BF00903306 Sterling, M. (2014). Timestore. Retrieved August 17, 2014, from http://www.mike- stirling.com/redmine/projects/timestore Sterling, M. (2014). Timestore. Retrieved August 17, 2014, from http://www.mike- stirling.com/redmine/projects/timestore St. John, J. (2014, August 18). Data Analytics: What Utilities Are Investing In Now. Retrieved September 27, 2015, from http://www.greentechmedia.com/articles/read/data- analytics-what-utilities-want-now St. John, J. (2014, August 18). Data Analytics: What Utilities Are Investing In Now. Retrieved September 27, 2015, from http://www.greentechmedia.com/articles/read/data- analytics-what-utilities-want-now Syrek, D. B., & Legislature, M. (1980). Michigan litter: after: a study of the impact of beverage container deposit legislation on street, roadside and recreation area litter in Michigan. Retrieved from http://books.google.com/books?id=PpIQAQAAMAAJ Syrek, D. B., & Legislature, M. (1980). Michigan litter: after: a study of the impact of beverage container deposit legislation on street, roadside and recreation area litter in Michigan. Retrieved from http://books.google.com/books?id=PpIQAQAAMAAJ 129 Todorovic, D. (2008). Gestalt principles. Scholarpedia, 3(12), 5345. http://doi.org/10.4249/scholarpedia.5345 Trinh, K. (2010). Temporal De-biasing of Behaviour in Residential Energy Consumption: Supporting Conservation Compliance Through Feedback Design (Unpublished Masters thesis). University of Toronto, Toronto. Trinh, K., & Jamieson, G. A. (2014). Feedback Design Heuristics for Energy Conservation. Ergonomics in Design: The Quarterly of Human Factors Applications, 22(2), 13–21. http://doi.org/10.1177/1064804613516761 Trope, Y., & Liberman, N. (2003). Temporal construal. Psychological Review, 110 Trope, Y., & Liberman, N. (2003). Temporal construal. Psychological Review, 110(3), 403–421. Tufte, E. R. (1983). References The Visual Display of Quantitative Information (Reprinted edition. Cheshire: CT: Graphics Press. Tweed, K. (2013, October 7). Can Low-Income and Multi-Family Households Benefit From Energy Efficiency? : Greentech Media. Retrieved August 17, 2014, from http://www.greentechmedia.com/%20articles/read/Can-Low-Income-and-Multi-Family- Households-Benefit-From-Energy-Efficiency Tweed, K. (2015, May 13). Opower Books $90M Contract With PG&E—Its Biggest Ever. Retrieved September 27, 2015, from http://www.greentechmedia.com/articles/read/opower-books-90m-contract-with-pge Tweed, K. (2015, May 13). Opower Books $90M Contract With PG&E—Its Biggest Ever. Retrieved September 27, 2015, from http://www.greentechmedia.com/articles/read/opower-books-90m-contract-with-pge http://www.greentechmedia.com/articles/read/opower-books-90m-contract-with-pge Verplanken, B., & Wood, W. (2006). Interventions to break and create consumer habits. Journal of Public Policy & Marketing, 90–103. Verplanken, B., & Wood, W. (2006). Interventions to break and create consumer habits. Journal of Public Policy & Marketing, 90–103. Verplanken, B., & Wood, W. (2006). Interventions to break and create consumer habits. Journal of Public Policy & Marketing, 90–103. Winett, R. A., Neale, M. S., & Grier, H. C. (1979). Effects of self-monitoring and feedback on residential electricity consumption. Journal of Applied Behavior Analysis, 12(2), 173– 184. Winett, R. A., Neale, M. S., & Grier, H. C. (1979). Effects of self-monitoring and feedback on residential electricity consumption. Journal of Applied Behavior Analysis, 12(2), 173– 184. 130
https://openalex.org/W3004859768	https://jurnal.ugm.ac.id/jgki/article/download/41185/27781	Indonesian	null	Pola konsumsi atlet sepakbola remaja di Indonesia	Jurnal gizi klinik Indonesia	2,019	cc-by-sa	6,250	Korespondensi: Mirza Hapsari Sakti Titis Penggalih, Departemen Gizi Kesehatan, Fakultas Kedokteran, Kesehatan Masyarakat, dan Keperawatan, Universitas Gadjah Mada, Jl. Farmako, Sekip Utara, Yogyakarta 55281, Indonesia, e-mail: mirza_hapsari@yahoo.com ABSTRACT Background: Some early research shows that nutrition in adolescent soccer athlete is still less than the recommended requirements. Indonesia still lack of studies that describe the pattern of consumption of teenage football athlete. Objective: Describe the consumption pattern in Indonesia teenage football athlete. Methods: This study is an observational study at soccer boarding school in Jakarta and Malang. The subjects involved 131 athletes with age 15-19 year. This research conduct on February – May 2016. Researchers examined the intake of food the subject inside the hostel and outside the hostel with 3x24 hour food recall and semi quantitative food frequencies method. Fulfillment nutrient intake of the subject compared with the standard requirements nutrition adequacy score (AKG 2013) according to age groups. Results: The intake of macro nutrients include energy and carbohydrate was significantly lower than recommendation (p<0.000), whereas protein intake beyond the amount recommended (p<0.000). A total of 33.6% of the subjects taking the supplement with varying types. Types of supplements most consumed supplements vitargo electrolyte (18 people), followed by multivitamins and vitamin C (15 people), calcium (13 people), and herbal supplements (9 people). Conclusions: The consumption pattern teenage soccer athletes for energy and carbohydrate nutrition recommendations do not meet the standards when compared with the needs and AKG 2013. Provision of education needs to be done to improve intake patterns athletes, so the nutritional needs can be met to get the physical quality and the best performance. KEYWORDS: adolescent; athlete; carbohydrate; energy; fats; nutrients; protein; soccer Consumsion pattern of football athlete in Indonesia Consumsion pattern of football athlete in Indonesia Mirza Hapsari Sakti Titis Penggalih1, Mohammad Juffrie2, Toto Sudargo1, Zaenal Muttaqien Sofr 1Departemen Gizi Kesehatan, Fakultas Kedokteran, Kesehatan Masyarakat, dan Keperawatan, Universitas Gadjah Mada 2Departemen Ilmu Kesehatan Anak, Fakultas Kedokteran, Kesehatan Masyarakat, dan Keperawatan, Universitas Gadjah Mada 3Departemen Fisiologi, Fakultas Kedokteran, Kesehatan Masyarakat, dan Keperawatan, Universitas Gadjah Mada Mirza Hapsari Sakti Titis Penggalih, dkk: Pola konsumsi atlet sepak bola remaja di Indonesia Jurnal Gizi Klinik Indonesia Vol 15 No 3 - Januari 2019 (101-110) ISSN 1693-900X (Print), ISSN 2502-4140 (Online) Online sejak Januari 2016 di https://jurnal.ugm.ac.id/jgki Mirza Hapsari Sakti Titis Penggalih, dkk: Pola konsumsi atlet sepak bola remaja di Indonesia Jurnal Gizi Klinik Indonesia Vol 15 No 3 - Januari 2019 (101-110) ISSN 1693-900X (Print), ISSN 2502-4140 (Online) Online sejak Januari 2016 di https://jurnal.ugm.ac.id/jgki Jurnal Gizi Klinik Indonesia Vol 15 No 3 - Januari 2019 (101-110) ISSN 1693-900X (Print), ISSN 2502-4140 (Online) Online sejak Januari 2016 di https://jurnal.ugm.ac.id/jgki KATA KUNCI: remaja; atlet; karbohidrat; energi; lemak; gizi; protein; sepakbola Korespondensi: Mirza Hapsari Sakti Titis Penggalih, Departemen Gizi Kesehatan, Fakultas Kedokt Mada, Jl. Farmako, Sekip Utara, Yogyakarta 55281, Indonesia, e-mail: mirza_hapsari@yahoo.com BAHAN DAN METODE Di Indonesia, perkembangan dunia sepak bola diikuti dengan perkembangan sekolah sepak bola yang melatih dan mendidik atlet sepak bola sejak usia dini. Program pelatihan atlet-atlet muda tidak hanya berfokus pada latihan fisik dan skill saja, tetapi juga harus memperhatikan kebutuhan gizi. Program gizi yang tepat dapat menunjang pencapaian pertumbuhan dan perkembangan fisik yang optimal dan membantu mencapai kualitas performa terbaik (3). Program diet pada atlet harus dapat memenuhi seluruh kebutuhan gizi yaitu kebutuhan energi, protein, lemak, karbohidrat, serta vitamin dan mineral. Penyusunan program diet pada atlet remaja menjadi perhatian khusus karena kelompok usia tersebut memiliki kebutuhan gizi yang lebih tinggi untuk menunjang masa pertumbuhan dan perkembangan fisik (4). Asupan makan yang tidak maksimal akan berdampak pada pemenuhan kebutuhan zat gizi, pembentukan komposisi tubuh, dan tingkat performa atlet yang kurang optimal (4-6). Penelitian ini merupakan penelitian observasional dengan desain cross-sectional. Peneliti mengkaji asupan makan subjek di dalam asrama maupun di luar asrama dengan metode 3x24 jam food recall dan semi quantitative food frequencies. Data yang digali meliputi asupan energi, karbohidrat, protein, lemak, serat pangan, kolesterol, vitamin, dan mineral serta konsumsi suplemen. Data yang didapatkan merupakan rerata asupan selama 4 bulan karena waktu minimal yang dibutuhkan untuk melihat pola konsumsi adalah 90 hari atau 3 bulan. Pemenuhan zat gizi dari asupan subjek dibandingkan dengan kebutuhan standar masing-masing individu dan kebutuhan berdasarkan rekomendasi angka kecukupan gizi (AKG 2013) sesuai kelompok umur. Penelitian ini merupakan penelitian observasional dengan desain cross-sectional. Peneliti mengkaji asupan makan subjek di dalam asrama maupun di luar asrama dengan metode 3x24 jam food recall dan semi quantitative food frequencies. Data yang digali meliputi asupan energi, karbohidrat, protein, lemak, serat pangan, kolesterol, vitamin, dan mineral serta konsumsi suplemen. Penelitian ini berlangsung selama 4 bulan dimulai dari bulan Februari sampai Mei 2016. Penelitian dilaksanakan di Asrama Atlet Remaja Ragunan, Kementerian Pemuda, dan Olahraga Republik Indonesia (Kemenpora RI) Jakarta dan Sekolah Sepak Bola (SSB) Aji Santoso International Football Academy (ASIFA) Malang. Populasi target adalah atlet sepakbola laki- laki dan populasi terjangkau pada penelitian ini adalah atlet sepakbola remaja. Kriteria inklusi subjek adalah atlet sepakbola remaja yang tinggal di asrama, aktif berlatih, berusia 12-19 tahun, dan bersedia mengikuti Beberapa penelitian terdahulu menyatakan bahwa pemenuhan gizi pada atlet sepakbola remaja masih kurang dari kebutuhan yang direkomendasikan. Hal ini berhubungan dengan pengetahuan dan pemahaman atlet yang kurang tentang pemenuhan kebutuhan gizi untuk mendapatkan performa yang baik (7). PENDAHULUAN Sepak bola merupakan cabang olahraga yang paling terkenal dan banyak penggemar dari seluruh kelompok usia di seluruh dunia (1). Atlet sepak bola membutuhkan performa terbaik dalam hal teknik, taktik, fisik, dan mental. Olahraga yang termasuk dalam tipe stop and go ini membutuhkan kekuatan yang eksplosif karena atlet harus melakukan sprint setiap 90 detik selama permainan, pergerakan yang kuat dan gesit untuk melakukan tackling, heading, dan cutting sebanyak 150-250 kali dalam setiap pertandingan, serta kemampuan bertahan untuk mengontrol bola. Olahraga ini membutuhkan kekuatan, kecepatan, dan ketahanan secara aerobik maupun anaerobik yang optimal (2). Perkembangan dunia sepak bola membuat pembibitan atlet-atlet sepak bola dimulai sejak usia dini untuk melatik fisik, mental, dan keahlian yang matang supaya menjadi pemain handal (1). ABSTRAK Latar belakang: Beberapa penelitian terdahulu menunjukkan bahwa pemenuhan gizi pada atlet sepakbola remaja masih kurang dari kebutuhan yang direkomendasikan. Di Indonesia, belum banyak studi yang meneliti pola konsumsi atlet sepakbola remaja. Tujuan: Penelitian ini bertujuan untuk mengetahui pola konsumsi atlet sepakbola remaja di Indonesia. Metode: Penelitian observasional pada 131 atlet berusia 12-19 tahun di asrama sepak bola di Jakarta dan Malang pada bulan Februari-Mei 2016. Peneliti mengkaji asupan makan subjek di dalam asrama maupun di luar asrama dengan metode 3x24 jam food recall dan semi quantitative food frequencies. Pemenuhan zat gizi dari asupan subjek dibandingkan dengan kebutuhan standar/individu dan angka kecukupan gizi (2013) sesuai kelompok umur. Hasil: Asupan zat gizi makro meliputi energi dan karbohidrat secara signifikan lebih rendah dari rekomendasi berdasarkan kebutuhan standar/individu maupun AKG (p<0,000) sedangkan asupan protein melebihi jumlah yang direkomendasikan (p<0,000). Sebanyak 33,6% subjek mengonsumsi suplemen dengan jenis yang bervariasi. Jenis suplemen yang paling banyak dikonsumsi adalah suplemen elektrolit Vitargo (18 orang), diikuti dengan multivitamin dan vitamin C (15 orang), kalsium (13 orang), dan suplemen herbal (9 orang). Simpulan: Pola konsumsi atlet sepakbola remaja untuk zat gizi energi dan karbohidrat belum memenuhi rekomendasi bila dibandingkan dengan kebutuhan individu maupun AKG. Pemberian edukasi perlu dilakukan untuk memperbaiki pola asupan atlet sehingga kebutuhan zat gizi dapat dipenuhi untuk mendapatkan kualitas fisik dan performa terbaik. KATA KUNCI: remaja; atlet; karbohidrat; energi; lemak; gizi; protein; sepakbola 101 Jurnal Gizi Klinik Indonesia, Vol. 15, No. 3, Januari 2019: 101-110 memperhitungkan faktor aktivitas harian, kebutuhan yang digunakan untuk latihan juga perlu diperhitungkan. Kebutuhan zat gizi untuk latihan dapat berbeda untuk setiap atlet karena perbedaan jenis latihan fisik yang dilakukan. Dukungan nutrisi yang kurang dari para pelatih, tim, dan orangtua menjadi faktor penyebab lainnya. Beberapa studi sudah pernah dilakukan untuk melihat asupan makan atlet (1,3,7), tetapi belum membandingkan dengan kebutuhan setiap individu karena setiap atlet memiliki target latihan yang berbeda sehingga kebutuhan zat gizi yang diperlukan juga akan berbeda. Oleh karena itu, penelitian ini dilakukan untuk mengetahui gambaran konsumsi atlet sepakbola remaja yang kemudian dibandingkan dengan kebutuhan zat gizi setiap individu dan kebutuhan secara umum dari angka kecukupan gizi (AKG) untuk melihat tingkat kecukupan pemenuhan zat gizi makro. BAHAN DAN METODE Kebutuhan zat gizi atlet berbeda dengan populasi normal karena selain 102 Mirza Hapsari Sakti Titis Penggalih, dkk: Pola konsumsi atlet sepak bola remaja di Indonesia penelitian dengan menandatangani informed consent. Kriteria eksklusi yaitu atlet yang mengalami cedera sehingga tidak dapat mengikuti program latihan harian selama lebih dari 3 hari. Populasi penelitian ini sebanyak 145 orang atlet sepakbola di ASIFA Malang dan 26 orang atlet sepakbola di asrama atlet Ragunan. Semua populasi penelitian diikutkan dalam penelitian. Jumlah total subjek yang mengikuti sampai akhir penelitian ini sebanyak 131 atlet sepakbola remaja. Perhitungan besar sampel menggunakan formula dari Lameshow et al (1997) dengan tingkat kepercayaan 95% dan power 90%. Perkiraan besar sampel yang digunakan yaitu 171 orang dari total populasi terjangkau di ASIFA Malang dan asrama atlet Ragunan. Teknik pemilihan sampel menggunakan metode purposive sampling yaitu semua atlet yang berada di kedua asrama apabila memenuhi kriteria diikutsertakan dalam penelitian ini. monitoring pemenuhan kebutuhan zat gizi melalui food recall 3x24 jam dan semi quantitative food frequencies setiap bulannya. Penggunaan kedua metode tersebut digunakan untuk mengkonfirmasi kembali asupan makan yang telah dikonsumsi. Data asupan yang diperoleh merupakan data rata-rata selama 4 bulan. Asupan makan subjek diolah menggunakan perangkat Nutrisurvey untuk mengetahui pemenuhan masing-masing zat gizi. Analisis perbandingan asupan makan subjek dengan kebutuhan individu dan AKG menggunakan independent sample t-test. Penelitian ini telah memperoleh izin dari Komisi Etik Penelitian Kedokteran dan Kesehatan, Fakultas Kedokteran, Universitas Gadjah Mada dengan nomor: KE/FK/102/ EC/2016 tertanggal 1 Februari 2016. HASIL Variabel yang diteliti yaitu: 1) Zat gizi makro: energi, karbohidrat, protein, dan lemak yang didapatkan dari hasil recall 3x24 jam dan semi quantitative food frequencies; 2) Kebutuhan zat gizi standar: kebutuhan zat gizi setiap subjek penelitian dengan memperhitungkan berat badan, tinggi badan, umur, dan jenis latihan yang dilakukan. Kebutuhan standar atau kebutuhan individu setiap atlet meliputi aspek BMR, energi aktivitas harian, energi untuk penyerapan makanan, dan energi untuk latihan fisik. Semua asupan zat gizi makro (energi, karbohidrat, protein, dan lemak) dan serat dari konsumsi harian dibandingkan dengan kebutuhan individu dan AKG. Kebutuhan standar atau individu untuk serat tidak dilakukan penghitungan sehingga perbandingan yang dilakukan hanya antara asupan subjek dengan AKG. Pengukuran berat badan menggunakan timbangan badan digital Karada Scan HBF-375. Perhitungan basal metabolic rate (BMR) diperoleh dari perhitungan menggunakan formula Harris-Benedict yang memperhitungkan tinggi badan, berat badan, usia, dan jenis kelamin. Kebutuhan protein 15%, lemak 20%, dan karbohidrat 65% dari total kebutuhan energi. Seluruh subjek merupakan atlet sepakbola laki- laki berusia 12-19 tahun yang berada di Asrama Atlet Remaja Ragunan, Kemenpora RI Jakarta dan SSB ASIFA Tabel 1. Karakteristik subjek penelitian (n=131) Tabel 1. Karakteristik subjek penelitian (n=131) Keterangan n % Umur (tahun) 12-15 90 68,7 15-19 41 31,3 Lama bergabung dengan tim (tahun) 0-4 96 73,3 5-8 24 18,3 9-12 11 8,4 Durasi mengikuti kompetisi (kali/tahun) 0-1 34 25,9 2-3 63 48,1 4 34 26 Jumlah klub yang pernah diikuti 0-1 83 63,4 2-3 43 32,8 4-5 5 3,8 Kebiasaan tidur (jam/hari) 6-8 103 78,6 8,5-10 22 16,8 >10 6 4,6 Jam sekolah (jam/hari) 3-4 87 66,4 4,5-5 43 32,8 >5 1 0,8 1Lama bergabung dengan tim sejak awal menjadi atlet (tahun) Pengambilan data awal subjek meliputi tinggi badan, berat badan, aktivitas latihan, dan food recall 3x24 jam. Selanjutnya data yang diperoleh digunakan untuk menghitung kebutuhan zat gizi masing-masing individu sesuai aktivitas yang dilakukan. Selama 3 bulan dilakukan 103 Jurnal Gizi Klinik Indonesia, Vol. 15, No. 3, Januari 2019: 101-110 Tabel 2. Jenis makanan yang disajikan Tabel 2. HASIL Jenis makanan yang disajikan Jenis makanan Asrama atlet Ragunan Asrama atlet ASIFA Karbohidrat Nasi putih Nasi putih, nasi merah Lauk hewani Ayam, telur, ikan, daging (dan olahannya) Ayam, telur, ikan, daging (dan olahannya) Lauk nabati Tahu, tempe Tahu, tempe Sayur Disajikan setiap waktu makan Disajikan setiap waktu makan Buah Disajikan setiap waktu makan Disajikan setiap waktu makan Susu Disajikan saat makan pagi dan malam Disajikan saat makan pagi Minuman lainnya Air putih, teh, sirup Air putih Tabel 3. Pemenuhan asupan zat gizi berdasarkan food recall 3x24 jam Parameter Minimal Maksimal X SD p1 Asupan zat gizi makro total Energi (kkal) 852,3 2510,1 1750,3 289,08 0,200 Karbohidrat (g) 120,33 349,13 243,2 56,57 0,006 Protein (g) 37,3 116,23 81,92 14,71 0,200 Lemak (g) 24,8 101,6 62,17 17,27 0,000 Serat pangan (g) 3,00 188,25 15,67 21,96 0,000 Asupan zat gizi dalam asrama Energi (kkal) 695,03 1873,57 1247,73 238,50 0,200 Karbohidrat (g) 89,62 294,20 168,16 43,8 0,000 Protein (g) 35,02 119,23 75,97 15,65 0,200 Lemak (g) 25,93 80,66 47,38 9,63 0,200 Serat pangan (g) 3,44 29,84 10,26 3,27 0,013 Asupan zat gizi dari luar asrama Energi (kkal) 78,01 1346,61 526,25 283,52 0,001 Karbohidrat (g) 10,64 227,30 70,22 43,48 0,000 Protein (g) 2,99 58,08 17,84 8,97 0,043 Lemak (g) 2,37 49,89 20,50 10,57 0,001 Serat pangan (g) 0,26 62,97 4,35 7,37 0,000 1Nilai p hasil uji normalitas Kolmogorov-Smirnov, data terdistribusi normal jika p>0,05 Tabel 3. Pemenuhan asupan zat gizi berdasarkan food recall 3x24 jam 1.750±289,07 kkal, jauh lebih rendah dari rerata kebutuhan energi standar (2.555,55±439,18 kkal) dan kebutuhan berdasarkan AKG (2.575 kkal) (p=0,000 dan p=0,000). Demikian juga dengan pemenuhan karbohidrat total (243±56,57 g) juga yang masih di bawah kebutuhan standar dan AKG (p=0,000 dan p=0,000). Sebaliknya, rerata pemenuhan protein dari asupan secara signifikan lebih tinggi dibandingkan kebutuhan standar dan AKG (p=0,000 dan p=0,000). Sementara itu, pemenuhan lemak dari asupan (62,16±17,27 g) lebih tinggi daripada kebutuhan standar (56,79±9,76 g), tetapi lebih rendah jika dibandingkan dengan kebutuhan berdasarkan AKG (86 g). Tujuan dilakukan perbandingan antara makanan luar asrama dengan kebutuhan standar dan AKG adalah untuk mengetahui seberapa besar kontribusi makanan Malang. Detail karakteristik subjek meliputi umur, lama bergabung dalam tim, durasi mengikuti kompetisi, jumlah klub yang pernah diikuti, kebiasaan tidur, dan jam sekolah yang ditampilkan dalam Tabel 1. Tabel 2 menunjukkan secara deskriptif jenis makanan yang disajikan di kedua asrama setiap kali waktu makan. HASIL Pemenuhan zat gizi subjek yang dianalisis yaitu energi, protein, lemak, karbohidrat, dan serat pangan ditampilkan pada Tabel 3. Hasil uji normalitas untuk asupan makro menunjukkan bahwa beberapa data asupan tidak terdistribusi normal (p<0,05). Malang. Detail karakteristik subjek meliputi umur, lama bergabung dalam tim, durasi mengikuti kompetisi, jumlah klub yang pernah diikuti, kebiasaan tidur, dan jam sekolah yang ditampilkan dalam Tabel 1. Tabel 2 menunjukkan secara deskriptif jenis makanan yang disajikan di kedua asrama setiap kali waktu makan. Pemenuhan zat gizi subjek yang dianalisis yaitu energi, protein, lemak, karbohidrat, dan serat pangan ditampilkan pada Tabel 3. Hasil uji normalitas untuk asupan makro menunjukkan bahwa beberapa data asupan tidak terdistribusi normal (p<0,05). Tabel 4 menunjukkan bahwa rerata pemenuhan zat gizi makro berbeda signifikan jika dibandingkan dengan standar kebutuhan individu dan AKG. Rerata pemenuhan energi berdasarkan asupan total sebesar 104 Mirza Hapsari Sakti Titis Penggalih, dkk: Pola konsumsi atlet sepak bola remaja di Indonesia Tabel 4. Perbandingan pemenuhan zat gizi makro berdasarkan standar kebutuhan individu dan AKG Zat gizi Asupan (X±SD) Standar (X±SD) AKG p Asupan dalam asrama Energi (kkal) 1.247,73 ± 238,50 2.555,55 ± 439,18 2575 0,0001 0,0002 Protein (g) 75,97 ± 15,65 66,58 ± 1,46 69 0,0001 0,0002 Lemak (g) 47,38 ± 9,63 56,79 ± 9,76 86 0,0001 0,0002 Karbohidrat (g) 168,16 ± 43,83 288,52 ± 58,32 354 0,0001 0,0002 Serat pangan (g) 10,26 ± 3,27 - 36 0,0002 Asupan luar asrama Energi (kkal) 526,25 ± 283,52 2.555,55 ± 439,18 2575 0,0001 0,0002 Protein (g) 17,84 ± 8,97 66,58 ± 13,46 69 0,0001 0,0002 Lemak (g) 20,50 ± 10,57 56,79 ± 9,76 86 0,0001 0,0002 Karbohidrat (g) 70,22 ± 43,48 288,52 ± 58,32 354 0,0001 0,0002 Serat pangan (g) 4,35 ± 7,37 - 36 0,0002 Asupan total Energi (kkal) 1.750 ± 289,07 2.555,55 ± 439,18 2575 0,0001 0,0002 Protein (g) 81,19 ± 14,7 66,58 ± 13,46 69 0,0001 0,0002 Lemak (g) 62,16 ± 17,27 56,79 ± 9,76 86 0,0421 0,0002 Karbohidrat (g) 243 ± 56,57 288,52 ± 58,32 354 0,0001 0,0002 Serat pangan (g) 15,67 ± 21,96 - 36 0,0002 AKG = angka kecukupan gizi; 1Tes non parametrik beda kelompok antara asupan dan standar, p<0,05 terdapat perbedaan signifikan 2One sample t-test antara asupan dan AKG, p<0,05 terdapat perbedaan signifikan Tabel 4. BAHASAN Kedua asrama sepakbola yang menjadi tempat penelitian ini memberikan pelayanan makan pada atlet dengan sistem yang sedikit berbeda. Pada asrama atlet remaja Ragunan, atlet bebas mengambil sendiri menu makanan yang telah disiapkan oleh pihak asrama. Sementara pada asrama ASIFA, porsi lauk, sayur, dan buah telah disiapkan oleh pramusaji pada plato makan sehingga atlet hanya perlu mengambil nasi sendiri sesuai porsi yang diinginkan. Makan pagi mulai disajikan pukul 06.00-08.00, makan siang pukul 11.00-13.00, dan makan malam pukul 17.00-19.00 di asrama atlet Ragunan. Sementara penyajian makan di asrama atlet ASIFA dimulai pukul 06.00-07.00 untuk makan pagi, makan siang pukul 12.00-13.00, dan makan malam pukul 18.00- 19.00. Jenis makanan dan minuman yang disajikan pada kedua asrama atlet juga sedikit berbeda. Jenis karbohidrat yaitu nasi merah tidak diberikan di asrama atlet Ragunan, sementara jenis minuman yang lebih banyak disediakan di Asrama atlet Ragunan yaitu air putih, susu, teh, dan sirup sedangkan di asrama atlet ASIFA hanya disediakan air putih dan susu. gan kebutuhan ergi untuk ngan AKG 2013 ergi, protein, dan u, hal ini karena u, asupan lemak individu karena dikonsumsi saat ruhan 131 subjek n kebutuhan gi untuk gan AKG 2013 gi, protein, dan hal ini karena asupan lemak ndividu karena dikonsumsi saat han 131 subjek Karbohidra Karbohidrat Gambar 2. Persentase pemenuhan asupan energi dan zat gizi makro dibandingkan dengan kebutuhan individu yang memperhitungkan BMR, energi aktivitas fisik, energi latihan, dan energi untuk penyerapan makanan ase pemenuhan asupan energi dan zat gizi makro dibandingkan memperhitungkan BMR, energi aktivitas fisik, energi latihan, d penyerapan makanan tase pemenuhan asupan energi dan zat gizi makro dibandingkan d memperhitungkan BMR, energi aktivitas fisik, energi latihan, dan penyerapan makanan apabila menggunakan kebutuhan individu, hal ini karena kebutuhan energi individu masih ada yang berada di bawah AKG 2013. Sementara itu, asupan lemak menunjukkan persen pemenuhan yang lebih tinggi jika dibandingkan dengan kebutuhan individu karena pembagian kebutuhan zat gizi lemak yang lebih kecil jika dibandingkan dengan AKG 2013. unjukkan persentase pemenuhan asupan makan subjek dibandingka an dengan kebutuhan individu ditampilkan pada Gambar 2. Asupa kkan hasil yang lebih sesuai apabila menggunakan kebutuhan ind dividu masih ada yang berada di bawah AKG 2013. Sementa pemenuhan yang lebih tinggi jika dibandingkan dengan kebut nunjukkan persentase pemenuhan asupan makan subjek dibandingkan gan dengan kebutuhan individu ditampilkan pada Gambar 2. Asupan kkan hasil yang lebih sesuai apabila menggunakan kebutuhan indiv ndividu masih ada yang berada di bawah AKG 2013. HASIL Perbandingan pemenuhan zat gizi makro berdasarkan standar kebutuhan individu dan AKG AKG = angka kecukupan gizi; 1Tes non parametrik beda kelompok antara asupan dan standar, p<0,05 terdapat perbedaan signifikan 2One sample t-test antara asupan dan AKG, p<0,05 terdapat perbedaan signifikan yang disediakan oleh asrama belum memenuhi kebutuhan atlet berdasarkan standar kebutuhan individu maupun AKG, jika tidak ditambahkan energi yang berasal dari makanan luar asrama. Pemenuhan protein yang berasal dari makanan dalam asrama ternyata sudah mampu memenuhi kebutuhan standar dan AKG sehingga penambahan asupan protein dari makanan luar asrama akan meningkatkan pemenuhan asupan zat gizi protein. luar asrama dalam memenuhi kebutuhan atlet. Tujuan lainnya yaitu untuk melihat rerata tingkat asupan dari makanan luar asrama yang menggambarkan kepatuhan atlet. Apabila konsumsi makanan dari luar asrama besar, maka perlu evaluasi penyelenggaraan makanan dalam asrama karena atlet lebih memilih jajan di luar daripada mengonsumsi makanan yang sudah disediakan di asrama. luar asrama dalam memenuhi kebutuhan atlet. Tujuan lainnya yaitu untuk melihat rerata tingkat asupan dari makanan luar asrama yang menggambarkan kepatuhan atlet. Apabila konsumsi makanan dari luar asrama besar, maka perlu evaluasi penyelenggaraan makanan dalam asrama karena atlet lebih memilih jajan di luar daripada mengonsumsi makanan yang sudah disediakan di asrama. yang disediakan oleh asrama belum memenuhi kebutuhan atlet berdasarkan standar kebutuhan individu maupun AKG, jika tidak ditambahkan energi yang berasal dari makanan luar asrama. Pemenuhan protein yang berasal dari makanan dalam asrama ternyata sudah mampu memenuhi kebutuhan standar dan AKG sehingga penambahan asupan protein dari makanan luar asrama akan meningkatkan pemenuhan asupan zat gizi protein. Asupan makan subjek penelitian berasal dari makanan dalam dan luar asrama. Jumlah total kalori makanan yang disediakan oleh pihak asrama sebesar 2.500 kkal, tetapi ternyata hasil analisis menunjukkan rerata konsumsi atlet yang berasal dari makanan asrama adalah 1.200 kkal. Total asupan energi dari makanan Gambar 1 menunjukkan persentase pemenuhan asupan makan subjek dibandingkan dengan AKG 2013 sedangkan perbandingan dengan kebutuhan individu ditampilkan pada Gambar 2. Asupan energi, protein, dan karbohidrat menunjukkan hasil yang lebih sesuai 105 Jurnal Gizi Klinik Indonesia, Vol. 15, No. 3, Januari 2019: 101-110 Vitargo (18 orang), diikuti dengan multivitamin dan vitamin C (15 orang), kalsium (13 orang), dan suplemen herbal (9 orang). Tabel 5. HASIL Daftar konsumsi suplemen Jenis suplemen n Multivitamin 8 Vitamin C 7 Kalsium 13 Herbal 9 Madu 1 BCAA 3 Vitargo (elektrolit/recovery) 18 Minyak ikan 5 Isogenik 4 Susu 2 Tidak menggunakan suplemen 87 gan AKG 2013 an AKG 2013 Gambar 1. Persentase pemenuhan asupan energi dan zat gizi makro dibandingkan dengan AKG 2013 ase pemenuhan asupan energi dan zat gizi makro dibandingkan 80,9 0 56,5 77,8 19,9 9,9 3,8 18,3 0 90,1 39,7 3,9 Energi Protein Lemak Karbohidrat Kurang Cukup Lebih ntase pemenuhan asupan energi dan zat gizi makro dibandingkan d 80,9 0 56,5 77,8 19,9 9,9 3,8 18,3 0 90,1 39,7 3,9 Energi Protein Lemak Karbohidrat Kurang Cukup Lebih Vitargo (18 orang), diikuti dengan multivitamin dan vitamin C (15 orang), kalsium (13 orang), dan suplemen herbal (9 orang). Tabel 5. Daftar konsumsi suplemen Jenis suplemen n Multivitamin 8 Vitamin C 7 Kalsium 13 Herbal 9 Madu 1 BCAA 3 Vitargo (elektrolit/recovery) 18 Minyak ikan 5 Isogenik 4 Susu 2 Tidak menggunakan suplemen 87 gan AKG 2013 an AKG 2013 Tabel 5. Daftar konsumsi suplemen Gambar 1. Persentase pemenuhan asupan energi dan zat gizi makro dibandingkan dengan AKG 2013 ase pemenuhan asupan energi dan zat gizi makro dibandingkan tase pemenuhan asupan energi dan zat gizi makro dibandingkan d Vitargo (18 orang), diikuti dengan multivitamin dan vitamin C (15 orang), kalsium (13 orang), dan suplemen herbal (9 orang). Vitargo (18 orang), diikuti dengan multivitamin dan vitamin C (15 orang), kalsium (13 orang), dan suplemen herbal (9 orang). Gambar 2. Persentase pemenuhan asupan energi dan zat gizi makro dibandingkan dengan kebutuhan individu yang memperhitungkan BMR, energi aktivitas fisik, energi latihan, dan energi untuk penyerapan makanan ase pemenuhan asupan energi dan zat gizi makro dibandingkan memperhitungkan BMR, energi aktivitas fisik, energi latihan, d penyerapan makanan 77,9 0 1,6 55,7 22,1 16,8 32,8 43,5 0 83,2 65,6 0,8 Energi Protein Lemak Karbohidrat Kurang Cukup Lebih tase pemenuhan asupan energi dan zat gizi makro dibandingkan d memperhitungkan BMR, energi aktivitas fisik, energi latihan, dan penyerapan makanan 77,9 0 1,6 55,7 22,1 16,8 32,8 43,5 0 83,2 65,6 0,8 Energi Protein Lemak Karbohidrat Kurang Cukup Lebih 77,9 0 1,6 55,7 22,1 16,8 32,8 43,5 0 83,2 65,6 0,8 Energi Protein Lemak Karbohidrat Kurang Cukup Lebih 77,9 0 1,6 55,7 22,1 16,8 32,8 43,5 0 83,2 65,6 0,8 Energi Protein Lemak Karbohidrat Kurang Cukup Lebih BAHASAN Makanan yang berasal dari dalam asrama ternyata belum mampu memenuhi kebutuhan zat gizi energi dan karbohidrat. Hal ini bukan berarti makanan yang disajikan sedikit, tetapi karena subjek tidak menghabiskan makanan yang sudah disajikan. Tingkat konsumsi makanan dalam asrama yang belum maksimal disebabkan oleh penggunaan siklus menu 4 hari sehingga pengulangan menu lebih sering terjadi dan atlet menjadi mudah merasa bosan dengan makanan yang disajikan. Temuan utama dalam penelitian ini adalah kebutuhan gizi makro energi dan karbohidrat atlet yang belum tercukupi jika dibandingkan dengan kebutuhan standar individu dan AKG 2013 sesuai kelompok umur. Pemenuhan energi dan karbohidrat dari asupan makan secara signifikan lebih rendah dari standar (p=0,000). Rerata pemenuhan energi total hanya memenuhi ±68,5% dari kebutuhan energi standar dan AKG, sementara asupan karbohidrat total memenuhi ±84% dari kebutuhan standar dan ±68% dari AKG. Penyelenggaraan makanan pada sebuah klub atau asrama atlet sangat mempengaruhi jumlah asupan yang dikonsumsi oleh atlet. Ketersediaan makanan yang sesuai dengan kebutuhan atlet baik dari segi jumlah dan jenisnya harus benar-benar diperhitungkan. Jumlah makanan yang disajikan untuk setiap atlet harus disesuaikan dengan kebutuhan gizi melalui koreksi aktivitas yang dilakukan. Jenis menu yang disajikan juga harus beragam untuk mencegah pengulangan menu yang disajikan dan membuat atlet bosan sehingga asupan makan tidak optimal. Review dari ahli gizi olahraga pada Olympic Games 2000 menyebutkan siklus menu 10 hari dinilai paling baik untuk mencegah timbulnya rasa bosan (10). Memenuhi kebutuhan energi merupakan hal yang esensial untuk menjaga massa otot, mengoptimalkan output dari program latihan yang dilakukan, serta memastikan tercukupinya kebutuhan zat gizi lainnya (5). Pada usia remaja, pemenuhan kebutuhan energi diperlukan untuk menunjang pertumbuhan dan perkembangan fisik serta menjaga kecukupan energi untuk melakukan aktivitas fisik atau latihan (8). Asupan energi yang kurang akan berdampak pada penurunan massa lemak, penurunan densitas tulang, mempercepat atlet mencapai kelemahan atau fatigue, dan meningkatkan risiko cedera (9). Sejauh ini, belum ada metode perhitungan kebutuhan energi pada usia remaja yang sederhana sehingga parameter tumbuh kembang atlet harus diperhatikan untuk memantau apakah energi total yang ditentukan telah memenuhi kebutuhan yang optimal (8). Jumlah makanan yang disajikan untuk setiap atlet harus disesuaikan dengan kebutuhan gizi melalui koreksi aktivitas yang dilakukan. Jenis menu yang disajikan juga harus beragam untuk mencegah pengulangan menu yang disajikan dan membuat atlet bosan sehingga asupan makan tidak optimal. Review dari ahli gizi olahraga pada Olympic Games 2000 menyebutkan siklus menu 10 hari dinilai paling baik untuk mencegah timbulnya rasa bosan (10). BAHASAN Sementara n pemenuhan yang lebih tinggi jika dibandingkan dengan kebutuh Lebih lanjut, jenis suplemen terbanyak yang dikonsumsi oleh atlet yaitu elektrolit yang dikonsumsi saat recovery (Tabel 5). Pengkajian terhadap konsumsi suplemen menunjukkan bahwa dari keseluruhan 131 subjek penelitian, hanya sebagian kecil (33,6%) yang mengonsumsi suplemen dan mayoritas (66,4%) tidak mengonsumsi suplemen. Konsumsi suplemen pada subjek cukup bervariasi, dengan jenis suplemen yang paling banyak dikonsumsi adalah suplemen elektrolit pemenuhan yang lebih tinggi jika dibandingkan dengan kebutu zat gizi lemak yang lebih kecil jika dibandingkan dengan AKG 2013 is suplemen terbanyak yang dikonsumsi oleh atlet yaitu elektrolit engkajian terhadap konsumsi suplemen menunjukkan bahwa dari ke bagian kecil (33,6%) yang mengonsumsi suplemen dan may n zat gizi lemak yang lebih kecil jika dibandingkan dengan AKG 2013. nis suplemen terbanyak yang dikonsumsi oleh atlet yaitu elektrolit ya Pengkajian terhadap konsumsi suplemen menunjukkan bahwa dari kes ebagian kecil (33,6%) yang mengonsumsi suplemen dan mayor Pengkajian pemenuhan zat gizi atlet dilakukan dengan membandingkan rerata pemenuhan zat gizi berdasarkan asupan dan standar kebutuhan gizi yang telah dihitung oleh ahli gizi sesuai dengan berat badan, tinggi badan, usia, dan aktivitas fisik yang dilakukan. ruhan 131 subjek s (66,4%) tidak (66,4%) tidak 106 Mirza Hapsari Sakti Titis Penggalih, dkk: Pola konsumsi atlet sepak bola remaja di Indonesia Kebutuhan protein ditentukan sebanyak 15%, karbohidrat sebesar 65%, dan lemak 20% dari energi total. Selain itu, perbandingan pemenuhan gizi juga dilakukan antara asupan dan AKG untuk orang Indonesia pada populasi laki-laki usia 13-18 tahun. Data pemenuhan asupan subjek dikumpulkan menggunakan metode 3x24 jam food recall dan semi quantitative food frequencies yang mengkaji asupan harian subjek di dalam asrama maupun di luar asrama. Jika dibandingkan antara standar kebutuhan individu dan AKG 2013 ternyata tidak berbeda jauh, walaupun menunjukkan standar kebutuhan individu lebih tinggi ±439 kkal. Hal ini menunjukkan bahwa aktivitas fisik yang dilakukan oleh atlet belum terlalu tinggi sehingga kebutuhan kalori subjek tidak berbeda jauh apabila dibandingkan dengan AKG 2013. atau recovery di antara jadwal pertandingan (1). Asupan karbohidrat yang cukup, bermanfaat untuk menjaga ketersediaan glukosa yang diperlukan sistem saraf pusat dan simpanan glikogen dalam otot yang kebutuhannya meningkat saat latihan (8). Kebutuhan karbohidrat bagi atlet bervariasi mulai dari 6-10 g/kg berat badan atau memenuhi setidaknya 45%-65% kebutuhan energi total. Karbohidrat yang dibutuhkan untuk atlet kategori sepakbola yaitu 65% dari kebutuhan energi total. Kebutuhan karbohidrat dipengaruhi juga oleh jenis kelamin, usia, tipe olahraga, dan kondisi lingkungan (1,6). BAHASAN Beberapa penelitian juga menunjukkan bahwa pemenuhan asupan karbohidrat dan energi untuk atlet masih kurang dari kebutuhan. Sebuah studi yang melakukan pengamatan selama tiga tahun pada atlet dengan kategori endurance, melaporkan bahwa kebutuhan zat gizi energi dan karbohidrat belum terpenuhi pada mayoritas subjek (80,8%) (11). Penelitian lainnya menyebutkan asupan makan terutama untuk karbohidrat pada atlet sepak bola masih tergolong rendah sedangkan asupan protein dan lemak cenderung lebih tinggi (12). Kecukupan asupan karbohidrat menjadi faktor utama yang diperlukan untuk memenuhi kebutuhan energi saat latihan dan mempercepat masa pemulihan 107 Jurnal Gizi Klinik Indonesia, Vol. 15, No. 3, Januari 2019: 101-110 yang direkomendasikan AKG, yaitu hanya memenuhi 15,67 g dari kebutuhan 36 g/hari (43,52%). Hasil analisis menunjukkan rerata pemenuhan kebutuhan protein total dari asupan makan atlet secara signifikan lebih tinggi dari kebutuhan standar dan AKG, yaitu mencapai 122% dan 118% secara berturut- turut (p=0,000). Temuan ini sejalan dengan hasil studi yang menyimpulkam bahwa seringkali asupan protein atlet lebih tinggi dari rekomendasi, sementara asupan karbohidrat lebih rendah (13). Rekomendasi kebutuhan protein bervariasi antara 1,2-1,7 g/kg berat badan atau setidaknya memenuhi 10-30% dari kebutuhan energi total. Kebutuhan protein untuk atlet cabang olahraga sepakbola yaitu 15%. Kecukupan kebutuhan protein penting untuk pembentukan dan perbaikan otot serta membantu menjaga kadar glukosa darah melalui proses glukoneogenesis di hati (6,9). Pemenuhan kebutuhan protein ternyata sudah bisa terpenuhi dari makanan yang disajikan dari dalam asrama. Hal ini menunjukkan bahwa atlet cenderung lebih menyukai lauk yang disajikan daripada menu karbohidrat utama. Belum tercukupinya sebagian besar kebutuhan zat gizi tersebut dipengaruhi oleh beberapa hal, salah satunya adalah faktor kesukaan dan preferensi atlet terhadap makanan (1). Berdasarkan hasil pengamatan, atlet di asrama Ragunan dibebaskan untuk mengambil makanan sendiri dan tidak terdapat ahli gizi atau petugas lain yang mengawasi. Kondisi tersebut membuat atlet mengambil makanan sesuai keinginan dan hasil pengamatan terhadap sisa makanan yang terbanyak adalah makanan jenis sayuran. Sementara itu, meskipun di asrama ASIFA porsi lauk dan sayur telah ditentukan oleh pramusaji, tidak adanya pengawasan dari ahli gizi membuat sebagian atlet tidak menghabiskan porsi lauk dan sayur yang telah ditentukan. Berdasarkan pengamatan, mudahnya akses bagi atlet untuk membeli makanan dari luar asrama juga mendorong atlet untuk mengonsumsi makanan yang tidak dianjurkan, seperti makanan yang tinggi lemak karena sebagian besar pedagang makanan disekitar asrama menjual makanan yang digoreng. Lebih lanjut, rerata pemenuhan kebutuhan lemak dari asupan total secara signifikan lebih tinggi dari kebutuhan standar (p=0,000), tetapi lebih rendah dari rekomendasi berdasarkan AKG (p=0,000). BAHASAN Asupan lemak subjek dari makanan yang disajikan oleh asrama belum bisa memenuhi kebutuhan standar dan AKG 2013. Makanan dari luar asrama ternyata mampu menyumbang lemak yang cukup untuk memenuhi kebutuhan standar, tetapi belum mencapai pemenuhan yang direkomendasikan oleh AKG 2013. Hal ini karena berbeda dengan AKG, standar penentuan kebutuhan lemak menggunakan perhitungan sebanyak 20% dari total kebutuhan energi. Pemenuhan kebutuhan lemak penting untuk absorbsi vitamin larut lemak (A, D, E, dan K), menyediakan asam lemak esensial, dan melindungi organ vital. Atlet direkomendasikan memenuhi kebutuhan lemak sebesar 20-30% dari energi total. Pemenuhan zat gizi bagi atlet merupakan hal yang sangat penting untuk diperhatikan. Tubuh memerlukan energi yang cukup untuk melakukan serangkaian metabolisme dalam menggerakan otot pada saat atlet beraktivitas. Pemenuhan energi dan zat gizi seharusnya bisa terpenuhi dari makanan yang disediakan oleh asrama dan atlet bisa meminimalkan konsumsi makanan dari luar asrama. Pendampingan ahli gizi kepada atlet untuk meningkatkan pengetahuan dan perilaku tentang pemenuhan asupan zat gizi menjadi penting dilakukan. Sebuah studi menyebutkan bahwa intervensi edukasi gizi pada atlet selama 5 minggu sudah mampu meningkatkan secara signifikan pengetahuan mengenai gizi (14). Hasil penelitian lainnya menunjukkan bahwa intervensi edukasi gizi selama 4 bulan oleh ahli gizi kepada atlet terbukti dapat meningkatan pemenuhan asupan energi dan zat gizi makro (15,16). Beberapa penelitian tersebut dapat digunakan sebagai acuan teknik intervensi maupun durasi yang dibutuhkan untuk penelitian intervensi lanjutan dengan tujuan meningkatkan pengetahuan dan pemenuhan asupan zat gizi pada atlet. Asupan lemak lebih dari 20% dapat menyebabkan peningkatan persentase lemak dan berat badan yang tidak diharapkan sehingga meningkatkan risiko kelebihan berat badan. Sebaliknya, asupan kurang dari 20% tidak dianjurkan karena dapat mempengaruhi pencapaian kualitas performa yang kurang optimal (9). Kebutuhan serat pangan juga ditemukan masih kurang dari jumlah 108 Mirza Hapsari Sakti Titis Penggalih, dkk: Pola konsumsi atlet sepak bola remaja di Indonesia Pendampingan ahli gizi pada atlet dibutuhkan untuk mengoptimalkan konsumsi makan bagi atlet. Pengkajian terhadap konsumsi suplemen menunjukkan bahwa sebanyak 33% dari seluruh subjek mengonsumsi suplemen dengan jenis yang bervariasi. Jenis suplemen yang paling banyak dikonsumsi adalah minuman elektrolit/recovery Vitargo, multivitamin, vitamin C, kalsium, dan suplemen herbal. Salah satu keterbatasan penelitian ini yaitu belum mencantumkan pemenuhan kebutuhan zat gizi mikro sedangkan kandungan zat gizi pada suplemen sebagian besar adalah vitamin dan mineral sehingga belum bisa diketahui berapa persen kontribusi suplemen terhadap kebutuhan zat gizi mikro atlet. RUJUKAN 1. Garcia-Roves PM, Garcia-Zapico P, Patterson AM, Iglesias-Gutierrez E. Nutrient intake and food habits of soccer players: analyzing the correlates of eating practice. Nutrients. 2014;6(7):2697-717. doi: 10.3390/nu6072697 2. Kwan Chan HC, Pui Fong DT, Yuk Lee JW, Ching Yau QK, Hang Yung PS, Ming Chan K. Power and endurance in Hong Kong professional football players. Asia Pac J Sports Med Arthrosc Rehabil Technol. 2016;5:1-5. doi: 10.1016/j.asmart.2016.05.001 3. Iglasias-Guitare E, Garcia-Roves PM, Garcia A, Patterson AM. Food preferences do not influence adolescent high- level athletes’ dietary intake. Appetite. 2008; 50(2-3):536- 43. doi: 10.1016/j.appet.2007.11.003 4. Jorge Molina L, Molina JM, Chirosa LJ, Florea D, Sáez L, Planells E, et al. Implementation of a nutrition education program in a handball team: consequences on nutritional status. Nutr Hosp. 2013;28(4):1065-76. doi: 10.3305/ nh.2013.28.4.6600 5. Murphy S, Jeanes Y. Nutritional knowledge and dietary intake of young professional football players. Nutrition & Food Science. 2006;36(5):343-8. doi: 10.1108/00346650610703199 5. Murphy S, Jeanes Y. Nutritional knowledge and dietary intake of young professional football players. Nutrition & Food Science. 2006;36(5):343-8. doi: 10.1108/00346650610703199 Pernyataan konflik kepentingan Penulis menyatakan tidak ada konflik kepentingan dengan pihak-pihak yang terkait dalam penelitian ini. Penulis menyatakan tidak ada konflik kepentingan dengan pihak-pihak yang terkait dalam penelitian ini. BAHASAN Penelitian terdahulu pada atlet remaja menggambarkan temuan yang serupa, yaitu suplemen yang paling sering dikonsumsi adalah multivitamin dan multimineral, protein atau asam amino, serta suplemen herbal (17). Konsumsi suplemen pada atlet perlu menjadi perhatian tersendiri karena selain membutuhkan biaya yang lebih, konsumsi suplemen yang tidak tepat justru dapat memberikan pengaruh negatif pada atlet atau efek doping yang tidak disadari (18). Pada dasarnya tidak ada rekomendasi khusus tentang konsumsi suplemen bagi atlet selama kebutuhan zat gizi dari diet telah tercukupi dengan baik. Penggunaan suplemen mungkin diperlukan untuk tujuan tertentu seperti pada masa pemulihan (recovery), pada atlet yang mengalami anemia, atau saat dalam keadaan sakit (17). Dengan demikian, atlet perlu mendapatkan edukasi mengenai konsumsi suplemen, meliputi fungsi berbagai jenis suplemen, batas aman penggunaannya, serta kapan suplemen perlu untuk dikonsumsi. UCAPAN TERIMA KASIH Peneliti mengucapkan terima kasih kepada pemberi bantuan dana penelitian melalui Hibah Penelitian Disertasi Fakultas Kedokteran Universitas Gadjah Mada tahun 2016. SIMPULAN DAN SARAN Pola konsumsi atlet sepakbola remaja untuk zat gizi energi dan karbohidrat belum memenuhi rekomendasi jika dibandingkan dengan kebutuhan standar maupun AKG. Makanan yang disediakan oleh asrama belum dikonsumsi secara maksimal oleh atlet dibuktikan dengan rerata jumlah asupan makan yang berasal dari dalam asrama sebesar 1.200 kkal sedangkan pihak asrama setiap hari menyediakan makanan sebesar 2.500 kkal. Pola konsumsi suplemen pada atlet bervariasi, dengan jenis suplemen yang paling banyak dikonsumsi adalalah minuman recovery Vitargo, multivitamin, dan vitamin. 6. Manore MM, Thompson JL. Energy requirements of the athlete: assessment and evidence of energy deficiency. In: Clinical Sports Nutrition (Burke L, Deakin V, Eds). Roseville, CA, USA: McGraw-Hill Book Company Australia; 2006. 7. Burke LM, Hawley JA, Wong SHS, Jeukendrup AE. Carbohydrates for training and competition. Journal of Sports Sciences. 2011;29(1):S17-S27. doi: 10.1080/02640414.2011.585473 8. American College of Sport Medicine. Nutrition and athletic performance: joint position statement. Canada: 8. American College of Sport Medicine. Nutrition and athletic performance: joint position statement. Canada: 109 Jurnal Gizi Klinik Indonesia, Vol. 15, No. 3, Januari 2019: 101-110 American College of Sports Medicine, American Dietetic Association, Dietitians of Canada; 2009. American College of Sports Medicine, American Dietetic Association, Dietitians of Canada; 2009. 14. Rivera-Brown AM, Gutierrez R, Gutierrez JC, Frontera WR, Bar-Or O. Drink composition, voluntary drinking, and fluid balance in exercising, trained, heat-acclimatized boys. J Appl Physiol. 1999;86(1):78-84. doi: 10.1152/ jappl.1999.86.1.78 9. Purcell LK. Sport nutrition for young athletes. Pediatric Child Health 2013;18(4):200-2. doi: 10.1093/pch/18.4.200 10. Pelly F, O’Connor H, Denyer G, Caterson I. Catering for the athletes village at the Sydney 2000 olympic games: the role of sports dietitians. Int J Sport Nutr Exerc Metab. 2009;19(4):340-54. doi: 10.1123/ijsnem.19.4.340 15. Molina-Lopez J, Molina JM, Chirosa LJ, Florea D, Sáez L, Planells E, et al. Implementation of a nutrition education program in a handball team; consequences on nutritional status. Nutr Hosp 2013;28(4):1065-76. doi: 10.3305/ nh.2013.28.4.6600 11. Baranauskas M, Stukas R, Tubelis L, Zagminas K, Surkiene G, Abaravicius JA, et al. Nutritional habits among haigh-performance endurance athletes. Medicina (Kaunas). 2015;51(6):351-62. doi: 10.1016/j.medici.2015.11.004 16. Valliant MW, Emplaincourt HP, Wenzel RK, Garner BH. Nutrition education by a registered dietitian improves dietary intake and nutrition knowledge of a NCAA volleyball team. Nutrients. 2012;4(6):506-16. doi: 10.3390/nu4060506 12. do Prado WL, Botero JP, Guerra RLF, Rodrigues CL, Cuvello LC, Damaso AR. Anthropometric profile and macronutrient intake in professional Brazilian soccer players according to their field positioning. Rev Bras Med Esporte 2006;12(2):61-5. SIMPULAN DAN SARAN doi: 10.1590/S1517- 86922006000200001 17. Lieberman HR, Marriott BP, Williams C, Judelson DA, Glickman EL, Mahoney CR, et al. Patterns of dietary supplement use among college students. Clinical Nutrition. 2015;34(5):976-85. doi: 10.1016/j.clnu.2014.10.010 13. Sousa M, Fernandes MJ, Carvalho P, Soares J, Moreira P, Teixeira VH. Nutritional supplements use in high- performance athletes is related with lower nutritional inadequacy from food. J Sport Health Sci. 2016;5(3):368- 74. doi: 10.1016/j.jshs.2015.01.006 18. Geyer H, Braun H, Burke LM, Stear SJ, Castell LM. A-Z of nutritional supplements: dietary supplements, sports nutrition foods and ergogenic aids for health and performance – part 22. Br J Sports Med. 2011;45(9):752-4. doi: 10.1136/bjsports-2011-090180 110
https://openalex.org/W1994846170	https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0036780&type=printable	English	null	The Prognostic Role of RASSF1A Promoter Methylation in Breast Cancer: A Meta-Analysis of Published Data	PloS one	2,012	cc-by	5,050	Yong Jiang1, Lin Cui1, Wen-de Chen2, Shi-hai Shen2, Li-dong Ding2 1 Department of Oncology, Jiangyan People’s Hospital, the Affiliated Hospital of Yangzhou University, Jiangyan, China, 2 Department of Scientific Research, Jiangyan People’s Hospital, the Affiliated Hospital of Yangzhou University, Jiangyan, China Abstract This is an open-access article distributed under the terms of the Creative Commons Attribution L restricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: The authors have no support or funding to report. Funding: The authors have no support or funding to report. Competing Interests: The authors have declared that no competing interests exist. Competing Interests: The authors have declared that no competing interests exist. Competing Interests: The authors have declared that no competing interests exist. E-mail: cuilintz@126.com PLoS ONE \| www.plosone.org May 2012 \| Volume 7 \| Issue 5 \| e36780 Abstract Purpose: Epigenetic alterations have been investigated as prognostic indicators in breast cancer but their translation into clinical practice has been impeded by a lack of appropriate validation. We present the results of a meta-analysis of the associations between RASSF1A promoter methylation status and both disease free survival (DFS) and overall survival (OS) in female breast cancer. Methods: Eligible studies were identified through searching the PubMed, Web of Science and Embase databases. Studies were pooled and summary hazard ratios (HR) with corresponding confidence intervals (CIs) were calculated. Funnel plots were also carried out to evaluate publication bias. Results: A total of 1795 patients from eight studies were included in the meta-analysis. There are eight studies which investigated DFS in 1795 cases. The relative hazard estimates ranged from 1.77–5.64 with a combined HR of 2.75 (95%CI 1.96–3.84). The HR of RASSF1A promoter methylation on DFS adjusted for other potential prognostic factors was 2.54 (95%CI 1.77–3.66). There has been five trials which analyzed the associations of RASSF1A promoter methylation status with OS in 1439 patients. The hazard estimates ranged from 1.21–6.90 with a combined random-effects estimates of 3.47 (95%CI 1.44–8.34). OS reported in multivariate analysis was evaluated in four series comprising 1346 cases and the summarized random-effects HR estimate was 3.35 (95%CI 1.14–9.85). Additionally, no publication bias was detected for both OS and DFS. Conclusion: The results of this meta-analysis suggest that RASSF1A promoter hypermethylation confers a higher risk of relapse and a worse survival in patients with breast cancer. Large prospective studies are now needed to establish the clinical utility of RASSF1A promoter methylation. Citation: Jiang Y, Cui L, Chen W-d, Shen S-h, Ding L-d (2012) The Prognostic Role of RASSF1A Promoter Methylation in Breast Cancer: A Meta-Analysis of Published Data. PLoS ONE 7(5): e36780. doi:10.1371/journal.pone.0036780 Editor: Amanda Ewart Toland, Ohio State University Medical Center, United States of America Received November 16, 2011; Accepted April 6, 2012; Published May 17, 2012 Copyright: 2012 Jiang et al. This is an open-access article distributed under the terms of the Creative Commons Attr unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. ng et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits tion, and reproduction in any medium, provided the original author and source are credited. opyright: 2012 Jiang et al. Meta-analysis The meta-analysis was carried out for the analyses of all studies on OS, DFS and their subgroups. The main results of the meta- analysis are summarized in Table 2. When all study populations combined, dismal survival outcomes on BC patients with hypermethylation of RASSF1A promoter were observed: for overall survival, summary HR = 3.47, 95%CI 1.44–8.34; I2 = 72.70%, random-effects model (Figure 1), and for disease free survival, summary HR = 2.75, 95% CI 1.96–3.84; I2 = 0.00%, fixed-effects model (Figure 2). Even by carrying out the meta-analysis using the HRs from Cox regression models only, we still observed significant pejorative impacts on OS (HR = 3.35, 95% CI 1.14–9.85; test for heterogeneity: I2 = 76.20%) (Figure 3) and DFS (HR = 2.54, 95% CI 1.77–3.66; test for heterogeneity: I2 = 0.00%) (Figure 4). Due to strong heterogeneity existed in the trials aggregated for overall survival, Galbraith plot was used to explore the heterogeneity. The heterogeneity disappeared after omitting one trial by Cho et al. (Chi-squared = 0.38, p = 0.945) [17]. In the subgroup analyses on overall survival, a significant prognostic role of RASSF1A methylation status was detected in the studies using MSP methods (HR = 4.26, 95%CI 1.65–10.98). However, no statistical significance reached in those using QMSP (HR = 3.28, 95%CI 0.94–11.50). When the differences of material reported for detecting RASSF1A promoter methylation levels were taken into consideration, the aggregated survival data showed an unfavorable survival prognosis using plasma (HR 6.03, 95% CI 2.77–13.11), but not tissue samples. Despite a number of individual studies performed in breast cancer patients, the prognostic value of RASSF1A promoter methylation status in breast cancer patient’s survival remains controversial. Therefore, we performed a systematic review of the literature with meta-analysis to obtain a more accurate evaluation of its prognostic value in breast cancer. In the subgroup analyses on disease-free survival, a subset of five studies (1525 patients) reporting the DFS for breast cancer patients using QMSP, and a subset of three studies (270 patients) reporting the DFS using MSP were pooled separately. The summary HR estimates for both groups showed inverse correlations with DFS (HR = 2.77, HR = 2.71, respectively). In addition, there was no difference when various materials used in detecting RASSF1A methylation status. Furthermore, no evidence of heterogeneity observed in these comparisons. RASSF1A Methylation and Breast Cancer Survival RASSF1A Methylation and Breast Cancer Survival Several potential tumor suppressor genes have been described as frequently silenced by hypermethylation in breast cancer. Among which, RAS-association domain family 1 (RASSF1A) is widely investigated. RASSF1A (http://www.ncbi.nlm.nih.gov/ epigenomics/view/genome/56289?term = Rassf1), which is locat- ed at 3p21.3, is functionally involved in cell cycle control, microtubule stabilization, cellular adhesion, motility, and apopto- sis [6]. Depletion of RASSF1A is reported to be associated with accelerated mitotic progression, an elevated risk for chromosomal defects, enhanced cellular motility, and increased tumor suscep- tibility in knockout mice [7,8,9]. Epigenetic inactivation of RASSF1A by hypermethylation of CpG islands in the promoter region [NC_000075.5 (107,453,580–107,454,373)] is observed in a considerable proportion of cancers and is associated with clinicopathological factors in various types of cancers, including breast cancer (see review [10]). Furthermore, RASSF1A promoter hypermethylation was reported as a prognostic indicator in renal cell carcinoma, non-small cell lung cancer, neuroblastoma, melanoma, endometrial cancer and breast cancer [11–18]. All of these findings suggested that it might play a pivotal role in the development of human cancer. Assessment of Publication Bias Visual assessment of the funnel plots provided no evidence of overt publication bias for studies in either of the two outcomes. Further evaluation using Egger’s linear regression test also failed to reveal any evidence for significant publication bias in OS (P = 0.36) and DFS (P = 0.34) study groups. The characteristics of retained 8 studies are listed in Table 1. The sample size of the included studies ranged from 78 to 670 patients (median sample size, 224 patients). The trials were conducted in 7 countries (Portugal, USA, Saudi Arabia, Tunisia, India, Greece, and Austria) and published between 2005 and 2011. There was 60.9% of BC patients had the methylated RASSF1A allele with a frequency ranging from 19.6 to 87.0% (median, 64.0%) in individual trials. The methylated RASSF1A levels were detected using either methylation specific PCR (MSP) [18,22,23] or quantitative methylation specific PCR (QMSP) [15,17,20,21,24]. The corresponding primer sequences of PCR are provided in a supplementary table (Table S1). DNA methylation status of RASSF1A promoter was assessed in plasma or tumor tissues. Except for one study that used fine-needle aspirate washings [15]. A HR on DFS and OS could be extracted from 5 and 8 of the studies, respectively. Most of the survival data for breast cancer were available in the form of multivariate analysis except for one study reported in univariate form (Kaplan– Meier survival curve) [18]. Study Selection and Characteristics Fifty-eight relevant citations were identified for initial review using search strategies as described previously. Of these, forty-six were initially excluded after read the titles and abstracts (13 not about breast cancer; 7 on cell lines; 11 review articles; 10 were on tumor biological behavior; 5 with other gene methylation). Investigators retrieved the remaining 12 citations for full text evaluation. Upon further review, three articles were eliminated on the basis of inadequate data for meta-analysis. Moreover, one was excluded for overlapping publication [19]. Ultimately, the systematic literature search yielded a total of 8 studies comprising 1795 patients for final analysis [15–18,20–24]. Meta-analysis These results suggest that breast cancer patients with RASSF1A promoter hypermethylation have a poor prognosis of relapse, irrespective of the detecting methods and samples. Introduction Although the current well-established clinical and histological factors and some other well-defined biological factors (e.g., hormone receptors and HER2 status) have been established and are assessed routinely in therapy decision-making and evaluating the prognosis, there are increasing concerns that these prognostic determinants are limited in their ability to capture the diversity of clinical behaviors of breast cancer and that they would be insufficient to predict the response to specific treatment strategies for individual patients. Recently, gene-expression-based prognostic assays are being used to predict breast cancer outcomes, but their prognostic validities are still undergoing evaluation [3]. Therefore, research efforts continue to focus on identifying more sensitive and specific indicators that could more reliably predict clinical outcomes and enhance treatment options. Breast cancer (BC) is the most commonly diagnosed cancer and the leading cause of cancer death in females worldwide, accounting for 23% (1.38 million) of the total new cancer cases and 14% (458,400) of the total cancer deaths in 2008 [1]. Because of early detection and effective adjuvant medical treatments, the survival rate of breast cancer has increased during the past decades. However, breast cancer is remarkably heterogeneous in histology and genetics, as well as in clinical behavior. Tradition- ally, pathologic determinations of tumor size, lymph node status, endocrine receptor status, histological grade, and human epider- mal growth factor receptor 2 (HER2) expression have driven prognostic predictions and, ultimately, adjuvant therapy recom- mendations for patients with breast cancer [2]. Nonetheless, these prognostic and predictive factors are relatively crude measures and it poses a great challenge for clinicians regarding the choice of optimum adjuvant treatment. It is of great importance to avoid overtreatment in patients who only receive a modest benefit, while suffering from more toxic side effects. On the other hand, undertreatment or incorrect treatment has to be avoided as well. Bulks of epidemiological and experimental studies have verified epigenetic and genetic changes involved in the development and progression of breast cancer (see review [4]). Recently, changes in the status of DNA methylation, known as epigenetic alterations, have turned out to be one of the most common molecular alterations in human malignancies, including breast cancer [5]. May 2012 \| Volume 7 \| Issue 5 \| e36780 1 PLoS ONE \| www.plosone.org PLoS ONE \| www.plosone.org Discussion For proper management of patients with cancer, accurate prognostic and predictive factors are necessary. Such factors are particularly important in breast cancer that has widely varying outcomes and for which systemic adjuvant therapy may be beneficial. Prognostic factors may help us to differentiate those patients with indolent from those with more aggressive disease. Patients with aggressive disease may then be candidates for treatment with systemic adjuvant therapy, while those with indolent disease may be spared the toxic side-effects and costs of this treatment. The accumulating evidence for epigenetic defects in breast cancer may be potentially useful in cancer progression. Aberrant DNA methylation of CpG islands within 5-prime of genes occurs almost in every type of cancer and easy to measure. Potential of gene-specific DNA methylation as a predictor of important clinical features has been explored in a number of studies now. PLoS ONE \| www.plosone.org May 2012 \| Volume 7 \| Issue 5 \| e36780 May 2012 \| Volume 7 \| Issue 5 \| e36780 2 RASSF1A Methylation and Breast Cancer Survival Table 1. Baseline characteristics of eligible studies evaluating RASSF1A hypermethylation and OS or DFS in breast cancer patients. Table 1. Baseline characteristics of eligible studies evaluating RASSF1A hypermethylation and OS or DFS in breast cancer patients. First Author Year Country Methods M/N (%) N Stage Grade Materials OS DFS HR (95%CI) HR (95%CI) Martins [15] 2011 Portugal QMSP 86 178 0–IV 1–3 fine-needle aspirate washings NA 2.53 (1.09–5.87) Cho [17] 2011 USA QMSP 85.2 670 I–IV NA formalin fixed paraffin-embedded tissues1.21 (0.76–1.93) 1.77 (0.86–3.67) Gobel [20] 2011 Austria QMSP 21.8 428 0–IV 1–3 peripheral blood-plasma 5.60 (2.10–14.50) 3.40 (1.60–7.30) Kioulafa [18] 2009 Greece MSP 57 93 I–II 1–3 formalin fixed paraffin-embedded tissues4.31 (0.92–7.58) 3.47 (1.24–9.32) Buhmeida [21] 2011 Saudi Arabia QMSP 65 100 I–IV 1–3 formalin fixed paraffin-embedded tissuesNA 5.64 (1.23–25.81) Karray-Chouayekh [22] 2010 Tunisia MSP 87 78 I–IV 1–3 fresh-frozen specimens NA 7.33 (1.37–37.72) Sharma [23] 2009 India MSP 63 100 I–III NA formalin fixed paraffin-embedded tissues4.05 (0.47–34.92) 1.80 (0.79–4.09) Fiegl [24] 2005 Austria QMSP 19.6 148 I–III 1–3 peripheral blood-plasma 6.90 (1.90–25.90) 5.10 (1.30–19.80) FFPE, formalin fixed paraffin-embedded; PBP, peripheral blood-plasma; FF, fresh-frozen; FNAW, fine-needle aspirate washings; MSP, methylation specific PCR; QMSP, quantitative methylation specific PCR. doi:10.1371/journal.pone.0036780.t001 FFPE, formalin fixed paraffin-embedded; PBP, peripheral blood-plasma; FF, fresh-frozen; FNAW, fine-needle aspirate washings; MSP, methylation specific PCR; QMSP, quantitative methylation specific PCR. May 2012 \| Volume 7 \| Issue 5 \| e36780 Discussion doi:10.1371/journal.pone.0036780.t001 Among which, the tumor suppressor gene RASSF1A promoter methylation was reported to be valuable as a prognostic indicator for breast cancer. Due to relatively small samples of individual study and controversial conclusions, we performed this meta-analysis of the literature to analyze whether RASSF1A hypermethylation could readily be harnessed as clinically useful predictive biomarker for breast cancer. This is the first meta-analysis of published studies to evaluate the association between RASSF1A promoter methylation and breast cancer prognosis in 1795 cases. Our results using the summarized HR of OS and DFS indicated that hypermethylation of RASSF1A is associated with both DFS and OS (pooled HR estimates of 2.75 and 3.47 for DFS and OS, respectively). These effects were slightly attenuated but still significant in multivariate analyses (adjusted HRs of 2.54 and 3.35, respectively), showing that its effect is independent Table 2. Main results of eligible studies evaluating RASSF1A hypermethylation and OS/DFS in breast cancer patients. N. of studies/cases HR (95% CI) Heterogeneity x2 p I2 Overall Survival (OS) All studies Fixed effects 5/1439 2.10 (1.45–3.03) 14.67 0.005 72.70% Random effects 5/1439 3.47 (1.44–8.34) 14.67 0.005 72.70% Cox regression model 4/1346 3.35 (1.14–9.85) 12.63 0.006 76.20% Testing methods QMSP 3/1246 3.28 (0.94–11.50) 12.14 0.002 83.5% MSP 2/192 4.26 (1.65–10.98) 0.00 0.959 0.00% Testing materials Plasma 2/576 6.03 (2.77–13.11) 0.06 0.801 0.00% Tissue samples 3/863 2.27 (0.82–6.27) 5.46 0.065 63.4% Disease-Free Survival (DFS) All studies Fixed effects 8/1795 2.75 (1.96–3.84) 6.01 0.539 0.00% Cox regression model 6/1624 2.54 (1.77–3.66) 4.28 0.51 0.00% Testing methods QMSP 5/1525 2.77 (1.84–4.15) 3.44 0.487 0.00% MSP 3/270 2.71 (1.49–4.91) 2.57 0.277 22.1% Testing materials Plasma 2/576 3.74 (1.93–7.26) 0.26 0.610 0.00% Tissue samples 5/1041 2.54 (1.57–4.13) 4.62 0.328 13.5% MSP, methylation specific PCR; QMSP, quantitative methylation specific PCR. doi:10.1371/journal.pone.0036780.t002 igible studies evaluating RASSF1A hypermethylation and OS/DFS in breast cancer patients. esults of eligible studies evaluating RASSF1A hypermethylation and OS/DFS in breast cancer patients. May 2012 \| Volume 7 \| Issue 5 \| e36780 PLoS ONE \| www.plosone.org RASSF1A Methylation and Breast Cancer Survival Figure 1. Forest plot showing the association between RASSF1A methylation and overall survival (OS) of breast cancer. The summary HR and 95% CIs were shown (according to the random-effects estimations). d i 10 1371/j l 0036780 001 Figure 1. Forest plot showing the association between RASSF1A methylation and overall survival (OS) of breast cancer. Definitions and Data Extraction Some limitations of this meta-analysis should be discussed. First, this analysis was performed at the study level, which limited ability to explore the potential for confounding by various demographic and clinical factors (e.g., ethnicity, hormone receptor status, disease stage, differentiation and treatment regimes). Second, this study was predominately based on the findings of observational studies, which inherently contain greater potential for confounding than randomized controlled trials. Third, potential risk bias was a concern, as published studies are often positive and so the omission of unpublished studies may lead to exaggeration of the summary HR. Although publication bias evaluation did not suggest any bias in the pooled OS and DFS studies, we identified studies only from limited databases, the total number of included studies and the total sample size were relatively small; which might influence the validity of our analysis to some extent. Fourth, the quality of pooled studies influences the level of confidence of meta- analysis remarkably. Published articles often lack sufficient in- formation to allow adequate assessment of the quality of the study or the generalisability of the study results. So REMARK criteria were recommended when reporting tumor markers [26]. Only one involved study reported the prognostic role of RASSF1A methyl- ation in BC using REMARK criteria [17]. Finally, most of studies included in the pooled analyses of breast cancer outcomes were carried out in European populations, it is possible that the results of these analyses are not readily generalizable to other populations. Because of these limitations existing in the identified studies and the current meta-analysis, our results should be interpreted with caution and likewise, the conclusions of this meta-analysis should also be drawn carefully. Overall survival was defined as the interval between the medical treatment (including surgical excision, chemotherapy or radio- therapy) and the death of patients or the last observation. Disease free survival was measured from the date of treatment until the detection of recurrence or the last follow-up assessment. The following data from all eligible publications was extracted respectively by two reviewers (Cui L and Chen WD) with a standardized data extraction form: first author’s surname, year of publication, patient source, sample size, disease stage, tumor grade, methylation status detecting method, positive ratio, and prognostic outcomes of interest (DFS and OS, including the information whether the outcomes were tested by multivariate analysis). Disagreements were resolved by discussion. RASSF1A Methylation and Breast Cancer Survival RASSF1A Methylation and Breast Cancer Survival of lymph node status, tumor size and tumor grade as well as a range of other biological variables on multivariate analysis. clinical trials, are now urgently needed to test whether hyper- methylation of RASSF1A can provide prognostic information in addition to currently used standards and also to establish if it has clinical utility. When the five studies reported the HR of overall survival were pooled, a considerable degree of interstudy heterogeneity was noticed (I2 = 72.7%). We applied Galbraith plot which is visualized in identifying the heterogeneous studies to explore the heteroge- neity. When one study by Cho et al. was excluded, the hazard size remains significant but the heterogeneity disappeared. The heterogeneity was probably due to the difference in the baseline characteristics of patients (age, tumor stage, race or country), the detecting methods, testing materials, the duration of follow-up or others. For example, when we stratified them according to detecting methods, heterogeneity disappeared in MSP subgroup. Strong heterogeneity still existed in quantitative methylation- specific PCR subgroup. Some techniques features regarding QMSP may partially explain this heterogeneity. First, lack of clear hypermethylation cut-off definition, it should be made about the cut-off value of RASSF1A methylation level for increased survival risk. To date, the researchers use median or self-defined value in their laboratory as the cut-off value and the accurate value was different. In addition, testing materials may also contribute to the heterogeneity, in this subgroup, methylation level detecting using tissue samples was marked (I2 = 63.4%). We postulated that the timing from resection to fixation or the process of fixation itself may potentially alter methylation status in paraffin-embedded tumors. One study observed that methylation status varied when different fixation techniques used [25]. We addressed the issue of heterogeneity by a rigorous methodological approach that used a random-effects model for more conservative estimates. Never- theless, there is no definitive explanation for the heterogeneity. PLoS ONE \| www.plosone.org Publication Selection A comprehensive literature search was carried out by two independent reviewers (Jiang Y and Cui L) using the PubMed, Web of Science and Embase databases. The search ended on 9 September 2011. The following keywords were used in various combinations: ‘breast cancer’, ‘biomarkers’, ‘molecular markers’, ‘survival’, ‘prognosis’, ‘RAS-association domain family 19 and ‘RASSF1A’. The search was performed without langue restriction. Reference lists from relevant primary studies and review articles were also checked for additional relevant publications. To be eligible for inclusion, studies had to meet the following criteria: (1) evaluating the association between RASSF1A promoter methyla- tion status and the prognosis of breast cancer patients, e.g., disease free survival (DFS) and/or overall survival (OS); (2) hazard ratio (HR) for OS or DFS according to RASSF1A methylation status either had to be reported or could be calculated from the data presented; (3) studies should be with full text not only abstracts for relevant information extraction; (4) when the same patient population reported in several publications, only the most recent report or the most complete one was included in this analysis to avoid overlapping between cohorts. Discussion The summary HR and 95% CIs were shown (according to the random-effects estimations). doi:10.1371/journal.pone.0036780.g001 Figure 1. Forest plot showing the association between RASSF1A methylation and overall survival (OS) of breast cancer. The summary HR and 95% CIs were shown (according to the random-effects estimations). doi:10.1371/journal.pone.0036780.g001 Figure 2. Forest plot showing the association between RASSF1A methylation and disease-free survival (DFS) of breast cancer. The summary HR and 95% CIs were shown (according to the fixed-effects estimations). doi:10.1371/journal.pone.0036780.g002 Figure 2. Forest plot showing the association between RASSF1A methylation and disease-free survival (DFS) of breast cancer. The summary HR and 95% CIs were shown (according to the fixed-effects estimations). doi:10.1371/journal.pone.0036780.g002 May 2012 \| Volume 7 \| Issue 5 \| e36780 PLoS ONE \| www.plosone.org RASSF1A Methylation and Breast Cancer Survival Figure 3. Forest plot showing the association between RASSF1A methylation and overall survival (OS) of breast cancer calculating from the data of multivariate Cox regression analyses. doi:10.1371/journal.pone.0036780.g003 Figure 3. Forest plot showing the association between RASSF1A methylation and overall survival (OS) of breast cancer calculating from the data of multivariate Cox regression analyses. doi:10.1371/journal.pone.0036780.g003 Figure 4. Forest plot showing the association between RASSF1A methylation and disease-free survival (DFS) of breast cancer calculating from the data of multivariate Cox regression analyses. doi:10.1371/journal.pone.0036780.g004 Figure 4. Forest plot showing the association between RASSF1A methylation and disease-free survival (DFS) of breast cancer calculating from the data of multivariate Cox regression analyses. doi:10.1371/journal.pone.0036780.g004 May 2012 \| Volume 7 \| Issue 5 \| e36780 5 RASSF1A Methylation and Breast Cancer Survival References Karray-Chouayekh S, Trifa F, Khabir A, Boujelbane N, Sellami-Boudawara T, et al. (2010) Aberrant methylation of RASSF1A is associated with poor survival in Tunisian breast cancer patients. J Cancer Res Clin Oncol 136: 203–210. 9. Song MS, Song SJ, Ayad NG, Chang JS, Lee JH, et al. (2004) The tumour suppressor RASSF1A regulates mitosis by inhibiting the APC-Cdc20 complex. Nat Cell Biol 6: 129–137. 23. Sharma G, Mirza S, Yang YH, Parshad R, Hazrah P, et al. (2009) Prognostic relevance of promoter hypermethylation of multiple genes in breast cancer patients. Cell Oncol 31: 487–500. 10. Hesson LB, Cooper WN, Latif F (2007) The role of RASSF1A methylation in cancer. Dis Markers 23: 73–87. p 24. Fiegl H, Millinger S, Mueller-Holzner E, Marth C, Ensinger C, et al. (2005) Circulating tumor-specific DNA: a marker for monitoring efficacy of adjuvant therapy in cancer patients. Cancer Res 65: 1141–1145. 11. Wang J, Wang B, Chen X, Bi J (2011) The prognostic value of RASSF1A promoter hypermethylation in non-small cell lung carcinoma: a systematic review and meta-analysis. Carcinogenesis 32: 411–416. py p 25. Hamilton MG, Roldan G, Magliocco A, McIntyre JB, Parney I, et al. (2011) Determination of the methylation status of MGMT in different regions within glioblastoma multiforme. J Neurooncol 102: 255–260. 12. Tanemura A, Terando AM, Sim MS, van Hoesel AQ, de Maat MF, et al. (2009) CpG island methylator phenotype predicts progression of malignant melanoma. Clin Cancer Res 15: 1801–1807. 26. McShane LM, Altman DG, Sauerbrei W, Taube SE, Gion M, et al. (2006) REporting recommendations for tumor MARKer prognostic studies (RE- MARK). Breast Cancer Res Treat 100: 229–235. 13. Jo H, Kim JW, Kang GH, Park NH, Song YS, et al. (2006) Association of promoter hypermethylation of the RASSF1A gene with prognostic parameters in endometrial cancer. Oncol Res 16: 205–209. 27. Tierney JF, Stewart LA, Ghersi D, Burdett S, Sydes MR (2007) Practical methods for incorporating summary time-to-event data into meta-analysis. Trials 8: 16. 14. Misawa A, Tanaka S, Yagyu S, Tsuchiya K, Iehara T, et al. (2009) RASSF1A hypermethylation in pretreatment serum DNA of neuroblastoma patients: a prognostic marker. Br J Cancer 100: 399–404. 28. DerSimonian R (1996) Meta-analysis in the design and monitoring of clinical trials. Stat Med 15: 1237–1248; discussion 1249–1252. 15. Martins AT, Monteiro P, Ramalho-Carvalho J, Costa VL, Dinis-Ribeiro M, et al. References 1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, et al. (2011) Global cancer statistics. CA Cancer J Clin 61: 69–90. 16. Kawai Y, Sakano S, Suehiro Y, Okada T, Korenaga Y, et al. (2010) Methylation level of the RASSF1A promoter is an independent prognostic factor for clear- cell renal cell carcinoma. Ann Oncol 21: 1612–1617. 2. Rakha EA, Reis-Filho JS, Baehner F, Dabbs DJ, Decker T, et al. (2010) Breast cancer prognostic classification in the molecular era: the role of histological grade. Breast Cancer Res 12: 207. 17. Cho YH, Shen J, Gammon MD, Zhang YJ, Wang Q, et al. (2011) Prognostic significance of gene-specific promoter hypermethylation in breast cancer patients. Breast Cancer Res Treat. 3. Kim C, Paik S (2010) Gene-expression-based prognostic assays for breast cancer. Nat Rev Clin Oncol 7: 340–347. p 18. Kioulafa M, Kaklamanis L, Mavroudis D, Georgoulias V, Lianidou ES (2009) Prognostic significance of RASSF1A promoter methylation in operable breast cancer. Clin Biochem 42: 970–975. 4. Esteller M (2008) Epigenetics in cancer. N Engl J Med 358: 1148–1159. 5. Muller HM, Fiegl H, Widschwendter A, Widschwendter M (2004) Prognostic DNA methylation marker in serum of cancer patients. Ann N Y Acad Sci 1022: 44–49. 19. Muller HM, Widschwendter A, Fiegl H, Ivarsson L, Goebel G, et al. (2003) DNA methylation in serum of breast cancer patients: an independent prognostic marker. Cancer Res 63: 7641–7645. 6. Agathanggelou A, Cooper WN, Latif F (2005) Role of the Ras-association domain family 1 tumor suppressor gene in human cancers. Cancer Res 65: 3497–3508. 20. Gobel G, Auer D, Gaugg I, Schneitter A, Lesche R, et al. (2011) Prognostic significance of methylated RASSF1A and PITX2 genes in blood- and bone marrow plasma of breast cancer patients. Breast Cancer Res Treat. 7. Dallol A, Cooper WN, Al-Mulla F, Agathanggelou A, Maher ER, et al. (2007) Depletion of the Ras association domain family 1, isoform A-associated novel microtubule-associated protein, C19ORF5/MAP1S, causes mitotic abnormal- ities. Cancer Res 67: 492–500. 21. Buhmeida A, Merdad A, El-Maghrabi J, Al-Thobaiti F, Ata M, et al. (2011) RASSF1A Methylation is Predictive of Poor Prognosis in Female Breast Cancer in a Background of Overall Low Methylation Frequency. Anticancer Res 31: 2975–2981. 8. Tommasi S, Dammann R, Zhang Z, Wang Y, Liu L, et al. (2005) Tumor susceptibility of Rassf1a knockout mice. Cancer Res 65: 92–98. 22. Table S1 The primer sequences of detecting RASSF1A promoter methylation status of the eligible studies. (DOC) Table S1 The primer sequences of detecting RASSF1A promoter methylation status of the eligible studies. (DOC) Statistical Analysis Meta-analysis techniques were used to compute a summary estimate of the hazard ratio (HR) and 95% confidence intervals (CIs) for recurrence or death with breast cancer. Survival outcome data were synthesized using the time-to-event HR as the effective measure. When HR was not provided directly, estimated value was derived indirectly from other presented data using the methods described by Tierney et al. [27]. Moreover, when univariate and multivariate analyses of OS and/or DFS were both available, the latter was selected to be combined because survival response variable is influenced by multiple factors. Heterogeneity between the studies was tested using Q-statistics. It was considered statistically significant if p value less than 0.10 and was also quantified using the I2 metric (I2,25%, no heterogeneity; I2 = 25–50%, moderate heterogeneity; and I2.50%, strong heterogeneity) [28,29]. If the heterogeneity was existed, we used a random-effects model in place of a fixed-effects model and the Galbraith plot was used to provide a graphical display to get a visual impression of the amount of heterogeneity from a meta-analysis [30]. By convention, an observed HR.1 In conclusion, hypermethylation of RASSF1A promoter was found to be independently associated with decreased survival of breast cancer patients. The promoter methylation of the RASSF1A gene is potentially useful biomarker for predicting prognosis in breast cancer. Large studies, both observational cohorts and May 2012 \| Volume 7 \| Issue 5 \| e36780 6 RASSF1A Methylation and Breast Cancer Survival implied a worse survival for the group with RASSF1A hyper- methylation. This impact of RASSF1A on survival was considered as statistically significant if the corresponding 95% CI for the summary HR did not overlap 1 unit. Publication bias was assessed by funnel plots and Egger’s linear regression. All p values were two sided. Statistical calculations were all performed using STATA version 11.0, College Station TX. Author Contributions Conceived and designed the experiments: JY CL. Performed the experiments: JY CDW. Analyzed the data: CL SSH. Contributed reagents/materials/analysis tools: JY. Wrote the paper: DLD JY. Conceived and designed the experiments: JY CL. Performed the experiments: JY CDW. Analyzed the data: CL SSH. Contributed reagents/materials/analysis tools: JY. Wrote the paper: DLD JY. References (2011) High RASSF1A promoter methylation levels are predictive of poor prognosis in fine-needle aspirate washings of breast cancer lesions. Breast Cancer Res Treat 129: 1–9. 29. Higgins JP, Thompson SG, Deeks JJ, Altman DG (2003) Measuring inconsistency in meta-analyses. BMJ 327: 557–560. 30. Bax L, Ikeda N, Fukui N, Yaju Y, Tsuruta H, et al. (2009) More than numbers: the power of graphs in meta-analysis. Am J Epidemiol 169: 249–255. PLoS ONE \| www.plosone.org May 2012 \| Volume 7 \| Issue 5 \| e36780 7
https://openalex.org/W4281556372	https://www.frontiersin.org/articles/10.3389/fendo.2022.835880/pdf	English	null	The Effect of Inactivated SARS-CoV-2 Vaccines on TRAB in Graves’ Disease	Frontiers in endocrinology	2,022	cc-by	6,636	The Effect of Inactivated SARS- CoV-2 Vaccines on TRAB in Graves’ Disease LingHong Huang 1,2, ZhengRong Jiang 2, JingXiong Zhou 2, YuPing Chen 2 and HuiBin Huang 2* LingHong Huang 1,2, ZhengRong Jiang 2, JingXiong Zhou 2, YuPing Chen 2 and HuiBin Huang 2* 1 The Second Clinical Medical College of Fujian Medical University, Quanzhou, China, 2 Department of Endocrinology, The Second Afﬁliated Hospital of Fujian Medical University, Quanzhou, China Background: The ongoing coronavirus disease 2019 (COVID-19) pandemic has forced the development of vaccines. Reports have suggested that vaccines play a role in inducing autoimmune diseases (AIDs). Scattered cases have reported that severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccines may promote thyroid disease, including Graves’ disease (GD). However, the effect of inactivated SARS-CoV-2 vaccine on GD remains unclear. The aim of the present study was to investigate the response of thyrotropin receptor antibody (TRAB) to inactivated SARS-COV-2 vaccines. ORIGINAL RESEARCH published: 16 May 2022 doi: 10.3389/fendo.2022.835880 ORIGINAL RESEARCH published: 16 May 2022 doi: 10.3389/fendo.2022.835880 ORIGINAL RESEARCH published: 16 May 2022 doi: 10.3389/fendo.2022.835880 Edited by: Jose Sgarbi, Faculdade de Medicina de Marı´lia, Brazil Reviewed by: Lorenzo Scappaticcio, University Hospital “Luigi Vanvitelli”, Italy Ilaria Muller, University of Milan, Italy Giusy Elia, University of Pisa, Italy Methods: We conducted a retrospective study to observe the differences in thyroid function and TRAB trends between pre-vaccination (n=412) and post-vaccination (n=231) groups at an interval of 2 months. We then retrospectively observed the differences in serum thyroid function and TRAB levels at 3 months before (n=280), 1 month before (n=294), 1 month after (n=306), and 3 months after (n=250) vaccination. Subsequently, 173 GD patients who were not vaccinated with inactivated SARS-COV-2 vaccines were selected for a prospective study. Thyroid function and TRAB assessment were performed before 3 and 1 months and 1 and 3 months after the ﬁrst dose of vaccination and were then compared by repeated measures ANOVA to explore their dynamic changes. *Correspondence: HuiBin Huang huibinhuang@aliyun.com Specialty section: This article was submitted to Thyroid Endocrinology, a section of the journal Frontiers in Endocrinology Results: A retrospective study preliminarily observed that the trend of TRAB post- vaccination was opposite of that pre-vaccination (p=0.000), serum TRAB levels decreased before vaccination and increased after vaccination. In this prospective study, repeated measures ANOVA indicated signiﬁcant differences in serum FT3 (p=0.000), FT4 (p=0.000), TSH (p=0.000), and TRAB (p=0.000) levels at different time points before and after vaccination. Serum TRAB levels showed dynamic changes that decreased signiﬁcantly at 1 month before vaccination (p=0.000), no signiﬁcant differences at 1 month after vaccination (p=0.583), and reﬂected an upward trend at 3 months after vaccination (p=0.034). Serum FT3 and FT4 levels showed similar trends to serum TRAB levels before and after vaccination. Instead, the serum TSH levels showed a continuous upward trend over time. Received: 15 December 2021 Accepted: 04 April 2022 Published: 16 May 2022 BACKGROUND phenomenon of ASIA (9). Furthermore, one woman with a history of controlled GD developed ocular symptoms and signs after the mRNA COVID-19 vaccine, which, combined with elevated thyroid-stimulating immunoglobulin and orbital imaging, was consistent with a diagnosis of active Graves ophthalmopathy (GO) (10). Among the outpatients in our hospital, there were also cases of AITD induced by inactivated SARS-CoV-2 vaccines, including GD and Hashimoto’s thyroiditis (HT), which mainly manifested as hyperthyroidism, such as a swollen neck, palpitations, and weight loss (this case report is in the process of publication). In the aforementioned reports, there were both newly diagnosed and recurrent or aggravated cases. The coronavirus disease 2019 (COVID-19) pandemic, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has infected more than 20 million individuals and caused more than 5 million deaths worldwide by December 12, 2021. COVID- 19 can cause both pulmonary and systemic inﬂammation, potentially leading to multi-organ dysfunction. Thyroid diseases, including thyrotoxicosis, hypothyroidism, and non- thyroid disease syndromes, can also be caused by COVID-19 (1). To date, there is no speciﬁc treatment for SARS-CoV-2, and vaccination is a basic and effective way to prevent the spread of this virus. Currently, the types of vaccines in use include inactivated virus vaccines, live attenuated virus vaccines, nucleic acid vaccines, recombinant viral vector vaccines, and recombinant subunit vaccines (2). In China, more than 1 billion people have been vaccinated with inactivated SARS-CoV-2, and mass vaccination continues. Until now, little was known about the effect of inactivated SARS-CoV-2 vaccines on GD. As a marker of diagnosis and evaluation of treatment and remission, TRAB was the entry point of the present study. This study aimed to explore the response of TRABs in GD after inactivated SARS-CoV-2 vaccines in retrospective and prospective studies to further investigate the factors that may modulate these responses. Graves’ disease (GD) is an organ-speciﬁc autoimmune disease (AID) which is characterized by thyrotropin receptor antibody (TRAB). Genetic factors account for 80% of the risk of developing GD, whereas the other 20% are related to environmental risk factors (3). For example, autoimmune thyroid disease (AITD) is a common side effect of alemtuzumab therapy in patients with multiple sclerosis (4). These factors contribute to the onset of GD in genetically susceptible individuals by breaking down the mechanisms that lead to immune tolerance. The immunopathogenesis of GD is complex, and TRAB is the ultimate cause of hyperthyroidism (5). Citation: Huang L, Jiang Z, Zhou J, Chen Y and Huang H (2022) The Effect of Inactivated SARS-CoV-2 Vaccines on TRAB in Graves’ Disease. Front. Endocrinol. 13:835880. doi: 10.3389/fendo.2022.835880 May 2022 \| Volume 13 \| Article 835880 Frontiers in Endocrinology \| www.frontiersin.org Huang et al. Inactivated SARS-CoV-2 Vaccines on TRAB Conclusion: Based on the results obtained in both retrospective and prospective studies, we concluded that serum TRAB levels decreased less after inactivated SARS- CoV-2 vaccination and showed an upward trend, which may be related to humoral immunity induced by vaccination. Keywords: inactivated SARS-CoV-2 vaccine, Graves’ disease (GD), thyrotropin receptor antibody (TRAB), autoimmune disease (AID), autoimmune thyroid disease (AITD) Keywords: inactivated SARS-CoV-2 vaccine, Graves’ disease (GD), thyrotropin receptor antibody (TRAB), autoimmune disease (AID), autoimmune thyroid disease (AITD) BACKGROUND It binds to the thyroid-stimulating hormone (TSH) receptor on the surface of thyroid follicular cells, resulting in persistent and uncontrolled thyroid stimulation, leading to abnormal overproduction of thyroid hormones and hyperthyroidism. Frontiers in Endocrinology \| www.frontiersin.org Study Population These retrospective studies were performed to observe the effects of vaccination on thyroid function before and after vaccination. These retrospective studies were performed to observe the effects of vaccination on thyroid function before and after vaccination. We included all GD patients in the Endocrinology Department of the Second Afﬁliated Hospital of Fujian Medical University from January to August 2021. The patients included conformed to the diagnostic and treatment criteria of GD of the European Thyroid Association (ETA) (2018). Patients with GO, GD after I131 or surgical treatment, thyroid cancer, pregnant and suckling period females, or other autoimmune diseases were excluded (Figure 1). All enrolled patients were treated with methimazole (MMI) combined with levothyroxine (L-T4) to avoid drug- induced hypothyroidism according to the ATD, which is the ﬁrst-line treatment for GD. All of them were biochemically consistent with hyperthyroidism and were TRAB-positive, and treatment regimens remained unchanged during the study period. According to the time point of the ﬁrst vaccination, the included population was divided into pre-vaccination (n=412) and post-vaccination (n=231) groups. Pre-vaccination referred to patients with GD who had not yet been vaccinated with inactivated SARS-CoV-2 vaccine. Post-vaccination, the GD was vaccinated with inactivated SARS-CoV-2 vaccine. We collected data on thyroid function and TRAB levels, which were measured Vaccines have long been suspected to play a role in inducing AIDs (6). There have been isolated case reports of arthritis, vasculitis, and central or peripheral nervous system symptoms following vaccination. Although these cases tend to be very infrequent, there have been reports of AID after SARS-CoV-2 vaccination, including AITD. SARS-CoV-2 vaccines-induced thyroid disease is not a single report. Recently, Alberto et al. and Zettinig et al. successively reported four cases of GD with positive TRAB induced by SARS-Cov2 RNA vaccination, which met the diagnostic criteria for autoimmune/inﬂammatory syndrome induced by adjuvants (ASIA) (7, 8). Another study reported that three women developed anterior neck pain after inactivated SARS-CoV-2 vaccine and were diagnosed with subacute thyroiditis, which is also thought to be a May 2022 \| Volume 13 \| Article 835880 Frontiers in Endocrinology \| www.frontiersin.org 2 Inactivated SARS-CoV-2 Vaccines on TRAB Huang et al. FIGURE 1 \| Flow charts of retrospective study population. FIGURE 1 \| Flow charts of retrospective study population. willingness and the control of hyperthyroidism, they are advised to receive inactivated SARS-COV-2 vaccines. The exclusion criteria were as follows: 1. GO; 2. Study Population patients who needed to change treatment due to illness during the test; 3. GD after I131 or surgical treatment; 4. thyroid cancer; 5. pregnant and suckling period females; 6. GD accompanied by serious medical diseases, liver and kidney dysfunction, or granulocytopenia; and 7. other autoimmune diseases. Finally, we enrolled 173 GD patients who had not received inactivated SARS-COV-2 vaccines. All subjects provided informed consent to participate in the study, which was approved by the local ethical committee. at 2-month intervals within a speciﬁed time through the clinical system. We then retrospectively observed the differences in serum thyroid function and TRAB levels at different times before and after vaccination. We included all GD patients who received the ﬁrst dose of inactivated SARS-CoV-2 vaccine in the Endocrinology Department of the Second Afﬁliated Hospital of Fujian Medical University from January to October 2021 (n=482). Patients with GO (n=1), GD after I131 (n=8) or surgical treatment (n=1), thyroid cancer (n=2), pregnant and suckling period females (n=0), or other autoimmune diseases (n=0) were excluded. Thyroid function and TRAB levels of the included GD patients were reviewed 3 months before, 1 month before, 1 month after, and 3 months after vaccination according to the time point of the ﬁrst vaccination. After excluding patients with missing checklists at all the above time points (n=156), 314 patients were included and divided into four groups according to time points: pre-vaccination -3 month and -1 month and post-vaccination +1 month and +3 months. Due to the absence of follow-up at some time points, the cases in each group were as follows: pre- vaccination -3 months (n=280) and -1 month (n=294) and post-vaccination +1 month (n=306) and +3 months (n=250). Vaccination We reviewed the vaccination information in the Fujian Health Code to conﬁrm and collect the type and date of vaccination and vaccine manufacturers of the participants to ensure the accuracy of information collection. They were also asked about their discomfort after the vaccination. The vaccine manufacturers of the enrolled patients included SINOVAC, Beijing Bio, and Chengdu Bio, which produced inactivated SARS-COV- 2 vaccines. Frontiers in Endocrinology \| www.frontiersin.org The Change Trend of TRAB in Post- Vaccination Was Opposite of That in Pre-Vaccination In this retrospective study, we reviewed changes in thyroid function and TRAB in populations at different stages of vaccination, including pre-vaccination (n=412) and post- vaccination (n=231), to determine the effect of inactivated SARS-CoV-2 vaccines on TRAB. The baseline clinical data of the two groups were analyzed statistically to exclude other inﬂuencing factors after vaccination for GD. No signiﬁcant differences were observed between the groups in terms of sex, age, medication, or other clinical characteristics. Serum FT3 (p=0.000), FT4 (p=0.000), and TRAB (p=0.000) levels were signiﬁcantly lower after 2 months than before pre-vaccination. There were no differences in thyroid function and TRAB between the 2-month post-vaccination intervals. The t-test Assays Subsequently, we conducted a prospective study on GD patients who were admitted to the Endocrinology Department of our hospital between March and May 2021 and had not been vaccinated. The enrolled population also met the diagnostic and treatment criteria of GD in ETA (2018), and the treatment regimen followed ﬁrst-line treatment, which remained unchanged during the study period. According to their Venous blood samples were collected before 3 and 1 months and 1 and 3 months after the ﬁrst dose of inactivated SARS-COV-2 vaccines. Serum free triiodothyronine 3 (FT3), free thyroxine 4 (FT4), thyroid-stimulating hormone (TSH), and TRAB levels were measured using a competitive electrochemiluminescence immunoassay (ECLIA) according to the manufacturer’s May 2022 \| Volume 13 \| Article 835880 Frontiers in Endocrinology \| www.frontiersin.org 3 Inactivated SARS-CoV-2 Vaccines on TRAB Huang et al. instructions (Roche COBAS-E601). Normal ranges of these parameters were as follows: FT3 (3.1-6.8pmol/L), FT4 (12.0- 22.0pmol/L), TSH (0.27-4.20mIU/L), TRAB (0.00-1.75IU/L). The samples were analyzed in routine clinical laboratories at the Second Afﬁliated Hospital of Fujian Medical University. instructions (Roche COBAS-E601). Normal ranges of these parameters were as follows: FT3 (3.1-6.8pmol/L), FT4 (12.0- 22.0pmol/L), TSH (0.27-4.20mIU/L), TRAB (0.00-1.75IU/L). The samples were analyzed in routine clinical laboratories at the Second Afﬁliated Hospital of Fujian Medical University. instructions (Roche COBAS-E601). Normal ranges of these parameters were as follows: FT3 (3.1-6.8pmol/L), FT4 (12.0- 22.0pmol/L), TSH (0.27-4.20mIU/L), TRAB (0.00-1.75IU/L). The samples were analyzed in routine clinical laboratories at the Second Afﬁliated Hospital of Fujian Medical University. showed that the TRAB change was statistically different between the two groups (p=0.000). In contrast to the pre- vaccination values, the TRAB change trend post-vaccination (1.290IU/L vs. 0.060IU/L) (p=0.000) was the opposite (Table 1 and Figure 2). The Changes in Serum TRAB Levels Decreased Before Vaccination and Increased After Vaccination All analyses were performed using Statistical Package for the Social Sciences software version 23 (SPSS Inc., Chicago, IL). Continuous variables were normally distributed and shown as mean ± standard deviation (x̅ ± s) and irregularly distributed data were expressed as medium (interquartile range). The Kolmogorov-Smirnov test was used for variables with skewed distributions. We used Blom’s formula to transform the skewed distribution into a normal distribution. The paired t-test and post hoc one-way analysis of variance (ANOVA) were used to assess the statistical signiﬁcance of differences among the groups. Repeated measures ANOVA was performed to compare dynamic changes in thyroid function and TRAB levels in this prospective study. Statistical signiﬁcance was set at p < 0.05. Increased After Vaccination To detect changes in serum thyroid function and TRAB levels at different time points before and after vaccination, we compared thyroid function and TRAB levels at 3 months, 1 month before and 1 month, 3 months after the ﬁrst dose of vaccine. The baseline clinical data of the groups were analyzed statistically to exclude other inﬂuencing factors after vaccination for GD. No signiﬁcant differences were observed between the groups in terms of sex, age, medication, and other clinical characteristics. The ANOVA analysis showed that there was a statistically signiﬁcant difference in TRAB among the groups (p=0.019). The change trend of TRAB decreased before vaccination (5.880IU/L vs. 4.275IU/L) (p=0.009) and increased after vaccination (4.345IU/L vs. 4.475IU/L) (p=0.509). The change trends of FT3 and FT4 were both similar to those of TRAB, which decreased before vaccination (FT3: 5.220 pmol/L vs. 4.905 pmol/L, p=0.002; FT4: 16.415 pmol/L vs. 16.050 pmol/L, p=0.164) and increased after vaccination (FT3: 4.860 pmol/L vs. 4.990 pmol/L, p=0.247; FT4: 16.375 pmol/L vs. 16.840 pmol/L, p=0.271). The difference was that TSH levels continued to rise (0.218 mIU/L vs. 0.548 mIU/L vs. 0.817 mIU/L vs. 1.070 mIU/L) (p=0.000) (Table 2). Baseline Characteristics of GD in Prospective Study To further investigate the relationship between vaccination and serum TRAB levels, 173 GD who had received inactivated SARS- COV-2 vaccines were enrolled in a prospective study to compare the dynamic changes of serum TRAB levels before and at 3 and 1 months after vaccination. The baseline characteristics and clinical parameters of GD in this prospective study are summarized in Table 3. The mean age of the subjects in this TABLE 1 \| Comparison of clinical characteristics and thyroid function between groups. Pre-vaccination (n = 412) Post-vaccination (n = 231) P Gender (female%) 77.910 76.190 0.618 Age (year) 39.020 ± 11.298 38.830 ± 10.895 0.941 Duration (month) 12.067 (5.433,20.725) 15.167 (9.300,25.600) 0.000 Thiamazole (mg) 15.127± 4.902 14.946 ± 5.563 0.844 Letrox (ug) 54.854 ± 36.674 56.522 ± 37.830 0.585 2 Months Before 2 Months After P 2 Months Before 2 Months After P – FT3 (pmol/L) 5.295 (4.490,7.540) 4.765 (4.110,5.790) 0.000 4.770 (4.210,5.810) 4.950 (4.490,5.750) 0.756 – FT4 (pmol/L) 16.490 (13.400,21.678) 15.765 (12.913,19.380) 0.000 16.050 (13.320,19.410) 16.740 (14.380,20.120) 0.246 – TSH (mIU/L) 0.134 (0.005,1.758) 0.524 (0.006,2.170) 0.308 0.815 (0.017,2.850) 1.300 (0.035,3.170) 0.419 – TRAB (IU/L) 6.300 (3.075,13.145) 4.455 (2.313,9.893) 0.000 4.450 (2.260,10.930) 4.470 (2.410,10.340) 0.237 – TRAB Change (IU/L) 1.290 (0.440,2.770) 0.060 (-0.600,1.010) 0.000 Data are presented as mean±standard error (x±s) or median (interquartile range). Categorical outcomes were shown as absolute and relative prevalence of complications (%). FT3 free triiodothyronine 3, FT4 free thyroxine 4, TSH thyroid-stimulating hormone, TRAB thyrotropin receptor antibody. TABLE 1 \| Comparison of clinical characteristics and thyroid function between groups. Data are presented as mean±standard error (x±s) or median (interquartile range). Categorical outcomes were shown as absolute and relative prevalence of complications (%). FT3 free triiodothyronine 3, FT4 free thyroxine 4, TSH thyroid-stimulating hormone, TRAB thyrotropin receptor antibody. May 2022 \| Volume 13 \| Article 835880 4 Frontiers in Endocrinology \| www.frontiersin.org Inactivated SARS-CoV-2 Vaccines on TRAB Huang et al. FIGURE 2 \| Changes of TRAB levels and amplitude between pre-vaccination and post-vaccination groups. FIGURE 2 \| Changes of TRAB levels and amplitude between pre-vaccination and post-vaccination groups. vaccination. Serum TRAB levels showed dynamic changes that decreased signiﬁcantly at 1 month before vaccination (5.450IU/L vs. 3.950 IU/L) (p=0.000), no signiﬁcant differences at 1 month after vaccination (3.95 0IU/L vs. 3.700 IU/L) (p=0.583), and a slight change and reﬂected an upward trend at 3 months after vaccination (3.700 IU/L vs. 4.100 IU/L) (p=0.034) (Table 4 and Figure 3). Baseline Characteristics of GD in Prospective Study Serum FT3 and FT4 levels showed similar trends to serum TRAB levels before and after vaccination. However, the slight differences were that FT3 and FT4 showed an upward trend at 1 month after vaccination, although there was no difference in the change of FT3 (4.630pmol/L vs. 4.740 pmol/ L) (p=0.095) and FT4 (15.490 pmol/L vs. 16.220 pmol/L) (p=0.068) 1 month before and after vaccination, and the study was 39.168 years, and 73.988% were women. The subjects had a median disease duration of 16.333 months, and their medication was MMI 14.957 mg and LT-4 58.960 µg on average. study was 39.168 years, and 73.988% were women. The subjects had a median disease duration of 16.333 months, and their medication was MMI 14.957 mg and LT-4 58.960 µg on average. The baseline median serum FT3, FT4, TSH, and TRAB levels were 5.230 pmol/L, 15.930 pmol/L, 0.375 mIU/L, and 5.450 IU/L, respectively. The baseline median serum FT3, FT4, TSH, and TRAB levels were 5.230 pmol/L, 15.930 pmol/L, 0.375 mIU/L, and 5.450 IU/L, respectively. Upward Trend Gender (female%) 73.988 Age (year) 39.168 ±10.713 Duration (month) 16.333 (9.617,26.309) Thiamazole (mg) 14.957±5.293 Letrox (ug) 58.960 ±37.246 FT3 (pmol/L) 5.230 (4.500,6.570) FT4 (pmol/L) 15.930 (13.530,20.300) TSH (mIU/L) 0.375 (0.005,2.020) TRAB (IU/L) 5.450 (2.555,11.350) Data are presented as mean±standard error (x±s) or median (interquartile range). Categorical outcomes were shown as absolute and relative prevalence of complications (%). FT3 free triiodothyronine 3, FT4 free thyroxine 4, TSH thyroid-stimulating hormone, TRAB thyrotropin receptor antibody. TABLE 3 \| Demographic and clinical characteristics of enrolled patients (n = 173). Inactivated SARS-CoV-2 vaccines are made by taking live viral samples from multiple patients and replicating them in Vero cells of the African green monkey cell line, which is susceptible to infection (12). Then, the strain with the best proliferation and lowest mutagenesis rate is isolated, further proliferated, inactivated, and absorbed onto aluminum hydroxide to generate (13). In preclinical studies, inactivated SARS-CoV-2 vaccines provided complete protection against SARS-CoV-2 infection by triggering effective humoral immune responses and inducing SARS-CoV-2-speciﬁc neutralizing antibodies in serrated animals and nonhuman primates (14). Furthermore, vaccines have been found to elicit a rapid humoral response in healthy individuals to be tolerable and immunogenic (15). Data are presented as mean±standard error (x±s) or median (interquartile range). Categorical outcomes were shown as absolute and relative prevalence of complications (%). FT3 free triiodothyronine 3, FT4 free thyroxine 4, TSH thyroid-stimulating hormone, TRAB thyrotropin receptor antibody. Adjuvants are compounds added to vaccines to enhance immunogenicity, which could lead to practical advantages, including dose-sparing and inducing a more rapid, strong, and long-lasting immune response (16, 17). Aluminum compounds are the most widely used adjuvants for human vaccines. Aluminum traps soluble antigens, interacts with dendritic cells, enhances antigen presentation and complement and eosinophil activation, promotes an inﬂux of neutrophils, enhances the secretion of pro-inﬂammatory cytokines and chemokines, and reduces immunopathology, elevating protective immunity levels to the threat of homologous viruses (18). While everything has two sides, the more effective it is, the higher is the risk. However, adjuvants are not completely free of side effects. In genetically predisposed individuals, adjuvants may induce ASIA by disrupting the host’s immunological balance through molecular simulations, triggering polyclonal activation of B lymphocytes, or other similar etiological mechanisms (19). Upward Trend Adjuvants can trigger generalized autoimmune reactions, resulting in multiple autoantibodies, and contribute to the development of autoimmune diseases including AITD, rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), Sjogren’s syndrome (SS), Guillain Barre syndrome (GBS), and multiple sclerosis (18). upward trend was more signiﬁcant 3 months after vaccination (FT3: 4.740 pmol/L vs. 5.020 pmol/L, p=0.001; FT4: 16.220 pmol/L vs. 16.610 pmol/L, p=0.012). Instead, serum TSH levels showed a continuous upward trend over time (0.375 mIU/L vs. 0.948 mIU/L vs. 1.110 mIU/L vs. 1.420 mIU/L) (p=0.000). Upward Trend Repeated measures ANOVA indicated signiﬁcant differences in serum FT3 (p=0.000), FT4 (p=0.000), TSH (p=0.000), and TRAB (p=0.000) levels at different time points before and after TABLE 2 \| Variation in thyroid function at different time points before and after vaccination. Pre-vaccination Post-vaccination P Post Hoc 3 Months Before (n = 280) 1 Month Before (n = 294) 1 Month After (n = 306) 3 Months After (n = 250) Gender (female%) 79.286 76.871 77.124 76.400 0.857 – Age (year) 40.181 ±10.913 40.252 ±10.838 40.101 ±10.886 39.606±10.678 0.904 – Duration (month) 11.150 (4.467,21.050) 13.917 (6.508,23.075) 15.283 (8.500,25.600) 18.367 (10.292,29.233) 0.000 – Thiamazole (mg) 15.384±4.667 15.145±4.901 15.204±4.852 15.350±4.476 0.955 – Letrox (ug) 53.214 ±37.309 52.976 ±38.665 51.797 ±38.476 53.200±37.555 0.968 – FT3 (pmol/L) 5.220 (4.468,7.108) 4.905 (4.220,6.000) 4.860 (4.268,5.950) 4.990 (4.485,5.878) 0.006 0.002a 0.997b 0.247c FT4 (pmol/L) 16.415 (13.313,21.478) 16.050 (13.300,19.413) 16.375 (13.410,20.118) 16.840 (14.465,19.915) 0.361 0.164a 0.999b 0.271c TSH (mIU/L) 0.218 (0.005,2.018) 0.548 (0.006,2.283) 0.817 (0.012,2.595) 1.070 (0.017,3.133) 0.000 0.169a 0.047b 0.318c TRAB (IU/L) 5.880 (2.708,13.070) 4.275 (2.203,10.310) 4.345 (2.288,10.110) 4.475 (2.453,9.853) 0.019 0.009a 0.865b 0.509c Data are presented as mean±standard error (x±s) or median (interquartile range). Categorical outcomes were shown as absolute and relative prevalence of complications (%). FT3 free triiodothyronine 3, FT4 free thyroxine 4, TSH thyroid-stimulating hormone, TRAB thyrotropin receptor antibody. a3 Months Before versus 1 Month Before. b1 Month Before versus 1 Month After. c1 Month After versus 3 Months After. May 2022 \| Volume 13 \| Article 835880 Frontiers in Endocrinology \| www.frontiersin.org Frontiers in Endocrinology \| www.frontiersin.org 5 Inactivated SARS-CoV-2 Vaccines on TRAB Huang et al. TABLE 3 \| Demographic and clinical characteristics of enrolled patients (n = 173). Gender (female%) 73.988 Age (year) 39.168 ±10.713 Duration (month) 16.333 (9.617,26.309) Thiamazole (mg) 14.957±5.293 Letrox (ug) 58.960 ±37.246 FT3 (pmol/L) 5.230 (4.500,6.570) FT4 (pmol/L) 15.930 (13.530,20.300) TSH (mIU/L) 0.375 (0.005,2.020) TRAB (IU/L) 5.450 (2.555,11.350) Data are presented as mean±standard error (x±s) or median (interquartile range). Categorical outcomes were shown as absolute and relative prevalence of complications (%). FT3 free triiodothyronine 3, FT4 free thyroxine 4, TSH thyroid-stimulating hormone, TRAB thyrotropin receptor antibody. TABLE 3 \| Demographic and clinical characteristics of enrolled patients (n = 173). DISCUSSION In the present study, we combined retrospective and prospective studies to investigate the effect of inactivated SARS-CoV-2 vaccines on TRAB in patients with GD. The results of inter- group comparison and repeated measures ANOVA indicated that serum TRAB levels decreased less after inactivated SARS- CoV-2 vaccination and showed an upward trend. Similarly, the serum FT3 and FT4 levels increased after vaccination. To the best of our knowledge, this is the ﬁrst comparative assessment of serum TRAB levels with inactivated SARS-CoV-2 vaccines in GD since the COVID-19 outbreak. SARS-CoV-2 is spreading rapidly worldwide with high numbers of conﬁrmed cases and fatality rates and limited treatment options. Widespread vaccination against COVID-19 is a crucial tool to control the pandemic. In China, most citizens are vaccinated with inactivated SARS-CoV-2. Inactivated vaccines are a mature technology with highly efﬁcient proliferation and high genetic stability and are widely used for the prevention and control of emerging infectious diseases (11). GD is the most common cause of hyperthyroidism. Under normal conditions, TSH receptors (TSHRs) located on the surface of thyroid cells bind to TSH, which activates adenylate cyclase and phosphoinositol-dependent signaling pathways to produce thyroid hormones (20). Hyperthyroidism is caused by the growth and reproduction of thyroid cells and persistent and uncontrolled thyroid stimulation resulting from the interaction TABLE 4 \| Variation in thyroid function of prospective subjects (n = 173). Pre-vaccination Post-vaccination P Pairwise Comparison 3 Months Before 1 Month Before 1 Month After 3 Months After FT3 (pmol/L) 5.230 (4.500,6.570) 4.630 (4.095,5.515) 4.740 (4.235,5.760) 5.020 (4.530,5.735) 0.000 0.000a 0.095b 0.001c FT4 (pmol/L) 15.930 (13.530,20.300) 15.490 (12.910,18.760) 16.220 (13.540,19.145) 16.610 (14.660,19.400) 0.000 0.006a 0.068b 0.012c TSH (mIU/L) 0.375 (0.005,2.020) 0.948 (0.017,2.700) 1.110 (0.042,2.765) 1.420 (0.204,3.225) 0.000 0.002a 0.064b 0.138c TRAB (IU/L) 5.450 (2.555,11.350) 3.950 (2.080,8.780) 3.700 (2.000,7.835) 4.100 (2.360,8.965) 0.000 0.000a 0.583b 0.034c Data are presented as median (interquartile range). Categorical outcomes were shown as absolute and relative prevalence of complications (%). FT3 free triiodothyronine 3, FT4 free thyroxine 4, TSH thyroid-stimulating hormone, TRAB thyrotropin receptor antibody. a3 Months Before versus 1 Month Before. b1 Month Before versus 1 Month After. c1 Month After versus 3 Months After. May 2022 \| Volume 13 \| Article 835880 Frontiers in Endocrinology \| www.frontiersin.org Frontiers in Endocrinology \| www.frontiersin.org 6 Inactivated SARS-CoV-2 Vaccines on TRAB Huang et al. FIGURE 3 \| Line charts of variation in thyroid function of prospective subjects. Frontiers in Endocrinology \| www.frontiersin.org CONCLUSION Taken together, based on retrospective and prospective studies, the data presented here demonstrate that serum TRAB levels decreased less after inactivated SARS-CoV-2 vaccination and showed an upward trend, and FT3 and FT4 were consistent with it. This may be related to humoral immunity induced by vaccination. This ﬁnding suggests that humoral immunity induced by inactivated SARS-CoV-2 vaccine may affect autoimmunity. The advantages and disadvantages of vaccination should be weighed according to the applicable population. Clinicians should be aware that TRAB levels may stop declining following vaccination. Based on the results obtained in both retrospective and prospective studies, we concluded that serum TRAB levels decreased less after vaccination and showed an upward trend. Thinking along the lines above, the association between inactivated SARS-CoV-2 vaccines and serum TRAB levels may be related to humoral immunity. After vaccination, antibodies were generated through humoral immunity, which could stimulate B cells and promote the synthesis of TRABs, changing the original declining trend of TRABs. Another explanation may be that adjuvants added to inactivated vaccines may disrupt the host immune balance and stimulate B cell cloning, affecting the original trend of TRAB. DATA AVAILABILITY STATEMENT The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. DISCUSSION Serum TRAB levels showed dynamic changes that decreased signiﬁcantly at 1 month before vaccination, showed no signiﬁcant differences at 1 month after vaccination, and changed slightly and reﬂected an upward trend at 3 months after vaccination. Serum FT3 and FT4 levels showed trends similar to those of serum TRAB levels before and after vaccination, but the nuances were that their upward trend moved forward and appeared 1 month after vaccination. Surprisingly, the changes in TSH after vaccination 1-3 months were inconsistent with those in FT3 and FT4, showing a continuous upward trend. We speculate that this phenomenon may be related to the sensitivity of TSH, which usually changes earlier than FT3 and FT4 levels. It may be that thyroid hormone is about to decline at the time point, but we have not captured their decreased levels due to the short follow-up time. The data presented here demonstrate that inactivated SARS- CoV-2 vaccines may affect TRAB trends in GD patients. Nevertheless, we must emphasize that the initial goal of vaccination is to protect the population from infection and reduce infection and mortality, and inactivated SARS-CoV-2 vaccines are not contraindicated in patients with GD (22). COVID-19 has led to millions of disabilities and deaths worldwide, especially in men, the elderly, and those with previous health problems; therefore, we believe that the risks of COVID-19 outweigh the minor risks of the vaccine in these populations. However, autoimmune diseases, especially GD, predominantly affect young women, who have a signiﬁcantly reduced risk of severe Covid-19 disease. Thus, a careful analysis of the risk/beneﬁt ratio should be continuously applied and revised according to the new scientiﬁc data that are produced daily. The results of this study provide evidence for clinical management and clinicians should be aware that TRAB levels may stop declining after vaccination. DISCUSSION To further explore the dynamic changes in TRAB before and after vaccination, we expanded a prospective study that measured the levels of serum thyroid function and TRAB at 3 and 1 months before and after vaccination in GD and then performed repeated measurement ANOVA. Lifestyle and treatment of the enrolled subjects were unaltered during the study. Consistent with the results of this retrospective study, there were signiﬁcant differences in serum FT3, FT4, and TRAB levels at different time points before and after vaccination. Serum TRAB levels showed dynamic changes that decreased signiﬁcantly at 1 month before vaccination, showed no signiﬁcant differences at 1 month after vaccination, and changed slightly and reﬂected an upward trend at 3 months after vaccination. Serum FT3 and FT4 levels showed trends similar to those of serum TRAB levels before and after vaccination, but the nuances were that their upward trend moved forward and appeared 1 month after vaccination. Surprisingly, the changes in TSH after vaccination 1-3 months were inconsistent with those in FT3 and FT4, showing a continuous upward trend. We speculate that this phenomenon may be related to the sensitivity of TSH, which usually changes earlier than FT3 and FT4 levels. It may be that thyroid hormone is about to decline at the time point, but we have not captured their decreased levels due to the short follow-up time. Generally, serum TRAB levels should decline over time with standard treatment, which was not observed in our study. However, an unexpected increase in TRAB levels was observed after vaccination. The expected trend of TRAB titer is to decrease before vaccination, which contrasts with the TRAB titer plateau immediately after vaccination (+1-month time point) and the subsequent inverted and unexpected trend of TRAB increase at +3-months time-point after vaccination. To further explore the dynamic changes in TRAB before and after vaccination, we expanded a prospective study that measured the levels of serum thyroid function and TRAB at 3 and 1 months before and after vaccination in GD and then performed repeated measurement ANOVA. Lifestyle and treatment of the enrolled subjects were unaltered during the study. Consistent with the results of this retrospective study, there were signiﬁcant differences in serum FT3, FT4, and TRAB levels at different time points before and after vaccination. DISCUSSION FIGURE 3 \| Line charts of variation in thyroid function of prospective subjects. recurrence. We conducted a retrospective study to observe the differences in TRAB trends between pre-vaccination and post- vaccination. Before vaccination, serum thyroid function and TRAB levels decreased signiﬁcantly at intervals of 2 months under standard treatment, indicating an effective treatment. In contrast to pre-vaccination, no signiﬁcant differences were found in serum thyroid function and TRAB levels post-vaccination, which indicated that there was no improvement in thyroid function and TRAB after vaccination under the same treatment regimen. Simultaneously, we compared serum TRAB levels at speciﬁed time points before and after vaccination. The results showed that the TRAB increased after vaccination, which also supports the previous prediction. The differences in the duration of the retrospective study were due to the overlap among subgroups of GD enrolled during the research, with the course of disease apparently longer post-vaccination than before. of TRAB with TSHR in genetically predisposed individuals with GD. The pathogenesis of GD involves the destruction of thyroid immune tolerance, with the most pathogenic antibody being TRAB, and the immune mechanism is complex. A large amount of evidence has shown that the active phase of GD is associated with an immune prevalence of the Th1 immune response, whereas the inactive or later phases of GD are associated with a switch from Th1 to Th2 immune prevalence (21). It has been speculated that the mechanism of immune tolerance disruption is the maladjustment of autoreactive B cells that switch to plasma cells that produce pathogenic immunoglobulin G (20). Moreover, increasing the stimulating effect of Th2 cells on B cells promotes the production of more TRABs, which is considered another mechanism of GD occurrence (20). TRABs were explored as an entry point of this study as an important indicator to evaluate treatment efﬁcacy and May 2022 \| Volume 13 \| Article 835880 Frontiers in Endocrinology \| www.frontiersin.org Inactivated SARS-CoV-2 Vaccines on TRAB Huang et al. Generally, serum TRAB levels should decline over time with standard treatment, which was not observed in our study. However, an unexpected increase in TRAB levels was observed after vaccination. The expected trend of TRAB titer is to decrease before vaccination, which contrasts with the TRAB titer plateau immediately after vaccination (+1-month time point) and the subsequent inverted and unexpected trend of TRAB increase at +3-months time-point after vaccination. Frontiers in Endocrinology \| www.frontiersin.org REFERENCES Best Pract Res Clin Endocrinol Metab (2020) 34 (1):101388. doi: 10.1016/j.beem.2020.101388 7. Vera-Lastra O, Ordinola Navarro A, Cruz Domiguez MP, Medina G, Sanchez Valadez TI, Jara LJ. Two Cases of Graves’ Disease Following SARS-CoV-2 Vaccination: An Autoimmune/Inﬂammatory Syndrome Induced by Adjuvants. Thyroid (2021) 31(9):1436–9. doi: 10.1089/thy.2021.0142 22. Ku CR, Jung KY, Ahn CH, Moon JS, Lee JH, Kim EH, et al. COVID-19 Vaccination for Endocrine Patients: A Position Statement From the Korean Endocrine Society. Endocrinol Metab (Seoul) (2021) 36(4):757–65. doi: 10.3803/EnM.2021.404 8. Zettinig G, Krebs M. Two Further Cases of Graves’ Disease Following SARS- Cov-2 Vaccination. J Endocrinol Invest (2022) 45(1):227–8. doi: 10.1007/ s40618-021-01650-0 Conﬂict of Interest: The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. Conﬂict of Interest: The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. 9. Iremli BG, Sendur SN, Unluturk U. Three Cases of Subacute Thyroiditis Following SARS-CoV-2 Vaccine: Postvaccination ASIA Syndrome. J Clin Endocrinol Metab (2021) 106(9):2600–5. doi: 10.1210/clinem/dgab373 10. Rubinstein TJ. Thyroid Eye Disease Following COVID-19 Vaccine in a Patient With a History Graves’ Disease: A Case Report. Ophthalmic Plast Reconstruct Surg (2021) 37(6):e221–e3. doi: 10.1097/iop.0000000000002059 Publisher’s Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their afﬁliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. 11. Wang H, Zhang Y, Huang B, Deng W, Quan Y, Wang W, et al. Development of an Inactivated Vaccine Candidate, BBIBP-CorV, With Potent Protection Against SARS-CoV-2. Cell (2020) 182(3):713–21.e9. doi: 10.1016/ j.cell.2020.06.008 12. Murray J, Todd KV, Bakre A, Orr-Burks N, Jones L, Wu W, et al. A Universal Mammalian Vaccine Cell Line Substrate. PLoS One (2017) 12(11):e0188333. doi: 10.1371/journal.pone.0188333 Copyright © 2022 Huang, Jiang, Zhou, Chen and Huang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). REFERENCES 14. Gao Q, Bao L, Mao H, Wang L, Xu K, Yang M, et al. Development of an Inactivated Vaccine Candidate for SARS-CoV-2. Science (New York NY) (2020) 369(6499):77–81. doi: 10.1126/science.abc1932 1. Scappaticcio L, Pitoia F, Esposito K, Piccardo A, Trimboli P. Impact of COVID-19 on the Thyroid Gland: An Update. Rev Endocr Metab Disord (2021) 22(4):803–15. doi: 10.1007/s11154-020-09615-z 1. Scappaticcio L, Pitoia F, Esposito K, Piccardo A, Trimboli P. Impact of COVID-19 on the Thyroid Gland: An Update. Rev Endocr Metab Disord (2021) 22(4):803–15. doi: 10.1007/s11154-020-09615-z 15. Xia S, Zhang Y, Wang Y, Wang H, Yang Y, Gao GF, et al. Safety and Immunogenicity of an Inactivated SARS-CoV-2 Vaccine, BBIBP-CorV: A Randomised, Double-Blind, Placebo-Controlled, Phase 1/2 Trial. Lancet Infect Dis (2021) 21(1):39–51. doi: 10.1016/s1473-3099(20)30831-8 2. Dong Y, Dai T, Wei Y, Zhang L, Zheng M, Zhou F. A Systematic Review of SARS-CoV-2 Vaccine Candidates. Signal Transduct Target Ther (2020) 5 (1):237. doi: 10.1038/s41392-020-00352-y 16. Guimaraes LE, Baker B, Perricone C, Shoenfeld Y. Vaccines, Adjuvants and Autoimmunity. Pharmacol Res (2015) 100:190–209. doi: 10.1016/j.phrs.2015.08.003 3. Boelaert K, Visser WE, Taylor PN, Moran C, Leger J, Persani L. ENDOCRINOLOGY IN THE TIME OF COVID-19: Management of Hyperthyroidism and Hypothyroidism. Eur J Endocrinol (2020) 183(1): G33–G9. doi: 10.1530/EJE-20-0445 17. Pellegrino P, Clementi E, Radice S. On Vaccine’s Adjuvants and Autoimmunity: Current Evidence and Future Perspectives. Autoimmun Rev (2015) 14(10):880–8. doi: 10.1016/j.autrev.2015.05.014 18. Ruiz JT, Lujan L, Blank M, Shoenfeld Y. Adjuvants- and Vaccines-Induced Autoimmunity: Animal Models. Immunol Res (2017) 65(1):55–65. doi: 10.1007/s12026-016-8819-5 4. Scappaticcio L, Castellana M, Virili C, Bellastella G, Centanni M, Cannavo S, et al. Alemtuzumab-Induced Thyroid Events in Multiple Sclerosis: A Systematic Review and Meta-Analysis. J Endocrinol Invest (2020) 43 (2):219–29. doi: 10.1007/s40618-019-01105-7 19. Bragazzi NL, Hejly A, Watad A, Adawi M, Amital H, Shoenfeld Y. ASIA Syndrome and Endocrine Autoimmune Disorders. Best Pract Res Clin Endocrinol Metab (2020) 34(1):101412. doi: 10.1016/j.beem.2020.101412 5. Bartalena L. Diagnosis and Management of Graves Disease: A Global Overview. Nat Rev Endocrinol (2013) 9(12):724–34. doi: 10.1038/ nrendo.2013.193 20. Chistiakov DA. Thyroid-Stimulating Hormone Receptor and its Role in Graves’ Disease. Mol Genet Metab (2003) 80(4):377–88. doi: 10.1016/ j.ymgme.2003.09.001 6. Toussirot É, Bereau M. Vaccination and Induction of Autoimmune Diseases. Inﬂamm Allergy Drug Targets (2015) 14(2):94–8. doi: 10.2174/ 1871528114666160105113046 21. Antonelli A, Fallahi P, Elia G, Ragusa F, Paparo SR, Rufﬁlli I, et al. Graves’ Disease: Clinical Manifestations, Immune Pathogenesis (Cytokines and Chemokines) and Therapy. ETHICS STATEMENT Our study has two limitations. First, the current study was conducted in a single-center cohort of patients, with the possibility of selection bias. Second, the short follow-up period in our study resulted in unclear trends in serum TRAB levels 3 months after vaccination. The follow-up period should be extended to validate the time point of the peak of its increase. The studies involving human participants were reviewed and approved by The Second afﬁliated Hospital of Fujian Medical University. The patients/participants provided their written informed consent to participate in this study. May 2022 \| Volume 13 \| Article 835880 Frontiers in Endocrinology \| www.frontiersin.org Inactivated SARS-CoV-2 Vaccines on TRAB Huang et al. AUTHOR CONTRIBUTIONS Research Project of Fujian Province (2018-CX-33), and High- Level Talent Program of Science and Technology Project of Quanzhou (2018C044R). All authors planned the concept of this report and wrote and revised the ﬁnal manuscript. All authors contributed to the article and approved the submitted version. FUNDING This work was supported by Science and Technology Project of Fujian Provincial Department (2019J01166), Innovative Medical The authors would like to thank Prof. HuiBin Huang for his valuable suggestions helping us to improve the manuscript. REFERENCES The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. 13. Calvo Fernandez E, Zhu LY. Racing to Immunity: Journey to a COVID-19 Vaccine and Lessons for the Future. Br J Clin Pharmacol (2021) 87(9):3408– 24. doi: 10.1111/bcp.14686 May 2022 \| Volume 13 \| Article 835880 Frontiers in Endocrinology \| www.frontiersin.org
https://openalex.org/W3163501001	https://pure.eur.nl/ws/files/58646105/Do_educational_reforms_increase_or_decrease_health_inequalities_A_matter_of_methods.pdf	English	null	Do educational reforms increase or decrease health inequalities: A matter of methods?	Social science & medicine	2,021	cc-by	6,451	A R T I C L E I N F O Keywords: Causal inference Difference-in-differences Health inequalities Microsimulation Impact assessment Evaluating whether social policies reduce health inequalities is complicated by the fact that these upstream determinants may also change the socioeconomic distribution. Failure to account for these compositional changes may severely bias the effect estimation procedure. In this article, we illustrate how a health inequality impact assessment of a policy that (also) changes the socioeconomic distribution may produce biased results. First, we show why analyses that do not account for compositional changes fail to estimate the correct coun terfactual outcome. This problem most notably occurs when using repeated cross-sectional data, often the only available option to evaluate the health effect of large-scale policies. Second, we conducted a microsimulation study to estimate the magnitude of the bias under various conditions. The results showed that the actual impact of the policy on health inequalities is often underestimated and may even produce results that are in the opposite direction of the actual causal effect of the policy. Future studies should explore new strategies, such as simulation methods, to assess the impact of policies that (also) cause changes in the socioeconomic composition of the population, to enable researchers to accurately estimate their effect on health inequalities. Do educational reforms increase or decrease health inequalities: A matter of methods? Joost Oude Groeniger a,b,, M´arta K. Rad´o c,a, Frank J. van Lenthe a Joost Oude Groeniger a,b,, M´arta K. Rad´o c,a, Frank J. van Lenthe a a Department of Public Health, Erasmus University Medical Centre, PO Box 2040, 3000 CA, Rotterdam, the Netherlands b Department of Public Administration and Sociology, Erasmus University Rotterdam, PO Box 1738, 3000 DR, Rotterdam, the Netherlands c Division of Neonatology, Department of Paediatrics, Erasmus MC — Sophia Children’s Hospital, University Medical Centre Rotterdam, Rotterdam, Netherlands * Corresponding author. Department of Public Health, Erasmus University Medical Centre, PO Box 2040, 3000 CA, Rotterdam, the Netherlands, E-mail addresses: j.oudegroeniger@erasmusmc.nl (J. Oude Groeniger), m.rado@erasmusmc.nl (M.K. Rad´o), f.vanlenthe@erasmusmc.nl (F.J. van Lenthe). Available online 7 May 2021 0277-9536/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) * Corresponding author. Department of Public Health, Erasmus University Medical Centre, PO Box 2040, 3000 CA, Rotterdam, the Netherlands, E-mail addresses: j.oudegroeniger@erasmusmc.nl (J. Oude Groeniger), m.rado@erasmusmc.nl (M.K. Rad´o), f.vanlenthe@erasmusmc.nl (F.J. van Lenthe). https://doi.org/10.1016/j.socscimed.2021.114003 Received in revised form 30 April 2021; Accepted 4 May 2021 Available online 7 May 2021 0277-9536/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) https://doi.org/10.1016/j.socscimed.2021.114003 Received in revised form 30 April 2021; Accepted 4 May 2021 Social Science & Medicine 279 (2021) 114003 Do educational reforms increase or decrease health inequalities: A matter of methods? Joost Oude Groeniger a,b,, M´arta K. Rad´o c,a, Frank J. van Lenthe a a Department of Public Health, Erasmus University Medical Centre, PO Box 2040, 3000 CA, Rotterdam, the Netherlands b Department of Public Administration and Sociology, Erasmus University Rotterdam, PO Box 1738, 3000 DR, Rotterdam, the Netherlands c Division of Neonatology, Department of Paediatrics, Erasmus MC — Sophia Children’s Hospital, University Medical Centre Rotterdam, Rotterdam, Netherlands Contents lists available at ScienceDirect Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed Social Science & Medicine 279 (2021) 114003 Do educational reforms increase or decrease health inequalities: A matter of methods? Joost Oude Groeniger a,b,, M´arta K. Rad´o c,a, Frank J. van Lenthe a a Department of Public Health, Erasmus University Medical Centre, PO Box 2040, 3000 CA, Rotterdam, the Netherlands b Department of Public Administration and Sociology, Erasmus University Rotterdam, PO Box 1738, 3000 DR, Rotterdam, the Netherlands c Division of Neonatology, Department of Paediatrics, Erasmus MC — Sophia Children’s Hospital, University Medical Centre Rotterdam, Rotterdam, Netherlands Contents lists available at ScienceDirect Social Science & Medicine journal homepage: www.elsevier.com/locate/socscimed Social Science & Medicine 279 (2021) 114003 2. Assessing the equity impact of policies that change the socioeconomic distribution Imagine a hypothetical population consisting of 8 individuals, 4 of which are less-educated and 4 of which are more-educated (Table 1). In this population, the less-educated have a mean life expectancy of 73 years, whereas the more-educated have a mean life expectancy of 82 years. Let’s assume that at one particular point in time, the members of this population decide that it would be beneficial if a higher proportion of them would be more-educated. In order to achieve this, they imple ment a particular policy that encourages more students to attain a higher educational degree. Although not the primary goal of the policy, someone suggests that the policy may also be an effective strategy to tackle health inequalities. How would one decide whether or not this policy was (also) an effective means to reduce educational inequalities in health (i.e. whether or not the policy had an equity-positive impact)? Because we are unable to actually observe the counterfactual educational levels, we must rely on a comparison between a population that has been exposed to the policy and a population that has not been exposed to the policy. Doing so, however, also implies that in the exposed population, the observed educational distribution will always include any compositional changes brought on by the policy. Conse quently, the actual equity effect of the policy cannot be identified without any correction for these compositional changes. Returning to our hypothetical example: the policy has, in fact, improved the health of the less-educated, which, in this population, is the health of Hector, Laodice, Polydamas and Cassandra. That this result is actually achieved by raising Polydamas’ and Cassandra’s educational level doesn’t negate the fact that – from a counterfactual perspective – implementing the policy caused a decrease in educational inequalities in health. To illustrate our argument, imagine that we know the pre-policy and post-policy life expectancy from all individuals in this hypothetical population. Also imagine that their life expectancy would not have changed if the policy had not been implemented. This is known as their potential outcome or counterfactual outcome: the outcome that would have been observed if, counter to the fact, the policy had not been implemented (Hernan, 2004). Table 1 shows that most individuals were not affected by the policy: the pre-policy life expectancy of Hector, Laodice, Paris, Andromache, Aeneas and Glaucus is equal to their post-policy life expectancy. 3. Compositional changes lead to ambiguous causal inference Identification of the causal effect of the policy becomes ambiguous when the compositional changes in educational level caused by the policy are used to ‘re-classify’ the less and more-educated groups: in dividuals that have obtained a higher educational level due to the policy are classified as less-educated in the pre-policy period and classified as more-educated in the post-policy period (Table 2). In our hypothetical population, this re-classification would result in an observed post-policy life-expectancy of 71 years for the less-educated (now only Hector and Laodice) and 81 years for the more-educated (which now includes Pol ydamas and Cassandra). Crucially, because the post-policy more- educated group includes individuals who would have remained less- educated in the absence of the policy, the comparison between observed outcomes is no longer made between the same groups of in dividuals. Calculating the policy effect in this way would lead us to falsely conclude that the policy caused a 1-year increase in the difference in life expectancy between the less and more-educated (Table 2). 2. Assessing the equity impact of policies that change the socioeconomic distribution However, Polydamas and Cassandra were affected by the policy: their life expectancy increased by 4 years because they became more-educated. Note, however, that from a health equity perspective, it may also be relevant to consider how the policy impacts the health gap that remains between those that are not upwardly mobile (i.e. Hector and Laodice) and the others, but this is a different research objective. Since the policy is not targeting any intermediate factor, but rather educational attain ment itself, it is much less likely that it will also increase the health of the non-mobile. Comparing the pre-policy and post-policy life expectancies shows that the policy caused a 1-year increase in mean life expectancy (Table 1). Moreover, this gain in life expectancy occurred exclusively among (initially) less-educated individuals (Polydamas and Cassandra). Whereas the difference in life expectancy between the less and more- educated was 9 years before the policy was implemented, it was only 7 years after the policy had been implemented (Table 2). Hence, the equity impact of the policy was a 2-year reduction in the difference in 1. Introduction For example, policies and programs that reduce the number of early school leavers, allocate subsidies to low income groups, or change the labor market, may substantially improve population health by lowering the proportion of people exposed to disadvantageous social Social Science & Medicine 279 (2021) 114003 J. Oude Groeniger et al. life expectancy between those with a (pre-policy) low level of education and those with a (pre-policy) high level of education. findings could also be the result of increased educational mobility: comprehensive schooling reforms reduce the probability that children drop out of school early (Van de Werfhorst and Mijs, 2010), which may positively affect their health. Since repeated cross-sectional data are not able to factor in compositional changes in the population, the true equity impact of the reform remains unknown. In the next section, we illustrate why an analysis that does not account for compositional changes fails to identify the causal effect of a policy on health inequalities. Subse quently, we use microsimulations to estimate the magnitude of this bias under various conditions. 1. Introduction positions. Obtaining an accurate assessment of the equity impact of these policies requires that evaluation studies factor in these socioeco nomic shifts (Harper and Lynch, 2006). Failure to do so, may severely bias the results and even produce results that are in the opposite di rection of the actual causal effect of the policy. Despite ongoing efforts to reduce health inequalities in Western so cieties, the scientific evidence-base for effective measures to tackle so cioeconomic inequalities in health is still limited. This is usually attributed to the fact that little is known about the effects of macro-level determinants of health, such as social policies and institutions (Brave man et al., 2011; Lorenc et al., 2013; Petticrew et al., 2004). While these upstream determinants probably have the greatest potential to reduce health inequalities, changes in these determinants are also hard to evaluate. They require quasi-experimental methods and adequate con trol groups for assessing causal effects (Basu et al., 2017; Craig et al., 2017). An increasing number of studies propose to tackle health inequalities not by using traditional health care reforms, but by relying on educa tional reforms (Cohen and Syme, 2013; Low et al., 2005; Walsemann et al., 2013). To evaluate whether these policies actually have an impact on educational inequalities in health, one needs to disentangle the direct health effects of these policies from the health effects that occur via increasing social mobility. As studies often fail to distinguish these ef fects, they might be subject to bias. For example, a recent study used a difference-in-differences design on repeated cross-sectional data to investigate the effects of comprehensive school reforms (increasing the age of early tracking) on educational inequalities in self-rated health (Delaruelle et al., 2019). They found that middle and high educated had better, and early school leavers worse health after the reform, and interpreted this as evidence that comprehensive school reforms were unable to reduce educational inequalities in health. However, these Assessing the impact of social policies on socioeconomic inequalities in health (i.e. the equity impact) is complicated by the fact that these upstream determinants may also define the nature of stratification in a society. a Counterfactual educational level and counterfactual life expectancy is equal to pre-policy level. 4. Bias in difference-in-differences analysis To further illustrate our argument, we consider the example of a difference-in-differences (DiD) analysis. This estimation procedure is a common approach to evaluate the health effects of social policies (Basu et al., 2017; Saeed et al., 2019; Wing et al., 2018). If a suitable control population is available to fulfil the counterfactual assumptions of the DiD approach, the design allows researchers to estimate the total causal effect of a policy. However, when the equity effect of the policy is of interest, any change in the distribution of socioeconomic position associated with a change in exposure (i.e. the policy) violates the com mon trends assumption of the DiD model (Stuart et al., 2014). (Note that estimating the equity effect of policies is complicated even further by the fact that it may also induce collider bias (Cole et al., 2010; Elwert and Winship, 2014; Hernan et al., 2004). This bias occurs because condi tioning on SES (e.g. by stratifying the analysis) introduces a non-causal association between SES and health by opening up a backdoor path via any unmeasured confounder of this relationship (e.g. parental SES, ethnic background or genetic factors). Because our paper aims to spe cifically address the problems associated with compositional changes, we assume the absence of collider bias.) Table 1 Educational level and life expectancy of all individuals in the hypothetical population. Educational level (pre- policy – post- policya) Pre-policy life expectancy Post-policy life expectancya Policy effect on life expectancy Hector Low – Low 70 70 0 Laodice Low – Low 72 72 0 Polydamas Low – High 74 78 4 Cassandra Low – High 76 80 4 Paris High – High 80 80 0 Andromache High – High 82 82 0 Glaucus High – High 82 82 0 Aeneas High – High 84 84 0 Mean life expectancy total population 77.5 78.5 1 a Counterfactual educational level and counterfactual life expectancy is equal to pre-policy level. Table 1 Educational level and life expectancy of all individuals in the hypothetical population. Table 2 d Polydamas, Cassandra, Paris, Andromache, Glaucus, Aeneas. outcome for the intervention group (estimated from the trend in the control group) cannot be estimated. treated and j = 0 if untreated) at time t. Fig. 1A illustrates how the DiD approach estimates the effect of the policy for the total population. For the sake of simplicity, we only demonstrate the scenario for two time points, but the same argument is applicable when multiple time points are considered. In Fig. 1, the lowest (dashed) line depicts the change in outcome among the (untreated) control group and the highest (solid) line depicts the trajectory among the (treated) intervention group. To estimate the causal effect of the policy, the DiD approach uses the trend in the control group as the unobserved counterfactual trend that would have occurred in the intervention group in the absence of the policy (the dashed line in the middle). Consequently, the effect of the policy is calculated as the difference between the change in the outcome before and after the policy in the intervention group and the change in outcome in the corresponding time period in the control group. treated and j = 0 if untreated) at time t. Fig. 1A illustrates how the DiD approach estimates the effect of the policy for the total population. For the sake of simplicity, we only demonstrate the scenario for two time points, but the same argument is applicable when multiple time points are considered. In Fig. 1, the lowest (dashed) line depicts the change in outcome among the (untreated) control group and the highest (solid) line depicts the trajectory among the (treated) intervention group. To estimate the causal effect of the policy, the DiD approach uses the trend in the control group as the unobserved counterfactual trend that would have occurred in the intervention group in the absence of the policy (the dashed line in the middle). Consequently, the effect of the policy is calculated as the difference between the change in the outcome before and after the policy in the intervention group and the change in outcome in the corresponding time period in the control group. Table 1 Let yj t denote the average life expectancy of treatment group j (j = 1 if 2 Social Science & Medicine 279 (2021) 114003 J. Oude Groeniger et al. Table 2 Causal and observed effect of the policy by level of education. Mean life expectancy Policy effect Pre- policy Post-policy (based on counterfactual educational levels) Post-policy (based on observed educational levels) Causal effect Observed effect Low education 73a 75a 71c 2 −2 High education 82b 82b 81d 0 −1 Difference high-low education 9 7 10 −2 1 a Hector, Laodice, Polydamas, Cassandra. b Paris, Andromache, Glaucus, Aeneas. c Hector, Laodice. d Polydamas, Cassandra, Paris, Andromache, Glaucus, Aeneas. Table 2 Causal and observed effect of the policy by level of education. 5. Microsimulation model description Illustration of a biased difference-in-differences approach to estimate the effect of a policy for different socioeconomic groups (the grey arrow depicts a compositional change caused by the policy). J. Oude Groeniger et al. Social Science & Medicine 279 (2021) 114003 J. Oude Groeniger et al. Fig. 2. Illustration of a biased difference-in-differences approach to estimate the effect of a policy for different socioeconomic groups (the grey arrow depicts a compositional change caused by the policy). Fig. 2. Illustration of a biased difference-in-differences approach to estimate the effect of a policy for different socioec compositional change caused by the policy). upwardly mobile had a mean LE of 76.3–76.8 (depending on the probability of upward mobility) prior to policy implementation. and standard deviation 13, reflecting the actual life expectancies of less and more-educated individuals in The Netherlands (Volksgezondhei denzorg.info, 2021). In the second stage, we simulated a policy inter vention that caused some of the initially less-educated individuals, to become more-educated. We let the probability of upward mobility (which corresponds to the proportion of initially less-educated in dividuals that become more-educated due to the policy) range from 0 to 0.40 in steps of 0.02. We chose this wide range of probabilities and small incremental step for the purposes of illustration, rather than specifically targeting one real-life example. We simulated four scenarios: and standard deviation 13, reflecting the actual life expectancies of less and more-educated individuals in The Netherlands (Volksgezondhei denzorg.info, 2021). In the second stage, we simulated a policy inter vention that caused some of the initially less-educated individuals, to become more-educated. We let the probability of upward mobility (which corresponds to the proportion of initially less-educated in dividuals that become more-educated due to the policy) range from 0 to 0.40 in steps of 0.02. We chose this wide range of probabilities and small incremental step for the purposes of illustration, rather than specifically targeting one real-life example. We simulated four scenarios: Last, we also varied the gain in LE acquired by the upwardly mobile individuals by calculating their post-policy LE as a percentage of the LE of more-educated individuals (randomly estimated from its distribution) and the remainder based on the individual’s pre-policy LE. We let the percentage of life expectancy gained via upward mobility range from 0% to 100% in steps of 10% (again using a wide range for illustrative pur poses). 5. Microsimulation model description The previous paragraphs illustrate that the impact assessments of policies and interventions on socioeconomic inequalities in health may be biased by compositional changes brought on by these same policies and interventions. However, the magnitude of this bias depends on several factors, such as the degree to which the policy is able to cause upward mobility, whether or not this upward mobility is related to health (e.g. are those who are already more healthy also more likely to be upwardly mobile), and the amount of health gained by upward mobility. To estimate how these various conditions affect the results of an impact assessment, we conducted a microsimulation study. When the same approach is used to estimate the effect of the policy on educational inequalities in health, however, the effect should be separately identifiable for all educational groups considered. For example, if two groups are considered (i.e. ‘high SES’ and ‘low SES’), the policy effect is calculated as the difference in the DiD estimate for the two groups (Δinequality = Δhigh SES – Δlow SES). Fig. 2 illustrates how the DiD approach estimates these separate effects and why this approach is invalid if the policy also changes the educational composition of the population. In this scenario, half of the initially-less educated (‘low SES’) individuals among the intervention group become more-educated (‘high SES’) due to the intervention. In the control group, however, these compositional changes do not occur. Consequently, the counterfactual Our microsimulation model followed a synthetic population of 10.000 individuals divided into two educational groups and consisted of two stages. In the first stage, the baseline characteristics of the popula tion were set up. Each member of the population was categorized as either less-educated or more-educated. In the first set of simulations, 20% of the synthetic population was less-educated; in the second set of simulations, 50% of the synthetic population was less-educated, and in the third set of simulations 80% of the synthetic population was less- educated. The more-educated group’s life expectancy (LE) followed a normal distribution with mean 84 and standard deviation 13, while the less-educated group’s LE followed a normal distribution with mean 79 Fig. 1. Illustration of a difference-in-differences approach to estimate the effect of a policy for the total population. Fig. 1. Illustration of a difference-in-differences approach to estimate the effect of a policy for the total population. Social Science & Medicine 279 (2021) 114003 Fig. 2. 5. Microsimulation model description In this scenario, the upwardly mobile had a mean LE of 81.5–82.1 (depending on the probability of upward mobility) prior to policy implementation. 3) We simulated an additional scenario where the probability of up ward mobility was three times as high among less-educated in dividuals with an above-average LE than among less-educated individuals with a below-average LE. In this scenario, the upwardly mobile had a mean LE of 81.5–82.1 (depending on the probability of upward mobility) prior to policy implementation. 5. Microsimulation model description These parameters correspond to a scenario where there is no causal effect of education on LE (i.e. the gain in LE for upwardly mobile individuals is 0% because becoming more-educated does not increase LE) to a scenario where the association between education and LE is completely causal (i.e. those who become more-educated obtain 100% of the LE of the high educated). 1) Upward mobility was assumed to be completely random. In other words, those whose educational level is increased by the policy (i.e. the upwardly mobile) had a mean LE of 79 prior to policy imple mentation (calculated from the first stage of the simulation), similar to those who are not upwardly mobile. For the sake of simplicity, we assumed that the policy did not cause any downward mobility and only impacted life expectancy via its impact on educational attainment. Making these assumptions underestimates the magnitude of the bias, except for the unrealistic scenarios where downward mobility has a positive effect on health or where the positive (negative) health effect of the policy is stronger (weaker) among those who remain less-educated. 2) Social mobility is often not random, but associated with health (i.e. direct selection) or determinants that are relevant for health (i.e. indirect selection). In other words, those who are already more healthy or have individual characteristics conducive to good health are more likely to benefit from a policy that increases their proba bility to become more-educated. To investigate the impact of this selection effect we simulated a scenario where the probability of upward mobility was twice as high among less-educated individuals with an above-average LE than among less-educated individuals with a below-average LE. In this scenario, the upwardly mobile had a mean LE of 80.5–81.0 (depending on the probability of upward mobility) prior to policy implementation. For each simulation, we calculated the actual causal effect of the policy on educational inequalities in LE (based on the pre-policy clas sification of education only) and the observed effect of the policy if compositional changes in education caused by the policy are used to re- classify low and high educated individuals (based on the post-policy classification of education). We used 1.000 iterations for each simulation. 3) We simulated an additional scenario where the probability of up ward mobility was three times as high among less-educated in dividuals with an above-average LE than among less-educated individuals with a below-average LE. 6. Results Results from the other microsimulations also showed that the probability of observing an effect in the opposite direction of the actual causal effect is larger when the prevalence of low education is smaller or when the probability of upward mobility is larger. A complete overview of the results of all microsimulations is provided in the Supplementary Material. impact of income redistribution or active labor market policies on in come inequalities in health. Whenever the policy under evaluation also affects the socioeconomic indicator that is used in the study, the equity impact assessment may be biased. Results from our microsimulations suggest that this bias may be substantial. One solution to the problem of compositional changes is to use a socioeconomic indicator that is not affected by the policy: in the case of an educational reform, studies could examine the impact of the policy on health inequalities by parental SES (Ravesteijn et al., 2017). Since parental SES will not be affected by the educational reform, the evalu ation study will be not be biased by compositional changes. Following the same argument, studies can examine the impact of social security and labor market policies on educational inequalities in health. How ever, the downside is that this may also change the study’s substantive research question. A different solution would be to use health inequality measures that are able to account for compositional changes in the so cioeconomic distribution. The Slope Index of Inequality and the Relative Index of Inequality are often used for these purposes (Harper and Lynch, 2006; Mackenbach and Kunst, 1997), although a recent study suggests that these measures are also not able to sufficiently factor in socioeco nomic shifts (Renard et al., 2019). In addition, using these measures does also not allow for pairwise comparisons between specific socio economic groups (e.g. early school leavers), because they aggregate information from the entire socioeconomic distribution. Future studies should explore new strategies to assess the equity impact of policies that (also) cause changes in the socioeconomic distribution. Micro simulations, such as the one we used in this study, may be a useful tool to estimate complex and long-term health equity impacts (Abraham, 2013; Epstein, 2008). These models can be recalibrated to different contexts and certain parameters (i.e. the probability of upward mobility) can be quantified to reflect real-life asymmetries and facilitate data provision for effective policy implementation. 7. Discussion Scholars increasingly recognize the need to address the most up stream determinants of health to effectively tackle health inequalities. Promising strategies to tackle these ‘causes of the causes’ include the implementation of social policies that directly address people’s social and economic opportunities, such as educational policies, social security policies and labor market policies. Evaluating whether these social policies affect health inequalities, however, will be biased if no correc tion is made for the compositional changes brought on by these policies. This is an especially pressing issue in case of policies that were not specifically designed to decrease health inequalities, but rather to in crease social mobility or reduce social inequalities in general. While we used the example of educational reforms and its impact on educational inequalities in health, the same argument applies to, for example, the In conclusion, empirical analyses attempting to estimate if a social policy decreases health inequalities may be severely biased if they do not account for compositional changes brought on by the policy. Generally, policies that reduce social inequality are also beneficial from a health equity perspective; we should be careful not to convince ourselves otherwise. Fig. 3. Estimated changes in inequality in life ex pectancy in a synthetic population in which 50% was less-educated and 20% of those were upwardly mo bile, by mean life expectancy of these upwardly mo bile persons prior to the policy implementation. The dashed lines represent the causal effect of the policy accounting for compositional changes (depicted as ‘actual’). The solid lines represent the effect of the policy not accounting for compositional changes (depicted as ‘observed’). The darkest lines represent the scenarios where the upwardly mobile obtain 100% of the life expectancy of the more-educated, the lightest lines represent the scenarios where the up wardly mobile obtain 0% of the life expectancy of the more-educated. The y-axis displays the estimated change in inequalities in life expectancy (a negative score indicates a decrease in health inequalities). The x-axis displays different LE’s of the upwardly mobile prior to the policy implementation (i.e., their mean LE in the first phase of the simulation). 6. Results Finally, these models can also be used to conduct sensitivity analyses and estimate the magnitude of po tential biases in real-life applications (Epstein, 2008). 6. Results The microsimulations showed that, in most scenarios, the actual impact of the policy on educational inequalities in life expectancy was severely underestimated if compositional changes were not accounted for, and in many conditions even produced results that are in the opposite direction of the actual causal effect of the policy. To illustrate how the different parameters affect the magnitude and direction of the 4) Although a much less realistic scenario, we also simulated a scenario where the probability of upward mobility was half as high among less-educated individuals with an above-average LE than among less- educated individuals with a below-average LE. In this scenario, the 4) Although a much less realistic scenario, we also simulated a scenario where the probability of upward mobility was half as high among less-educated individuals with an above-average LE than among less- educated individuals with a below-average LE. In this scenario, the 4 Social Science & Medicine 279 (2021) 114003 J. Oude Groeniger et al. bias, we plotted the results of the microsimulations where 50% of the synthetic population was less-educated and 20% of those were upwardly mobile in Fig. 3. It shows that the difference between the actual causal effect of the policy accounting for compositional changes and the observed effect of the policy not accounting for compositional changes was strongly dependent on the health gain acquired via upward mobility: the greater the actual increase in LE (i.e. the more effective a policy is in reducing health inequalities), the lower the observed change in inequality in LE. In other words, the actual effect and the observed effect are inversely related to each other and depend on the extent to which an increase in education leads to an increase in LE. Moreover, the results from the microsimulation showed that the observed change in inequality in LE may even be in the opposite direction of the actual causal effect of the policy (i.e. show an increase in health inequalities) when upward mobility is positively related to health. Given the high likelihood that policies that increase upward mobility do so especially among those with (characteristics conducive to) better health, it is conceivable that studies may actually conclude that a particular policy increases health inequalities, while in fact it does the opposite. Appendix A. Supplementary data , ,i p g J. Epidemiol. Community Health 58, 265–271. Hernan, M.A., Hernandez-Diaz, S., Robins, J.M., 2004. A structural approach to selection i J. Epidemiol. Community Health 58, 265–271. H M A H d Di S R bi J M 2004 A t t l h t l ti i J. Epidemiol. Community Health 58, 265–271. Hernan, M.A., Hernandez-Diaz, S., Robins, J.M., 2004. A structural approach to selection bias. Epidemiology 15, 615–625. Hernan, M.A., Hernandez-Diaz, S., Robins, J.M., 2004. A structural approach to selection bias. Epidemiology 15, 615–625. Supplementary data to this article can be found online at https://doi. org/10.1016/j.socscimed.2021.114003. Lorenc, T., Petticrew, M., Welch, V., Tugwell, P., 2013. What types of interventions generate inequalities? Evidence from systematic reviews. J. Epidemiol. Community Health 67, 190–193. Low, M.D., Low, B.J., Baumler, E.R., Huynh, P.T., 2005. Can education policy Be health policy? Implications of research on the social determinants of health. J. Health Polit. Policy Law 30, 1131–1162. Basu, S., Meghani, A., Siddiqi, A., 2017. Evaluating the health impact of large-scale public policy changes: classical and novel approaches. Annu. Rev. Publ. Health 38, 351–370. 7. Discussion This indicates to what extent the probability of upward mobility is related to a person’s LE: a score of 79 (similar to the mean LE of all less-educated individuals) indicates that the upwardly mobile had the same mean LE prior to the policy implementation as the other less-educated, a mean LE above 79 years indicates that the upwardly mobile already had a higher mean LE prior to the policy implementation than the other less-educated, and a mean LE below 79 indicates that the upwardly mobile had a lower mean LE prior to the policy implementation than the other less-educated. p mean LE of all less-educated individuals) indicates that the upwardly mobile had the same mean LE prior to the policy implementation as the other less-educated, a mean LE above 79 years indicates that the upwardly mobile already had a higher mean LE prior to the policy implementation than the other less-educated, and a mean LE below 79 indicates that the upwardly mobile had a lower mean LE prior to the policy implementation than the other less-educated. mean LE of all less educated individuals) indicates that the upwardly mobile had the same mean LE prior to the policy implementation as the other less-educated, a mean LE above 79 years indicates that the upwardly mobile already had a higher mean LE prior to the policy implementation than the other less-educated, and a mean LE below 79 indicates that the upwardly mobile had a lower mean LE prior to the policy implementation than the other less-educated. 5 J. Oude Groeniger et al. Social Science & Medicine 279 (2021) 114003 Acknowledgements g , Elwert, F., Winship, C., 2014. Endogenous selection bias: the problem of conditioning on a collider variable. Annu. Rev. Sociol. 40, 31–53. Elwert, F., Winship, C., 2014. Endogenous selection bias: the problem of conditioning on a collider variable. Annu. Rev. Sociol. 40, 31–53. E i J M 2008 Wh d l? J A if S S Si l 11 12 , , p, , g p g a collider variable. Annu. Rev. Sociol. 40, 31–53. Epstein J M 2008 Why model? J Artif Soc Soc Simulat 11 12 JOG thanks Famke M¨olenberg and Roel Bottema for helpful discussion. , Epstein, J.M., 2008. Why model? J. Artif. Soc. Soc. Simulat. 11, 12. Epstein, J.M., 2008. Why model? J. Artif. Soc. Soc. Simulat. 11, 12. Harper, S., Lynch, J., 2006. Measuring inequalities in health. In: Oa d h d l d l Harper, S., Lynch, J., 2006. Measuring inequalities in health. In: Oakes, J., Kaufman, J. (Eds.), Methods in Social Epidemiology. Jossey-Bass, San Francisco. Hernan, M.A., 2004. A definition of causal effect for epidemiological research. Hernan, M.A., 2004. A definition of causal effect for epidemiological research. J. Epidemiol. Community Health 58, 265–271. Funding Ravesteijn, B., van Kippersluis, H., Avendano, M., Martikainen, P., Vessari, H., Van Doorslaer, E., 2017. The Impact of Later Tracking on Mortality by Parental Income in Finland. Tinbergen Institute Discussion Paper. No. 17-030/V. The study was supported by a grant from the Netherlands Organi zation for Health Research and Development (ZonMw) (project number 531003013). The funders had no role in the study design or the analysis and interpretation of the data. All authors and their institutions reserve intellectual freedom from the funders. Renard, F., Devleesschauwer, B., Speybroeck, N., Deboosere, P., 2019. Monitoring health inequalities when the socio-economic composition changes: are the slope and relative indices of inequality appropriate? Results of a simulation study. BMC Publ. Health 19, 662. Saeed, S., Moodie, E.E.M., Strumpf, E.C., Klein, M.B., 2019. Evaluating the impact of health policies: using a difference-in-differences approach. Int. J. Publ. Health 64, 637–642. The authors declare that they have no competing interests. Delaruelle, K., van de Werfhorst, H., Bracke, P., 2019. Do comprehensive school reforms impact the health of early school leavers? Results of a comparative difference-in- diff d i S S i M d 239 112542 Delaruelle, K., van de Werfhorst, H., Bracke, P., 2019. Do comprehensive school reforms impact the health of early school leavers? Results of a comparative difference-in- difference design. Soc. Sci. Med. 239, 112542. l i hi C 2014 d l i bi h bl f di i i Acknowledgements Authors’ contributions JOG conceived the study. JOG and MKR ran the microsimulation study. All authors contributed to writing the manuscript. All authors read and approved the final manuscript. Mackenbach, J.P., Kunst, A.E., 1997. Measuring the magnitude of socio-economic inequalities in health: an overview of available measures illustrated with two examples from Europe. Soc. Sci. Med. 44, 757–771. Petticrew, M., Whitehead, M., Macintyre, S.J., Graham, H., Egan, M., 2004. Evidence for public health policy on inequalities: 1: the reality according to policymakers. J. Epidemiol. Community Health 58, 811–816. Abraham, J.M., 2013. Using microsimulation models to inform U.S. Health policy making. Health Serv. Res. 48, 686–695. Braveman, P., Egerter, S., Williams, D.R., 2011. The social determinants of health: coming of age. Annu. Rev. Publ. Health 32, 381–398. Cole, S.R., Platt, R.W., Schisterman, E.F., Chu, H., Westreich, D., Richardson, D., et al., 2010. Illustrating bias due to conditioning on a collider. Int. J. Epidemiol. 39, 417–420. Cohen, A.K., Syme, S.L., 2013. Education: a missed opportunity for public health intervention. Am. J. Publ. Health 103, 997–1001. Declaration of competing interest Craig, P., Katikireddi, S.V., Leyland, A., Popham, F., 2017. Natural experiments: an overview of methods, approaches, and contributions to public health intervention research. Annu. Rev. Publ. Health 38, 39–56. The authors declare that they have no competing interests. References Stuart, E.A., Huskamp, H.A., Duckworth, K., Simmons, J., Song, Z., Chernew, M., et al., 2014. Using propensity scores in difference-in-differences models to estimate the effects of a policy change. Health Serv. Outcome Res. Methodol. 14, 166–182. d fh ij hi i li d h i i i l Van de Werfhorst, H.G., Mijs, J.J.B., 2010. Achievement inequality and the institutional structure of educational systems: a comparative perspective. Annu. Rev. Sociol. 36, 407–428. Volksgezondheidenzorg.info. www.volksgezondheidenzorg.info, 2021. W l K M G G C R A 2013 Ed i l i i Walsemann, K.M., Gee, G.C., Ro, A., 2013. Educational attainment in the context of social inequality:new directions for research on education and health. Am. Behav. Sci. 57, 1082–1104. Wing, C., Simon, K., Bello-Gomez, R.A., 2018. Designing difference in difference studies: best practices for public health policy research. Annu. Rev. Publ. Health 39, 453–469.
https://openalex.org/W2110822101	http://hrcak.srce.hr/file/150253	English	null	An Enchiridion of Supramolecular Thermodynamics: Calix[N]arene (N=4,5,6) Tertiary Amide Derivatives and their Ionic Recognition	Croatica chemica acta	2,013	cc-by	16,052	† This article belongs to the Special Issue devoted to the 85th anniversary of Croatica Chemica Acta. * Author to whom correspondence should be addressed. (E-mail: a.danil-de-namor@surrey.ac.uk) CROATICA CHEMICA ACTA CCACAA, ISSN 0011-1643, e-ISSN 1334-417X Croat. Chem. Acta 86 (1) (2013) 1–19. http://dx.doi.org/10.5562/cca2170 CROATICA CHEMICA ACTA CCACAA, ISSN 0011-1643, e-ISSN 1334-417X Croat. Chem. Acta 86 (1) (2013) 1–19. http://dx.doi.org/10.5562/cca2170 Feature Article RECEIVED SEPTEMBER 3, 2012; REVISED JANUARY 24, 2013; ACCEPTED FEBRUARY 15, 2013 RECEIVED SEPTEMBER 3, 2012; REVISED JANUARY 24, 2013; ACCEPTED FEBRUARY 15, 2013 Abstract. The importance of detailed thermodynamic studies in assessing the selective behaviour of macrocyclic receptors for one species relative to another in a given solvent and the medium effect on complexation processes involving ionic species are emphasised. Factors to be considered in the determi- nation of thermodynamic parameters of complexation in non-aqueous solvents are highlighted. Particular reference is made to the need for considering the bulk of information available in the literature on the so- lution properties of electrolytes in non-aqueous medium in the selection of the solvent for ion complexation processes involving macrocycles. A detailed thermodynamic study on the interaction of p-tert-butyl calix[n]arene (n = 4−6) tertiary amide derivatives with uni- and bivalent cations in protic (methanol) and dipolar aprotic (acetonitrile) media is reported. It is demonstrated that as the number of phenyl units in the macrocycle increases, the vital fea- ture of the cyclic tetramer receptor for selective recognition of cations decreases significantly for the cy- clic pentamer and almost disappears for the hexamer. Concluding remarks are included. (doi: 10.5562/cca2170) Keywords: thermodynamics, supramolecular chemistry, calix[n]arenes, cation complexation Angela F. Danil de Namor,* Tomas T. Matsufuji-Yasuda, Katherine Zegarra-Fernandez, Oliver A. Webb, and Abdelaziz El Gamouz Laboratory of Thermochemistry, Department of Chemistry, University of Surrey, Guildford, Surrey GU2 7XH, United Kingdom INTRODUCTION functionalities with alkali-metal cations involving ther- modynamic, structural and computational studies has been reported.30 Even though the receptor does not involve a tertiary amide it is worth mentioning this paper given that an integrated approach has been taken for complexation studies. This is encouraging in view of the fact that in most cases only qualitative data sets have been reported on calix[n]arene tertiary amide deriva- tives and metal cations. The main point of interest in this paper is to discuss the research so far carried out on p-tert-butyl calix[n]arene tertiary amide derivatives L1, L2 and L3 (Figure 1) and ionic species in solution and to provide results on the basis of recent investigations aiming to assess the factors contributing to the thermo- dynamics of ion complexation processes in moving from the cyclic tetramer to the hexamer. functionalities with alkali-metal cations involving ther- modynamic, structural and computational studies has been reported.30 Even though the receptor does not involve a tertiary amide it is worth mentioning this paper given that an integrated approach has been taken for complexation studies. This is encouraging in view of the fact that in most cases only qualitative data sets have been reported on calix[n]arene tertiary amide deriva- tives and metal cations. The main point of interest in this paper is to discuss the research so far carried out on p-tert-butyl calix[n]arene tertiary amide derivatives L1, L2 and L3 (Figure 1) and ionic species in solution and to provide results on the basis of recent investigations aiming to assess the factors contributing to the thermo- dynamics of ion complexation processes in moving from the cyclic tetramer to the hexamer. Acetonitrile, (HPLC grade, Fisher UK Scientific International) was placed over CaH2, refluxed under a nitrogen atmosphere and the middle fraction collected for use in the experiments.31 The water content deter- mined by Karl Fisher titration was found to be less than 0.01 %. Toluene, methanol, and dichloromethane (Fisher UK Scientific International) were purified as described in the literature.32,33 Solvents used for NMR measurements were deuterated acetonitrile, CD3CN, chloroform, CDCl3 and methanol, CD3OD Cambridge Isotope Laboratories, Inc. INTRODUCTION tions in the field of calixarene chemistry have been largely significant.5−9 Although emphasis has been made about the selectivity issue,10−12 few contributions are found in the literature reporting detailed thermodynam- ics on these systems in their interaction with neutral or ionic species. It is indeed from thermodynamics that a quantitative evaluation of the selective behaviour of a receptor for one species relative to another in a given solvent at a given temperature can be obtained from stability constant data for the systems involved. Unlike calix[n]arene esters13−26 and to some extent ca- lix[n]arene ketones,27−29 no detailed thermodynamics have been reported on calix[n]arene tertiary amide de- rivatives and ionic species (by detailed thermodynamics the authors refer to processes in which the speciation in solution, the solution thermodynamics for reactants and product and the composition of the complex are investi- gated and the scope and limitations of the methodology used are considered in the determination of stability constants and enthalpies of complexation as discussed below). Recently a detailed study on the interactions of a lower rim calix[4]arene containing secondary amide Supramolecular Chemistry and Nanoscience/Nano- technology have grown extensively in a relatively short period of time. These fields of research are closely re- lated. In fact in the middle eighties the use of molecules as building blocks to construct nanoscale devices and machines emerged within the framework of Supramolecular Chemistry. The latter involves intermo- lecular interactions which result in larger structures while the molecular approach to Nanotechnology is concerned with the building up of devices within the nanoscale from atoms and molecules. It has been recent- ly shown that calixarenes, products of the condensation reaction of p-substituted phenol and formaldehyde in alkaline medium and their derivatives are powerful tools for nano-technological developments.1 Based on metal ion complexes, self assembled systems have been pro- duced. Within this context it is relevant to emphasise the importance of thermodynamics in assessing the stability and selectivity of calixarenes for metal ions.2−4 The synthetic and structural advances as well as the applica- A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 2 Figure 1. Structure of Receptors. Figure 1. Structure of Receptors. several days before use. The absence of a water signal on the NMR spectra upon the addition of metal cation salts indicated the removal of water from these salts. EXPERIMENTAL SECTION p-tert-Butyl calix[4]arene (5 g), 18-crown-6 (0.53 g), potassium carbonate (10.66 g) and MeCN (250 ml) were mixed in a 500 ml three-necked-round bottom flask equipped with a condenser.34 The mixture was stirred for one hour under a nitrogen atmosphere. Then 2-chloro-N,N-diethylacetamide (6.35 ml) was added and the temperature was increased to 80 °C for ten hours. The reaction time was followed by TLC using a DCM:MeOH (9:1) mixture. The product was rotary- evaporated to extract the solvent. Then DCM (100 ml) was added to the crude product and washed several times with a saturated solution of NaHCO3 and hydro- chloric acid (100 ml, 0.2 molar). The organic phase was extracted using a separatory funnel, and anhydrous magnesium sulphate (50 g), was added to remove any water residue left in DCM. The mixture was then fil- 1H NMR Studies 1 1H NMR measurements were recorded at 298 K using a Bruker AC-300E pulse Fourier transform NMR spec- trometer. Typical operating conditions for routine pro- ton measurements involved a “pulse” or flip angle of 30º, spectral frequency (SF) of 300 MHz, delay time of 1.60 s, acquisition of the ligand (≈0.5 ml, 6.0 × 10−3 mol dm−3) in deuterated acetonitrile, CD3CN, were placed in 5 mm NMR tubes using TMS (tetramethylsilane) as the internal reference. 1H NMR (CDCl3): δ/ppm 6.79(s, 2H, Ar-H), 5.22 (d, 1H, Ar-CH2-AR), 5.047 (s, 2H, O-CH2-CO), 3.35 (q, 4H, N-CH2-CH3), 3.21 (d, 1H, Ar-CH2-AR), 1.156 (t, 6 H, CH2-CH3), 1.091 (s, 9H, 4-tert-but). Microanalysis was carried out at the University of Surrey, calcd. % C, 74.14; H, 9.15; N, 5.09; Found: % C, 74.02; H, 9.38; N, 5.00. Stepwise addition of the metal cation salt in CD3CN (≈1.0 × 10−2 mol dm−3) were made until no further chem- ical shift changes were observed. Similar experiments were carried out in deuterated methanol, CD3OD. Calorimetric Titrations Calorimetric titrations (direct and competitive) were carried out in order to determine the stability constant (log Ks) and the enthalpy of complexation (ΔcH). An isoperibol calorimeter (Tronac 450) was used for this purpose. Calibration of the equipment was performed in order to ensure a good reliability of the data. For this, tris(hydroxymethyl)amino methane (THAM) was titrat- ed into a solution of HCl (0.1 mol dm−3) at 298.15 K.39 The obtained value, −47.58 ± 0.08 kJ mol−1, was com- pared with the ones reported in the literature40(−47.47 kJ mol−1) and found to be in good agreement. 1H NMR (CDCl3) δ/ppm: 6.79 (s, 2 H), 5.22 (d, 1 H), 5.047 (s, 2 H), 3.35 (q, 4 H), 3.21 (d, 1 H), 1.156 (t, 6 H), 1.091 (s, 9 H). Microanalysis was carried out at the University of Surrey, calcd. % C, 74.14; H, 9.15; N, 5.09; Found: % C, 74.02; H, 9.38; N, 5.00). Conductance Measurements For these measurements, a Wayne Kerr B642 Autobalance Universal Bridge type B642 was used. p p p-tert-butyl calix[5]arene (1.00 g, 1.23 mmol), 18- crown-6 (0.1 g, 0.38 mmol), potassium carbonate (2.04 g, 14.79 mmol) and MeCN (200 ml) were mixed in a 500 ml three-necked-round bottom flask equipped with a condenser. The mixture was stirred for an hour under a nitrogen atmosphere. Then, 2-chloro-N,N-diethyla- cetamide (1.36 ml, 9.86 mmol) was added and the tem- perature was increased to 70 °C for two days. The reac- tion was followed by TLC using hexane: ethyl acetate (4:1) mixture as the developing solvent. The product was rota-evaporated to remove the solvent. DCM was added to the crude product and washed several times with a saturated solution of NaHCO3 and hydrochloric acid (100 ml, 0.2 molar). The organic phase was ex- tracted using a separatory funnel. Anhydrous magnesi- um sulphate (10 g) was added to remove any water residue left in DCM. The mixture was then gravitation- ally filtered and rotary-evaporated until dry. The crude oil was dissolved in methanol and refluxed. Small white crystals were formed upon cooling and after the slow evaporation of solvent. The yield obtained was 60−70 %. The product was dried at 80 °C under vacuum and stored in a desiccator containing calcium chloride. The conductivity cell constant was determined by the stepwise addition of an aqueous solution of KCl (0.1 mol dm−3) to the cell containing deionised water.37 The molar conductance for each addition was calculated using the Lind, Zwolenik and Fuoss equation.38 From the molar conductances of KCl, the cell constant was calculated. Conductometric titrations were carried out to de- termine the composition of the ligand/metal cation complex. Solutions of the ligand (≈0.7−1 × 10−3 mol dm−3) and metal cation salts (≈1 × 10−4 mol dm−3) were prepared in the corresponding solvent (MeCN and MeOH). The metal cation salt solution was added inside the cell and left to reach equilibrium (298.15 K). Thus the ligand solution was added stepwise into the metal salt solution and the molar conductance for each titra- tion recorded. The procedure was repeated until the molar conductance of the solution showed no significant change. Chemicals p-tert-Butylphenol (98 %), p-formaldehyde (95 %), potassium-tert-butoxide (95 %), purchased from Aldrich were dried over P4O10 under vacuum for several days before use. The p-tert-butyl-calix[4]arene, 18-crown-6, potassium carbonate (K2CO3) and 2-chloro-N,N- diethylacetamide were purchased from Aldrich and were used without further purification. Metal cation salts as perchlorates, LiClO4.H2O, NaClO4.H2O, KClO4.H2O, RbClO4.H2O, CsClO4.H2O, AgClO4.H2O, Mg(ClO4)2.2H2O, Ca(ClO4)2.4H2O, Sr(ClO4)2.2H2O, Ba(ClO4)2.nH2O, Ni(ClO4)2.6H2O, Pb(ClO4)2.nH2O, Cu(ClO4)2.6H2O, Zn(ClO4)2.6H2O, Cd(ClO4)2.H2O, Hg(ClO4)2.H2O and (C4H9)4NClO4 (Aldrich) were dried over P4O10 under vacuum for Croat. Chem. Acta 86 (2013) 1. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 3 tered gravitationally and rotary-evaporated until dry. The crude oil was dissolved in MeOH and refluxed. Small white crystals started to form upon cooling of the solvent. Further slow evaporation was required to give the resulting yield of 55−65 %. The product was dried at 80 °C under vacuum and stored in a desiccator contain- ing calcium chloride. out in order to verify the absence of non-volatile impuri- ties. Solvate formation was verified by exposing the solid to a saturated atmosphere of the appropriate sol- vent for several days. Synthesis of 5,11,17,23,29-penta-tert-butyl- 31,32,33,34,35-pentadiethylacetamide calix[5]arene (L2) The parent calix[5]arene was synthesised according to the procedure reported in the literature.35,36 Conductance Measurements For these measurements, a Wayne Kerr B642 Autobalance Universal Bridge type B642 was used. Potentiometric Titrations These were also performed to determine the stability constant of the ligands with Na+ and Ag+ cations using sodium and silver selective electrodes respectively as previously described.44 M (s) L(s) M L (s) s K n an a b a b     (1) (1) If this effect is observed,60,61 other processes be- sides complexation are taking place in solution. As a result, the data should be referred to as ‘apparent’ data where ions and ion pairs are present in solution. To fulfil the fundamental issue of electroneutrality, a neu- tral receptor complexing either a cation or an anion must have a counter-ion, both may be present in solu- tion as ions, ion-pairs or a mixture of both. This is de- pendent on the nature of the species involved but mainly on the permittivity of the medium and its solvating power. Such receptor cannot be regarded as an ion pair receptor. An ion pair receptor is that with the capability of complexing the cation and the anion by offering different active sites. A representative example for a calix[4]pyrrole amide derivative is shown in Figure 2 where the pyrrole NH proton interacts with the anion while the amide functionalities interact with the cation, ii) The behaviour of the ligand in solution. Hardly any attempts have been made to establish it. It is currently assumed that the monomer is predominant in solution iii) The composition of the complex needs to be estab- lished particularly when dealing with calixarene recep- tors where it is often found that the stoichiometry of the complex is altered by moving from one solvent to an- other. RESULTS AND DISCUSSION Thermodynamics of Complexation. Fundamental Concepts Solubility Measurements Saturated solutions of L1 and L2 were prepared by the addition of an excess amount of the ligand into the sol- vents. The saturated solutions were left to reach equilib- rium in a thermostatic bath at 298.15 K for several days. Aliquots of the solutions were taken and analysed grav- imetrically in triplicate. Blank experiments were carried For complexes with high stability constants (log Ks > 6), competitive calorimetric titrations were Croat. Chem. Acta 86 (2013) 1. 4 A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics Figure 2. Structure of Receptors with two binding sites. performed. A solution of the metal cation (M2 q+) salt was titrated into the vessel containing the ligand-metal ion (M1Ln+) complex. The following process takes places: 1 ( ) M Ln s  s(ov) 2 ( ) 2 (s) 1 (s) M M L M K q q n s       1 M n  s1 1 L M L K n     s2 2 ( ) ( ) 2 M L M L K q q s s        1 ( ) M Ln s  s(ov) 2 ( ) 2 (s) 1 (s) M M L M K q q n s       1 M n  s1 1 L M L K n     s2 2 ( ) ( ) 2 M L M L K q q s s        Stability constants and enthalpy of complexation were also determined by microcalorimetric titrations using a four channel heat conduction microcalorimeter (Ther- mal Activity Monitor 2277)41 Electrical (static and dy- namic) and chemical calibrations were carried out to check the reliability of the TAM 2277.42,43 The calori- metric titrations were performed by adding a metal cation salt solution into the vessel containing the ligand solution. Corrections for the enthalpy of dilution of the titrant into the solvent were carried out in all cases. The data were recorded and processed using Digitam 4.1 for Windows from Thermometric AB and select software AB Sweden. Figure 2. Structure of Receptors with two binding sites. requires the free and the complex cations to be pre- dominantly in their ionic form in solution. Quite clearly the counter-ion should not have any effect on the complexation process. Croat. Chem. Acta 86 (2013) 1. Thermodynamics of Complexation. Fundamental Concepts Prior to the discussion of the research related to calixarene tertiary amide derivatives and their comple- xation with metal cations some fundamental concepts regarding the thermodynamics of ion complexation processes involving neutral ligands are first discussed. Calix[n]arene Tertiary Amide Derivatives. Solution Studies Under the umbrella of thermodynamics, stability constants, enthalpies and entropies of complexation in methanol were reported.61 Given that these data are dependent on the counter-ion, the data reported are considered as ‘apparent’ values where besides complexation other processes are taking place as previously discussed.2 This is not surprising when dealing with 1:3 electrolytes which are much more likely to undergo ion-pair formation than 1:2 and 1:1 electrolytes even at low concentrations. As far as L3 is concerned an interesting paper was published by Meier and Detellier75 in which the complexation of this receptor with the caesium cation in mixed solvents was investigated. The results obtained show the formation of caesium complexes of different composition (Cs(I): L3 = 1:1, 2:1, 3:1). X ray diffraction studies provided in- formation regarding the active sites of complexation of this receptor with this cation for the 2:1 complex. Thermodynamics of Ion Complexation Processes: Methodology Selection of the Solvent S f S The solvent plays a crucial role in the derivation of thermodynamic data for ion complexation processes. Many papers regarding the medium effect on the complexation process involving macrocycleshave been published by Danil de Namor and co- workers.2,24−26,45−53 In selecting the solvent there are a number of issues which need to be carefully addressed such as i) The behaviour of electrolytes in non-aqueous solvents. There is a great deal of information in the literature addressing this issue.54−58 Conductance stud- ies have been carried out and successfully used to de- termine the ion-pair formation constants of electrolytes in different solvents.59 The complexation process in- volving a cation Mn+ and a receptor L in a solvent (s) to give the metal ion complex, MLn+ described in Eq. 1 Another important issue related to the solvent is its solvating power on the reactants and the products. As A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 5 derivatives (L1, L2 and L3) and their interaction with metal cations. far as processes involving 1:1 electrolytes are con- cerned, transfer Gibbs energies provide invaluable in- formation regarding the differences in solvation of these electrolytes from one solvent to another.62−66 Even more relevant are data for single-ion values.67,68 Although these are based on an extra-thermodynamic convention it allows selecting the appropriate solvent to carry out the experimental work, particularly if the solvation properties of the receptor and the new electrolytes (met- al-ion complex salt) are available from solubility meas- urements from which transfer Gibbs energies can be obtained. These data can be used to predict the relative strength of complexation in moving from one medium to another. Apart from the work reported by ourselves this approach has been hardly used in the general areas of calixarene and calixpyrrole chemistry. As previously stated2 a great deal of experimental work assessing the counter-ion effect on cation extraction processes by calixarenes and derivatives could have been avoided by the use of single-ion transfer values, particularly in sol- vent systems such as water – dichloromethane for which these data are readily available.69 Needless to mention comparative studies being made between complex stabil- ities and extraction data70,71 without taking into account the various processes involved in the latter (phase trans- fer and ion-pair processes) relative to the former.72 Calix[n]arene Tertiary Amide Derivatives. Solution Studies Calix[n]arene Tertiary Amide Derivatives. Solution Studies The synthesis and characterisation of a number of ca- lix[n]arene (n = 4,5,6,8) and their upper and lower rim derivatives including the ones containing amide func- tional groups in the narrow rim have been discussed in several books, reviews and papers.5−9,73 The thermody- namics of complexation of calix[n]arene derivatives and metal cations in non-aqueous solvents have been inves- tigated and reviewed in 1998.2 As far as calix[4]arene amide derivatives are concerned until then only quanti- tative data for p-tert-butyl-tetraacetamide, L1 and alka- li-metal cations, Ag (I) and Ba (II) in methanol and Rb (I) Cs (I) and Ag (I) in acetonitrile have been reported.74 From the information given for Ca(II) and Sr(II) in methanol (log Ks ≥ 9), Li(I), Na(I) and K(I) (log Ks ≥ 8.5) in acetonitrile at 298.15 K it can be concluded that high stability complexes are formed in these solvents. Since then, the only published data on receptor L1 and cations is that involving lanthanides. Under the umbrella of thermodynamics, stability constants, enthalpies and entropies of complexation in methanol were reported.61 Given that these data are dependent on the counter-ion, the data reported are considered as ‘apparent’ values where besides complexation other processes are taking place as previously discussed.2 This is not surprising when dealing with 1:3 electrolytes which are much more likely to undergo ion-pair formation than 1:2 and 1:1 electrolytes even at low concentrations. As far as L3 is concerned an interesting paper was published by Meier and Detellier75 in which the complexation of this receptor with the caesium cation in mixed solvents was investigated. The results obtained show the formation of caesium complexes of different composition (Cs(I): L3 = 1:1, 2:1, 3:1). X ray diffraction studies provided in- formation regarding the active sites of complexation of this receptor with this cation for the 2:1 complex. From the information given for Ca(II) and Sr(II) in methanol (log Ks ≥ 9), Li(I), Na(I) and K(I) (log Ks ≥ 8.5) in acetonitrile at 298.15 K it can be concluded that high stability complexes are formed in these solvents. Since then, the only published data on receptor L1 and cations is that involving lanthanides. Croat. Chem. Acta 86 (2013) 1. Thermodynamics of Ion Complexation Processes: Methodology Among the various techniques used to determine the complex stability; titration calorimetry offers ad- vantages in that the stability constant (hence the Gibbs energy) and the enthalpy of complexation can be deter- mined from a single titration. Therefore the entropy of complexation can be calculated from these data. Alt- hough this technique is concentration dependent and therefore has limitations when dealing with very strong complexes, these limitations can be overcome by carry- ing out competitive calorimetric titrations. The results can be checked by potentiometry which has been proved to be a suitable technique for the determination of high- ly stable complexes. This technique is based on the Nernst equation and therefore it is a function of the logarithmic scale of activity. Other techniques used are spectrometry, conductometry, NMR as well as calorimetry, all of them are concentration dependent and as such, their usefulness for obtaining accurate stability constant data depends on the magnitude of the stability constant.50 Awareness on the scope and limitations of the techniques used is required in the determination of the stability constants of ion-macrocycle complexes. Having highlighted some important issues regarding the determination of thermodynamic parameters of complexation involving neutral ligands and ionic spe- cies, the next section addresses calixarene tertiary amide Considering that the solvation of the receptor plays a role in the complexation process but nothing is known regarding the solution thermodynamics of L1, L2 and L3 in different solvents, these studies were un- dertaken. Thus Table 1 reports the solubility of these receptors in various solvents. For systems in which the composition of the solid in equilibrium with the saturat- ed solution is the same, the standard Gibbs energies of solution, ∆sGº in the various solvents at 298.15 K were calculated and the results are also included in Table 1. The role of solvation in complexation processes has been emphasized in several papers.2,4,76,77 Therefore given that the ∆sGº values involves the contribution of solvation and crystal lattice Gibbs energies, the latter is eliminated by the calculation of the transfer Gibbs ener- Croat. Chem. Acta 86 (2013) 1. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 6 Table 1. Solubilities and standard Gibbs energy of solution of L1 and L2 in different solvents at 298.15 K. Standard Transfer Gibbs energy from acetonitrile to other solvent Table 1. Thermodynamics of Ion Complexation Processes: Methodology In general terms, higher solubilities in the alcohols are found for the ca- lix[n]arene derivatives than for the parent compounds. This is mainly due to the strong hydrogen bond for- mation between the OH of the parent calix[n]arenes (intramolecular hydrogen bonding) while the calixarene amide derivatives may interact with the alcohols through hydrogen bond formation with the carbonyl or ethereal oxygens of the amide moiety (intermolecular hydrogen bonding). It is known that the amide group is more basic than the ketone or the ester. The basicity of the functional groups present in the pendant arms of the receptors is likely to have an impact on their solubility. A representative example of this statement is provided by solubility data and standard Gibbs energies of solu- tion of calix[4]arene ester, ketone and amide derivatives given in Table 2. Thus the amide derivative has the highest solubility while the one with the lowest basicity (ketone) shows the lowest solubility in methanol. 2 1 s 1 s 2 ( ) ( ) s s K s S K s  (2) (2) Another aspect to emphasise is that related with the number of substituted phenol units in the macrocycle. In moving from L1 to L3 the solubility decreases in MeOH. These findings are likely to be related with the increase in the crystal lattice energy as the number of substituted phenol units increase. As expected this is reflected in the melting points of these 1 2 1 2 1 2 1 2 ( ) ( ) [M ]( ) [L]( ) [ML ]( ) n c c t n t t G s G s G s s G s s G s s                 (3) (3) In fact this information leads to the assessment of the Croat. Chem. Acta 86 (2013) 1. Table 2. Solubilities and standard Gibbs energy of solution for calix[4]arene derivatives in methanol at 298.15 K Ligand 3 Solubility mol dm s 1 º kJ mol G   Calix[4]arene ketone2 4.62 × 10−4 19.04 Calix[4]arene ester21 3.65 × 10−3 13.91 Calix[4]arene amide (L1)(a) 6.20 × 10−3 12.80 (a) this work Table 2. Thermodynamics of Ion Complexation Processes: Methodology Solubilities and standard Gibbs energy of solution of L1 and L2 in different solvents at 298.15 K. Standard Transfer Gibbs energy from acetonitrile to other solvent Solvent(a) 3 Solubility mol dm s 1 º kJ mol G     t 1 º MeCN s kJ mol G    L1 L2 L1 L2 L1 L2 DMF (2.8 ± 0.1) × 10−2 ----(b) 9.0 ± 0.2 ---- −8.7 ---- PC (3.8 ± 0.1) × 10−3 2.6 × 10−3 13.8 ± 0.1 14.8 −3.5 1.1 n-Hex ---- 4.3 × 10−4 ---- 19.4 ---- 5.7 MeCN (8.1 ± 0.3 ) × 10−4 4.1 × 10−3 17.9 ± 0.3 13.7 0 0 Tol ---- 2.8 × 10−2 ---- 9.1 ---- −4.6 EtOH (4.1 ± 0.2) × 10−3 3.3 × 10−3 13.9 ± 0.1 16.6 −4 2.9 MeOH (6.2 ± 0.2) × 10−3 2.7 × 10−3 12.8 ± 0.1 14.8 −5 1.1 (a) Abbreviations for solvents: DMF, N,N-dimethylformamide; PC, propylene carbonate; n-Hex, Hexane; MeCN, acetonitrile; Tol, toluene; EtOH, ethanol; MeOH, Methanol. (b) Solvate formation. Table 1. Solubilities and standard Gibbs energy of solution of L1 and L2 in different solvents at 298.15 K. Standard Transfer Gibbs energy from acetonitrile to other solvent (a) Abbreviations for solvents: DMF, N,N-dimethylformamide; PC, propylene carbonate; n-Hex, Hexane; MeCN, acetonitrile; Tol, toluene; EtOH, ethanol; MeOH, Methanol. (b) Solvate formation. receptors (L1= 196 ºC, L2= 241 ºC, L3= 254 ºC). The same pattern observed in MeOH is found in EtOH, although the changes in solubility are less pronounced in the latter solvent. Like for the parent cyclic pentamer in acetonitrile, a higher solubility is found for the ca- lix[5]arene amide. Undoubtedly the different confor- mations adopted by these receptors in solution appear to play a role in the different solvation trends observed. However it should be emphasised that although the ∆tGº values reflect the difference in solvation of a receptor in one solvent relative to another, they do not provide any direct information regarding the sites of ligand-solvent interactions. Having stated it from the availability of ∆tGº values for the reactants (receptor and guest) and the product (complex) in two solvents, the ratio between the stability constants, Ks, in two solvents , s1 and s2 (Eq. 2) can be calculated from the thermodynamic cycle (Eq. 3) introduced by us in 1977.46 gies, ∆tGº using acetonitrile as the reference solvent. Data are shown in Table 1. Thermodynamics of Ion Complexation Processes: Methodology This is an important aspect to consider. From the information given in Table 1, solvents selected for these studies are now discussed. As far as L1 is concerned, relative to methanol, acetonitrile is the poorest solvator for this ligand. In addition this solvent is also a poor cation solvator as assessed from the sin- gle-ion values for the transfer of cations from water to acetonitrile (data based on the Ph4AsPh4B conven- tion).63 Therefore from the point of view of the reac- tants, acetonitrile offers the most suitable complexation medium. Regarding L2, n-Hex is a poor solvator for this ligand. However this solvent together with toluene are not suitable for complexation for two reasons; i) the poor solubility of metal cation salts in these media ii) in the absence of solubility limitations, their unsuitability still remains due to the tendency of metal cations to undergo strong ion pair formation in these media. Therefore either methanol or propylene carbonate are the solvents to select for complexation studies. Given that i) most literature data on calix[4]arenes are in ace- tonitrile and ii) it was considered of interest to investi- gate the complexation behaviour of these systems in a protic and a dipolar aprotic solvent, these studies were carried out in acetonitrile and methanol so comparative studies on the ability of these receptors to interact with metal cations can be made. 1H NMR data for L1 and L2 in three deuterated sol- vents, acetonitrile, CD3CN, methanol, CD3OD and chlo- roform, CDCl3 (reference solvent) were obtained and are listed in Table 3. In all cases, the ‘cone’ conformation is found. Thus the axial and equatorial hydrogens (H-3 and H-4) of the methylene bridge are non-equivalent and appear as a pair of doublets. However the confor- mational characteristic of the ‘cone’ in calixarenes can be assessed from the difference between the chemical shifts of the axial and the equatorial protons (Δδax-eq).7 As far as L1 is concerned these values are 2.01, 2.02 and 1.78 ppm in CDCl3, CD3CN and CD3OD re- spectively. However for a system in a perfect ‘cone’ conformation the Δδax-eq value is generally around 0.90 ppm. These shift differences indicate the presence of a distorted ‘cone’ conformation for L1 in these solvents. Thermodynamics of Ion Complexation Processes: Methodology Solubilities and standard Gibbs energy of solution for calix[4]arene derivatives in methanol at 298.15 K Ligand 3 Solubility mol dm s 1 º kJ mol G   Calix[4]arene ketone2 4.62 × 10−4 19.04 Calix[4]arene ester21 3.65 × 10−3 13.91 Calix[4]arene amide (L1)(a) 6.20 × 10−3 12.80 (a) thi k Table 2. Solubilities and standard Gibbs energy of solution for calix[4]arene derivatives in methanol at 298.15 K and standard Gibbs energy of solution for calix[4]arene derivatives in methanol at 298.15 K Croat. Chem. Acta 86 (2013) 1. Croat. Chem. Acta 86 (2013) 1. 7 A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics Table 3. 1H NMR data for L1 and L2 in various solvents at 298 K CDCl3 (a) CD3CN(a) CD3OD(a) L1 L2 L1 L2 L1 L2 H-1 1.09 1.02 1.19 1.14 1.09 1.03 H-2 6.79 6.90 7.13 7.14 6.82 6.95 H-3 5.22 4.99 5.26 5.05 4.96 4.85 H-4 3.21 3.32 3.24 3.27 3.18 3.31 H-5 5.05 4.75 5.01 4.85 4.96 4.73 H-6 3.35 3.40 3.38 3.38 3.38 3.42 H-7 1.16 1.11 1.1 1.18 1.18 1.14 (a) Deuterated solvents. H O O N 5 H H O O N 4 H H-1 H-2 H-4 H-3 H-5 H-6 H-6' H-7' H-7 H-1 H-2 H-4 H-3 H-5 H-6 H-6' H-7' H-7 1 2 Table 3. 1H NMR data for L1 and L2 in various solvents at 298 K Table 3. H NMR data for L1 and L2 in various solvents at 298 K CDCl3 (a) CD3CN(a) CD3OD(a) L1 L2 L1 L2 L1 L2 H-1 1.09 1.02 1.19 1.14 1.09 1.03 H-2 6.79 6.90 7.13 7.14 6.82 6.95 H-3 5.22 4.99 5.26 5.05 4.96 4.85 H-4 3.21 3.32 3.24 3.27 3.18 3.31 H-5 5.05 4.75 5.01 4.85 4.96 4.73 H-6 3.35 3.40 3.38 3.38 3.38 3.42 H-7 1.16 1.11 1.1 1.18 1.18 1.14 (a) Deuterated solvents. H O O N 5 H H O O N 4 H H-1 H-2 H-4 H-3 H-5 H-6 H-6' H-7' H-7 H-1 H-2 H-4 H-3 H-5 H-6 H-6' H-7' H-7 1 2 H O O N 5 H H-1 H-2 H-4 H-3 H-5 H-6 H-6' H-7' H-7 2 H O O N 4 H H-1 H-2 H-4 H-3 H-5 H-6 H-6' H-7' H-7 1 1 2 (a) Deuterated solvents. 1H NMR of L1 and L2 in Various Solvents - The Medium Effect medium effect on the complexation process. Thermodynamics of Ion Complexation Processes: Methodology Such distortion can be attributed to new effects on the methylene protons leading to a greater shielding of the equatorial protons (signal moves upfield) and a less shielding of the axial (signal moves downfield). The net result is an increase in the Δδax-eq. This distortion seems to be more pronounced in CD3CN and CDCl3 than in CD3OD. Again the distortion of the ‘cone’ is more pro- nounced in L1 than in L2. For the latter, Δδax-eq values are 1.67, 1.78 and 1.54 pm in CDCl3, CD3CN and Croat. Chem. Acta 86 (2013) 1. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 8 Table 4. 1H NMR chemical shift changes of L1 and L2 upon complexation with monovalent metal cations in CD3CN and CD3OD at 298 K Table 4. 1H NMR chemical shift changes of L1 and L2 upon complexation with monovalent metal cations in CD3CN and CD3OD at 298 K Metal cation L1 (CD3CN) Δδ/ppm L2 (CD3CN) Δδ/ppm H-2 H-3 H-4 H-5 H-6’ H-2 H-3 H-4 H-5 H-6’ Li+ 0.21 −0.66 0.17 −0.28 −0.17 0.18 −0.85 0.15 −0.28 −0.20 Na+ 0.23 −0.77 0.14 −0.45 −0.24 0.26 −0.78 0.17 −0.34 −0.23 K+ 0.25 −0.63 0.17 −0.38 −0.18 −0.06 −0.37 0.09 −0.16 −0.14 Rb+ 0.23 −0.63 0.14 −0.29 −0.18 −0.07 −0.41 0.11 −0.18 −0.13 Cs+ 0.24 −0.79 0.17 −0.42 −0.20 −0.02 −0.47 0.11 −0.21 −0.13 Ag+ 0.49 −0.34 0.28 −0.36 −0.18 0.20 −0.82 0.26 −0.10 −0.19 L1 (CD3OD) Δδ/ppm L2 (CD3OD) Δδ/ppm H-2 H-3 H-4 H-5 H-6’ H-2 H-3 H-4 H-5 H-6’ Li+ 0.41 −0.24 0.23 −0.13 −0.17 0.32 −0.44 overlap −0.02 −0.11 Na+ 0.48 −0.40 0.24 −0.34 −0.21 0.38 −0.51 overlap 0.08 −0.13 K+ 0.49 −0.27 0.23 −0.28 −0.21 0.12 −0.60 overlap 0.04 −0.03 Rb+ 0.37 −0.18 0.15 −0.17 −0.19 0.12 −0.10 overlap 0.03 −0.05 Cs+ 0.03 −0.02 0.00 −0.01 −0.06 0.12 −0.21 overlap 0.01 −0.04 Ag+ 0.49 −0.34 0.28 −0.36 −0.18 --- --- --- --- --- Table 4. 1H NMR chemical shift changes of L1 and L2 upon complexation with monovalent metal cations in CD3CN and CD3OD at 298 K CD3OD respectively. However, the most remarkable feature of the data is the downfield chemical shift changes observed for the aromatic protons (H-2) (0.34 ppm for L1 and 0.24 ppm for L2) in CD3CN relative to CDCl3. Thermodynamics of Ion Complexation Processes: Methodology A similar behaviour has been previously found for other calix[4]arene derivatives.2,14,15,22 This has been attributed to an interaction of the macrocycle through its hydrophobic cavity with the solvent. Given that this cavity is larger for the cyclic pentamer than for the te- tramer, this interaction is expected to be greater for the latter relative to the former. As previously shown the presence of acetonitrile in the hydrophobic cavity of the ligand22 has implications on the cation complexing ability of these macrocycles in this solvent relative to others. ylene bridge (H-5) and the terminal protons (H-7). Thus strong shielding effects were observed for protons H-3, H-5 and H-7 possibly due to the inclusion of the metal cation inside the hydrophilic cavity forcing the pendant arms to move towards the inner region. However strong deshielding effects were observed in the aromatic and equatorial protons (H-2 and H-4 respectively) and this is attributed to the interaction of the metal cation with the lower rim (functionalised sites) and possibly a higher penetration of the solvent into the hydrophobic cavity of the complex ligand. Given that the conformational char- acteristics of the receptor can be assessed from Δδax-eq values, these are now considered. The degree of flattening of the ‘cone’ reflected in the Δδax-eq value of the metal cation complex, shows a decrease in the distortion of the cone relative to that of the free ligand. Thus the Δδax-eq value (1.11 ppm) for the Na (I) complex is closer to that of a calix[4]arene in a perfect ‘cone’ conformation (Δδax-eq = 0.90 ppm). As for L1 upon complexation with Li (I), K(I) , Rb (I), Cs(I) and Ag (I) metal cations, Δδax-eq values (1.20, 1.22, 1.24, 1.51 and 1.40 ppm respectively) show a more distorted conformation than that for the Na (I) cation complex but lesser distortion as compared to that for the free ligand. As far as L2 is concerned, different patterns were observed for Li (I), Na (I) and Ag (I) relative to the other monovalent metal cations in CD3CN. This is observed in the deshielding of the aromatic protons (H- 2) while the rest of the alkali-metal cations (K (I), Rb (I) and Cs (I)) appear to induce a slight shielding effect for Croat. Chem. Acta 86 (2013) 1. Interaction of L1 and L2 with Metal Cations 1H NMR studies of L1 and L2 with cations in CD3CN and CD3OD This section will be discussed under the following headings 1H NMR studies of L1 and L2 with cations in CD3CN and CD3OD This section will be discussed under the following headings 1H NMR studies of L1 and L2 with cations in CD3CN and CD3OD 1H NMR studies of L1 and L2 with cations in CD3CN and CD3OD This section will be discussed under the following headings i) 1H NMR studies of L1 and L2 with univalent cat- ions in CD3CN and CD3OD at 298 K (Table 4) ii) 1H NMR studies of L1 and L2 with bivalent cat- ions in CD3CN and CD3OD at 298 K (Table 5) i) Chemical shift changes (Δδ, ppm) of L1 with mono- valent metal cations observed in CD3CN relative to the free ligand show significant changes in the aromatic (H- 2), the axial and the equatorial (H-3 and H-4), the meth- Croat. Chem. Acta 86 (2013) 1. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 9 Table 5. 1H NMR chemical shift changes of L1 and L2 upon complexation with monovalent metal cations in CD3CN and CD3OD at 298 K Table 5. 1H NMR chemical shift changes of L1 and L2 upon complexation with monovalent metal cations in CD3CN and CD3OD at 298 K Table 5. Interaction of L1 and L2 with Metal Cations 1H NMR chemical shift changes of L1 and L2 upon complexation with monovalent metal cations in CD3CN and CD3OD at 298 K Δδ/ppm 1H Mg2+ complex Ca2+ Sr2+ Ba2+ Pb2+ Zn2+ Cd2+ Hg2+ H-1 0.04 0.05 0.05 0.08 0.07 0.06 0.06 0.07 H-2 0.24 0.35 0.36 0.36 0.37 0.29 0.29 0.33 H-3eq −1.23 −1.11 −1.05 −1.05 −1.01 −1.19 −1.19 −1.11 H-4ax overlap 0.34 −0.37 −0.29 0.38 0.26 0.26 0.33 H-5 overlap −0.22 −0.24 −0.26 −0.12 −0.36 −0.36 overlap H-6 overlap overlap 0.11 0.18 0.14 0.15 0.15 overlap H-6’ −0.13 overlap −0.17 −0.18 −0.12 −0.11 −0.11 −0.12 H-7 overlap overlap --- overlap overlap overlap overlap overlap H-7’ 0.10 0.08 0.09 0.07 0.10 0.11 0.09 0.10 Δδax-eq --- 0.77 0.68 0.76 0.63 0.93 0.93 0.78 Δδ/ppm 1H Mg2+ Ca2+ Sr2+ Ba2+ Pb2+ Zn2+ Cd2+ Hg2+ H-1 −0.08 0.02 0.04 0.08 0.01 −0.02 0.01 0.01 H-2 −0.04 0.18 0.24 0.40 0.20 −0.04 0.17 0.20 H-3eq 0.20 0.22 0.21 0.16 0.26 0.18 0.17 0.26 H-4ax −0.77 −0.65 −0.65 −0.85 −0.82 −0.82 −0.68 −0.82 H-5 −0.08 −0.09 −0.09 −0.29 −0.10 −0.12 −0.12 −0.10 H-6 0.05 0.05 0.04 0.04 0.07 0.05 0.05 0.07 H-6’ −0.12 −0.14 −0.16 −0.18 −0.17 −0.10 −0.11 −0.15 H-7 overlap overlap --- overlap overlap overlap overlap overlap H-7’ 0.01 0.01 0.01 0.28 0.08 0.07 0.01 0.08 Δδax-eq 0.80 0.90 0.91 0.76 0.70 0.81 0.93 0.70 O N O 5 1 2 3 4 5 6 7 6' 7' O N O 5 1 2 3 4 5 6 7 6' 7' these protons. The values for the chemical shift changes for the axial and equatorial protons of Li (I), Na (I) and Ag (I) (Δδax-eq = 0.78, 0.83, and 0.70 ppm) show that L2 adopts a slightly flattened conformation close to a per- fect ‘cone’. However with larger metal cations (K (I), Rb (I) and Cs (I)), L2 tends to adopt a more distorted conformation (Δδax-eq = 1.32, 1.26 and 1.20 ppm). al protons are overlapped and therefore the Δδax-eq can- not be assessed. ii) Chemical shift changes of L1 and L2 upon complexation with bivalent cations in CD3CN and CD3OD at 298 K listed in Table 5 show that in the case of L1 and bivalent cations in CD3CN, chemical shift changes are more pronounced than those observed for monovalent ions in this solvent, particularly for protons 2 and 7’. Croat. Chem. Acta 86 (2013) 1. Interaction of L1 and L2 with Metal Cations As far as Δδax-eq values are concerned, these are lower than 0.90 ppm suggesting that these complex- es adopt a flattened ’cone’ conformation in solution. A similar pattern is observed in CD3OD (Table 6) where these values vary in the 0.50−0.80 ppm range. However The Ba (II) and Sr (II) complexes adopt a perfect ‘cone’ conformation in this solvent (Δδax-eq = 0.90 ppm). This conformation is also observed for L2 and Ca (II), Sr (II) and Cd (II) in CD3CN. In addition, in this solvent a lineal relationship is shown between the chemical shift change of H-6’ and the ionic radius of the bivalent cati- on (Figure 3) suggesting a size effect on the chemical shift of this proton in acetonitrile. Unfortunately it was ( q pp ) As far as the complexation process involving L1 and monovalent cations in CD3OD is concerned, the Δδax-eq values decrease in moving from the free ligand (1.78 ppm) to the cation complexes. Thus values of 1.31, 1.14, 1.28, 1.45 and 1.16 ppm are found for Li (I), Na (I), K (I), Rb (I) and Ag (I) respectively. The axial protons move upfield while the equatorial ones down- field. As a result, the Δδax-eq values decrease relative to the free ligand and therefore the distortion of the ‘cone’ is reduced upon complexation. Similar 1H NMR inves- tigations were carried out with the same cations and L2 in CD3OD. These data are also shown in Table 4. Strong shielding effects of the axial protons are observed in the presence of the monovalent metal cations. The equatori- Croat. Chem. Acta 86 (2013) 1. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 10 Table 6. Interaction of L1 and L2 with Metal Cations 1H NMR chemical shift changes of L1 and L2 upon complexation with bivalent metal cations in CD3OD at 298 K Δδ/ppm 1H Mg2+ complex Ca2+ Sr2+ Ba2+ Pb2+ Zn2+ Cd2+ Hg2+ H-1 0.13 0.13 0.12 0.12 0.13 0.11 0.13 0.14 H-2 0.01 0.61 0.58 0.58 0.62 0.52 0.59 0.61 H-3eq Overlap −0.73 −0.58 −0.58 −0.64 −0.73 −0.78 −0.72 H-4ax Overlap 0.47 0.40 0.40 0.49 Overlap 0.45 0.47 H-5 Overlap Overlap Overlap −0.13 0.04 Overlap Overlap Overlap H-6 Overlap 0.18 0.23 0.23 0.21 0.20 −0.07 −0.05 H-6’ Overlap Overlap −0.16 −0.17 0.15 0.14 −0.13 −0.11 H-7 Overlap Overlap overlap Overlap Overlap 0.12 0.13 0.16 H-7’ Overlap Overlap 0.03 0.03 overlap Overlap Overlap Overlap Δδ/ppm 1H Mg2+ complex Ca2+ Sr2+ Ba2+ Pb2+ Zn2+ Cd2+ Hg2+ H-1 0.08 0.08 0.11 0.19 0.13 0.07 0.10 0.07 H-2 0.25 0.25 0.36 0.58 0.34 0.23 0.30 0.31 H-3eq Overlap Overlap Overlap Overlap Overlap 0.24 Overlap Overlap H-4ax −0.26 −0.26 −0.36 −0.52 −0.34 Overlap −0.42 −0.38 H-5 0.26 0.26 0.15 0.30 0.29 Overlap 0.19 0.23 H-6 −0.05 −0.05 −0.07 −0.09 −0.03 −0.06 −0.05 −0.06 H-6’ 0.12 0.12 0.16 0.11 0.15 0.08 0.05 0.12 H-7 0.07 0.07 0.05 0.05 0.09 0.08 0.07 0.08 O N O 5 1 2 3 4 5 6 7 6' 7' 4 O N O 5 1 2 3 4 5 6 7 6' 7' 1H NMR chemical shift changes of L1 and L2 upon complexation with bivalent metal cations in CD3OD at 298 K Table 6. 1H NMR chemical shift changes of L1 and L2 upon complexation with bivalent metal cations in CD3 not possible to calculate Δδax-eq values for these cations in CD3OD due to the interferences of the residual peaks for the solvent. these cations and L1 in the same solvent. The next sec- tion discusses the conductance behaviour of uni- and bivalent salts in acetonitrile and methanol upon addition of calix[n]arene amide derivatives in these solvents. In summary from these NMR investigations in- volving uni- and bivalent cations it is relevant to em- phasise the significant conformational differences ob- served in the Δδax-eq values for Li (I) and Na (I) upon complexation with L2 in CD3CN relative to i) other univalent cations in this solvent and ii) the values for Croat. Chem. Acta 86 (2013) 1. Conductometric Titrations In most cases, Λm values de- crease as the ligand is added to the metal-ion salt due to the reduction in the mobility of the complex relative to the free cation salt. In moving from acetonitrile to meth- anol the composition of the complexes remains the same but the stability decreases for Li (I) and Rb (I) while increases for Ag (I) showing the effect of cation solva- tion on the strength of complexation as previously dis- cussed.2,4 changes in conductance. The formation of 1:2 complex- es between receptor L2 and Li (I) and Na (I) in MeCN explain the differences found in the conformational changes reflected in the Δδax-eq values for these systems relative to i) other cations and this receptor in this sol- vent and ii) the same cations and L2 in MeOH. For other univalent cations (K (I), Rb (I), Cs (I), and Ag (I)) well defined break points were found revealing that 1:1 complexes of moderate stability are formed in acetoni- trile. In MeOH only 1:1 complexes were found for uni- valent cations and L2. This is a typical example in which the medium effect plays a key role in the complexation process. Indeed acetonitrile is a poor cation solvator while methanol is a strong solvator63 to the extent that the hosting capacity of the receptor is enhanced as to host two of the smallest cations per unit of receptor in acetonitrile while L2 interacts with only one cation in methanol. The solvation of the functional groups in the pendant arms of the receptor capable of interacting with a protic solvent such as MeOH is likely to contribute to the lower hosting ability of this receptor for these cations in MeOH relative to MeCN. As far as bivalent cations and L2 in MeCN and MeOH are con- cerned, complexes of 1:1 stoichiometry were found with all alkaline-earth metals and heavy metals tested (Hg (II), Pb (II), Cd (II)) and Zn (II)) although weaker com- plexes appear to be formed in MeOH relative to MeCN. Conductometric titration curves for L1 and biva- lent metal cations in MeCN showed the formation of 1:1 complexes with alkaline-earth , heavy metals (Pb(II), Cd(II), Hg(II), Cu (II)) as well as with Zn(II) and Ni(II) in both solvents. Conductometric Titrations Conductometric titrations for metal cation salts in ace- tonitrile and methanol at 298.15 K were carried out by titrating the corresponding ligand (L1, L2 or L3) onto the metal-ion salt. The molar conductance, Λm, for each addition was calculated and plotted against the ligand / metal cation (L / Mn+) concentration ratio (molar scale). Figure 4 shows representative data for L1 and the sodi- um cation in acetonitrile and methanol respectively. In all cases the results show that: Figure 3. Chemical shift changes (ppm) of H-6’ for L2 against of the ionic radii, (Å) of bivalent metal cations in CD3CN at 298 K. i) As expected the Λm value for the free metal cation salt (L/Mn+ = 0) would be slightly lower than the limiting molar conductance, Λºm, of the corresponding salt as reported in the literature.78 ii) When ligand-metal cation interaction occurs, com- plexes of 1:1 stoichiometry are formed between L1 and the relevant cations in these solvents. Figure 3. Chemical shift changes (ppm) of H-6’ for L2 against of the ionic radii, (Å) of bivalent metal cations in CD3CN at 298 K. Croat. Chem. Acta 86 (2013) 1. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 11 Figure 4. Conductometric curve for the titration of sodium (as perchlorate) and L1 in acetonitrile and methanol at 298.15 K. Figure 5. Conductometric curve for the titration of sodium (as perchlorate) and L2 in acetonitrile and methanol at 298.15 K. Figure 5. Conductometric curve for the titration of sodium (as perchlorate) and L2 in acetonitrile and methanol at 298.15 K. Figure 4. Conductometric curve for the titration of sodium (as perchlorate) and L1 in acetonitrile and methanol at 298.15 K. Figure 5. Conductometric curve for the titration of sodium (as perchlorate) and L2 in acetonitrile and methanol at 298.15 K. iii) From the intensity of the curvature of the conductometric titration curve it is observed that in acetonitrile relatively stable complexes are formed be- tween L1 with Na (I). Moderate changes in the curva- ture were found for K (I). The change of intensity in the curvature for the conductometric titrations of L1 and Li (I) and Rb (I), showed the formation of weak complex- es. A 1:1 complex was also found between this receptor and Ag (I) in acetonitrile. Conductometric Titrations As far as L2 is concerned, Figure 5 shows repre- sentative titration curves for this ligand with the sodium cation in acetonitrile and methanol at 298.15 K. Com- plexes of moderate stability and 1:1 (ligand:metal cati- on) stoichiometry are found between L2 and alkali- metal and silver cations in methanol. In the case of acetonitrile, complexes of 1:2 (ligand:metal cation) composition are found when this ligand interacts with lithium and sodium in this solvent. The titration curves for these cations and L2 in MeCN showed a gradual decrease in conductance with a change in the curvature at the L:Mn+ molar ratio of 0.5 indicating the formation of 1:2 complexes. Further addition of the ligand led to a further decrease in conductance and a clear break at the ligand:metal cation ratio of 1 was found reflecting that strong 1:1 complexes with these cations are formed. Completion of the reaction was shown by the small The solubility of L3 in acetonitrile (Table 1) was found to be very low as to proceed with conductance studies in this solvent. Therefore conductance studies were performed in MeOH. No changes in conductance were observed for alkali-metal cations and this receptor in the alcohols. Only a slight break at the 1:1 ligand: metal cation ratio was found for Ag (I) and L3 in MeOH indicating the formation of a very weak complex in this solvent. Among bivalent metal cations, 1:1 com- plexes of moderate stability are found for Ca (II), Sr (II), Ba (II), Cu (II), Cd (II), Hg (II) and Pb (II). How- ever a 1:2 L3:Mg (II) complex is formed in MeOH as Croat. Chem. Acta 86 (2013) 1. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 12 Figure 6. Conductometric curve for the titration of magnesi- um (as perchlorate) and L3 in methanol at 298.15 K. two ligands is much greater than that involving one ligand. Therefore the mobility of the former (Eq. 6) is expected to be lower than that for the latter (Eq. 5). Consequently the expected decrease in conductance for the process involving two receptors is not corrobo- rated with the increase in conductance observed from B to C. No changes in conductance are observed from C to D. Thermodynamics of Complexation of L1, L2 and L3 with Metal Cations in Acetonitrile and Methanol shown in the conductometric titration curve (Figure 6). Thus two breaks are observed. The first one occurs at a L3:Mg (II) ratio of 0.50 when an excess of the salt rela- tive to the receptor is present in solution. Under these experimental conditions two metal cations are taken up by unit of receptor and a decrease in conductance is observed. Further addition of the receptor may result in the transport of one cation from the complex to the receptor. Consequently the conductance increases from B to C until the ligand: metal cation ratio reaches a value of 1 indicating the possible formation of a 1:1 [MgL3]2+ complex. We have no experimental evidence as to suggest the mode by which L3 can hold two cati- ons. Quite clearly cations must be separated from each other to avoid electrostatic repulsion. It is known that several conformations can be found in solution for the amide derivative of the cyclic hexamer and the possibil- ity of hosting two caesium cations per unit of receptor has been demonstrated earlier by Meier and Detellier75 in solution and in the solid state. shown in the conductometric titration curve (Figure 6). Thus two breaks are observed. The first one occurs at a L3:Mg (II) ratio of 0.50 when an excess of the salt rela- tive to the receptor is present in solution. Under these experimental conditions two metal cations are taken up by unit of receptor and a decrease in conductance is observed. Further addition of the receptor may result in the transport of one cation from the complex to the receptor. Consequently the conductance increases from B to C until the ligand: metal cation ratio reaches a value of 1 indicating the possible formation of a 1:1 [MgL3]2+ complex. We have no experimental evidence as to suggest the mode by which L3 can hold two cati- ons. Quite clearly cations must be separated from each other to avoid electrostatic repulsion. It is known that several conformations can be found in solution for the amide derivative of the cyclic hexamer and the possibil- ity of hosting two caesium cations per unit of receptor has been demonstrated earlier by Meier and Detellier75 in solution and in the solid state. Conductometric Titrations Having (i) determined the composition of metal- ion complexes involving L1 and L2 in acetonitrile and methanol, and (ii) previous information regarding the concentration range at which these salts are predomi- nantly in their ionic forms in these solvents, we pro- ceeded with the thermodynamic characterisation of these systems and these are now discussed. Figure 6. Conductometric curve for the titration of magnesi- um (as perchlorate) and L3 in methanol at 298.15 K. Croat. Chem. Acta 86 (2013) 1. Thermodynamics of Complexation of L1, L2 and L3 with Metal Cations in Acetonitrile and Methanol Although Tables 7 and 8 report thermodynamic data for calix[n]arenes (n=4,5,6) with uni- and bivalent metal cations in acetonitrile and methanol at 298.15 K the discussion will be carried out under the following head- ings: i. Thermodynamics of complexation of L1, L2 and L3 with univalent metal cations in acetoni- trile and methanol at 298.15 K. ii. Thermodynamics of complexation of L1, L2 and L3 with bivalent metal cations in acetoni- trile and methanol at 298.15 K. i) The complexation of L1 with univalent cations in MeCN shows the formation of very strong complexes in this solvent. Their high stability is enthalpically controlled and entropy destabilised except for Li (I) and Ag (I) which are entropically favoured. Data for L1 and these cations in MeOH show large differences with literature values for Na (I) and K (I) in this sol- vent but good agreement between the two methods used by us (competitive titration calorimetry and potentiometry). Like in MeCN among the alkali-metal cations this receptor is selective for Na (I). However if the stability of the Ag (I) complex in MeOH is consid- ered, this receptor is unable to distinguish between Ag (I) and Na (I) as reflected by the standard deviation of the data. No interaction between L1 and Cs (I) was found in MeOH. In most cases the process is enthalpy controlled with unfavourable entropy except for Li (I) (enthalpy and entropy favoured). The medium effect on the complexation of L1 with univalent cations can be assessed from the ratio between the stability con- stant for this receptor and a given cation in one solvent relative to another (Eq. 7). The processes involved are shown in Eqs. 4 and 5 2 4 2 2Mg (MeOH) L3(MeOH) [Mg L3] (MeOH)     (4) 4 2 2 g L3] (MeOH) L3(MeOH) 2[MgL3] (MeOH)     ( Having stated it, emphasis must be made about the fact that the break at the ligand:cation ratio of 1 does not necessarily imply the formation of a 1:1 complex. In- deed if the number of ligand units interacting equals that of the cation, a ligand:cation ratio of 1 will be obtained. The possibility of the process shown in Eq. 6 cannot be excluded 4 4 2 2 2 [Mg L3] (MeOH) L3(MeOH) [Mg L3 ] (MeOH)     (6) 1 1 s MeCN MeCN s (M ) (M ) n s n s K S K    (7) 1 1 s MeCN MeCN s (M ) (M ) n s n s K S K    (7) However this process (Eq. 6) seems unlikely to occur given that the size of the metal ion complex containing Croat. Chem. Acta 86 (2013) 1. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 13 Table 7. Thermodynamic Parameters of Complexation of L1 and L2 with monovalent metal cations in acetonitrile and methanol at 298.15 K MeCN log Ks c 1 º kJ mol G   c 1 º kJ mol H   c 1 1 º J K mol S    Cation L1 L2 L1 L2 L1 L2 L1 L2 Li+ 9.6(b) 4.8 (1:1)(a) −54.9 −27.4 −51.7(b) −76.4(a) 11 −164 2.5 (1:2)(a) −14.3 −63.3(a) −164 Na+ 9.8(b) 5.6 (1:1)(a) −55.9 −32.0 −67.6(b) −39.2(a) −39 −24 3.0 (1:2)(a) −17.1 −57.2(a) −134 K+ 7.2(b) 5.9(a) −41.1 −33.7 −52.8(b) −37.8(a) −39 −14 Rb+ 5.53(a) 5.8 (a) −31.57 −33.1 −35.7(a) −29.8(a) −14 11 Cs+ - 5.6(a) - −32.0 - −21.1(a) - 37 Ag+ 6.10c 4.5(a) −34.82 −25.7 −28.7(a) - 21(a) - MeOH Cation L1 L2 L1 L2 L1 L2 L1 L2 Li+ 4.1(a) 5.2(a) −23.4 −29.7 −9.0(a) −17.0(a) 48 43 Na+ 6.2(b) 5.1(a) −35.4 −29.1 −46.5(b) −31.0(a) −37 −6 K+ 4.73(a) 5.4(a) −27.0 −30.8 −33.8(a) −57.9(a) −23 −91 Rb+ 3.50(a) 5.8(a) −20.0 −33.1 −22.4(a) −69.3(a) −8 −121 Cs+ - 5.5(a) - −31.4 - −55.7(a) - −82 Ag+ 6.60(c) 5.9(a) −37.7 −33.7 −47(a),(b) −41.5(a) −31 −26 (a) Direct calorimetry. Standard deviations for log Ks values given with one decimal place are in the range ± 0.1−0.2. 2 4 2 2Mg (MeOH) L3(MeOH) [Mg L3] (MeOH)     (4) For log Ks values given with two decimal places, the standard deviations are ± 0.03−0.06. For ΔcHº values are in the 0.1−0.5 kJ mol−1 range. (b) Competitive titration calorimetry. Standard deviation in log Ks values is in the range ± 0.2−0.3. Standard deviation in the ΔcHº values are in the ±. 0.5−0.7 kJ mol−1 range. (c) Potentiometry. Table 7. Thermodynamic Parameters of Complexation of L1 and L2 with monovalent metal cations in acetonitrile and methanol at 298.15 K rmodynamic Parameters of Complexation of L1 and L2 with monovalent metal cations in acetonitrile and methanol (a) Direct calorimetry. Standard deviations for log Ks values given with one decimal place are in the range ± 0.1−0.2. For log Ks values given with two decimal places, the standard deviations are ± 0.03−0.06. For ΔcHº values are in the 0.1−0.5 kJ mol−1 range. (b) Competitive titration calorimetry. Standard deviation in log Ks values is in the range ± 0.2−0.3. Standard deviation in the ΔcHº values are in the ±. 0.5−0.7 kJ mol−1 range. (c) Potentiometry. Data presented in Table 9 shows that the selectivity of L1 towards alkali metal cations in methanol relative to acetonitrile increases as the size of the metal cation increases. Data for the transfer Gibbs energy of metal cations from MeCN to MeOH show the level of solva- tion of these systems in one solvent relative to the oth- er.63 Data presented in Table 9 shows that the selectivity of L1 towards alkali metal cations in methanol relative to acetonitrile increases as the size of the metal cation increases. Data for the transfer Gibbs energy of metal cations from MeCN to MeOH show the level of solva- tion of these systems in one solvent relative to the oth- er.63 of these cations in the same solvent increases. This find- ing is concomitant with previous statements by Danil de Namor and co-workers47 in that the most suitable complexation medium is that which is a poor solvator for the reactants and a good solvator for the product. 2 4 2 2Mg (MeOH) L3(MeOH) [Mg L3] (MeOH)     (4) This lineal relationship does not necessarily imply that the solvation of the receptor and the metal-ion complex do A plot of the transfer Gibbs energy, MeOH MeCN tG   (da- ta based on the Ph4AsPh4B convention), of alkali metal cations and silver cation from acetonitrile to methanol against the log ( MeOH MeCN S ) (Figure 7) shows a linear regres- sion of the type: Figure 7. A plot of the transfer Gibbs energy of monovalent metal cations from MeCN to MeOH at 298.15 K (data based in the Ph4AsPh4B convention) against the selectivity of L1 (logarithm scale), for these metal cations in MeOH relative to MeCN at 298.15 K. MeOH MeCN (M )(MeCN MeOH) log n tG a b S       (8) MeOH MeCN (M )(MeCN MeOH) log n tG a b S       (8) The positive values for the transfer Gibbs energies of Ag (I), Rb (I) and K (I) cations from acetonitrile to methanol ( MeOH t MeCN G   ) show that these cations are more solvated in acetonitrile.63 The negative values of MeOH t MeCN G   found for Li (I) and Na (I) reflect a better solvation of these cations in methanol relative to acetonitrile. The selectivity of L1 towards metal cations in methanol relative to acetonitrile is strongly dependent on the solvation of the cations in these solvents. Therefore the selectivity of L1 for univa- lent metal cations in methanol decreases as the solvation Figure 7. A plot of the transfer Gibbs energy of monovalent metal cations from MeCN to MeOH at 298.15 K (data based in the Ph4AsPh4B convention) against the selectivity of L1 (logarithm scale), for these metal cations in MeOH relative to MeCN at 298.15 K. Croat. Chem. Acta 86 (2013) 1. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 14 Table 8. MeOH MeCN (M )(MeCN MeOH) log n tG a b S       (8) Thermodynamic Parameters of Complexation of L1 and L2 with bivalent metal cations in acetonitrile and methanol at 298.15 K MeCN log Ks c 1 º kJ mol G   c 1 º kJ mol H   c 1 1 º J K mol S    L1 L2 L1 L2 L1 L2 L1 L2 Mg2+ 9.9(b) 6.1(b) −56.5 −34.8 −37.1(b) −61.3(b) 65 −89 Ca2+ 14.5(b) 8.9(b) −82.8 −50.8 −105.7(b) −78.3(b) −77 −92 Sr2+ 12.7(b) 10.1(b) −72.5 −57.7 −78.3(b) −85.6(b) −19 −94 Ba2+ 10.4(b) 11.1(b) −59.4 −63.4 −48.8(b) −122.5(b) 36 −198 Pb2+ 11.0(b) 7.7(b) −62.8 −44.0 −102.6(b) −75.6(b) −133 −106 Zn2+ 5.3(a) 4.4(a) −30.3 −25.1 −37.8(a) −105.9(a) −25 −271 Cd2+ 11.5(b) 5.3(a) −65.6 −30.3 −96.4(b) −113.5(a) −103 −279 Hg2+ 9.9(b) 5.8(a) −56.5 −33.1 −90.3(b) −122(a) −113 −298 Cu2+ 5.30(a) 4.8(a) −30.3 −27.4 −51.3(a) −56.1(a) −70 −96 Co2+ 5.50(a) 4.5(a) −31.4 −25.7 −48.8(a) −35.5(a) −58 −33 Ni2+ 5.58(a) 5.5(a) −31.9 −31.4 4.9(a) −35.6(a) 123 −14 MeOH log Ks c 1 º kJ mol G   c 1 º kJ mol H   c 1 1 º J K mol S    L1 L2 L3 L1 L2 L3 L1 L2 L3 L1 L2 L3 Mg2+ 3.4(a) 3.0(a) -- −19.4 −17.2 --- −1.2(a) 34.8(a) --- 61 174 --- Ca2+ 11.3(b) 5.1(a) 4.11(a) −65 −29.2 −23.6 −31(b) −26.0(a) 25.5(a) 114 11 165 Sr2+ 9.1(b) 7.9b 5.20(a) −51.9 −45.2 −29.7 −5.6(b) −19.8(b) 18.3(a) 155 85 161 Ba2+ 6.1(b) 9.3b 4.20(a) −35.4 −52.8 −24.2 0.5(b) −17.8(b) 28.9(a) 120 117 178 Co2+ 5.70(a) --- −32.5 −29.2 --- 7.4(a) −31.2(a) --- 133 −7 --- Cu2+ 4.30(a) 5.1(a) 4.10(a) −24.5 −29.1 −23.2 13.2(a) −42.4(a) 32.7(a) 126 −45 187 Cd2+ 6.44(b) 4.9(a) 5.70(a) −36.7 −27.7 −32.5 −21.4(b) −38.7(a) 11.7(a) 51 −37 148 Zn2+ 5.91(a) 5.0(a) −33.7 −28.2 5.3(a) −32.9(a) --- 131 −16 --- Hg2+ 4.6(a) --- 5.34(a) −26.3 --- −30.5 −33.4(a) ---- 13.9(a) −24 --- 149 Pb2+ 4.53(a) 4.9(a) --- −25.9 −28 --- −23.5(a) 16.9(a) --- 8 151 --- Ni2+ --- 5.2(a) --- --- −29.7 --- --- −35.6(a) --- --- −20 --- (a) Direct calorimetry. Standard deviations for log Ks values given with one decimal place are in the range ± 0.1–0.2. For log Ks values given with two decimal places, the standard deviations are ± 0.03–0.06. For ΔcHº values are in the 0.1–0.5 kJ mol-1 range. (b) Competitive titration calorimetry. Standard deviation in log Ks values is in the range ± 0.2–0.3. MeOH MeCN (M )(MeCN MeOH) log n tG a b S       (8) Standard deviation in the ΔcHº values are in the ± 0.2–0.9 kJ /mol range. Table 8. Thermodynamic Parameters of Complexation of L1 and L2 with bivalent metal cations in acetonitrile and methanol at 298.15 K ermodynamic Parameters of Complexation of L1 and L2 with bivalent metal cations in acetonitrile and methanol at (a) Direct calorimetry. Standard deviations for log Ks values given with one decimal place are in the range ± 0.1–0.2. For log Ks values given with two decimal places, the standard deviations are ± 0.03–0.06. For ΔcHº values are in the 0.1–0.5 kJ mol-1 range. (b) Competitive titration calorimetry. Standard deviation in log Ks values is in the range ± 0.2–0.3. Standard deviation in the ΔcHº values are in the ± 0.2–0.9 kJ /mol range. hosting capacity for these cations in MeCN than in MeOH. Computer modelling simulation was carried out not contribute to the complexation but may compensate each other. A representative example is given by Eq. 9 expressed in the following thermodynamic cycle shown below where the higher solvation of the receptor in MeOH relative to MeCN is almost compensated by the higher solvation of the complex in MeCN than in MeOH. Table 9. Selectivity of L1 for univalent metal cations in meth- anol relative to acetonitrile at 298.15 K Table 9. Croat. Chem. Acta 86 (2013) 1. MeOH MeCN (M )(MeCN MeOH) log n tG a b S       (8) Selectivity of L1 for univalent metal cations in meth- anol relative to acetonitrile at 298.15 K Metal cation MeOH MeCn S Li+ 3.2 × 105 Na+ 4.0 × 103 K+ 3.0 × 102 Rb+ 1.1 × 102 Ag+ 3.2 × 10−1 O N O 5 1 2 3 4 5 6 7 6' 7' 4 1 1 55 kJ mol 1 1 1 23 kJ mol Li (MeCN) L1(MeCN) Li L1(MeCN) 23.64 kJ mol 5 kJ mol 2.96 kJ mol Li (MeOH) L1(MeOH) Li L1(MeOH)                   (9) 1 1 55 kJ mol 1 1 1 23 kJ mol Li (MeCN) L1(MeCN) Li L1(MeCN) 23.64 kJ mol 5 kJ mol 2.96 kJ mol Li (MeOH) L1(MeOH) Li L1(MeOH)                   (9) (9) As for L2, complexation with univalent metal cations in acetonitrile at 298.15 K shows complex formation of 1:2 and 1:1 (ligand: metal cation) stoichiometries for Na (I) and Li (I) in this solvent. Thus L2 shows a higher Croat. Chem. Acta 86 (2013) 1. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 15 Figure 8. Molecular modelling of L2-2Na+ (1:2) complex as determined by molecular simulation studies, the hydrogen atoms were removed to increase the clarity of the structure. Figure 9. Plot for the enthalpies of complexation against the entropies of complexation of L2 with univalent metal cations in methanol at 298.15 K. Figure 9. Plot for the enthalpies of complexation against the entropies of complexation of L2 with univalent metal cations in methanol at 298.15 K. Figure 8. Molecular modelling of L2-2Na+ (1:2) complex as determined by molecular simulation studies, the hydrogen atoms were removed to increase the clarity of the structure. ing L2 and other univalent metal cations (Na (I), K (I), Rb (I) and Cs (I)), the processes were found to be enthalpically controlled and entropically unfavoured. Complexation of L2 with the Ag (I) cation in ace- tonitrile and methanol shows that like L1 this ligand forms stronger complexes in the protic (MeOH) than in the dipolar aprotic solvent (MeCN). Thus the processes involving the complexation of both ligands with this cation in MeOH are enthalpically controlled and entropically unfavoured. MeOH MeCN (M )(MeCN MeOH) log n tG a b S       (8) In acetonitrile, the complexation process is enthalpically controlled and entropically favoured. The latter parameter reflects the strong cation desolvation upon complexation in this solvent. The lineal correlation between the transfer Gibbs energy, MeOH MeCN tG   (data based on the Ph4AsPh4B convention)63, of alkali metal cations and silver cation from acetonitrile to methanol against the log( MeOH MeCN S ) found for L1 is not followed for L2 particularly for Ag (I) due to the higher solvation of the silver complex in MeCN relative to MeOH as shown in the thermodynam- ic cycle (Eq. 10). using HyperChem 5.0, in vacuum conditions. These studies show that the arrangement of minimum energy for this system is that in which one Na (I) is held by 6 coordination sites provided by the oxygen donor atoms and the solvent molecule (MeCN), while the second Na (I) is coordinated to 5 oxygen donor atoms in the com- plex, as shown in Figure 8. A detailed investigation by X-ray diffraction studies on heavy metal cations (Cd (II) and Pb (II))22 complexes with a calix[4]arene ester showed that the solvent is hosted in the hydrophobic cavity, thus, the solvent can provide a coordinating site for the cation through its nitrogen donor atom. Such is the case for the cadmium complex and this may be the case here. Moderate stabilities were found for the other uni- valent metal cations (K (I), Rb (I), Cs (I) and Ag (I)) and L2 in acetonitrile as shown in Table 7. As for L2 with alkali metal cations in methanol at 298.15 K, the highest stability constant was found for Rb+ (log Ks = 5.8). However the standard deviation of the stability constant data (in logarithm terms) (± 0.2) shows the low discrimination of this ligand towards these metal cations. The similar standard Gibbs energy values for L2 and alkali-metal cations give evidence of entropy-enthalpy compensation processes which Grunwald and Steel79 attributed to solvent reorganisa- tion. This is better depicted in the slope of 317.3 K calculated from a plot of enthalpies against entropies of complexation of L2 with univalent metal cations in methanol which is not far from the temperature of 298.15 K (Figure 9). The complexation of L2 with the lithium cation in methanol is enthalpically controlled and entropically favoured. MeOH MeCN (M )(MeCN MeOH) log n tG a b S       (8) As far as L2 and bivalent cations in MeCN are concerned the complex stability results from the enthalpy contribution given that the processes are entropically unfavoured to the extent that the maximum exothermicity is that for the Ba (II) cation and this receptor in this solvent. In MeOH, the enthalpy and entropy associated with the complexation of Mg (II) and L2 is typical of systems involving a high- ly solvated cation for which the energy required for desolvation overcomes that of cation-receptor binding. As a result the process is endothermic and entropy con- trolled. For all other alkaline-earth metal cations the complex stability is enthalpy and entropy favoured. As far as transition and heavy metal cations are concerned the enthalpy controls the stability of these complexes while the entropy has a destabilising effect. Figure 10. Plot of the standard Gibbs energies of hydration of the metal cations against the standard Gibbs energy of complexation of L1 and L2 with bivalent metal cations in acetonitrile at 298.15 K. Contribution of Cation Solvation and Cation- receptor Binding to Complex Formation of L1, L2 and L3 with Bivalent Cations in Acetonitrile and Methanol The information regarding transfer Gibbs energies of bivalent cations from one solvent to another is very lim- ited. Therefore to assess the cation solvation upon complexation of bivalent cations and these receptors in these solvents attempts are made to establish whether or not a correlation is found between the thermodynamic parameters of complexation and corresponding data for the hydration of these ions.80−83 Hydration rather than solvation is considered due to the limited amount of solv- ation data in these solvents. This is justified given that the trend observed in solvation is expected to be the same as that of hydration. In the complexation of macrocycles and metal cations in a given solvent, two main processes are taking place. These are cation desolvation and receptor binding. To obtain some information regarding which of these processes predominate, thermodynamic data of complexation are plotted against the standard Gibbs ener- gies of hydration of the bivalent cations involved. This is shown in Figure10. The pattern observed is similar to that found for these systems in MeOH. This figure shows that the minimum Gibbs energy (higher stability constant) is that for CaL2 (II). It is clear that the extent of complexation increases from Ba (II) to Ca (II). MeOH MeCN (M )(MeCN MeOH) log n tG a b S       (8) As for the processes concern- 1 1 25.5 kJ mol 1 1 1 33 kJ mol Ag (MeCN) L2(MeCN) Ag L2(MeCN) 30.1 kJ mol 1.10kJ mol 23.6kJ mol Ag (MeOH) L2(MeOH) Ag L2(MeOH)                   (10) (10) ii) Data for bivalent cations and L1 in MeCN (Table 8) show that this receptor forms very strong complexes with alkaline-earth and heavy metal (Hg (II), Pb (II) and Cd (II) cations in this solvent. Moderate stabilities are observed for Zn (II), Cu (II), Ni (II) and Co (II) com- plexes. For Mg (II) and Ba (II) these processes are enthalpically controlled and entropically favoured. For Ca (II), Sr (II), Pb (II), Zn (II), Cd (II), Hg (II), Cu (II) and Co (II) the processes are enthalpically stabilised and ii) Data for bivalent cations and L1 in MeCN (Table 8) show that this receptor forms very strong complexes with alkaline-earth and heavy metal (Hg (II), Pb (II) and Cd (II) cations in this solvent. Moderate stabilities are observed for Zn (II), Cu (II), Ni (II) and Co (II) com- plexes. For Mg (II) and Ba (II) these processes are enthalpically controlled and entropically favoured. For Ca (II), Sr (II), Pb (II), Zn (II), Cd (II), Hg (II), Cu (II) and Co (II) the processes are enthalpically stabilised and Croat. Chem. Acta 86 (2013) 1. 16 A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics Figure 10. Plot of the standard Gibbs energies of hydration of the metal cations against the standard Gibbs energy of complexation of L1 and L2 with bivalent metal cations in acetonitrile at 298.15 K. entropically unfavoured. However for Ni (II) an inverted result is observed. The process is entropically controlled and enthalpically unfavoured. As far as MeOH is con- cerned, the only available stability constant value is that for Ba (II) and this differs from the value reported here derived from potentiometry and competitive titration calorimetry. Very strong complexes are found for Ca (II), Sr (II) and Ba (II) in this solvent. The highest stabil- ity constant in MeOH and MeCN is that for Ca (II). In MeOH, both parameters, enthalpy and entropy contribute favourably to the high stability observed for this system in this solvent with a slightly higher contribution from enthalpy. Croat. Chem. Acta 86 (2013) 1. MeOH MeCN (M )(MeCN MeOH) log n tG a b S       (8) Data listed in Table 8 for Mg (II), Sr (II), Ba (II), Co (II), Cu (II) and Zn (II) show that these processes are entropically controlled and enthalpically unfavoured, except for Mg (II) and Sr (II) which are slightly enthalpically favoured. For Hg (II) and Pb (II) the pro- cesses are enthalpy controlled but for Hg (II) entropically unfavoured. For Cd (II) complexation is favoured by both, enthalpy and entropy. As far as L2 and bivalent cations in MeCN are concerned the complex stability results from the enthalpy contribution given that the processes are entropically unfavoured to the extent that the maximum exothermicity is that for the Ba (II) cation and this receptor in this solvent. In MeOH, the enthalpy and entropy associated with the complexation of Mg (II) and L2 is typical of systems involving a high- ly solvated cation for which the energy required for desolvation overcomes that of cation-receptor binding. As a result the process is endothermic and entropy con- trolled. For all other alkaline-earth metal cations the complex stability is enthalpy and entropy favoured. As far as transition and heavy metal cations are concerned the enthalpy controls the stability of these complexes while the entropy has a destabilising effect. entropically unfavoured. However for Ni (II) an inverted result is observed. The process is entropically controlled and enthalpically unfavoured. As far as MeOH is con- cerned, the only available stability constant value is that for Ba (II) and this differs from the value reported here derived from potentiometry and competitive titration calorimetry. Very strong complexes are found for Ca (II), Sr (II) and Ba (II) in this solvent. The highest stabil- ity constant in MeOH and MeCN is that for Ca (II). In MeOH, both parameters, enthalpy and entropy contribute favourably to the high stability observed for this system in this solvent with a slightly higher contribution from enthalpy. Data listed in Table 8 for Mg (II), Sr (II), Ba (II), Co (II), Cu (II) and Zn (II) show that these processes are entropically controlled and enthalpically unfavoured, except for Mg (II) and Sr (II) which are slightly enthalpically favoured. For Hg (II) and Pb (II) the pro- cesses are enthalpy controlled but for Hg (II) entropically unfavoured. For Cd (II) complexation is favoured by both, enthalpy and entropy. REFERENCES Calixarenes: A Versatile Class of Macrocycles Compounds; J. Vicens, V. Böhmer (Eds), Dordrecht, 1991. 7. C. D. Gutsche, Aldrichim. Acta 28 (1995) 3−9. 8. C. D. Gutsche, Calixarenes Revisited, J. F. Stoddart (Eds), The Royal Society of Chemistry, Cambridge, UK, 1998. 9. Calixarenes 2001, Z. Asfari, V. Bohmer, J. M. Harro Vicens (Eds), Kluwer Academic Publishers, 2001. 10. J.-M. Lehn, Angew. Chem. Int. Ed. Engl. 27 (1988) 89-112. 11. B. G. Cox and H. Schneider, Coordination and Transport Prop- erties of Macrocyclic Compounds in Solution, Elsevier, New York, 1992. 12. J. W. Steed and J. L. Atwood, Supramolecular Chemistry, John Wiley & Sons, London, 2000. 13. A. F. Danil de Namor, M. C. Cabaleiro, B. M. Vuano, M. Salo- mon, O. I. Pieroni, D. A. Pacheco Tanaka, C. Y. Ng, M. A. Llosa Tanco, N. M. Rodriguez, J. D. Cardenas Garcia, and A. R. Casal, Pure & Appl. Chem. 66 (1994) 435−440. REFERENCES increases. In fact the same pattern applies in terms of enthalpy. This trend in Gibbs energies and enthalpies was also observed in MeOH, although binding energies are stronger in MeCN than in MeOH. The metal cations are more solvated in MeOH than in MeCN. The former sol- vent is a better cation solvator than MeCN. Therefore the energy required for cation desolvation upon complexation in MeOH is likely to be higher than in MeCN. This statement is corroborated by the endothermic character of the enthalpies associated with the complexation reactions involving L1 and Ba (II), Co (II), Cu (II) and Zn (II) cations in MeOH. 1. Calixarenes in the Nanoworld, J. Vicens, J. Harrowfield (Eds.), Springer, 2007. 2. A. F.Danil de Namor, R. M. Cleverley, and M. L. Zapata- Ormachea, Chem. Rev. 98 (1998) 2495−2525. 3. A. F. Danil de Namor, J.Wang, I. Gomez Orellana, F. J. Sueros Velarde, and D. A. Pacheco Tanaka, Thermodynamic and Elec- trochemical Aspects of p-tert-butyl-calix(n)arenes (n=4,6,8) and their Interactions with Amines in: J. Vicens, Z. Asfari, J. M. Har- rowfield (Eds), Calixarenes 50th Anniversary: Commemorative Volume, Kluwer Academic Publishers, Dordrecht, The Nether- lands, 1994, 371−378. 4. A. F. Danil de Namor, Thermodynamics of Calixarene-Ion Inter- actions,in: Z. Asfari, V. Bohmer, J. M. Harrowfield, J. Vicens (Eds), Chapter 19 in Calixarenes, 2001, Kluwer Academic Pub- lishers, 2001, 346. The selectivity peak observed when complexation data involving L1 were plotted against the cation hydra- tion Gibbs energies is not observed for L2 and these cations in MeCN or indeed in MeOH as shown in Figure 10. The same situation occurs with L3 and metal cations in MeOH. This ligand shows poor discrimination for these metal cations as shown in Figure 11. It seems that the binding process is overcome strongly by cation desolvation, given that for all the systems investigated the distinctive feature of the data is that in all cases the complexation process is endothermic and entropy con- trolled. It is quite clear from these results that as the num- ber of phenyl units in the structure of calix(n)arenes in- creases from n = 4 to n = 6, the flexibility of the ligand increases and therefore the receptors are unable to selec- tively recognise these metal cations in these solvents. 5. C. D. Gutsche,“Calixarenes, in Monographs in Supramolecular Chemistry”, J. F. Stoddart (Ed.), Royal Society of Chemistry: Cambridge, 1989, 1. 6. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 17 A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics Figure 11. Plot of the standard Gibbs energies of hydration of the metal cations against the standard Gibbs energy of complexation of L1, L2 and L3 with bivalent metal cations in methanol at 298.15 K. ionic guests and calls for the need of using an integrated approach, to ensure that the data reported are repre- sentative of the process taking place in solution so the concept of selectivity can be accurately addressed. The field of Supramolecular Chemistry requires a multidis- ciplinary approach within and outside Chemistry. Un- doubtedly a great deal of experimental work and mis- leading statements could have been avoided if early work in the general area of Physical Chemistry and particularly on electrochemical and thermodynamic studies of electrolytes and neutral species would have been considered. Within this context emphasis should be made about the fact that the remarkable growth of interest in macrocyclic chemistry involving neutral receptors and their interaction with ionic species has resulted in an extensive broadening in the field of solu- tion chemistry involving novel electrolytes and non- electrolytes and there is plenty of fascinating research waiting to be explored in this area. Figure 11. Plot of the standard Gibbs energies of hydration of the metal cations against the standard Gibbs energy of complexation of L1, L2 and L3 with bivalent metal cations in methanol at 298.15 K. MeOH MeCN (M )(MeCN MeOH) log n tG a b S       (8) However a drop in stability is observed from Ca (II) to Cu (II) as the standard hydration Gibbs energies of the ions in- crease. In conclusion the results show that from Ba (II) to Ca (II) the binding energy predominates over the energy required for cation desolvation. However from Ca (II) to Cu (II) the latter overcomes the former as cation solvation py g Given that L3 is slightly soluble in MeCN, thermo- dynamic data shown in Table 8 are referred to MeOH. Among alkaline-earth metal cations, the stability of the complex decreases in moving from the cyclic tetramer to the cyclic hexamer for Ca (II) and Sr (II) to the extent that no complexation (or extremely weak was found for Mg (II). For Ba (II) the stability increases from L1 to L2 and decreases from L2 to L3. In all cases the stability of complex formation is controlled by entropy given that these processes are endothermic. As expected different trends are observed for the complexation of alkaline-earth metal cations with L1 relative to L2. This may be at- tributed to the increase in the size of the hydrophilic pseudo-cavity of L2 and consequently on the number of donor atoms available for complexation. Thus among these cations L1 is selective for Ca (II) in both solvents while L2 shows the highest selectivity for Ba (II). In the following section the competition between cation desolvation (endothermic process) and cation-receptor binding (exothermic process) for systems involving biva- lent cations in MeCN and MeOH is discussed. CONCLUDING REMARKS 14. A. F. Danil de Namor, E. Gil, M. A. Llosa Tanco, D. A. Pacheco Tanaka, L. E.Pulcha Salazar, R. A. Schulz, and J. Wang, J. Phys. Chem. 99 (1995) 16776−16780. The emphasis in this article lies on fundamental aspects of thermodynamics involving macrocyclic receptors and 15. A. F. Danil de Namor, E. Gil, M. A. Llosa Tanco, D. A. Pacheco Tanaka, L. E. Pulcha Salazar, R. A. Schulz, and J.Wang, J. Phys. Croat. Chem. Acta 86 (2013) 1. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 18 A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics Chem. 99 (1995) 16781−16785. 45. J. L. Sessler, D. E. Gross, W. S. Cho, V. M. Lynch, F. P. Schmidtchen, G. W. Bates, M. E. Light, and P. A. Gale, J. Am. Chem. Soc. 128 (2006) 12281−12288. 16. A. F. Danil de Namor, M. L. Zapata-Ormachea, O. Jafou, and N. Al Rawi, J. Phys. Chem. B 101 (1997) 6772−6779. 17. A. F. Danil de Namor, F. J. Sueros Velarde, A. R. Casal, A. Pug- liese, M. T. Goitia, M. Montero, and F. Fraga Lopez, J. Chem. Soc. Faraday Trans. 93 (1997) 3955−3959. 46. M. H. Abraham, A. F. Danil de Namor , and W. H. Lee, J. Chem. Soc. Chem.Comm. (1977) 893−894. 47. A. F. Danil de Namor and L. Ghousseini, J. Chem. Soc. Faraday Trans I 80 (1984) 2349−2360. 18. A. F. Danil de Namor, M. A. Llosa Tanco, L. E. Pulcha Salazar, D. Kowalska, J. Villanueva Salas, and R. A. Schulz, J. Chem. Soc. Faraday Trans. 20 (1998) 3111−3115. 48. R. M. Izatt, K. Pawlak, J. S. Bradshaw, and R. L. Bruening, Chem. Rev. 91 (1991) 1721−2085. 49. R. M. Izatt, J. S. Bradshaw, K. Pawlak , R. L. Bruening, and B. J. Tarbet, Chem. Rev. 92 (1992) 1261−1354. 19. A. F. Danil de Namor, A. Pugliese, A. R. Casal, M. Barrios Llerena, P. J. Aymonino, and F. J. Sueros Velarde, Phys. Chem. Chem. Phys. 2 (2000) 4355−4360. 50. B. G. Cox and H. Schneider, Coordination and Transport Prop- erties of Macrocyclic Compounds in Solution, 1992, Elsevier Science Publishers, Amsterdam The Netherlands. 20. A. F. Danil de Namor, D. Kowalska, Y. Marcus, and J. Villanueva-Salas, J. Phys. Chem. B 105 (2001) 7542−7549. 51. A. F. Danil de Namor and M. Shehab, J .Phys. Chem. B 107 (2003) 6462−6468. 21. A. F. CONCLUDING REMARKS Danil de Namor, D. Kowalska, E. E. Castellano, O. E. Piro, F. J. Sueros Velarde, and J. Villanueva-Salas, Phys. Chem. Chem. Phys. 3 (2001) 4010−4021. 52. A. F. Danil de Namor, I. Abbas , and H. Hammud, J. Phys. Chem. B 110 (2006) 2142−2149. 22. A. F. Danil de Namor, S. Chahine, D. Kowalska, E. E. Castel- lano, and O. E. Piro, J. Am. Chem. Soc. 124 (2002) 12824−12836. 53. A. F. Danil de Namor and R. Khalife, Phys. Chem. Chem. Phys. 12 (2010) 753−760. 23. A. F. Danil de Namor, M. A. Pugliese, A. R. Casal, W. B. Apari- cio-Aragon, O. E. Piro, and E. E. Castellano, Phys. Chem. Chem. Phys. 6 (2004) 3286−3291. 54. Physical Chemistry of Organic Solvents, A. K. Covington, T. Dickinson, (Eds.) (1973) Plenum Press, New York. 55. O. Popovych and R. P. T. Tomkins, Nonaqueous Solution Chem- istry, (1981) John Wiley & Sons, New York. 24. A. F. Danil de Namor, S. Chahine, E. E. Castellano, and O. E. Piro, J. Phys. Chem. B 108 (2004) 11384−11389. 56. J. Koryta, J. Dvorak, and L. Kavan, Principles of Electrochemis- try, 2nd Ed., 1993, John Wiley & Sons, New York. 25. A. F. Danil de Namor and S. Chahine, J. Phys. Chem. B 109 (2005) 18096−18112. 57. M. Hojo, Pure Appl. Chem. 80 (7) (2008) 1539–1560. 26. A. F. Danil de Namor, S. Chahine, E. E. Castellano, O. E. Piro, and H. D. Jenkins, J. Chem. Soc. Chem. Comm. (2005) 3844−3846. 58. K. Izutsu, Electrochemistry in Nonaqueous Solutions, 2009, Wiley VCH, Weinheim. 59. G. J. Janz and R. P. Tomkins, Nonaqueous Electrolytes Hand- book,(1972), Academic Press, New York. 27. M. A. Pugliese, M. T. Goitia, M. E. Montero, A. R. Casal, and A. F. Danil de Namor, Supramol. Chem. 18 (2006) 575−580. 60. E. Dustin, F. P. Schmidtchen, W. Antonius, P. A. Gale, V. M. Lynch, and J. L. Sessler, Chemistry. A European J. 14 (2008) 7822−7827. 28. A. F. Danil de Namor and K. Zegarra Fernandez, J. Phys. Chem B 111 (2007) 7321−7330. 61. P. M. Marcos, J. P. Ascenso, M. A. Segurado, P. J. Cragg, P. S. Michel, V. Hubscher-Bruder, and F. Arnaud-Neu, Supramol. Chem. 23 (2011) 93−101. 29. A. S. de Araujo, O. E. Piro, E. E. Castellano, and A. F. Danil de Namor, J. Phys. Chem. A 46 (2008) 11885−11894. 30. G. Horvat, V. Stilinovic, B. Kaitner, L. Frkanec, and V. CONCLUDING REMARKS Tomisic, Inorg. Chem. 51 (2012) 6264−6278. 62. A. J. Parker, Chem. Rev. 69 (1969) 1−32. 31. A. I. Vogel, Revised by B. V. Smith, N. M. Waldron, 3rd ed, 87, 1980. 63. B. G. Cox, G. R. Hadwig, A. J. Parker, and D. W. Watts, Aus. J Chem. 27 (1974) 477−501. 64. A. J. Parker, U. Mayer, R. Schmid, and V. Gutmann, J. Org. Chem. 43 (1978) 1843−1854. 32. D. D. Perrin, W. L. F. Armarego, D. R. Perrin, Purification of Laboratory Chemicals, 2nd ed. Pergamon Press, UK, 1980. 65. G. Gritzner, Pure & Appl.Chem. 60 (1988) 1743−1766. 33. L. M. Harwood, C. J. Mody, Experimental Organic Chemistry, Principles and Practice, Blackwell Scientific Publications, 1980. 66. Y. Marcus, Ion Solvation, 1985, Wiley. 34. A. F. Danil de Namor, R. G. Hutcherson, F. J. Sueros Velarde, A. Alvarez-Larena, and J. L. Brianso, J. Chem. Soc. Perkin Trans. 1 (1998) 2933−2938. 67. A. J. Parker and R. Alexander, J. Am. Chem. Soc. 90 (1968) 3313−3319. 68. B. G. Cox, A. J. Parker, and W. E. Waghorne, J. Am. Chem. Soc. 95 (1973) 1010−1034. 35. D. R. Steward and C. D. Gutsche, Org. Prep. Proceed. Int. 25 (1993) 137−139. 69. A. F. Danil de Namor, R. Traboulssi, F. Fernandez Salazar, V. Dianderas de Acosta, I. Fernandez de Vizcardo, and J. Muñoz Portugal, J. Chem. Soc. Faraday Trans. I 85 (1989) 2705−2712. 36. K. Iwamoto and K. A. Shinkai, Bull. Chem. Soc. Jpn. 67 (1994) 1499−1502. 37. G. Jones and C. Bradshaw, J. Am. Chem. Soc. 55 (1933) 1780−1800. 70. M. A. McKervey, M. J. Schwing-Weill, and F. Arnaud-Neu in<. Comprehensive Supramolecular Chemistry., G. W. Gokel (Ed.), Vol. 1, 1996, Elsevier, Oxford. 38. J. E. Lind, J. J. Zwolenik, and R. M. Fuoss, J. Am. Chem. Soc. 81 (1957) 1557−1559. 39. J. J. Christensen, R. M. Izatt, and I. D. Hansen, Rev. Sci. Instrum. 36 (1965) 779−783. 71. F. Arnaud-Neu, S. Barboso, F. Berny, A. Casnati, N. Muzet, A. Pinalli, R. Ungaro, M. J. Schwing, and G. Wipff, J. Chem. Soc. Perkin Trans. 2 1727−1738 and Refs. within. 40. D. J. Eatough, J. J. Christensen, and R. M. Izatt in: Experimental in Thermometric, Titrimetry and Titration Calorimetry, 1974, Brigham Young University Press, Provo, Utah. 72. A. F. Danil de Namor, Chapter 19 in Calixarenes 2001, Z. Asfari, V. Bohmer, J. M. Harrowfield, J. Vicens (Eds) , Kluwer Academic Publishers, (2001). 41. J. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics Croat. Chem. Acta 86 (2013) 1. CONCLUDING REMARKS Suurkuusk and I. Wadso, Chim.Script 20 (1982) 155. 73. A. Arduini, E. Ghidini, A. Pochini, R. Ungaro, G. D. Andreetti, G. Calestani, F. Ugozzoli, J. Inclusion Phenom. 6 (1988) 119−134. 42. 2277 Thermal Activity Monitor, Instruction Manual, AB Brommo, Sweden, 1985. 43. L. E. Briggner and I. Wadso, J. Biochem. Biophys. Methods 22 (1991) 101−118. 74. F. Arnaud-Neu, G. Barrett, S. Fanni, D. Marrs, W. McGregor, M. A. McKervey, M. J. Schwing-Weill, V. Vetrogon, and S. Wechsler, J. Chem. Soc. Perkins Trans. 2 (1995) 453−461. 44. A. F. Danil de Namor and L. Ghousseini, J. Chem. Soc. Faraday Trans. 81 (1985) 781−791. 75. U. C. Meier and C. Detellier, Dalton Trans. (2003) 4574−4579. Croat. Chem. Acta 86 (2013) 1. 76. A. F. Danil de Namor, Coord. Chem. Rev. 190−192 (1999) 283−295. 77. A. F. Danil de Namor, M. L. Zapata Ormachea, and R. G. Hutcherson, J. Phys. Chem. 103 (1999) 366−371. 78. R. Fernandez Prini in: Physical Chemistry of Organic Solvents, A. K. Covington and T. Dickinson (Eds.), 1973, Plenum Press, New York. 9 G ld d C S l A Ch S 11 (199 ) 5687−5693. 80. J. Burgess, Metal Ions in Solution, Ellis Horwood, New York, 1978. 81. D.T. Richens, The Chemistry of Aqua Ions,Wiley, Chichester, 1997. 82. T. Ikeda, M. Boero, and K. Terakura, J. Chem. Phys. 126 (2007) 034501. 83 k d d Ch h 13 (2012) 041101 76. A. F. Danil de Namor, Coord. Chem. Rev. 190−192 (1999) 283−295. A. F. Danil de Namor et al., An Enchiridion of Supramolecular Thermodynamics 19 5687−5693. 80. J. Burgess, Metal Ions in Solution, Ellis Horwood, New York, 1978. 80. J. Burgess, Metal Ions in Solution, Ellis Horwood, New York, 1978. 81. D.T. Richens, The Chemistry of Aqua Ions,Wiley, Chichester, 1997. 81. D.T. Richens, The Chemistry of Aqua Ions,Wiley, Chichester, 1997. 82. T. Ikeda, M. Boero, and K. Terakura, J. Chem. Phys. 126 (2007) 034501. wald and C. Steel, J. Am. Chem. Soc. 117 (1995) 83. T. Ikeda and M. Boero, J. Chem. Phys. 137 (2012) 041101.
https://openalex.org/W2743364277	https://www.nature.com/articles/s41598-017-07909-y.pdf	English	null	Palladium nanoparticles entrapped in a self-supporting nanoporous gold wire as sensitive dopamine biosensor	Scientific reports	2,017	cc-by	7,007	Palladium nanoparticles entrapped in a self-supporting nanoporous gold wire as sensitive dopamine biosensor XinYi1,5 YuxuanWu4 GuoxinTan3 PengYu 2,5 Lei Zhou2,5 ZhengnanZhou2,5 JunqiChen2,5 Received: 13 April 2017 Accepted: 4 July 2017 Published: xx xx xxxx Received: 13 April 2017 Accepted: 4 July 2017 Published: xx xx xxxx Traced dopamine (DA) detection is critical for the early diagnosis and prevention of some diseases such as Parkinson’s, Alzheimer and schizophrenia. In this research, a novel self-supporting three dimensional (3D) bicontinuous nanoporous electrochemical biosensor was developed for the detection of dopamine by Differential Pulse Voltammetry (DPV). This biosensor was fabricated by electrodepositing palladium nanoparticles (Pd) onto self-supporting nanoporous gold (NPG) wire. Because of the synergistic effects of the excellent catalytic activity of Pd and novel structure of NPG wire, the palladium nanoparticles decorated NPG (Pd/NPG) biosensor possess tremendous superiority in the detection of DA. The Pd/ NPG wire biosensor exhibited high sensitivity of 1.19 μA μΜ−1, broad detection range of 1–220 μM and low detection limit up to 1 μM. Besides, the proposed dopamine biosensor possessed good stability, reproducibility, reusability and selectivity. The response currents of detection in the fetal bovine serum were also close to the standard solutions. Therefore the Pd/NPG wire biosensor is promising to been used in clinic. It is well-known that DA is an important catecholamine neurotransmtiter1–3. Abnormal level of DA will lead to some serious diseases such as Parkinson’s, Alzheimer and schizophrenia4, 5. The detection of DA is beneficial for early diagnosis and prevention. Therefore it is extremely necessary to develop DA biosensors with the bril- liant performance such as high sensitivity, broad detection range and excellent selectivity, and so on. Recently, high performance liquid chromatography6, liquid chromatography-electrospray tandem mass spectrometry7, surface-enhanced Raman scattering spectroscopy and fluorescence8 detection methods were applied to detect DA. Although these methods possessed some fine property, they were poor in certain aspects of property, such as instrument complexity, low sensitivity and time consuming, and so on. Fortunately, the characteristics of electro- chemical activity of DA make it detectable by electrochemical method9, which has attracted increasing attention, due to the advantages in sensitivity, detection time, selectivity and operation, and so on. However, the ascorbic acid (AA), uric acid (UA)10, norepinephrine (NE), epinephrine (EP)11 and catechol (CC)12 coexist with DA in a living organism and have some interferences during the detection of DA. Therefore, the electrochemical method used to detect DA must possess excellent selectivity on the basis of broad detection range, high sensitivity and low detection limit, and so on. www.nature.com/scientificreports www.nature.com/scientificreports www.nature.com/scientificreports Palladium nanoparticles entrapped in a self-supporting nanoporous gold wire as sensitive dopamine biosensor XinYi1,5 YuxuanWu4 GuoxinTan3 PengYu 2,5 Lei Zhou2,5 ZhengnanZhou2,5 JunqiChen2,5 Enzyme electrochemical biosensors with the function of biorecognition had the advantage in selectivity for detection of biomolecule13. The enzyme electrochemical biosensors also possessed advantages such as broad detection range, high sensitivity, low detection limit, rapid sensing and ease in miniaturization, and can become a good choice for the detection of DA. However, the limitation of enzymes was affected by the environment easily14, 15, which seriously restricted the application of enzyme electrochemical biosensors. In order to solve this 1School of Medicine, South China University of Technology, Guangzhou, China. 2School of Materials Science and Engineering, South China University of Technology, Guangzhou, China. 3Institute of Chemical Engineering and Light Industry, Guangdong University of Technology, Guangzhou, China. 4Department of Electronic Communication & Software Engineering, Nanfang College of Sun Yat-sen University, Guangzhou, China. 5Guangdong Key Laboratory of Biomedical Sciences and Engineering, South China University of Technology, Guangzhou, 510006, China. Correspondence and requests for materials should be addressed to P.Y. (email: conan424683@live.com) or C.N. (email: imcyning@scut.edu.cn) SCIENTIfIC REPOrTs \| 7: 7941 \| DOI:10.1038/s41598-017-07909-y 1 www.nature.com/scientificreports/ www.nature.com/scientificreports/ Figure 1. Schematic diagram for the fabrication of the biosensor. (a) Au wire. (b) 3D bicontinuous nanoporous gold wire fabricated by electrochemical alloying/dealloying. (c) Pd decorated NPG wire. Figure 1. Schematic diagram for the fabrication of the biosensor. (a) Au wire. (b) 3D bicontinuous nanoporous gold wire fabricated by electrochemical alloying/dealloying. (c) Pd decorated NPG wire. problem, researchers developed nonenzymatic electrochemical biosensors, which were promising to overcome the limitation of enzymes easily affected by the environment16, 17. The noble metal nanoparticles due to possessing some special property such as enhanced electrocatalytic activity18, excellent adsorption capacities19 and brilliant electron-transfer20, was a promising nonenzymatic catalysts21. Nevertheless, the inherent defect of nanoparti- cles aggregating easily22, prevented its application in sensing. For solving this problem, researchers attempted diverse methods, and supporting the noble metal nanoparticles by the solid supports obtained fascinating con- sequences23–25. It has reported that graphene26, and single-walled and multi-walled carbon nanotubes27–29 have been used to support the nanoparticles. These nanohybrids made the interface of electrode/electrolyte possess extraordinary large specific surface area28, but the nature of extremely poor charge transfer made these electrodes own an exceptionally poor electronic conductance weakening the sensing properties30. Besides, among the noble metal nanoparticles, conductive substance and current collector existed large contact resistances31, preventing their use in the detection area seriously. The low-dimensional nanostructure also weakened the superiority of these nanohybrids in sensing. Palladium nanoparticles entrapped in a self-supporting nanoporous gold wire as sensitive dopamine biosensor XinYi1,5 YuxuanWu4 GuoxinTan3 PengYu 2,5 Lei Zhou2,5 ZhengnanZhou2,5 JunqiChen2,5 y g In order to resolve this problem, we constructed a nanoporous structure, which not only prevent the aggregat- ing of nanoparticles, but also can take advantage of nanostructure. Lang et al. have reported that NPG supported cobalt oxide microelectrodes possess excellent performance in the detection of glucose32. Han et al. have reported that self-grown Ni(OH)2 layer on bimodal nanoporous AuNi alloys can enhanced electrocatalytic activity33. In this work, we report a unique self-supporting three dimensional (3D) NPG with self-supporting bicontinuous nanoporosity. In this structure, the NPG wire not only possesses brilliant conductivity, accelerating charge trans- fer34, 35, but also can support itself. Besides, the NPG wire also own bicontinuous nanoporous structure, which is beneficial for the mass transport36. In order to improve the sensing performance for DA further, we inserted Pd into the NPG wire, due to the Pd owning admirable electrocatalytic for DA37, 38, and fabricated the Pd/NPG wire DA biosensor. To the best of our knowledge, there is no report about the Pd/NPG wire biosensor used to detect DA. The synergistic effects of the excellent catalytic activity of Pd and novel structure of NPG wire make the Pd/ NPG wire biosensor possess tremendous superiority in the detection of DA. Also, the direct connection method decreased the contact resistances among Pd, nanoporous Au skeleton and current collector of solid Au wire at the same time, improving the sensing property. The unique structure of interconnected nanoporous channels and ligaments, not only offers fast DA transfer, but also enlarges the specific surface area of electrode/electrolyte interface, improving the performance of the Pd/NPG wire biosensor. All discussed above make the Pd/NPG wire biosensor own excellent sensing performance. Results and Discussion h i i f h Characterization of the Pd/NPG wire biosensor. Our strategy to improve the sensing performance of DA is to fabricate Pd/NPG wire biosensor with the interconnected nanoporous structure. Figure 1 shows the synthesis of Pd/NPG wire biosensor. Firstly, the NPG wire with the 3D bicontinuous nanoporous structure was fabricated by electrochemical alloying/dealloying. Secondly, Pd was electrodeposited onto NPG wire to improve the electrocatalytic activity for DA. Figure 2a shows the surface topography of NPG wire fabricated from the smooth Au wire with a diameter of 200 μm, and the NPG wire was composed of Au ligaments and connected nanopores (Fig. S1). The SEM images indicated that the NPG wire possessed uniform 3D bicontinuous nanoporous structure and the ligaments of NPG with the characteristic length of ~200 nm (Fig. 2b). This unique architecture possesses large specific surface area which is beneficial for the load of the Pd. i Figure 2c provides the SEM image of Pd/NPG wire, indicating the 3D bicontinuous nanoporous structure still exists after decorated by Pd, and this structure is advantageous for the transport of DA and electron. As Fig. 2d shown, the Pd distributed along with the ligaments of NPG wire uniformly and directly, and possessed the characteristic length of ~20 nm (Fig. S1d). Figure S2 showed the distributing of Pd. The red color represents the distributing of Pd, and the black color represents the nanoporous channels of NPG. From the Fig. S2, we can know that the Pd distributed along with the ligaments of NPG wire uniformly. The direct connection avoided the additional contact resistance between Pd and NPG wire, improving the performance of the biosensor. The SCIENTIfIC REPOrTs \| 7: 7941 \| DOI:10.1038/s41598-017-07909-y 2 www.nature.com/scientificreports/ Figure 2. Microstructure characterization of the biosensor. (a) Low-magnification and (b) high-magnification SEM images of NPG wire fabricated by electrochemical alloying/dealloying, showing the 3D bicontinuous nanoporous structure. (c) Low-magnification and (d) high-magnification SEM images of Pd/NPG wire prepared by Pd electrodepositing on NPG wire, indicating the Pd distributing along the ligament of NPG wire uniformly. (e) Energy dispersive spectrum of Pd/NPG wire. Figure 2. Microstructure characterization of the biosensor. (a) Low-magnification and (b) high-magnification SEM images of NPG wire fabricated by electrochemical alloying/dealloying, showing the 3D bicontinuous nanoporous structure. (c) Low-magnification and (d) high-magnification SEM images of Pd/NPG wire prepared by Pd electrodepositing on NPG wire, indicating the Pd distributing along the ligament of NPG wire uniformly. Results and Discussion h i i f h (e) Energy dispersive spectrum of Pd/NPG wire. incorporation of Pd with the diameter of 10 nm (Fig. S1d) further enlarges the interface of the electrode and electrolyte, which is advantage for the oxidation of DA. Figure 2e shows the EDS spectrum of the Pd/NPG wire, indicating the existence of Pd and Au from the corresponding peaks of two elements. The percentage of Pd and Au is can be obtained from Fig. 2e, the percentage of weight is 18.23% and 81.77% respectively, and the percentage of atomic is 29.22% and 70.78% respectively. This method makes the electrode own additional superiority such as large specific surface area, fast transports of DA and excellent electrocatalytic for DA, on the basis of brilliant electrical conductivity of bulk Au, improving the sensing property of the biosensor. Electrochemical features of the Pd/NPG wire biosensor. The Fig. 3a shows the CV curves of the Pd/ NPG wire biosensor with varying scan rate from 70 to 250 mV/s in the solution of phosphate buffered saline (PBS) containing 100 μΜ DA to research the performance of the electrode. It is obvious that the anodic peak value of DA become more positive potentials along with the increase of scan rate. This indicates that there are maybe electropolymerization of DA and it is more distinct at higher scan rate, and show that a low scan rate, for example 100 mV/s, is suitable for the detection of DA. The anodic peak current (Ipa) of DA increases with the square root of scan rate linearly (Fig. 3b). The function about Ipa of DA versus the square root of scan rate obtained from Fig. 3b is SCIENTIfIC REPOrTs \| 7: 7941 \| DOI:10.1038/s41598-017-07909-y 3 www.nature.com/scientificreports/ Figure 3. Electrochemical characterazations of the Pd/NPG wire biosensor. (a) The CV curves of the Pd/NPG wire biosensor at various scan rates (from (i) to (v): 70, 100, 120, 150 and 250 mV/s). (b) The calibration plot of the oxidation peak current of DA versus the square root of scan rate (70–250 mV/s). (c) The CV curves of Pd/NPG wire biosensor in the PBS containing 0 μM and 100 μΜ DA at the scan rate of 100 mV/s. (d) The CV curves of NPG wire biosensor and Pd/NPG wire biosensor in the PBS containing 100 μM DA at the scan rate of 100 mV/s. Figure 3. Electrochemical characterazations of the Pd/NPG wire biosensor. Results and Discussion h i i f h (a) The CV curves of the Pd/NPG wire biosensor at various scan rates (from (i) to (v): 70, 100, 120, 150 and 250 mV/s). (b) The calibration plot of the oxidation peak current of DA versus the square root of scan rate (70–250 mV/s). (c) The CV curves of Pd/NPG wire biosensor in the PBS containing 0 μM and 100 μΜ DA at the scan rate of 100 mV/s. (d) The CV curves of NPG wire biosensor and Pd/NPG wire biosensor in the PBS containing 100 μM DA at the scan rate of 100 mV/s. = . × − . = . I scanrate R 18 63 ( ) 84 26 ( 0 998), (1) pa 1/2 2 (1) which indicates that the oxidation of DA was controlled by the diffusion of DA. The surface concentration of ionic species of the NPG wire and Pd/NPG wire electrodes can get from Brown–Anson model (equation. (2))39. = I n F Av RT /4 p 2 2 = I n F Av RT /4 (2) p 2 2 (2) where Ip represent the peak current, n represent the number of electrons transferred (n = 1), F represent the Faraday constant (F = 96485 C/mol), γ represent the surface concentration of ionic species on NPG wire and Pd/NPG wire electrodes (mol/cm2), A represent the surface area of the electrode (A = 0.1 cm2), v represent the scan rate (v = 100 mV/s), R represent the gas constant (R = 8.314 J/mol·k) and T represent the room temperature (T = 25 °C). It is noted that the surface concentration of ionic species on Pd/NPG wire electrode (3.16 × 10−9 mol/ cm2) is higher as compared to the NPG wire electrode (2.27 × 10−9 mol/cm2). The diffusion co-efficient value (D) of electrolyte solution to the NPG wire and Pd/NPG wire electrode surface can be got from Randles–Sevcik equation (equation. (3))39. where Ip represent the peak current, n represent the number of electrons transferred (n = 1), F represent the Faraday constant (F = 96485 C/mol), γ represent the surface concentration of ionic species on NPG wire and Pd/NPG wire electrodes (mol/cm2), A represent the surface area of the electrode (A = 0.1 cm2), v represent the scan rate (v = 100 mV/s), R represent the gas constant (R = 8.314 J/mol·k) and T represent the room temperature (T = 25 °C). Results and Discussion h i i f h (b) Plots of the anodic peak currents versus the concentrations of DA of Pd/NPG wire biosensor (1, 5, 10, 30, 50, 100, 150, 200, 220 μΜ) and NPG wire biosensor (30, 50, 100, 150, 200 μΜ). of Pd/NPG wire biosensor show there is no definite anodic peak in the electrolyte without DA, indicating there is not the redox reaction occurrence of DA. However, in the electrolyte containing 100 μM DA, an obvious anodic peak of DA appears at 0.4 V, which represents the DA oxidized to dopamine quinone. And this anodic peak of 0.4 V is lower than others report, such as Zhao et al. (0.45 V) and Roychoudhury (1.1 V), which is that Pd/NPG wire biosensor has more high catalytic activity than that Besides, there are two cathodic peak in 0.3 V and 0.1 V, which are dopamine quinone reduced to DA and 5,6-dihydroxylindole reduced to 5,6-indolequinone respectively (dopamine quinone can form 5,6-dihydroxylindole during the electropolymerization)40. By the comparison of the response current, it demonstrates the anodic peak appearing at 0.4 V is due to the oxidation of DA existing in electrolyte.i y To confirm the Pd/NPG wire biosensor owning higher catalytic activity for DA than NPG wire biosensor, NPG wire biosensor and Pd/NPG wire biosensor were used to detect 100 μΜ, 50 μΜ and 150 μΜ DA in the PBS. As shown in Figs 3d and S3, The oxidation peak of DA on the Pd/NPG wire biosensor is higher than that obtained from the NPG wire biosensor, and the onset potential of the Pd/NPG wire biosensor is more negative than that of NPG wire biosensor obviously. These is because of the Pd/NPG wire biosensor owning higher electrocatalytic activities than the NPG wire biosensor, and the higher electrocatalytic activities is obtained from the excellent electrocatalytic of Pd to DA. Therefore, inserting Pd into the NPG wire enhances the detection ability for DA dramatically. Also, the higher current can prove the Pd electrodeposited on the NPG wire successfully. In order to study the effect of the Pd loading on the oxidation of DA, we have synthesized samples with different Pd loading. As shown Fig. Results and Discussion h i i f h S4f, the curve of i, ii, iii, iv and v represent the CV curve of samples with different Pd loading, which was fabricated by CV after five, six, seven, eight and nine cyclic and the oxidation peak current of iii is higher than others, and it owns the best electrochemical performance comparing with others. Along with the increase of Pd (Fig. S4a–e), the active site of DA is increase, which is benefit for the oxidation of DA. However, when the load of Pd is too big, it will decrease the space of channel of the NPG, which will hinder the transport of DA and the oxidation of DA.h The interface characteristic of electrode and electrolyte is related to the property of the biosensor, and it was detected by EIS. Figure S5 provides the nyquist diagrams of electrochemical impedance spectra of the NPG wire biosensor and the Pd/NPG wire biosensor. The semicircular segment at higher frequencies represents the charge transfer resistance (Rct) and the linear part at lower frequencies represents the diffusion process. As shown in Fig. S5, the Pd/NPG wire biosensor possesses the smaller semicircle diameter of the EIS, compared to the EIS of the NPG wire biosensor, indicating the decreased resistance. The incorporation of Pd decreases the impedance of the NPG wire biosensor. It demonstrates the formation of Pd/NPG wire biosensor again. Electrochemical analysis of DA on Pd/NPG wire biosensor. To estimate the detection property of Pd/NPG wire biosensor for DA, the detection tests are performed by DPV. Figure 4a shows the DPV curves of various concentrations of DA in the electrolyte of PBS. The anodic peak current of DPV increases along with the change of concentration from 1 to 220 μΜ, which is ascribed to the oxidation of more DA. The more detailed information of this change can learn from the calibration plot of Pd/NPG wire biosensor (Fig. 4b), and the line- arity relationship is = . × + . = . . I concentration R 1 19 23 32 ( 0 996) (5) pa 2 (5) The sensitivity of the Pd/NPG wire biosensor can obtain from the slope of calibration plot and it is 1.19 μA μΜ−1. Also, the detection range and the detection limit of the Pd/NPG wire biosensor is 1–220 μM and 1 μM respec- tively. For comparison, the calibration plot of NPG wire biosensor also was shown in Fig. Results and Discussion h i i f h It is noted that the surface concentration of ionic species on Pd/NPG wire electrode (3.16 × 10−9 mol/ cm2) is higher as compared to the NPG wire electrode (2.27 × 10−9 mol/cm2). The diffusion co-efficient value (D) of electrolyte solution to the NPG wire and Pd/NPG wire electrode surface can be got from Randles–Sevcik equation (equation. (3))39. = . × I n AD Cv (2 69 10 ) ) (3) p 5 3/2 1/2 1/2 (3) where, D represent the diffusion co-efficient and C represent the surface concentration in mol (C = 5 mM). The diffusion co-efficient value of Pd/NPG wire electrode (4.85 × 10−7 cm2/s) is higher than that of NPG elec- trode (2.57 × 10−7 cm2/s). The electroactive surface area (Ae) of Pd/NPG wire electrode has been obtained using Randles–Sevcik equation (equation. (4)) and calculated diffusion co-efficient value. = . × A S n CD /(2 99 10 ) (4) e 5 3/2 1/2 = . × A S n CD /(2 99 10 ) e 5 3/2 1/2 (4) where, S represents the slope of straight line obtained from equation. (1). The Ae value for Pd/NPG electrode has een estimated as 34.38 mm2.i where, S represents the slope of straight line obtained from equation. (1). The Ae value for Pd/NPG electrode has been estimated as 34.38 mm2. T fi h f DA d Pd/NPG b h d f d h To confirm the occurrence of DA redox reaction on Pd/NPG wire biosensor, the study was performed in the BS with 0 μM and 100 μM DA by the method of CV on Pd/NPG wire biosensor. As shown in Fig. 3c, the curve SCIENTIfIC REPOrTs \| 7: 7941 \| DOI:10.1038/s41598-017-07909-y 4 www.nature.com/scientificreports/ Figure 4. Detection of dopamine. (a) DPV curves of the Pd/NPG wire biosensor at various concentrations of DA (from 1 to 220 μΜ). (b) Plots of the anodic peak currents versus the concentrations of DA of Pd/NPG wire biosensor (1, 5, 10, 30, 50, 100, 150, 200, 220 μΜ) and NPG wire biosensor (30, 50, 100, 150, 200 μΜ). Figure 4. Detection of dopamine. (a) DPV curves of the Pd/NPG wire biosensor at various concentrations of DA (from 1 to 220 μΜ). Results and Discussion h i i f h S7, the response currents did not change obviously, and it was less than 9% (RSD value ~3–10%) (Table S2), with the interferences of AA (0.05 mM and 0.5 mM), UA (0.3 mM and 3 mM), NE (0.1 mM and 1 mM), EP (0.6 mM and 6 mM) and CC (0.5 mM and 5 mM)). This own to the efficient catalysis of Pd to DA at 0.4 V. The high selectivity makes this biosensor own the excellent reliability for detection of DA in blood serums. h g y y Considering the cost of gold, the repeated use of NPG wire form the composite of Pd/NPG wire biosensor is a feasible idea. The NPG wire can be obtained by immersing the composite of Pd/NPG wire biosensor into the concentrated HNO3 for 2 hour. The CV curves of Fig. 5b, shows the peak currents are 0.1029 mA and 0.09894 mA respectively, and that the response current retained 96.15% after recycle comparing with before recycle. Figure S8 shows the microstructure feature of NPG wire and Pd/NPG wire after recycle, and the 3D bicontinuous nanop- orous structure still exist, which also proves the superior reusability. The excellent reusability may be come from the good stability of NPG wire. g y The stability (Fig. 6a) of Pd/NPG wire biosensor was studied in the PBS containing 100 μΜ DA. The Fig. 6b shows CV curves of the 4th, 7th, 10th, 25th, 40th, 52th measurements. Figure 6b shows the response current of the 4th measurements still retain 100%, and the 40th measurements retain 94.23%, compared to the first time, which proves this biosensor of DA owning excellent stability at the first 40 times. However, the 52th not only has obvious decrease, but also has a clear shift of the dopamine oxidation process towards more positive potentials. This is because of dopamine being electropolymerized on the surface of the electrode, and shows that the use time should not over 40 times. Besides, the RSD is 9.74% of the 40th measurements. The solid support preventing the aggregation of Pd, may be the primary reason for brilliant stability of this biosensor. In order to ensure the reliability of the biosensor, the property of reproducibility is also need to examine. The reproducibility of this biosensor was studied by five biosensors fabricated at the same condition to test in the solution containing 100 μΜ DA. As show of Fig. 6c and Fig. Results and Discussion h i i f h 4b, the detection range is from 30 to 200 μΜ, and the detection range of Pd/NPG wire biosensor is much wider than NPG wire biosensor. Owing to the incorporation of Pd, Pd/NPG wire biosensor possesses higher catalytic activity than NPG wire SCIENTIfIC REPOrTs \| 7: 7941 \| DOI:10.1038/s41598-017-07909-y 5 www.nature.com/scientificreports/ www.nature.com/scientificreports/ Figure 5. Specificity and reusability of the Pd/NPG wire biosensor. (a) Response currents of the Pd/NPG wire biosensor in the solution of 100 μΜ DA in the presence of AA, UA, NE, EP and CC. (b) The CV curves of Pd/ NPG wire biosensor before and after recycle. Figure 5. Specificity and reusability of the Pd/NPG wire biosensor. (a) Response currents of the Pd/NPG wire biosensor in the solution of 100 μΜ DA in the presence of AA, UA, NE, EP and CC. (b) The CV curves of Pd/ NPG wire biosensor before and after recycle. biosensor for DA, and exhibits wider detection range. Besides, the current-time curves shown that the Pd/NPG wire biosensor owned excellent stability and short response time (Fig. S6).h y p g The superiority of this biosensor to others can be showed by the comparison with others (Table S1), in the range, sensitivity and limit of the detection. The sensitivity (1.19 μA μΜ−1) and the detection range (1–220 μΜ) of this work is superior to other reported such as Hu et al. (0.33 μA μΜ−1 and 1–6 μΜ), Bao et al. (0.27 μA μΜ−1 and 1–14 μΜ), Sheng et al. (0.03 μA μΜ−1 and 0.5–170 μΜ), Roychoudhury et al. (0.06 μA μΜ−1 and 2–100 μΜ), Yang et al. (0.48 μA μΜ−1 and 0.5–60 μΜ) and Yang et al. (1.04 μA μΜ−1 and 1–50 μΜ). Although the sensitivity of Yang et al. reported the sensitivity (2.31 μA μΜ−1) is higher than this work, the detection range (1–50 μΜ) is poor. The comparison reveals the high comprehensive property of the Pd/NPG wire biosensor, and proves the potentials application in the detection of DA.hi p pp The specificity of the Pd/NPG wire biosensor was studied by detection the DA with the concentration of 100 μΜ, in the presence of AA, UA, NE, EP and CC which are co-existed with DA in physiological fluids10. As shown of Fig. 5a and Fig. Conclusion A novel nonenzymatic electrochemical biosensor based on the Pd/NPG wire was fabricated for the detection of DA. The synergistic effects of the excellent electrocatalysis of Pd for DA and the 3D self-supporting bicontinuous nanoporous structure of NPG wire greatly enhance the response current and improve the electrochemical signal. Therefore, this biosensor possessing high sensitivity, broad detection range and excellent selectivity is promising for applications of DA in diagnosis and prevention. Results and Discussion h i i f h Serum sample analysis and the comparison with standard samples (n = 5). Results and Discussion h i i f h S9, the relative standard deviation (RSD) of results was 2.81%, indicating this biosensor possessing a fascinating reproducibility. The fascinating reproducibility is come from the superiority of electrochemical alloying/dealloying and electrochemical deposition.l y g y g p In order to demonstrate interference of other electrochemical active substances existing in biological fluids during the detection of DA, and estimate the performance of Pd/NPG wire biosensor in the application of diag- nosis and prevention, the assay of DA in fetal bovine serum samples is necessary. Briefly, diverse concentration of DA (10, 20, 30, 40, 50 μΜ) were added into the fetal bovine serum samples diluted 15 times by PBS (pH 7), comparing the response current with the response current of standard samples. The Table 1 shows that the results of the detection in fetal bovine serum samples and standard samples have a good accordance, and the variation range of recoveries was 97.3% to 103.8%. This result shows the Pd/NPG wire biosensor has a good anti-interference performance for the electrochemical activity substance, for example protein existing in biolog- ical fluids. The protein content which exist in fetal bovine serum sample is much large than in body fluids, so Pd/ NPG wire biosensor is promising to been used in the detection of DA in the area of diagnosis and prevention. SCIENTIfIC REPOrTs \| 7: 7941 \| DOI:10.1038/s41598-017-07909-y 6 www.nature.com/scientificreports/ Figure 6. Stability and reproducibility of the Pd/NPG wire biosensor. (a) Response currents of the Pd/ NPG wire biosensor under different measurements. (b) Stability study of the Pd/NPG wire biosensor. (c) Reproducibility study on Pd/NPG wire biosensor. Figure 6. Stability and reproducibility of the Pd/NPG wire biosensor. (a) Response currents of the Pd/ NPG wire biosensor under different measurements. (b) Stability study of the Pd/NPG wire biosensor. (c) Reproducibility study on Pd/NPG wire biosensor. Number Labeled (μΜ) Detected (μΜ) Recovery (%) RDS (%) 1 10.00 9.81± 98.10 4.61 2 20.00 19.49± 97.50 3.81 3 30.00 30.37± 101.20 3.52 4 40.00 41.53± 103.80 5.42 5 50.00 48.67± 97.30 4.76 Table 1. Serum sample analysis and the comparison with standard samples (n = 5). Number Labeled (μΜ) Detected (μΜ) Recovery (%) RDS (%) 1 10.00 9.81± 98.10 4.61 2 20.00 19.49± 97.50 3.81 3 30.00 30.37± 101.20 3.52 4 40.00 41.53± 103.80 5.42 5 50.00 48.67± 97.30 4.76 Table 1. Serum sample analysis and the comparison with standard samples (n = 5). Table 1. p Materials and reagents. p Materials and reagents. Gold wires with the diameter of 200 μm (purity of 99.99%) were purchased from Changshu Changhong Precious Metal Co., Ltd. (Jiangsu, China). Palladium dichloride (PdCl2), benzyl alcohol (BA), zinc chloride (ZnCl2), sodium dodecyl sulfate (SDS), DA, AA, UA, NE, EP, CC, potassium ferricyanide (K3Fe(CN)6), potassium ferrocyanide (K4Fe(CN)6), and potassium chloride (KCl) were purchased from Aladdin Reagent Company (Shanghai, China). Phosphate buffered saline was purchased from Shanghai Double-Helix Biotech co., Ltd. All other chemical reagents were analytical reagents grade and directly used without further purification. The length of all used gold wire is 1.5 cm, and the surface area is 0.09 cm2. Preparation of NPG wire. The NPG wire were fabricated by the method of electrochemical alloying/deal- loying, with a three electrode electrochemical system at the temperature of 120 °C in ZnCl2 solution41. Firstly, the alloy of Zn-Au was formed by electrochemical alloying. Then, the Zn atom was dissolved by electrochemical deal- loying, and the Au atom without any loss. The auxiliary electrode, the reference electrode and the working elec- trode were Zn plate, Zn wire and gold wire with diameters of 200 μm respectively. The electrolyte was prepared by SCIENTIfIC REPOrTs \| 7: 7941 \| DOI:10.1038/s41598-017-07909-y 7 www.nature.com/scientificreports/ heating the mixed solution of BA containing 1.5 M ZnCl2 to 80 °C stirring for several hours. The potential range used in the working electrode was from −0.72 to 1.8 V (vs. Zn), under the scan rate of 7 mV·s–1. After fifty cyclic, the NPG wire was fabricated with appropriate characteristic length. heating the mixed solution of BA containing 1.5 M ZnCl2 to 80 °C stirring for several hours. The potential range used in the working electrode was from −0.72 to 1.8 V (vs. Zn), under the scan rate of 7 mV·s–1. After fifty cyclic, the NPG wire was fabricated with appropriate characteristic length. Preparation of Pd/NPG wire. Pd was electrodeposited onto NPG wire according to certain literature meth- ods42. Briefly, Pd/NPG wire was formed by electrochemical deposition, in the mixed solution of 5 mM SDS and 2.5 mM palladium chloride. The auxiliary electrode, the reference electrode and the working electrode were plat- inum wire, saturated calomel electrode and NPG wire respectively. The potential range applied to the working electrode was from 0.2 to 1.2 V (vs. SCE) at the scanning rate of 20 mV s−1. p Materials and reagents. After certain cyclic, the Pd was depos- ited on the ligament of the NPG wire with appropriate thickness. Instruments and measurements. The scanning electron microscope (SEM) images and energy dispersive x-ray spectra (EDS) were obtained from a field emission scanning electron microscopy (FE-SEM, ZEISS Ultra 55, Germany). All electrochemistry measurements were carried out on an electrochemical workstation (Zennium Zahner Germany), with a three-electrode system: modified NPG wire electrode as the working electrode, a satu- rated calomel electrode (SCE) as the reference electrode, and a platinum wire electrode as the counter electrode. CV and DPV were performed in the electrolyte of phosphate buffered saline. Real sample analysis was performed in the solution of fetal bovine serum. The specificity of the biosensor was studied with the presence of AA, UA, NE, EP and CC. EIS was performed in the electrolyte of 5.0 mM [Fe(CN)6]3−/4−. All measurements were per- formed at room temperature. References 1. Kramer, S. G. Dopamine: a retinal neurotransmitter I. Retinal uptake, storage, and light-stimulated release of H3-dopamine in vivo. Investigative ophthalmology & visual science 10, 438–452 (1971).l 1. Kramer, S. G. Dopamine: a retinal neurotransmitter I. Retinal uptake, storage, and light-stimulated release of H3-dopamine in vivo. Investigative ophthalmology & visual science 10, 438–452 (1971). g p gy 2. Basu, S. & Dasgupta, P. S. Dopamine, a neurotransmitter, influences the immune system. Journal of neuroimmunology 102, 113–124 (2000). ( ) 3. Sansuk, S. et al. Ultrasensitive detection of dopamine using a carbon nanotube network microfluidic flow electrode. Analytica chemistry 85, 163–169 (2012). 4. Govindaraju, S., Ankireddy, S. R., Viswanath, B., Kim, J. & Yun, K. Fluorescent Gold Nanoclusters for Selective Detection of Dopamine in Cerebrospinal fluid. Scientific Reports 7 (2017).ff p pl fi p 5. Gagnon, D. et al. Striatal Neurons Expressing D-1 and D-2 Receptors are Morphologically Distinct and Differently Affected by Dopamine Denervation in Mice. Scientific Reports 7 (2017). p fi p ( ) 6. Lee, Y. H., Lin, Y. C., Feng, C. H., Tseng, W. L. & Lu, C. Y. A derivatization-enhanced detection strategy in mass spectrometry analysis of 4-hydroxybenzoates and their metabolites after keratinocytes are exposed to UV radiation. Scientific Reports 7 (2017). y y yt y p fi p ( ) 7. El‐Beqqali, A., Kussak, A. & Abdel‐Rehim, M. Determination of dopamine and serotonin in human urine samples utilizing microextraction online with liquid chromatography/electrospray tandem mass spectrometry. Journal of separation science 30, 421–424 (2007).li 8. Zhang, C. P. et al. Continuous fabrication of nanostructure arrays for flexible surface enhanced Raman scattering substrate. Scie Reports 7 (2017). p 9. Roychoudhury, A., Basu, S. & Jha, S. K. Dopamine biosensor based on surface functionalized nanostructured nickel oxide platform. Biosensors and Bioelectronics 84, 72–81 (2016). 10. Zhang, M., Gong, K., Zhang, H. & Mao, L. Layer-by-layer assembled carbon nanotubes for selective determination of dopami the presence of ascorbic acid. Biosensors and Bioelectronics 20, 1270–1276 (2005).l 11. Fotopoulou, M. A. & Ioannou, P. C. Post-column terbium complexation and sensitized fluorescence detection for the determination of norepinephrine, epinephrine and dopamine using high-performance liquid chromatography. Analytica Chimica Acta 462, 179–185 (2002). ( ) 12. Lv, M. et al. Electrochemical Detection of Catechol Based on As-Grown and Nanograss Array Boron-Doped Diamond Electrodes. Electroanalysis 22, 199–203 (2010). y 3. Gao, Z. D., Qu, Y. F., Li, T. T., Shrestha, N. K. & Song, Y. SCIENTIfIC REPOrTs \| 7: 7941 \| DOI:10.1038/s41598-017-07909-y Acknowledgementsh g This work was supported by the National High Technology Research and Development Program of China (863 Program) (Grant No. 2015AA033502), the National Natural Science Foundation of China (Grant Nos. 51372087, 51232002, 51541201), Science and Technology Planning Project of Guangdong Province, China (Grant No. 2014A010105048), the Natural Science Foundation of Guangdong Province (Grant No. 2015A030313493, 2016A030308014) and State Key Laboratory for Mechanical Behavior of Materials, China (Grant No. 20141607). www.nature.com/scientificreports/ Microchimica Acta 180, 1037–1042 (2013).fi p p p 9. Singh, J. et al. Highly Efficient Bienzyme Functionalized Biocompatible Nanostructured Nickel Ferrite-Chitosan Nanocomposite Platform for Biomedical Application. Journal of Physical Chemistry C 117, 8491–8502 (2013).f pp f y y 40. Bernsmann, F., Voegel, J. C. & Ball, V. Different synthesis methods allow to tune the permeability and permselectivity of dopamine- melanin films to electrochemical probes. Electrochimica Acta 56, 3914–3919 (2011).i 40. Bernsmann, F., Voegel, J. C. & Ball, V. Different synthesis methods allow to tune the perm melanin films to electrochemical probes. Electrochimica Acta 56, 3914–3919 (2011). i 41. Yu, C. F., Jia, F. L., Ai, Z. H. & Zhang, L. Z. Direct oxidation of methanol on self-supported nanoporous gold film electrodes with high catalytic activity and stability. Chemistry of Materials 19, 6065–6067 (2007). i Yu, C. F., Jia, F. L., Ai, Z. H. & Zhang, L. Z. Direct oxidation of methan y y y y f ( ) 2. Wang, X. et al. Simultaneous electrochemical determination of ascorbic acid, dopamine and uric acid using a palladium nanoparticle/graphene/chitosan modified electrode. Journal of Electroanalytical Chemistry 695, 10–16 (2013). Author Contributions C.Y.N. contributed to this work. X.Y., P.Y., L.Z., Z.N.Z., Z.G.W., G.X.T., and C.Y.N. conceived and designed the experiments. X.Y., Y.X.W. and J.S.P. performed the experiments. X.Y., and Y.X.W. performed the image processing procedures. All authors contributed to the written manuscript and have given approval to the final version of the manuscript. www.nature.com/scientificreports/ 8. Campuzano, S. & Wang, J. Nanobioelectroanalysis based on carbon/inorganic hybrid nanoarchitectures. Electroanalysis 23 1289–1300 (2011). 1289–1300 (2011). 29. Yan, N. et al. Fe2O3 Nanoparticles Wrapped in Multi-walled Carbon Nanotubes With Enhanced Lithium Storage Capability. S i tifi R p t 3 (2013) 9. Yan, N. et al. Fe2O3 Nanoparticles Wrapped in Multi-walled Carbon Nanotubes With Enhanced Lithium Storage Capability Scientific Reports 3 (2013). fi p ( ) 30. Léonard, F. & Talin, A. A. Electrical contacts to one-and two-dimensional nanomaterials. Nature nanotechnology 6, 773–783 (2011). 31. Gowda, S. R., Leela Mohana Reddy, A., Zhan, X., Jafry, H. R. & Ajayan, P. M. 3D nanoporous nanowire current collectors for thin film microbatteries. Nano letters 12, 1198–1202 (2012). fi p ( ) éonard, F. & Talin, A. A. Electrical contacts to one-and two-dimensional nanomaterials. Nature nanotechnology 6, 773–783 (2011). owda S R Leela Mohana Reddy A Zhan X Jafry H R & Ajayan P M 3D nanoporous nanowire current collectors for thin fi p ( ) 0. Léonard, F. & Talin, A. A. Electrical contacts to one-and two-dimensional nanomaterials. Nature nanotechnology 6, 773–783 (2011) , , gy , ( 31. Gowda, S. R., Leela Mohana Reddy, A., Zhan, X., Jafry, H. R. & Ajayan, P. M. 3D nanoporous nanowire current collectors for film microbatteries. Nano letters 12, 1198–1202 (2012). i 32. Lang, X.-Y. et al. Nanoporous gold supported cobalt oxide microelectrodes as high-performance electrochemical biosensors. Nature communications 4 (2013). 3. Han, G.-F. et al. Self-grown Ni (OH) 2 layer on bimodal nanoporous AuNi alloys for enhanced electrocatalytic activity and stability ACS applied materials & interfaces 6, 16966–16973 (2014). pp f 34. Chen, L. et al. Toward the theoretical capacitance of RuO2 reinforced by highly conductive nanoporous gold. Advanced Energy Materials 3, 851–856 (2013).h 5. Meng, F. & Ding, Y. Sub‐Micrometer‐Thick All‐Solid‐State Supercapacitors with High Power and Energy Densities. Advanced Materials 23, 4098–4102 (2011). 6. Lang, X., Hirata, A., Fujita, T. & Chen, M. Nanoporous metal/oxide hybrid electrodes for electrochemical supercapacitors. Nature Nanotechnology 6, 232–236 (2011).i gy ( ) 37. Tsierkezos, N. G. et al. Electrochemical analysis of ascorbic acid, dopamine, and uric acid on nobel metal modified nitrogen-doped carbon nanotubes. Sensors and Actuators B: Chemical 231, 218–229 (2016).i 38. Palanisamy, S., Ku, S. & Chen, S.-M. Dopamine sensor based on a glassy carbon electrode modified with a reduced graphene oxide and palladium nanoparticles composite. References Y. Development of Amperometric Glucose Biosensor Based on Prussian Blue Functionlized TiO2 Nanotube Arrays. Scientific Reports 4 (2014). y fi 4. Li, Z. Z., Chen, Y., Xin, Y. M. & Zhang, Z. H. Sensitive electrochemical nonenzymatic glucose sensing based on anodized CuO nanowires on three-dimensional porous copper foam. Scientific Reports 5 (2015). p pp fi p ( ) 15. Hsu, Y.-W. et al. Synthesis of CuO/graphene nanocomposites for nonenzymatic electrochemical glucose biosensor applications. Electrochimica Acta 82, 152–157 (2012).f 6. Kimmel, D. W., LeBlanc, G., Meschievitz, M. E. & Cliffel, D. E. Electrochemical sensors and biosensors. Analytical Chemistry 84 685–707 (2011).h 17. Wang, J., Thomas, D. F. & Chen, A. Nonenzymatic electrochemical glucose sensor based on nanoporous PtPb networks. Analytical Chemistry 80, 997–1004 (2008). y 8. Zhang, H. J., Wang, L. Q., Lu, L. L. & Toshima, N. Preparation and Catalytic Activity for Aerobic Glucose Oxidation of Crown Jewe Structured Pt/Au Bimetallic Nanoclusters. Scientific Reports 6 (2016). fi p 9. Wang, H. et al. Label-free immunosensor based on Pd nanoplates for amperometric immunoassay of alpha-fetoprotein. Biosensor and Bioelectronics 53, 305–309 (2014). 20. Gu, H.-Y., Yu, A.-M. & Chen, H.-Y. Direct electron transfer and characterization of hemoglobin immobilized on a Au colloid–cysteamine-modified gold electrode. Journal of Electroanalytical Chemistry 516, 119–126 (2001). yi g f y y 21. Chen, X. et al. Fabrication of Au-Ag nanocage@NaYF4@NaYF4: Yb,Er Core-Shell Hybrid and its Tunable Upconversion Enhancement. Scientific Reports 7 (2017).i fi p 2. Chen, S. & Kimura, K. Synthesis and characterization of carboxylate-modified gold nanoparticle powders dispersible in water Langmuir 15, 1075–1082 (1999).i g 23. Kros, A., Nolte, R. J. & Sommerdijk, N. Conducting polymers with confined dimensions: Track-etch membranes for amperometric biosensor applications. Advanced Materials 14, 1779–1782 (2002). 24. Besteman, K., Lee, J.-O., Wiertz, F. G., Heering, H. A. & Dekker, C. Enzyme-coated carbon nanotubes as single-molecule biosensors. Nano letters 3, 727–730 (2003). ( ) 25. Willner, I. & Willner, B. Biomolecule-based nanomaterials and nanostructures. Nano letters 10, 3805–3815 (2010). 26. Ye, D., Luo, L., Ding, Y., Liu, B. & Liu, X. Fabrication of Co 3 O 4 nanoparticles-decorated graphene composite for determination of L-tryptophan. Analyst 137, 2840–2845 (2012). yp p y 7. He, M. S. et al. Chiral-Selective Growth of Single-Walled Carbon Nanotubes on Lattice-Mismatched Epitaxial Cobalt Nanoparticles Scientific Reports 3 (2013). SCIENTIfIC REPOrTs \| 7: 7941 \| DOI:10.1038/s41598-017-07909-y 8 www.nature.com/scientificreports/ Additional Information Supplementary information accompanies this paper at doi:10.1038/s41598-017-07909-y Competing Interests: The authors declare that they have no competing interests. Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps an institutional affiliations. Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre- ative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not per- mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2017 SCIENTIfIC REPOrTs \| 7: 7941 \| DOI:10.1038/s41598-017-07909-y 9
https://openalex.org/W2804411129	https://www.dora.lib4ri.ch/psi/islandora/object/psi%3A4806/datastream/PDF/Chen-2018-Magnetic_hysteresis_in_self-assembled_monolayers-%28published_version%29.pdf	English	null	Magnetic hysteresis in self-assembled monolayers of Dy-fullerene single molecule magnets on gold	Nanoscale	2,018	cc-by	5,695	Cite this: Nanoscale, 2018, 10, 11287 Cite this: Nanoscale, 2018, 10, 11287 Received 18th January 2018, Accepted 22nd May 2018 DOI: 10 1039/c8nr00511g Received 18th January 2018, Accepted 22nd May 2018 DOI: 10.1039/c8nr00511g Fullerene single molecule magnets (SMMs) DySc2N@C80 and Dy2ScN@C80 are functionalized via a 1,3-dipolar cycloaddition with surface-anchoring thioether groups. The SMM properties of Dy-fullerenes are substantially aﬀected by the cycloaddition. Submonolayers of the physisorbed derivatives exhibit magnetic hysteresis on an Au(111) surface at 2 K as revealed by X-ray mag- netic circular dichroism. ing procedure. Functionalization of SMM molecules with surface-anchoring groups and formation of SAMs on metals has previously been studied for Mn12,7 TbPc2,8 Fe4,4a,b,9 and Fe3Cr.10 SAMs of Fe4 were the first SMMs to exhibit magnetic bistability on a metallic substrate.4a,b In this work, we explore the SAM approach for the depo- sition of magnetic fullerenes onto an Au(111) surface and study the magnetic properties of the derivatives and SAMs. Dy- based endohedral metallofullerenes (EMFs) have been found to be robust SMMs, with high anisotropy barriers or giant exchange interactions, and reasonably high blocking tempera- ture of magnetization in bulk materials.11 The reports on EMF-SAMs are limited so far to non-SMM Er3N@C80 and La@C82,12 but SAMs of empty fullerenes are relatively well known and have been studied in great detail.13 An STM study of a series of C60-based SAMs on Au(111) with thiol and thioether anchoring groups revealed that the latter provided physisorbed monolayers with more ordered packing of mole- cules than in chemisorbed thiol derivatives.14 Molecules with a bistable magnetic ground state and a slow relaxation of magnetization are known as single molecule magnets (SMMs).1 Information storage or spintronic appli- cations envisaged for SMMs require contacting the molecules to electrodes, and observing whether hysteresis with remanence is retained on metal is important for the electrically readable magnetic bits. The magnetic properties of SMMs on conduct- ing surfaces should therefore be well understood. However, this problem meets with serious diﬃculties in both bringing the molecules to metallic substrates and in studying the magnetic properties of monolayers.2 As a result, the number of molecules that have been studied on diﬀerent surfaces is very low compared to the number of published SMMs, and hysteresis of magnetization in monolayers deposited on metals has been observed so far only for TbPc2,3 Fe4,4 and Dy2ScN@C80.5 DySc2N@C80-Ih (1) and Dy2ScN@C80-Ih (2) were obtained as described before.11f,15 EMFs were functionalized with a thioether –S–CH3 group via 1,3-dipolar cycloaddition (Fig. †Electronic supplementary information (ESI) available. See DOI: 10.1039/ c8nr00511g aLeibniz Institute for Solid State and Materials Research (IFW), D-01069 Dresden, Germany. E-mail: a.popov@ifw-dresden.de bThe Division of Synchrotron Radiation Research, Lund University, SE-22100 Lund, Sweden cSwiss Light Source, Paul Scherrer Institute, CH-5232 Villigen PSI, Switzerland Open Access Article. Published on 23 May 2018. Downloaded on 2/12/2019 8:52:01 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Cite this: Nanoscale, 2018, 10, 11287 Cite this: Nanoscale, 2018, 10, 11287 (b) Magnetization curves of 1-R and 1 at T = 2 K; the inset shows determination of the blocking temperatures of magnetiza- tion TB, sweep rate 5 K min−1. (c) Magnetization curves of 2-R at tem- peratures 1.8–5 K, sweep rate 2.9 mT s−1; (d) magnetization curves of 2-R and 2 at T = 2 K. The inset shows determination of TB, sweep rate 5 K min−1. Open Access Article. Published on 23 May 2018. Downloaded on 2/12/2019 8:52:01 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Fig. 2 (a) Magnetization curves of 1-R at temperatures 2–8 K, sweep rate 2.9 mT s−1. (b) Magnetization curves of 1-R and 1 at T = 2 K; the inset shows determination of the blocking temperatures of magnetiza- tion TB, sweep rate 5 K min−1. (c) Magnetization curves of 2-R at tem- peratures 1.8–5 K, sweep rate 2.9 mT s−1; (d) magnetization curves of 2-R and 2 at T = 2 K. The inset shows determination of TB, sweep rate 5 K min−1. closes above 8 K, which is somewhat higher than in the pris- tine 1. Similarly, the blocking temperature of magnetization (TB), determined as a position of the peak in the magnetic sus- ceptibility of a zero-field cooled sample, is 8 K for 1-R and 7 K for 1 11g (Fig. 2b). The relaxation times of 1 and 1-R measured in a finite field of 0.2 T between 1.8 and 5 K are similar and vary from ca. 100 s near 5 K to 104 s near 2 K (Tables S1 and S2†). The temperature trend of relaxation times is slightly diﬀerent for 1 and 1-R, the latter showing a steeper tempera- ture decay (Fig. S6†). To summarize, based on the magnetiza- tion behaviour and relaxation times, we conclude that cyclo- addition improves low-temperature SMM properties of 1 (a similar eﬀect was also reported in ref. 18). Fig. 1 (a) Scheme of a Prato reaction to obtain EMF-R derivatives (EMF = Sc3N@C80, DySc2N@C80 (1), and Dy2ScN@C80 (2)); (b) HPLC curves measured at the end of the reaction (grey traces) and for puriﬁed derivatives (coloured curves). The absence of the pristine fullerenes (HPLC peaks near 60 min) in the puriﬁed derivatives can be clearly seen; (c) Vis-NIR absorption spectra of 1-R and 2-R in toluene. Insets show MALDI mass-spectra (1,1,4,4-tetraphenyl-1,3-butadiene as a matrix). Cite this: Nanoscale, 2018, 10, 11287 1) adopting a procedure developed for C60-SAMs.14 The optimal reaction time of 20 min was first determined for Sc3N@C80 (longer times resulted in the formation of bis-adducts) and then used for 1. 2 exhibited lower reactivity and was reacted for 40 minutes. Fig. 1b shows HPLC traces of the reaction mix- tures, as well as HPLC traces of isolated derivatives (denoted 1-R and 2-R hereafter). A partial formation of simple fullero- pyrrolidino adducts was also observed (denoted by asterisks in Fig. 1b). Evaporation is the most popular way to obtain monolayers of SMMs. Unfortunately, only a limited number of SMMs are sublimable and retain their structural integrity on a surfa- ce.4c,5,6 Chemical deposition from solution via covalent bonding to the substrate, such as realized in a self-assembled monolayer (SAM) approach, is a viable alternative to vapour deposition. It avoids the thermal stability problem and allows the formation of monolayers in a comparably simple self-limit- Prato addition to M3N@C80-Ih nitride clusterfullerenes with Sc3N and mixed-metal clusters gives predominantly [5,6]-fuller- opyrrolidino cycloadducts (i.e the addition proceeds across a pentagon/hexagon edge).16 The structural identity of isolated 1-R and 2-R as [5,6]-monoadducts was established by MALDI and vis-NIR absorption spectroscopy (Fig. 1c and d). For Sc3N@C80-R, the molecular structure was additionally con- firmed by 1D and 2D 1H NMR spectroscopy (Fig. S1–S3†). Nanoscale, 2018, 10, 11287–11292 \| 11287 This journal is © The Royal Society of Chemistry 2018 View Article Online Fig. 1 (a) Scheme of a Prato reaction to obtain EMF-R derivatives (EMF = Sc3N@C80, DySc2N@C80 (1), and Dy2ScN@C80 (2)); (b) HPLC curves measured at the end of the reaction (grey traces) and for puriﬁed derivatives (coloured curves). The absence of the pristine fullerenes (HPLC peaks near 60 min) in the puriﬁed derivatives can be clearly seen; (c) Vis-NIR absorption spectra of 1-R and 2-R in toluene. Insets show MALDI mass-spectra (1,1,4,4-tetraphenyl-1,3-butadiene as a matrix). Despite considerable fragmentation to pristine fullerenes, molecular ions of 1-R and 2-R can be clearly seen. Communication Open Access Article. Published on 23 May 2018. Downloaded on 2/12/2019 8:52:01 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Communication Fig. 2 (a) Magnetization curves of 1-R at temperatures 2–8 K, sweep rate 2.9 mT s−1. (b) Magnetization curves of 1-R and 1 at T = 2 K; the inset shows determination of the blocking temperatures of magnetiza- tion TB, sweep rate 5 K min−1. 11288 \| Nanoscale, 2018, 10, 11287–11292 This journal is © The Royal Society of Chemistry 2018 Cite this: Nanoscale, 2018, 10, 11287 (c) Magnetization curves of 2-R at tem- peratures 1.8–5 K, sweep rate 2.9 mT s−1; (d) magnetization curves of 2-R and 2 at T = 2 K. The inset shows determination of TB, sweep rate 5 K min−1. Nanoscale closes above 8 K, which is somewhat higher than in the pris- tine 1. Similarly, the blocking temperature of magnetization (TB), determined as a position of the peak in the magnetic sus- ceptibility of a zero-field cooled sample, is 8 K for 1-R and 7 K for 1 11g (Fig. 2b). The relaxation times of 1 and 1-R measured in a finite field of 0.2 T between 1.8 and 5 K are similar and vary from ca. 100 s near 5 K to 104 s near 2 K (Tables S1 and S2†). The temperature trend of relaxation times is slightly diﬀerent for 1 and 1-R, the latter showing a steeper tempera- ture decay (Fig. S6†). To summarize, based on the magnetiza- i b h i d l i i l d h l Fig. 1 (a) Scheme of a Prato reaction to obtain EMF-R derivatives (EMF = Sc3N@C80, DySc2N@C80 (1), and Dy2ScN@C80 (2)); (b) HPLC curves measured at the end of the reaction (grey traces) and for puriﬁed derivatives (coloured curves). The absence of the pristine fullerenes (HPLC peaks near 60 min) in the puriﬁed derivatives can be clearly seen; (c) Vis-NIR absorption spectra of 1-R and 2-R in toluene. Insets show MALDI mass-spectra (1,1,4,4-tetraphenyl-1,3-butadiene as a matrix). Despite considerable fragmentation to pristine fullerenes, molecular ions of 1-R and 2-R can be clearly seen. Fig. 2 (a) Magnetization curves of 1-R at temperatures 2–8 K, sweep rate 2.9 mT s−1. (b) Magnetization curves of 1-R and 1 at T = 2 K; the inset shows determination of the blocking temperatures of magnetiza- tion TB, sweep rate 5 K min−1. (c) Magnetization curves of 2-R at tem- peratures 1.8–5 K, sweep rate 2.9 mT s−1; (d) magnetization curves of 2-R and 2 at T = 2 K. The inset shows determination of TB, sweep rate 5 K min−1. Open Access Article. Published on 23 May 2018. Downloaded on 2/12/2019 8:52:01 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Fig. 2 (a) Magnetization curves of 1-R at temperatures 2–8 K, sweep rate 2.9 mT s−1. Open Access Article. Published on 23 May 2018. Downloaded on 2/12/2019 8:52:01 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. For the preparation of SAMs, Au(111) single crystals were immersed into toluene solutions of 1-R or 2-R for 24 hours, and then washed with toluene to remove excess molecules. The samples were then immediately inserted into the fast entry lock of the beam line’s vacuum system and subsequently studied by X-ray absorption spectroscopy (XAS), X-ray natural linear dichroism (XNLD), and X-ray magnetic circular dichro- ism (XMCD) at the Dy-M4,5 absorption edges as shown in Fig. 3(a and b) and Fig. S8–S11.† Both samples exhibit the multiplet structure typical for the Dy3+ state. The XAS signal is The XMCD signal at the strongest feature of the M5 edge of Dy was used to measure the magnetization of the samples at varying magnetic field. Fig. 3c and d show magnetization curves measured for SAMs of 1-R and 2-R at 2 K. In both samples, distinct hysteresis of magnetization is observed. Thus, magnetic bistability and slow relaxation of magnetiza- tion are observed in SAMs of 1-R and 2-R on Au(111). The hys- teresis of SAMs is narrower than in the bulk samples (see Fig. S11† for comparison), although the latter were studied with a considerably slower sweep rate. Faster relaxation of mag- netization in SAMs is presumably caused by direct contacts of fullerenes with the metallic substrate (see below), but may be also aﬀected by X-ray induced demagnetization.20 Fig. 3 (a, b) X-ray absorption spectra of 1-R (a) and 2-R (b) at the Dy- M4,5 absorption edge measured at 2 K in the magnetic ﬁeld of 6.5 T; I+ and I−denote right-hand and left-hand circular polarization of incom- ing X-rays. (c, d) Magnetization curves of sub-monolayers of 1-R (c) and 2-R (d) measured by XMCD at 2 K with a sweep rate of 2 T min−1 (aver- aging over ﬁve measured curves, error bars are standard deviations). To clarify the behaviour of 1-R and 2-R molecules on the gold surface, DFT calculations and DFT-based molecular dynamics (MD) simulations were performed for an Sc3N@C80-R′ molecule with a simplified structure of the linker R21 positioned on 3 atomic layers of gold. After the start of MD modelling22 at 300 K from the vertical position (such as shown in Fig. 4a, see also Fig. S12†), in 2 ps the molecule adopted a configuration with horizontal alignment of the linker and ver- tical position of the fullerene (Fig. Cite this: Nanoscale, 2018, 10, 11287 Despite considerable fragmentation to pristine fullerenes, molecular ions of 1-R and 2-R can be clearly seen. Paramagnetic temperature-dependent 1H NMR spectra were also measured for 1-R (Fig. S4†). Vis-NIR absorption spec- troscopy of the derivatives showed the characteristic pattern of [5,6]-adducts; the spectra of all three derivatives are similar except for the lowest-energy transition, the energy of which increases with the number of Dy atoms from 1.29 eV (960 nm) in Sc3N@C80-R to 1.38 eV (900 nm) in 1-R and 1.43 eV (865 nm) in 2-R (see also Fig. S5†). The influence of cycloaddition on 2 is opposite to that of 1. 2-R shows a considerably lower TB of 4 K (vs. 8 K in 2 11f), its hysteresis is narrower with the coercive field of 0.27 T at 2.0 K (vs. 0.70 T in 2) and is essentially closed above 5 K. Zero-field relaxation times of magnetization of 184 s and 55 s were deter- mined for 2-R at 1.8 and 2.0 K, respectively (at higher tempera- tures the times are too short for reliable measurement by dc- SQUID magnetometry). These values allow estimation of the universal SMM parameter TB100, the temperature at which the relaxation time is 100 s, at 1.9 K. In 2, relaxation times at 1.8 and 2.0 K are 5100 and 2360 s, respectively, and TB100 is near 5 K.11f Thus, derivatization substantially worsens the low- temperature SMM properties of 2. The magnetic properties of powder samples of 1-R and 2-R were first studied by SQUID magnetometry and compared to pristine 1 and 2. Both derivatives showed magnetic hysteresis at low temperatures (Fig. 2). 1-R exhibits the butterfly-shape hysteresis with an abrupt drop of magnetization in zero field, which is characteristic for quantum tunneling of magnetiza- tion and is often observed in single-ion Dy-SMMs, including 1.11d,e,17 At the field sweep rate of 2.9 mT s−1, hysteresis in 1-R The magnetization behaviour of 1-R and 2-R shows that although the magnetic cluster is hidden inside the carbon This journal is © The Royal Society of Chemistry 2018 11288 \| Nanoscale, 2018, 10, 11287–11292 View Article Online Nanoscale Communication Communication cage, exohedral derivatization of the fullerene induces pro- nounced eﬀects on the low-temperature SMM properties. Nanoscale, 2018, 10, 11287–11292 \| 11289 Cite this: Nanoscale, 2018, 10, 11287 Below 10 K, relaxation of magnetization in EMF-SMMs does not proceed via the Orbach mechanism involving crystal-field excited states of Dy ions, but rather involves low-energy exchange excited states and localized phonons strongly coupled to a spin system.11a,b,f,g Our results indicate that the cluster–cage interactions (which are altered by cycloaddition) have significant influence on the relaxation of magnetization in this low-temperature under-barrier regime. Cycloaddition changes the potential energy surface of the endohedral cluster and hence alters the frequencies of vibrations, corresponding to frustrated rotations and translations of the cluster. It is reasonable to suggest that such vibrations play an important role in the relaxation of magnetization.19 Understanding this factor will enable the design of better synthetic strategies for SMMs based on EMF derivatives. rather weak pointing to a low submonolayer coverage, which is determined to be 5–10% of a densely packed layer of unfunc- tionalized fullerene (Fig. S8†). XNLD measurements of 2-R did not show any preferential orientation of Dy–N units at room temperature (Fig. S10†). At 2 K and in the magnetic field of 6.5 T, magnetic circular dichroism can be clearly observed in both samples (Fig. 3; the magnetic field is parallel to X-rays). XMCD measurements with normal and grazing incidence of incoming X-rays gave essentially identical dichroism. In nitride clusterfullerenes such as 1 and 2, Dy ions have very large magnetic anisotropy due to the strong ligand field, mainly imposed by the nitride ion. As a result, the magnetic moments of Dy ions are aligned along the Dy–N bonds. The lack of the angular dependence of XMCD agrees with the negli- gible XNLD signal and points to the random, disordered orien- tations of Dy magnetic moments. In other words, it indicates that 1-R and 2-R molecules are not ordered on the Au(111) sub- strate, as opposed to the ordering found in the sub-monolayer of 2 on Rh(111).5 Open Access Article. Published on 23 May 2018. Downloaded on 2/12/2019 8:52:01 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. 4b), and after a further 5 ps this structure changed into the fully horizontal configuration (Fig. 4c) with both the fullerene and the linker touching the metal. At 300 K the latter configuration is found to be highly mobile. Over the next 20 ps the molecule exhibited both lateral motions on the surface and rotations of the fullerene core (Fig. S13†), resulting in multiple orientations of the endohe- dral cluster.21 Selected conformations along the MD trajectory were chosen for further structure optimization at the PBE-D level using the projector augmented-wave method and with the experimental structure of the linker R.23 The structures with fully horizontal alignment of the molecule parallel to the surface have the lowest energies, which span the range of ca. 0.0–0.3 eV (Fig. 4c shows the lowest energy configuration). The presence of many energy minima with similar energies Fig. 3 (a, b) X-ray absorption spectra of 1-R (a) and 2-R (b) at the Dy- M4,5 absorption edge measured at 2 K in the magnetic ﬁeld of 6.5 T; I+ and I−denote right-hand and left-hand circular polarization of incom- ing X-rays. (c, d) Magnetization curves of sub-monolayers of 1-R (c) and 2-R (d) measured by XMCD at 2 K with a sweep rate of 2 T min−1 (aver- aging over ﬁve measured curves, error bars are standard deviations). Nanoscale, 2018, 10, 11287–11292 \| 11289 This journal is © The Royal Society of Chemistry 2018 Nanoscale View Article Online Nanoscale View Article Online View Article Online Communication Fig. 4 Representative DFT-optimized conﬁguration of Sc3N@C80-R on gold: (a) vertical conﬁguration; (b) the structure with horizontal align- ment of the linker and vertical orientation of the fullerene cage; (c) the structure with fully horizontal alignment of the linker and the fullerene core. Also shown for each structure are relative energies (ΔEtot), dis- persion contribution (ΔEdisp), and the total charge transfer between the Sc3N@C80-R molecule and the substrate (ΔQ). Acknowledgements The authors thank M. Rosenkranz for the help with NMR measurements and M. Knupfer for the help with XPS measure- ments. The X-ray absorption measurements were performed at the X-Treme beam line at the Swiss Light Source, Paul Scherrer Institut.26 Computational resources were provided by the Center for High Performance Computing at the TU Dresden. The authors acknowledge funding from the European Union’s Horizon 2020 research and innovation programme, the European Research Council (grant agreement No 648295 to A. A. P.), the Marie Skłodowska-Curie action (grant agreements no. 701647 to M. S., no. 748635 to S. M. A., and the co-found- ing project INCA 600398 to R. W.). J. D. and M. S. acknowledge funding by the Swiss National Science Foundation (Grant No. 200021_165774/1); R. W. and C. B. acknowledge the Swedish Research Council (Grant No. 2015-00455). Analysis of the Bader atomic charges24 revealed that the Sc3N@C80-R molecule transfers 0.3–0.5 electrons to the metallic substrate (the values are referred versus the charges of the isolated molecule). The exact number depends on the configuration of the molecule. When Sc3N@C80 is in direct contact with the substrate, the fullerene-to-substrate charge transfer is the largest, reaching 0.30e in the structure shown in Fig. 4c. In the absence of direct contact (Fig. 4a and b), the charge transfer from the Sc3N@C80 core is reduced to ca. 0.15e. The linker also transfers ca. 0.15e to the metal (Table S3 and Fig. S14†). However, irrespective of the con- figuration of the molecule on the substrate and the magni- tude of the total charge transfer, the charge of the endohedral Sc3N cluster remains the same (within 0.01e) as in the iso- lated Sc3N@C80-R molecule. This shows that the π-system of the carbon cage eﬀectively screens the Sc3N cluster from the substrate and keeps its charge state intact. The “Faraday cage” eﬀect of the fullerene21,25 thus helps to preserve the magnetic state of the endohedral cluster even when the outer molecule experiences rather strong interactions with the metallic substrate. There are no conflicts to declare. indicates that freezing of the SAMs with a low degree of cov- erage will result in multiple orientations of fullerene mole- cules on the surface. The structures with a horizontal organic tail but vertical fullerene orientation are less stable by 1.7–1.8 eV, whereas the structures with vertical orien- tation of the molecule are the least stable with relative ener- gies of 2.8–2.9 eV. The main contribution to such a large variation of relative energies is caused by dispersion inter- actions (Table S3 and Fig. S14†). Note that the conclusion about the preference of the horizontal configuration was also reached for C60-based SAMs with analogous thioether linkers.14 Conclusions 1,3-Dipolar cycloaddition substantially aﬀects single molecule magnetism of nitride clusterfullerenes, but does so diﬀerently for DySc2N@C80 and Dy2ScN@C80. The low-temperature SMM properties of DySc2N@C80 are improved in the cycloadduct, whereas those of Dy2ScN@C80 are worsened as concluded from the changes in the blocking temperature of magnetiza- tion and relaxation times. Surface deposition onto Au(111) from toluene solution gives low-coverage SAMs of both fuller- enes with thioether anchoring groups. At 2 K both SAMs exhibit hysteresis of magnetization on gold. However, the strong interaction of the fullerene and the linker with the sub- strate results in the preference of horizontal configurations, featuring direct contact of the fullerene core with the metallic surface. The structures are mobile at room temperature and freeze in multiple orientations at low temperatures, leading to random orientation of the magnetic moments of endohedral Dy ions. Open Access Article. Published on 23 May 2018. Downloaded on 2/12/2019 8:52:01 AM. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Fig. 4 Representative DFT-optimized conﬁguration of Sc3N@C80-R on gold: (a) vertical conﬁguration; (b) the structure with horizontal align- ment of the linker and vertical orientation of the fullerene cage; (c) the structure with fully horizontal alignment of the linker and the fullerene core. Also shown for each structure are relative energies (ΔEtot), dis- persion contribution (ΔEdisp), and the total charge transfer between the Sc3N@C80-R molecule and the substrate (ΔQ). Conﬂicts of interest There are no conflicts to declare. Notes and references 6 (a) P. Stoll, M. Bernien, D. Rolf, F. Nickel, Q. Xu, C. Hartmann, T. R. Umbach, J. Kopprasch, J. N. Ladenthin, E. Schierle, E. Weschke, C. Czekelius, W. Kuch and K. J. Franke, Phys. Rev. B: Condens. Matter Mater. Phys., 2016, 94, 224426; (b) C. Gao, Q. Yang, B.-W. Wang, Z.-M. Wang and S. Gao, CrystEngComm, 2016, 18, 4165; (c) A. Ghirri, V. Corradini, V. Bellini, R. Biagi, U. del Pennino, V. De Renzi, J. C. Cezar, C. A. Muryn, G. A. Timco, R. E. P. Winpenny and M. Aﬀronte, ACS Nano, 2011, 5, 7090; (d) J. Dreiser, C. Wäckerlin, M. E. Ali, C. Piamonteze, F. Donati, A. Singha, K. S. Pedersen, S. Rusponi, J. Bendix, P. M. Oppeneer, T. A. Jung and H. Brune, ACS Nano, 2014, 8, 4662; (e) E. Moreno Pineda, T. Komeda, K. Katoh, M. Yamashita and M. Ruben, Dalton Trans., 2016, 45, 18417; (f) L. Rigamonti, M. Piccioli, L. Malavolti, L. Poggini, M. Mannini, F. Totti, B. Cortigiani, A. Magnani, R. Sessoli and A. Cornia, Inorg. Chem., 2013, 52, 5897. 6 (a) P. Stoll, M. Bernien, D. Rolf, F. Nickel, Q. Xu, C. Hartmann, T. R. Umbach, J. Kopprasch, J. N. Ladenthin, E. Schierle, E. Weschke, C. Czekelius, W. Kuch and K. J. Franke, Phys. Rev. B: Condens. Matter Mater. Phys., 2016, 94, 224426; (b) C. Gao, Q. Yang, B.-W. Wang, Z.-M. Wang and S. Gao, CrystEngComm, 2016, 18, 4165; (c) A. Ghirri, V. Corradini, V. Bellini, R. Biagi, U. del Pennino, V. De Renzi, J. C. Cezar, C. A. Muryn, G. A. Timco, R. E. P. Winpenny and M. Aﬀronte, ACS Nano, 2011, 5, 7090; (d) J. Dreiser, C. Wäckerlin, M. E. Ali, C. Piamonteze, F. Donati, A. Singha, K. S. Pedersen, S. Rusponi, J. Bendix, P. M. Oppeneer, T. A. Jung and H. Brune, ACS Nano, 2014, 8, 4662; (e) E. Moreno Pineda, T. Komeda, K. Katoh, M. Yamashita and M. Ruben, Dalton Trans., 2016, 45, 18417; (f) L. Rigamonti, M. Piccioli, L. Malavolti, L. Poggini, M. Mannini, F. Totti, B. Cortigiani, A. Magnani, R. Sessoli and A. Cornia, Inorg. Chem., 2013, 52, 5897. 12 (a) M. d. C. Gimenez-Lopez, J. A. Gardener, A. Q. Shaw, A. Iwasiewicz-Wabnig, K. Porfyrakis, C. Balmer, G. Dantelle, M. Hadjipanayi, A. Crossley, N. R. Champness, M. R. Castell, G. A. D. Briggs and A. N. Khlobystov, Phys. Chem. Chem. Phys., 2010, 12, 123; (b) N. Notes and references Crivillers, Y. Takano, Y. Matsumoto, J. Casado-Montenegro, M. Mas- Torrent, C. Rovira, T. Akasaka and J. Veciana, Chem. Commun., 2013, 49, 8145. 13 D. Bonifazi, O. Enger and F. Diederich, Chem. Soc. Rev., 2007, 36, 390. 14 M. del Carmen Gimenez-Lopez, M. T. Räisänen, T. W. Chamberlain, U. Weber, M. Lebedeva, G. A. Rance, G. A. D. Briggs, D. Pettifor, V. Burlakov, M. Buck and A. N. Khlobystov, Langmuir, 2011, 27, 10977. 14 M. del Carmen Gimenez-Lopez, M. T. Räisänen, T. W. Chamberlain, U. Weber, M. Lebedeva, G. A. Rance, G A D Briggs D Pettifor V Burlako M Buck and G. A. D. Briggs, D. Pettifor, V. Burlakov, M. Buck and A. N. Khlobystov, Langmuir, 2011, 27, 10977. 15 S. Yang, A. A. Popov, C. Chen and L. Dunsch, J. Phys. Chem. C, 2009, 113, 7616. 7 F. Moro, R. Biagi, V. Corradini, M. Evangelisti, A. Gambardella, V. De Renzi, U. del Pennino, E. Coronado, A. Forment-Aliaga and F. M. Romero, J. Phys. Chem. C, 2012, 116, 14936. 16 (a) N. Chen, E. Y. Zhang, K. Tan, C. R. Wang and X. Lu, Org. Lett., 2007, 9, 2011; (b) N. Chen, L. Z. Fan, K. Tan, Y. Q. Wu, C. Y. Shu, X. Lu and C. R. Wang, J. Phys. Chem. C, 2007, 111, 11823; (c) T. Cai, C. Slebodnick, L. Xu, K. Harich, T. E. Glass, C. Chancellor, J. C. Fettinger, M. M. Olmstead, A. L. Balch, H. W. Gibson and H. C. Dorn, J. Am. Chem. Soc., 2006, 128, 6486; (d) C. M. Cardona, A. Kitaygorodskiy and L. Echegoyen, J. Am. Chem. Soc., 2005, 127, 10448; (e) B. Wu, Y. Li, L. Jiang, C. Wang and T. Wang, J. Phys. Chem. C, 2016, 120, 6252. A. Forment-Aliaga and F. M. Romero, J. Phys. Chem. C, 2012, 116, 14936. 8 (a) U. Glebe, T. Weidner, J. E. Baio, D. Schach, C. Bruhn, A. Buchholz, W. Plass, S. Walleck, T. Glaser and U. Siemeling, ChemPlusChem, 2012, 77, 889; (b) G. Cucinotta, L. Poggini, A. Pedrini, F. Bertani, N. Cristiani, M. Torelli, P. Graziosi, I. Cimatti, B. Cortigiani, E. Otero, P. Ohresser, P. Sainctavit, A. Dediu, E. Dalcanale, R. Sessoli and M. Mannini, Adv. Funct. Mater., 2017, 27, 1703600. 17 J. Dreiser, J. Phys.: Condens. Matter, 2015, 27, 183203. 9 A. Cini, M. Mannini, F. Totti, M. Fittipaldi, G. Spina, A. Chumakov, R. Rüﬀer, A. Cornia and R. Sessoli, Nat. Notes and references 1 (a) R. Sessoli, D. Gatteschi, A. Caneschi and M. A. Novak, Nature, 1993, 365, 141; (b) D. N. Woodruﬀ, R. E. P. Winpenny and R. A. Layfield, Chem. Rev., 2013, 113, 5110; (c) D. Gatteschi, R. Sessoli and J. Villain, Molecular Nanomagnets, Oxford University Press, New York, 2006; (d) Introduction to Molecular Magnetism: From Transition Metals to Lanthanides, ed. C. Benelli and D. Gatteschi, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 2015. 2 (a) N. Domingo, E. Bellido and D. Ruiz-Molina, Chem. Soc. Rev., 2012, 41, 258; (b) A. Cornia, M. Mannini, P. Sainctavit 2 (a) N. Domingo, E. Bellido and D. Ruiz-Molina, Chem. Soc. Rev., 2012, 41, 258; (b) A. Cornia, M. Mannini, P. Sainctavit This journal is © The Royal Society of Chemistry 2018 11290 \| Nanoscale, 2018, 10, 11287–11292 View Article Online Nanoscale Communication Communication and R. Sessoli, Chem. Soc. Rev., 2011, 40, 3076; (c) R. J. Holmberg and M. Murugesu, J. Mater. Chem. C, 2015, 3, 11986. 10 E. Tancini, M. Mannini, P. Sainctavit, E. Otero, R. Sessoli and A. Cornia, Chem. – Eur. J., 2013, 19, 16902. 11 (a) F. Liu, D. S. Krylov, L. Spree, S. M. Avdoshenko, N. A. Samoylova, M. Rosenkranz, A. Kostanyan, T. Greber, A. U. B. Wolter, B. Büchner and A. A. Popov, Nat. Commun., 2017, 8, 16098; (b) C.-H. Chen, D. S. Krylov, S. M. Avdoshenko, F. Liu, L. Spree, R. Yadav, A. Alvertis, L. Hozoi, K. Nenkov, A. Kostanyan, T. Greber, A. U. B. Wolter and A. A. Popov, Chem. Sci., 2017, 8, 6451; (c) K. Junghans, C. Schlesier, A. Kostanyan, N. A. Samoylova, Q. Deng, M. Rosenkranz, S. Schiemenz, R. Westerström, T. Greber, B. Büchner and A. A. Popov, Angew. Chem., Int. Ed., 2015, 54, 13411; (d) R. Westerström, J. Dreiser, C. Piamonteze, M. Muntwiler, S. Weyeneth, K. Krämer, S.-X. Liu, S. Decurtins, A. Popov, S. Yang, L. Dunsch and T. Greber, Phys. Rev. B: Condens. Matter Mater. Phys., 2014, 89, 060406; (e) R. Westerström, J. Dreiser, C. Piamonteze, M. Muntwiler, S. Weyeneth, H. Brune, S. Rusponi, F. Nolting, A. Popov, S. Yang, L. Dunsch and T. Greber, J. Am. Chem. Soc., 2012, 134, 9840; (f) D. S. Krylov, F. Liu, S. M. Avdoshenko, L. Spree, B. Weise, A. Waske, A. U. B. Wolter, B. Büchner and A. A. Popov, Chem. Commun., 2017, 53, 7901; (g) D. Krylov, F. Liu, A. Brandenburg, L. Spree, V. Notes and references Bon, S. Kaskel, A. Wolter, B. Buchner, S. Avdoshenko and A. A. Popov, Phys. Chem. Chem. Phys., 2018, 20, 11656. 11 2015, 3, 11986. 3 (a) L. Margheriti, D. Chiappe, M. Mannini, P. E. Car, P. Sainctavit, M.-A. Arrio, F. B. de Mongeot, J. C. Cezar, F. M. Piras, A. Magnani, E. Otero, A. Caneschi and R. Sessoli, Adv. Mater., 2010, 22, 5488; (b) C. Wäckerlin, F. Donati, A. Singha, R. Baltic, S. Rusponi, K. Diller, F. Patthey, M. Pivetta, Y. Lan, S. Klyatskaya, M. Ruben, H. Brune and J. Dreiser, Adv. Mater., 2016, 28, 5195. 4 (a) M. Mannini, F. Pineider, C. Danieli, F. Totti, L. Sorace, P. Sainctavit, M. A. Arrio, E. Otero, L. Joly, J. C. Cezar, A. Cornia and R. Sessoli, Nature, 2010, 468, 417; (b) M. Mannini, F. Pineider, P. Sainctavit, C. Danieli, E. Otero, C. Sciancalepore, A. M. Talarico, M.-A. Arrio, A. Cornia, D. Gatteschi and R. Sessoli, Nat. Mater., 2009, 8, 194; (c) L. Malavolti, V. Lanzilotto, S. Ninova, L. Poggini, I. Cimatti, B. Cortigiani, L. Margheriti, D. Chiappe, E. Otero, P. Sainctavit, F. Totti, A. Cornia, M. Mannini and R. Sessoli, Nano Lett., 2015, 15, 535. 3 (a) L. Margheriti, D. Chiappe, M. Mannini, P. E. Car, P. Sainctavit, M.-A. Arrio, F. B. de Mongeot, J. C. Cezar, F. M. Piras, A. Magnani, E. Otero, A. Caneschi and R. Sessoli, Adv. Mater., 2010, 22, 5488; (b) C. Wäckerlin, F. Donati, A. Singha, R. Baltic, S. Rusponi, K. Diller, F. Patthey, M. Pivetta, Y. Lan, S. Klyatskaya, M. Ruben, H. Brune and J. Dreiser, Adv. Mater., 2016, 28, 5195. 4 (a) M. Mannini, F. Pineider, C. Danieli, F. Totti, L. Sorace, P. Sainctavit, M. A. Arrio, E. Otero, L. Joly, J. C. Cezar, A. Cornia and R. Sessoli, Nature, 2010, 468, 417; (b) M. Mannini, F. Pineider, P. Sainctavit, C. Danieli, E. Otero, C. Sciancalepore, A. M. Talarico, M.-A. Arrio, A. Cornia, D. Gatteschi and R. Sessoli, Nat. Mater., 2009, 8, 194; (c) L. Malavolti, V. Lanzilotto, S. Ninova, L. Poggini, I. Cimatti, B. Cortigiani, L. Margheriti, D. Chiappe, E. Otero, P. Sainctavit, F. Totti, A. Cornia, M. Mannini and R. Sessoli, Nano Lett., 2015, 15, 535. 5 R. Westerström, A.-C. Uldry, R. Stania, J. Dreiser, C. Piamonteze, M. Muntwiler, F. Matsui, S. Rusponi, H. Brune, S. Yang, A. Popov, B. Büchner, B. Delley and T. Greber, Phys. Rev. Lett., 2015, 114, 087201. This journal is © The Royal Society of Chemistry 2018 11292 \| Nanoscale, 2018, 10, 11287–11292 Notes and references Commun., 2018, 9, 480. 18 Y. Li, T. Wang, H. Meng, C. Zhao, M. Nie, L. Jiang and C. Wang, Dalton Trans., 2016, 45, 19226. This journal is © The Royal Society of Chemistry 2018 Nanoscale, 2018, 10, 11287–11292 \| 11291 View Article Online Communication Communication Nanoscale 19 (a) A. Lunghi, F. Totti, R. Sessoli and S. Sanvito, Nat. Commun., 2017, 8, 14620; (b) A. Lunghi, F. Totti, S. Sanvito and R. Sessoli, Chem. Sci., 2017, 8, 6051; (c) C. A. P. Goodwin, F. Ortu, D. Reta, N. F. Chilton and D. P. Mills, Nature, 2017, 548, 439. 22 J. Hutter, M. Iannuzzi, F. Schiﬀmann and J. VandeVondele, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 2014, 4, 15. 23 J. Hafner, J. Comput. Chem., 2008, 29, 2044. 24 G. Henkelman, A. Arnaldsson and H. Jónsson, Comput. Mater. Sci., 2006, 36, 354. 25 P. Delaney and J. C. Greer, Appl. Phys. Lett., 2004, 84, 431. 20 J. Dreiser, R. Westerström, C. Piamonteze, F. Nolting, S. Rusponi, H. Brune, S. Yang, A. Popov, L. Dunsch and T. Greber, Appl. Phys. Lett., 2014, 105, 032411. 26 C. Piamonteze, U. Flechsig, S. Rusponi, J. Dreiser, J. Heidler, M. Schmidt, R. Wetter, M. Calvi, T. Schmidt, H. Pruchova, J. Krempasky, C. Quitmann, H. Brune and F. Nolting, J. Synchrotron Radiat., 2012, 19, 661. 21 S. M. Avdoshenko, J. Comput. Chem., 2018, DOI: 10.1002/ jcc.25231.
https://openalex.org/W2038385014	https://www.frontiersin.org/articles/10.3389/fcimb.2014.00076/pdf	English	null	Forward genetic approaches for elucidation of novel regulators of Lyme arthritis severity	Frontiers in cellular and infection microbiology	2,014	cc-by	9,750	CELLULAR AND INFECTION MICROBIOLOGY published: 05 June 2014 doi: 10.3389/fcimb.2014.00076 Correspondence: Correspondence: Janis J. Weis, Department of Pathology, University of Utah, 15 North Medical Drive East #2100, Salt Lake City, UT 84112-5650, USA e-mail: janis.weis@path.utah.edu INTRODUCTION TO LYME DISEASE understanding the pathogenic mechanisms of acute clinical dis- ease and in characterizing predisposing features for chronic disease. This review will assess past and ongoing studies that have provided insight into genetic susceptibility to Lyme arthri- tis, with particular emphasis on studies using Forward Genetic approaches. Lyme Disease, caused by infection with the tick borne spiro- chete Borrelia burgdorferi, is a growing societal concern, espe- cially in endemic regions of the United States and Europe. Approximately 30,000 case reports are ﬁled by physicians each year in the United States (C.D.C, 2013a), while the CDC has upwardly revised their best estimate of the total incidence to 300,000/year, based on several complementary lines of evidence (Kuehn, 2013). Part of the societal concern is rooted in the uncertainty surrounding pathological outcomes associated with B. burgdorferi infection. A large percentage (70%) of infected individuals develop the characteristic bulls-eye rash erythema migrans at the site of the infected tick bite, with progression to further clinical complications following dissemination of the spirochete. Arthritis, the most common symptom occurs in 30–60% of infected individuals, while Bell’s palsy and other neu- rological symptoms are seen in 10–12% of patients (Wormser et al., 2006; C.D.C, 2013b). Carditis has been considered a rare complication (<1%), however, 3 recent deaths with docu- mented B. burgdorferi in autopsied heart tissue strongly argue for increased vigilance in detecting infection of this tissue (C.D.C, 2013c). Kenneth K.C. Bramwell 1, Cory Teuscher 2 and Janis J. Weis 1* 1 Department of Pathology, University of Utah, Salt Lake City, UT, USA 2 Department of Medicine, University of Vermont, Burlington, VT, USA Patients experiencing natural infection with Borrelia burgdorferi display a spectrum of associated symptoms and severity, strongly implicating the impact of genetically determined host factors in the pathogenesis of Lyme disease. Herein, we provide a summary of the host genetic factors that have been demonstrated to inﬂuence the severity and chronicity of Lyme arthritis symptoms, and a review of the resources available, current progress, and added value of a forward genetic approach for identiﬁcation of novel genetic regulators. Reviewed by: Reviewed by: Janakiram Seshu, The University of Texas at San Antonio, USA Reviewed by: Janakiram Seshu, The University of Texas at San Antonio, USA Philip E. Stewart, National Institute of Allergy and Infectious Disease, USA Keywords: Lyme disease, Lyme arthritis, forward genetics, pathogenesis, genome-wide association studies, innate immunity and responses, inﬂammation, beta-Glucuronidase Frontiers in Cellular and Infection Microbiology REVIEW ARTICLE published: 05 June 2014 doi: 10.3389/fcimb.2014.00076 REVIEW ARTICLE published: 05 June 2014 doi: 10.3389/fcimb.2014.00076 REVIEW ARTICLE Edited by: Edited by: Tanja Petnicki-Ocwieja, Tufts University School of Medicine and Tufts Medical Center, USA Reviewed by: Janakiram Seshu, The University of Texas at San Antonio, USA Philip E. Stewart, National Institute of Allergy and Infectious Disease, USA Edited by: Tanja Petnicki-Ocwieja, Tufts University School of Medicine and Tufts Medical Center, USA Tanja Petnicki-Ocwieja, Tufts University School of Medicine and Tufts Medical Center, USA GENOME WIDE ASSOCIATION STUDIES AS A METHODOLOGY FOR STUDYING COMPLEX GENETIC TRAITS IN HUMANS - SUCCESSES AND LIMITATIONS et al., 2011). More recently, the Collaborative Cross was devel- oped, which represents an ambitious community effort by mouse geneticists to develop approximately 1000 additional recombi- nant inbred mouse strains with deﬁned genetic composition. These recombinant mice were derived from 8 parental inbred and wild-derived strains through an intricate directed breeding pro- cess (Churchill et al., 2004). Due to the increased genetic diver- sity of mouse strains that can be interrogated simultaneously, the Collaborative Cross is expected to provide additional power to forward genetics screens for many diseases. Together, these more recent developments are expected to provide greater pre- dictive power for identiﬁcation of regulatory intervals underlying complex multigenic traits. Genome Wide Association Studies were ﬁrst proposed in the mid- 1990s as a way to study association between human genetic poly- morphisms and complex, multigenic traits. Rather than measure all genetic variation present in each individual, approximately one million Single Nucleotide Polymorphisms (SNPs) are assayed as genetic landmarks, and scored for association with the trait of interest. Based on linkage disequilibrium, genes in close proximity to associated SNP landmarks are potential candidates for further investigation. This technique is well-suited to identify susceptibility genes that are of intermediate prevalence in the population, with a Mean Allele Frequency of greater than 0.05 (Risch and Merikangas, 1996). Although no such studies of Lyme arthritis severity have been conducted, GWAS has been extensively used to investigate genetic modulators of other inﬂammatory conditions including rheumatoid arthritis (RA). Thus far, almost ﬁfty susceptibility loci have been identiﬁed for RA, accounting for approximately one-half of the total genetic variation expected in populations of European ancestry. Nineteen of these loci have been reﬁned to a single candidate gene association, and the underlying causal poly- morphism has been predicted for seven of these loci (Eyre et al., 2012). Prior to the availability of these more sophisticated modern resources, a seminal study by Dr. Stephen Barthold recognized that inbred mouse strains exhibit distinct genetic susceptibilities to Lyme arthritis, recapitulating the range of arthritis severity seen in patients (Barthold et al., 1990). However, mice do not reca- pitulate the full depth and breadth of symptoms experienced by human patients. This is evident in the inability of B. burgdor- feri to elicit neurological symptoms or erythema migrans in mice (Garcia-Monco and Benach, 2013). WHAT IS FORWARD GENETICS? Forward genetics is an unbiased genetic approach that begins with a heritable trait of interest and attempts to determine the alle- les responsible for the observed variability through a process of genetic mapping. In contrast, reverse genetics is a hypothesis- driven scientiﬁc approach that begins with a gene of interest and attempts to determine the phenotypic impacts caused by experimental manipulations of that gene. Classically, forward genetic studies were ﬁrst performed through random mutage- nesis screens. More recently, forward genetics has been used to map pre-existing genetic variations in human populations or experimental animal models. In practice, forward genetic stud- ies involve three key steps: (1) Each individual in a population of mixed genetic composition is surveyed for the trait of inter- est; (2) The genetic makeup of each individual is assessed; (3) Statistical calculations predict the strength of association in the study population between the measured trait and each genetic locus in the genome. Forward genetic screens often produce a map across the entire genome with peaks and valleys denoting areas with strong or weak statistical association with the trait, respectively. The wide variation in Lyme disease symptoms and sever- ity observed within the patient population is thought to reﬂect unique features of individual B. burgdorferi isolates that inﬂu- ence invasive potential, as well as heritable factors in the patient population that contribute to clinical severity. Furthermore, although most patients resolve infection with appropriate antibi- otic therapy, a small percentage of treated patients with severe clinical symptoms fail to resolve and develop a chronic dis- ease termed Post Treatment Lyme Disease (Steere and Glickstein, 2004). Thus, there are compelling reasons to identify host genes that determine the severity of Lyme disease, both in Scientiﬁc investigations into heritable genetic risk factors that contribute to complex disease have been conducted using a variety of approaches, including Genome Wide Association Studies (GWAS) and forward genetic screens in tractable animal models. June 2014 \| Volume 4 \| Article 76 \| 1 www.frontiersin.org www.frontiersin.org www.frontiersin.org Forward genetics of lyme arthritis Bramwell et al. GENOME WIDE ASSOCIATION STUDIES AS A METHODOLOGY FOR STUDYING COMPLEX GENETIC TRAITS IN HUMANS - SUCCESSES AND LIMITATIONS Despite these limitations, the ﬁnding that C3H mice develop severe arthritis and cardi- tis at reproducible times following intradermal infection with B. burgdorferi cultured in the laboratory was very important. The C3H mouse has been used extensively for studies of severe dis- ease, and the involvement of a variety of cell types and signaling pathways have been evaluated. Equally important was the obser- vation that C57BL/6 (B6) mice consistently develop less severe disease despite being equally susceptible to infection and having similar numbers of bacteria in joints. Since many mutant alle- les have been crossed onto the B6 background, this has allowed identiﬁcation of the contribution of numerous immunologically important genes to both host defense and modulation of arthritis and carditis severity. Recent studies with RA and juvenile RA have involved cohorts of up to 10,000 patients and controls, pointing out the require- ment for large populations of well-characterized patients for GWAS analysis (Hinks et al., 2013). If sufﬁciently large sample sizes could be achieved, these ﬁndings suggest that GWAS could be a successful strategy to investigate Lyme arthritis susceptibil- ity loci, but also indicate that additional approaches are needed to capture the signiﬁcant fraction of variation likely to be left unaccounted for. It is also important to recognize that identiﬁcation of regula- tory loci is not the ultimate goal of a forward genetics study, but is only a ﬁrst step. While association of a speciﬁc genetic landmark to disease susceptibility may have potential relevance to clinical diagnosis, there is added value in the formal investigation of can- didate genes and predicted causal polymorphisms, and in further understanding the underlying mechanisms of pathogenesis. This type of mechanistic investigation frequently involves the use of animal models. Mice and other small mammals are essential reservoir species for B. burgdorferi in nature, and greater than 90% of trapped wild mice in some Lyme endemic areas have tested seropositive for infection (Bunikis et al., 2004; Radolf et al., 2012). Wild mice are generally resistant to Lyme arthritis, although recent work has shown that natural variants of known innate immune reg- ulatory genes may be correlated with the prevalence of Lyme infection within the wild rodent population (Tschirren et al., 2013). These considerations make the mouse an excellent model for assessment of genetic factors contributing to Lyme arthritis development. Frontiers in Cellular and Infection Microbiology ROLE OF THE MHC IN IMMUNE RESPONSE TO B. BURGDORFERI AND IN ARTHRITIS SEVERITY: HUMAN AND MOUSE STUDIES Several important studies have discovered natural genetic alle- les that inﬂuence Lyme arthritis severity. Steere et al. ﬁrst reported the inﬂuence of the human major histocompatibil- ity complex (MHC) on Lyme arthritis severity, and provided important early evidence that Lyme arthritis has an immuno- genetic basis (Steere et al., 1990). This pioneering work iden- tiﬁed increased incidence of clinical Lyme arthritis, particularly that lasting longer that 12 months in a single joint, as associ- ated with two serologically deﬁned Class II alleles, HLA-DR4 and HLA-DR2. Importantly, the association of Class II alle- les with Lyme arthritis was not supported by studies inclu- sive of all outcomes of Lyme arthritis patients. The advent of molecular characterization of Class II alleles allowed more pre- cise analysis of associations with disease phenotype, and led to the conclusion that MHC alleles are not major determi- nants of early Lyme disease severity, a distinction from rheuma- toid arthritis (Feng et al., 1995; Klempner et al., 2005). More recently, Steere and colleagues have conﬁrmed the association of two Class II alleles (DRB1∗0101 and 0401) for the subgroup of patients with treatment refractory Lyme disease but not in the larger group of patients that respond to antibiotic treat- ment, and have proposed an auto-immune mechanism in this treatment refractory group (Steere et al., 2006; Drouin et al., 2013). Oosting et al. found that N248S and S602I polymorphisms in TLR1 were associated with reduced in vitro responsiveness to B. burgdorferi and TLR1/TLR2 agonist stimulation (Oosting et al., 2011a). Using a similar experimental approach, the same group also reported that peripheral blood mononuclear cells (PBMCs) for individuals bearing an IL-23R Arg381Gln polymorphism exhibited a reduced Th17 response following in vitro stimulation with B. burgdorferi (Oosting et al., 2011b). However, there was no association between the IL-23R polymorphism and the per- sistence of symptoms among patients in the study population, arguing against a role for this SNP in disease pathogenesis. Strle et al. recently described the frequency and impact of sev- eral polymorphisms in the TLR1 gene within a cohort of Lyme disease patients (Strle et al., 2012). This study found a skewed inheritance pattern of TLR1 1805GG polymorphisms within an antibiotic-refractory Lyme arthritis patient population. They also recognized a synergy between inheritance of this host polymor- phism and infection with a particular invasive isolate (termed RST1) of B. burgdorferi. USE OF ANIMAL MODELS FOR IDENTIFICATION OF GENES REGULATING DISEASE SEVERITY Animal models provide an alternative approach for identiﬁcation of genes that regulate disease development. Inbred mouse lines are powerful genetic resources, which have been widely used to identify genes associated with disease severity. Visionary scientists began the breeding of inbred mice over a century ago. Each mod- ern inbred line has ﬁxed genetic composition, while the plentitude of inbred strains collectively capture a large amount of genetic variation. Several advances in the past decade have signiﬁcantly added to their value, particularly the publication of the mouse genome, coupled with various efforts to deﬁne the genetic vari- ation between inbred mouse strains (Gregory et al., 2002; Keane Barthold’s ﬁndings were corroborated and expanded upon by others. Many groups then addressed speciﬁc facets of the immune response and pathogenesis of Lyme arthritis, using the mouse models developed by Barthold, and that important work contin- ues today. Many such studies rely on reverse-genetic approaches, such as targeted genetic deletion, gene silencing, treatment with inhibitory or stimulatory molecules, or transgenic manipulation. For example, the importance of the innate immune response in Lyme pathogenesis was demonstrated by Schaible, Barthold, and Brown who collectively observed that mice with severe com- bined immunodeﬁciency (scid/Rag−), lacking B and T cells, June 2014 \| Volume 4 \| Article 76 \| 2 Frontiers in Cellular and Infection Microbiology www.frontiersin.org www.frontiersin.org www.frontiersin.org Bramwell et al. Forward genetics of lyme arthritis ROLE OF THE MHC IN IMMUNE RESPONSE TO B. BURGDORFERI AND IN ARTHRITIS SEVERITY: HUMAN AND MOUSE STUDIES Importantly, patients carrying TLR1 1805GG exhibited higher serum levels of CXCL9 and CXCL10 chemokines, consistent with a functional role for this polymor- phism. This effect was reproduced through in vitro activation of PBMCs with a B. burgdorferi RST1 isolate, arguing that height- ened production of these IFNγ-inducible chemokines may set the stage for antibiotic refractory arthritis. A number of investigators found association of MHC hap- lotypes with antibody recognition of individual B. burgdorferi antigens using MHC congenic mouse lines. However, use of MHC congenics in our studies and in those of other investiga- tors led to the conclusion that MHC alleles were not determinants for the differences in arthritis severity found 4 weeks follow- ing infection in C3H-H2k, C57BL/6-H2b, and DBA-H2d mice (Yang et al., 1992; Brown and Reiner, 2000). Thus, studies with mice are consistent with patient studies failing to show associa- tion with early Lyme arthritis. Interestingly, mice expressing the H2k allele do not develop collagen-induced arthritis, a contrast with their development of severe Lyme arthritis (Wooley et al., 1981). Frontiers in Cellular and Infection Microbiology IDENTIFICATION OF TLR1/TLR2 IN THE HOST RESPONSE TO B. BURGDORFERI IN HUMANS AND MICE retained the differential genetic severities in arthritis and carditis observed between inbred mouse strains (Schaible et al., 1989; Barthold et al., 1992; Brown and Reiner, 1999). This ﬁnding set a lasting framework for future studies into various facets of the innate immune response. The reverse genetic techniques used in these and other studies are powerful and conclusive, and have resulted in the identiﬁcation of many genes with doc- umented importance in the pathogenesis of Lyme arthritis and host defense to B. burgdorferi. However, these approaches are by nature biased to genes with known function and are not suitable for global analysis of the potential genetic contribution to dis- ease. Selection of a candidate gene necessarily involves assessment of pathways that are suspected to inﬂuence the disease process, resulting in the rejection or delay of other non-candidate genes for study. Early seminal studies into the host-pathogen interaction of B. burgdorferi revealed the potential of the spirochete and its lipoproteins to induce inﬂammatory cytokine production in a variety of human and mouse cell types (Radolf et al., 1991; Wooten et al., 1996; Sellati et al., 1998). The association of NF-κB with these inﬂammatory responses directed numerous laboratories to investigate the involvement of Toll-like receptors as these molecules were discovered as central components of inﬂammatory responses to microbial pathogens (Wooten et al., 1996; Sellati et al., 1998). These studies documented the interac- tions between B. burgdorferi lipoproteins with TLR2 and TLR1, both with mouse knock-out and cell culture transfection studies and in patients, and established a critical role for TLR signal- ing through MyD88 in host defense to this pathogen (Aliprantis et al., 1999; Brightbill et al., 1999; Hirschfeld et al., 1999, 2000; Alexopoulou et al., 2002). More recent studies by Schroder et al. identiﬁed a human variant in TLR2, Arg753Gln, with reduced pro-inﬂammatory signaling in patient samples (Schroder et al., 2005). Cells from mice heterozygous for this variant also dis- played reduced inﬂammatory responses to B. burgdorferi lysate. Notably, this TLR2 allele was signiﬁcantly underrepresented within a cohort of late stage Lyme disease patients, suggesting that it has a protective effect. THE POWER OF FORWARD GENETICS Intercross populations of B6 and C3H mice were used for identiﬁcation of Quantitative Trait Loci (QTL) regulating Borrelia burgdorferi associated arthritis and other responses related to infection (Bbaa). Arthritis and other metrics of host response were assessed at 4 weeks of infection. A total genome scan was performed for each infected mouse (n = 450) and threshold permutation analysis identiﬁed loci associated with disease and response. FIGURE 1 \| Forward Genetics approach for Lyme arthritis severity. Intercross populations of B6 and C3H mice were used for identiﬁcation of Quantitative Trait Loci (QTL) regulating Borrelia burgdorferi associated arthritis and other responses related to infection (Bbaa). Arthritis and other metrics of host response were assessed at 4 weeks of infection. A total genome scan was performed for each infected mouse (n = 450) and threshold permutation analysis identiﬁed loci associated with disease and response. Other successes include the identiﬁcation of genes impor- tant in the regulation of animal models of rheumatoid arthritis and autoimmunity by comparing disease susceptible and dis- ease resistant mouse strains (Ma et al., 2002; Glant et al., 2004; Wicker et al., 2005; Ahlqvist et al., 2009). In some cases, QTL mapping efforts have bridged gaps between seemingly distinct experimental models of autoimmune and other inﬂammatory diseases through the identiﬁcation of shared immunopathology loci (Teuscher, 1985; Meeker et al., 1995; Teuscher et al., 1996, 1997, 1998; Del Rio et al., 2008; Spach et al., 2009, 2010) and iden- tiﬁcation of the relevant functional polymorphisms (Sudweeks et al., 1993; Ma et al., 2002). C3H/HeN (C3H) mice, with 150 total male and female mice included in the cohort. Each individual was assessed for seven quantitative traits and for genetic composition, which was deter- mined using 195 microsatellite markers distributed throughout the genome. Permutation threshold analysis was then used for the entire cohort to determine the degree of association between these quantitative traits and the parental derivation of speciﬁc loci. Four distinct regions on chromosomes 4, 5, and 11 were found to regulate arthritis severity traits as measured by caliper mea- surement of ankle swelling and by blinded scoring of a number of microscopically assessed histopathology traits. Five additional loci on chromosomes 6, 9, 11, 12, and 17 were found to reg- ulate B. burgdorferi-speciﬁc humoral IgM and IgG responses independently of arthritis severity. THE POWER OF FORWARD GENETICS The ﬁrst step of QTL analysis in mice is the direct inter- breeding of two strains of interest to generate a large cohort of genetically distinct individuals (Figure 1). F1 hybrids are geneti- cally identical, carrying one copy of each chromosome from each parental line. These hybrids can be backcrossed to either parental strain (BC1), or interbred to generate F2 hybrids. In each case, genetic variability among the offspring is generated by random recombination events between sister chromatids during meiosis. Each interbreeding strategy can identify regulatory alleles with a dominant, codominant, or additive effect. F2 intercross popula- tions allow the identiﬁcation of alleles acting in recessive fashion that are capable of “standing alone,” whereas the BC1 populations have the added advantage of allowing identiﬁcation of genetic alleles whose effect is most apparent in the genetic context of a particular inbred background. However, hybrids backcrossed to a parental strain are not expected to detect any phenotype from recessive alleles bred back to a dominant parent. This foundational study was followed up by Roper et al. with additional QTL experiments using reciprocal F1 × B6 and F1 × C3H backcrosses and a (BALB/c × C3H) F1 × C3H intercross that found 12 new QTL on Chromosomes 1, 2, 4, 6, 7, 9, 10, 12, 14, 15, 16, and 17 regulating a variety of traits (Roper et al., 2001). A total of twenty-three QTL were identiﬁed that regulate metrics of arthritis severity (Figure 2, red) or other traits related to the humoral response, inﬂammatory response, or host defense (Figure 2, blue). As predicted from previous MHC congenic stud- ies, none of the arthritis-associated QTL identiﬁed in the three B6:C3H intercrosses identiﬁed the MHC locus on chromosome 17. Interestingly, ankle swelling did associate with this region in a single backcross (BALB/c × C3H) F1 × BALB/c, with a lod score of 3.1, predicting an association with one or more of the numerous class I, class II, or class III genes in this region. Seven of the 23 QTL were reproduced in multiple crosses (Bbaa2, Bbaa6, Bbaa8, Bbaa10, Bbaa12, Bbaa14, and Bbaa15). The Bbaa2 QTL on Chromosome 5 was reproduced in all four intercross exper- iments, and in every case the arthritis severity originated from the C3H parental strain. The lod scores identifying Bbaa2 ranged from 3.5 to 10.2 for the four intercross populations, with the 10.2 THE POWER OF FORWARD GENETICS Forward Genetic approaches attempt to determine which genetic loci are responsible for a phenotype of interest. In general, indi- viduals are generated with genotypes that have been altered in an unbiased way, followed by analysis to map inheritance of the phe- notype of interest to speciﬁc genetic loci. This was made possible June 2014 \| Volume 4 \| Article 76 \| 3 Frontiers in Cellular and Infection Microbiology www.frontiersin.org www.frontiersin.org www.frontiersin.org Bramwell et al. Forward genetics of lyme arthritis by the development of genetic maps of microsatellite landmarks evenly distributed throughout the genome, the utility of which Paterson et al. ﬁrst demonstrated for Quantitative Trait Locus (QTL) mapping in plants, later followed by Todd et al. in mice (Paterson et al., 1988; Todd et al., 1991). Model organisms are often studied through QTL analysis followed by the breeding of recombinant inbred congenic lines to isolate regulatory loci, and the Collaborative Cross is a more expansive modern variation of this theme that combines these two steps together. QTL map- ping of disease susceptibility in mice has the potential to yield a veritable avalanche of information about complementary facets of disease initiation and pathogenesis. For example, efforts by Edward Wakeland and others to determine differential suscepti- bility to systemic lupus erythematosus (sle) between resistant and acutely lupus-prone inbred mouse strains led to the identiﬁcation of Ly108 and other SLAM family members as key modulators of B cell tolerance (Kumar et al., 2006), lack of proper Fcgr2b upregulation as a potentiator of IgG production (Rahman et al., 2007), Cr2 or other closely linked genes as mediators of autore- active B- and T-cell production (Chen et al., 2005; Tchepeleva et al., 2010), and hemostatic kallikreins as important regulators of kidney pathogenesis (Liu et al., 2009). FIGURE 1 \| Forward Genetics approach for Lyme arthritis severity. Intercross populations of B6 and C3H mice were used for identiﬁcation of Quantitative Trait Loci (QTL) regulating Borrelia burgdorferi associated arthritis and other responses related to infection (Bbaa). Arthritis and other metrics of host response were assessed at 4 weeks of infection. A total genome scan was performed for each infected mouse (n = 450) and threshold permutation analysis identiﬁed loci associated with disease and response. FIGURE 1 \| Forward Genetics approach for Lyme arthritis severity. CURRENT PROGRESS: MAPPING QUANTITATIVE TRAIT LOCI THAT REGULATE LYME ARTHRITIS SEVERITY IN MICE Janis Weis, Cory Teuscher, and their collaborators performed the ﬁrst murine Lyme arthritis QTL analysis (Weis et al., 1999). This initial study used an F2 intercross between C57BL/6N (B6) and June 2014 \| Volume 4 \| Article 76 \| 4 Frontiers in Cellular and Infection Microbiology www.frontiersin.org www.frontiersin.org Bramwell et al. Forward genetics of lyme arthritis FIGURE 2 \| Location of predicted Quantitative Trait Loci, summarized from Weis et al. (1999) and Roper et al. (2001). Black and gray bands on each chromosome denote an idealized representation of the Giemsa banding pattern found in a normal mouse karyotype. Red shaded regions denote QTL regulating arthritis severity (ankle swelling, histopathology score, tendon sheath). Blue shaded regions denote QTL regulating humoral immune response (Total/Speciﬁc IgM levels, Total/Speciﬁc IgG levels, serum IL-6) or host defense (B. burgdorferi bacterial burden). regulating arthritis severity (ankle swelling, histopathology score, tendon sheath). Blue shaded regions denote QTL regulating humoral immune response (Total/Speciﬁc IgM levels, Total/Speciﬁc IgG levels, serum IL-6) or host defense (B. burgdorferi bacterial burden). FIGURE 2 \| Location of predicted Quantitative Trait Loci, summarized from Weis et al. (1999) and Roper et al. (2001). Black and gray bands on each chromosome denote an idealized representation of the Giemsa banding pattern found in a normal mouse karyotype. Red shaded regions denote QTL resistant B6 background, while B6 derived Bbaa2Bbaa3 conferred reduced severity to susceptible C3H mice, in a reciprocal fashion. This publication also described a polymorphism carried by C3H mice in the Ncf1 gene, but ruled out this candidate with a variety of studies, including the ﬁnding that B6 Ncf1−/−mice exhibited no increase in arthritis severity relative to wild type B6 controls. lod value detected in the (BALB/c × C3H) F1 × C3H intercross. This study also predicted that the combined Bbaa2Bbaa3 locus contains at least four distinct regulatory genes. It is possible that some of these loci may be implicated in other QTL studies, but the use of different inbred strains of mice and the extensive poly- morphism of this region of the genome among strains confounds the ability to directly extrapolate between studies (Lindvall et al., 2009). Ma et al. reported the generation of additional B6xC3H recip- rocal congenic lines for ﬁve intervals identiﬁed in the foun- dational QTL study (Bbaa1, Bbaa2Bbaa3, Bbaa4, and Bbaa6), plus another pair of reciprocal congenic lines for an interval on Chromosome 1 (Bbaa12) (Ma et al., 2009). Frontiers in Cellular and Infection Microbiology CURRENT PROGRESS: MAPPING QUANTITATIVE TRAIT LOCI THAT REGULATE LYME ARTHRITIS SEVERITY IN MICE Through marker- assisted selection over the course of seven iterative backcrosses, these intervals were isolated and found to be free of genetic contamination on other chromosomes, with congenic intervals ranging from 25 to 146 megabases in size. Bbaa2Bbaa3 and Bbaa4 were found to reciprocally transfer the ankle swelling and histopathology phenotypes, while the B6 allele of Bbaa6 trans- ferred protection from ankle swelling and histopathology to the C3H background. Other congenic intervals conferred no change in arthritis severity or gave inconsistent results. This study also demonstrated the added utility of congenic lines as an experi- mental resource through comparative microarray gene expression proﬁling. Based on these and other data, several subsequent studies generated congenic mouse strains to isolate putative regulatory loci in the context of an otherwise uniform genetic background. This laborious and time-intensive process is essential to convert the statistically predicted loci derived from QTL analysis into physical genetic boundaries. Congenic lines can also be used to formally interrogate potential candidate genes, by determining if the phenotype of interest is retained after such a candidate gene is excluded from the congenic interval. The presence of a strongly penetrant phenotype within a congenic interval is a strong predictor of success in further steps of positional cloning. Based on these and other data, several subsequent studies generated congenic mouse strains to isolate putative regulatory loci in the context of an otherwise uniform genetic background. This laborious and time-intensive process is essential to convert the statistically predicted loci derived from QTL analysis into physical genetic boundaries. Congenic lines can also be used to formally interrogate potential candidate genes, by determining if the phenotype of interest is retained after such a candidate gene is excluded from the congenic interval. The presence of a strongly penetrant phenotype within a congenic interval is a strong predictor of success in further steps of positional cloning. Subsequent studies used congenic mice based on the initial QTL assignments. Crandall et al. described the phenotype of two B6 × C3H congenic lines, and the evaluation of a speciﬁc can- didate gene (Crandall et al., 2005). B6 × C3H F1 mice were backcrossed seven times onto each parental background, produc- ing reciprocal congenic lines B6.C3H-Bbaa2Bbaa3 and C3H.B6- Bbaa2Bbaa3. Of note, the congenic nomenclature in mice differs from other systems, with the background strain listed ﬁrst, fol- lowed by the donor strain, followed by the introgressed locus (Jackson Laboratory, 2000). CURRENT PROGRESS: MAPPING QUANTITATIVE TRAIT LOCI THAT REGULATE LYME ARTHRITIS SEVERITY IN MICE Bbaa2Bbaa3 from the C3H donor strain was found to confer increased Lyme arthritis severity on a Subsequent studies used congenic mice based on the initial QTL assignments. Crandall et al. described the phenotype of two B6 × C3H congenic lines, and the evaluation of a speciﬁc can- didate gene (Crandall et al., 2005). B6 × C3H F1 mice were backcrossed seven times onto each parental background, produc- ing reciprocal congenic lines B6.C3H-Bbaa2Bbaa3 and C3H.B6- Bbaa2Bbaa3. Of note, the congenic nomenclature in mice differs from other systems, with the background strain listed ﬁrst, fol- lowed by the donor strain, followed by the introgressed locus (Jackson Laboratory, 2000). Bbaa2Bbaa3 from the C3H donor strain was found to confer increased Lyme arthritis severity on a In the process of further reﬁning these congenic intervals, Bramwell et al. described the implementation of high throughput SNP genotyping and high resolution melting analysis (Bramwell et al., 2012). This improved genotyping methodology took advan- tage of the recently published Sanger high-resolution sequence of the C3H mouse, allowing enhanced comparison with the previ- ously published genome of the B6 reference strain (Keane et al., June 2014 \| Volume 4 \| Article 76 \| 5 www.frontiersin.org www.frontiersin.org www.frontiersin.org Forward genetics of lyme arthritis Bramwell et al. the K/BxN serum transfer model (Monach et al., 2008). Disease severity in this model is induced by autoantibodies generated against glucose-6-phosphate isomerase, a ubiquitous glycolytic enzyme. Importantly, transfer of this serum induces a joint spe- ciﬁc inﬂammatory arthritis that occurs independently of the MHC haplotype of the recipient and reﬂects the effector phase of arthritis development. Much greater arthritis severity was observed in Gusbh congenic mice than in wild type B6 control animals, revealing a common mechanism for the pathogenesis of Lyme arthritis and rheumatoid arthritis. Thus, the identiﬁcation of genes important in Lyme arthritis also illuminated previously unrecognized pathways in RA. This linkage to a gene associ- ated with Sly syndrome, an overt congenital lysosomal storage disease (LSD), strongly implicated a common pathogenic mech- anism involving accumulation of undigested glycosaminoglycans (Tomatsu et al., 2009). This possibility was conﬁrmed by detec- tion of pronounced Alcian blue staining of sulfated GAGs in the inﬂamed joint tissues of B. burgdorferi infected and K/BxN treated mice with partial or severe Gusb deﬁciencies (Figure 5). CURRENT PROGRESS: MAPPING QUANTITATIVE TRAIT LOCI THAT REGULATE LYME ARTHRITIS SEVERITY IN MICE The association of Gusbh with increased disease severity in both Lyme- associated and rheumatoid arthritis identiﬁes Gusb as a shared immunopathology disease gene (Teuscher, 1985; Sudweeks et al., 1993; Ma et al., 2002). 2011). The Sanger database revealed the precise location of thou- sands of SNPs distinguishing B6 and C3H genomic sequences within the 20 Mbp Bbaa2 interval, exponentially increasing the ability to discriminate donor sequences and deﬁne boundaries of congenic mice, and allowing the genetic composition of the congenic lines to be tested with greater precision (Figure 3). As an added beneﬁt, the screening process was accelerated, help- ing to reduce expenses by allowing litters to be screened prior to weaning age. g g This line of investigation has recently culminated in the iden- tiﬁcation of the ﬁrst deﬁnitive natural regulator of Lyme arthritis severity in laboratory mice. With further backcrossing and reﬁne- ment of the B6.C3H-Bbaa2 congenics, Bramwell et al. describe the generation of 14 new advanced congenics that delimit the bound- aries of several regulatory sub-intervals (Figure 4) (Bramwell et al., 2014). Notably, these intervals bear striking resemblance to several of the maximal linkage peaks predicted previously (Roper et al., 2001). One narrow 1.5 Mb C3H-derived interval, surrounding and including the highest peak of linkage predicted by QTL analysis at D5Mit30 on Chromosome 5, was able to independently confer an increased arthritis severity phenotype in the context of a resistant B6 genetic background. Close scrutiny of this interval revealed only a single coding-non-synonymous polymorphism between B6 and C3H mice. This point mutation in the lysosomal enzyme beta-Glucuronidase leads to a par- tially hypomorphic allele (Gusbh) in the C3H, AKR, and CBA/J inbred strains. Peromyscus mice, which do not exhibit Lyme arthritis but serve as important reservoir hosts for B. burgdor- feri in nature, appear to carry the wild-type B6 allele of Gusb (GenBank Accession XM_006971357). The exacerbated Lyme arthritis effect conferred by Gusbh was recapitulated in a spon- taneous Gusb mutant mouse line (GusbNull), and transgenic overexpression of wild type Gusbb in C3H mice (GusbTg) pro- foundly reduced ankle swelling and histopathology. The Gusbh congenic line was further tested in an experimental model of RA, The novelty of the beta-Glucuronidase polymorphism high- lights the power and added value of forward genetic approaches. CURRENT PROGRESS: MAPPING QUANTITATIVE TRAIT LOCI THAT REGULATE LYME ARTHRITIS SEVERITY IN MICE Gusb is most often cited in the recent scientiﬁc literature as a housekeeping gene, primarily used as a reference to study some- thing more interesting, making it a most unlikely candidate for a hypothesis-driven reverse genetics study. Allelic variants of the Gusb gene were found not to be differentially expressed under FIGURE 4 \| Regulatory loci identiﬁed by analysis of advanced congenic lines, modiﬁed from Bramwell et al. (2014). (A) Each of the top 8 horizontal bars represents one B6.C3H-Bbaa2 sub-interval congenic mouse line. The horizontal axis represents the position of Bbaa2 on mouse Chromosome 5 (120.3–141.2 Mb). The black portions of each row are derived from the C3H genetic background and the white portions are derived from the B6 background. Colored boxes indicate the position boundaries and predicted effect (+/−) of multiple regulatory intervals identiﬁed by advanced congenic lines. The lower 4 rows indicate arthritis severity QTL intervals predicted by Roper et al. (2001) for the backcross/intercross populations listed on the left vertical axis. Horizontal hatching denotes QTL for ankle swelling, vertical hatching denotes QTL for histopathology or tendon sheath thickening, cross hatching denotes overlap of multiple predicted QTL. (B) Ankle swelling measurements for the eight congenic lines listed in A, with signiﬁcance assessed relative to B6 negative control. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗∗P < 0.0001. FIGURE 3 \| A SNP genotyping methodology improves congenic mapping precision, adapted from Bramwell et al. (2012). ∗- Location of all 11 microsatellite markers that can differentiate between B6 and C3H DNA across Bbaa2. # - Of the thousands of SNPs available (blue line), the 28 positions that were developed into SNP genotyping assays for improved discrimination of sub-interval congenic lines (Bramwell et al., 2014). Regulatory loci identiﬁed by analysis of advanced congenic FIGURE 4 \| Regulatory loci identiﬁed by analysis of advanced congenic lines modiﬁed from Bramwell et al (2014) (A) Each of the top 8 FIGURE 4 \| Regulatory loci identiﬁed by analysis of advanced congenic lines, modiﬁed from Bramwell et al. (2014). (A) Each of the top 8 horizontal bars represents one B6.C3H-Bbaa2 sub-interval congenic mouse line. The horizontal axis represents the position of Bbaa2 on mouse Chromosome 5 (120.3–141.2 Mb). The black portions of each row are derived from the C3H genetic background and the white portions are derived from the B6 background. CURRENT PROGRESS: MAPPING QUANTITATIVE TRAIT LOCI THAT REGULATE LYME ARTHRITIS SEVERITY IN MICE Colored boxes indicate the position boundaries and predicted effect (+/−) of multiple regulatory intervals identiﬁed by advanced congenic lines. The lower 4 rows indicate arthritis severity QTL intervals predicted by Roper et al. (2001) for the backcross/intercross populations listed on the left vertical axis. Horizontal hatching denotes QTL for ankle swelling, vertical hatching denotes QTL for histopathology or tendon sheath thickening, cross hatching denotes overlap of multiple predicted QTL. (B) Ankle swelling measurements for the eight congenic lines listed in A, with signiﬁcance assessed relative to B6 negative control. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗∗P < 0.0001. FIGURE 4 \| Regulatory loci identiﬁed by analysis of advanced congenic lines, modiﬁed from Bramwell et al. (2014). (A) Each of the top 8 horizontal bars represents one B6.C3H-Bbaa2 sub-interval congenic mouse line. The horizontal axis represents the position of Bbaa2 on mouse Chromosome 5 (120.3–141.2 Mb). The black portions of each row are derived from the C3H genetic background and the white portions are derived from the B6 background. Colored boxes indicate the position boundaries and predicted effect (+/−) of multiple regulatory intervals identiﬁed by advanced congenic lines. The lower 4 rows indicate arthritis severity QTL intervals predicted by Roper et al. (2001) for the backcross/intercross populations listed on the left vertical axis. Horizontal hatching denotes QTL for ankle swelling, vertical hatching denotes QTL for histopathology or tendon sheath thickening, cross hatching denotes overlap of multiple predicted QTL. (B) Ankle swelling measurements for the eight congenic lines listed in A, with signiﬁcance assessed relative to B6 negative control. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗∗P < 0.0001. FIGURE 3 \| A SNP genotyping methodology improves congenic mapping precision, adapted from Bramwell et al. (2012). ∗- Location of all 11 microsatellite markers that can differentiate between B6 and C3H DNA across Bbaa2. # - Of the thousands of SNPs available (blue line), the 28 positions that were developed into SNP genotyping assays for improved discrimination of sub-interval congenic lines (Bramwell et al., 2014). Frontiers in Cellular and Infection Microbiology Frontiers in Cellular and Infection Microbiology June 2014 \| Volume 4 \| Article 76 \| 6 www.frontiersin.org www.frontiersin.org Forward genetics of lyme arthritis Bramwell et al. Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) (to Cory Teuscher and Janis J. Weis). Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) (to Cory Teuscher and Janis J. Weis). REFERENCES Genes outside the major histocompatibility complex control resistance and susceptibility to experimental Lyme arthritis. Med. Microbiol. Immunol. 189, 85–90. doi: 10.1007/s004300000044 Bunikis, J., Tsao, J., Luke, C. J., Luna, M. G., Fish, D., and Barbour, A. G. (2004). Borrelia burgdorferi infection in a natural population of Peromyscus Leucopus mice: a longitudinal study in an area where Lyme Borreliosis is highly endemic. J. Infect. Dis. 189, 1515–1523. doi: 10.1086/382594 baseline conditions, and no changes in Gusb expression were detected by microarray analysis of joint tissue from naïve and infected C3H and B6 mice (Crandall et al., 2006; Bramwell et al., 2014). Thus, Gusb and other similar genes associated with LSD are not likely to be picked up by a microarray or RNA-Seq study in Lyme arthritis patients. Gusb was also not included in the ImmunoChip used in human RA and juvenile RA studies, because it had not yet been identiﬁed as a potential regulator (Eyre et al., 2012; Hinks et al., 2013). As mentioned earlier, recent devel- opment of an expanded group of recombinant inbred strains incorporating 8 strains of laboratory and wild mice would appear to be a powerful resource for investigating regulators of Lyme arthritis severity. However, none of the three inbred strains carry- ing the Gusbh polymorphism were included in the Collaborative Cross, so it could not have been identiﬁed through this approach. Against all odds, the identiﬁcation of Gusb is a prime example of how relentlessly following the phenotype throughout a process of unbiased genetic reﬁnement can overcome preconception and bias to lead the way to truly novel and unexpected discoveries. C.D.C. (2013a). Reported Cases of Lyme Disease by Year, United States, 2003–2012. [Online]. CDC. Available online at: http://www.cdc.gov/lyme/stats/ chartstables/casesbyyear html (Accessed April 9 2014) chartstables/casesbyyear.html (Accessed April 9, 2014). C.D.C. (2013b). Clinical Manifestations of Conﬁrmed Lyme Disease Cases, United States, 2001–2010 [Online]. CDC. Available online at: http://www.cdc.gov/lyme/ stats/chartstables/casesbysymptom html (Accessed April 9 2014) C.D.C. (2013b). Clinical Manifestations of Conﬁrmed Lyme Disease Cases, United States, 2001–2010 [Online]. CDC. Available online at: http://www.cdc.gov/lyme/ stats/chartstables/casesbysymptom.html (Accessed April 9, 2014 C.D.C. (2013c). Three sudden cardiac deaths associated with Lyme carditis - United States, November 2012-July 2013. Morb. Mortal. Wkly. Rep. 62, 993–996. C.D.C. (2013c). Three sudden cardiac deaths associated with Lyme carditis - Unit States, November 2012-July 2013. Morb. Mortal. Wkly. Rep. 62, 993–996. Chen, Y., Perry, D., Boackle, S. A., Sobel, E. S., Molina, H., Croker, B. REFERENCES Ahlqvist, E., Hultqvist, M., and Holmdahl, R. (2009). The value of animal mod- els in predicting genetic susceptibility to complex diseases such as rheumatoid arthritis. Arthritis Res. Ther. 11, 226. doi: 10.1186/ar2600 Alexopoulou, L., Thomas, V., Schnare, M., Lobet, Y., Anguita, J., Schoen, R. T., et al. (2002). Hyporesponsiveness to vaccination with Borrelia burgdorferi OspA in humans and in TLR1- and TLR2-deﬁcient mice. Nat. Med. 8, 878–884. doi: 10.1038/nm732 Aliprantis, A. O., Yang, R. B., Mark, M. R., Suggett, S., Devaux, B., Radolf, J. D., et al. (1999). Cell activation and apoptosis by bacterial lipoproteins through Toll-like receptor-2. Science 285, 736–739. Barthold, S. W., Beck, D. S., Hansen, G. M., Terwilliger, G. A., and Moody, K. D. (1990). Lyme borreliosis in selected strains and ages of laboratory mice. J. Infect. Dis. 162, 133–138. Barthold, S. W., Sidman, C. L., and Smith, A. L. (1992). Lyme borreliosis in genet- ically resistant and susceptible mice with severe combined immunodeﬁciency. Am. J. Trop. Med. Hyg. 47, 605–613. Bramwell, K. K., Ma, Y., Weis, J. H., Chen, X., Zachary, J. F., Teuscher, C., et al. (2014). Lysosomal beta-glucuronidase regulates Lyme and rheumatoid arthritis severity. J. Clin. Invest. 124, 311–320. doi: 10.1172/JCI72339 Bramwell, K. K., Ma, Y., Weis, J. H., Teuscher, C., and Weis, J. J. (2012). High- throughput genotyping of advanced congenic lines by high resolution melt- ing analysis for identiﬁcation of Bbaa2, a QTL controlling Lyme arthritis. BioTechniques 52, 183–190. doi: 10.2144/000113838 FIGURE 5 \| Alcian blue staining reveals excess deposition of GAGs within severely arthritic ankle joints, modiﬁed from Bramwell et al. (2014). Top left panel: Ankle joint section from a day 7 K/BxN treated B6 mouse at ×4 magniﬁcation. Remaining panels: Ankle joint section from a day 7 K/BxN treated B6.C3H-Gusbh mouse. Original magniﬁcation ×4, ×20, and ×40. Scale bars: 500, 100, 50 μm, respectively. Boxes on the top right and lower left images indicate the location of the ﬁeld magniﬁed in subsequent images. Brightbill, H. D., Libraty, D. H., Krutzik, S. R., Yang, R. B., Belisle, J. T., Bleharski, J. R., et al. (1999). Host defense mechanisms triggered by microbial lipoproteins through Toll-like receptors. Science 285, 732–736. g Brown, C. R., and Reiner, S. L. (1999). Genetic control of experimental Lyme arthritis in the absence of speciﬁc immunity. Infect. Immun. 67, 1967–1973. Brown, C. R., and Reiner, S. L. (2000). REFERENCES P., et al. (2005). Several genes contribute to the production of autoreactive B and T cells in the murine lupus susceptibility locus Sle1c. J. Immunol. 175, 1080–1089. doi: 10.4049/jimmunol.175.2.1080 Churchill, G. A., Airey, D. C., Allayee, H., Angel, J. M., Attie, A. D., Beatty, J., et al. (2004). The collaborative cross, a community resource for the genetic analysis of complex traits. Nat. Genet. 36, 1133–1137. doi: 10.1038/ng1104-1133 Crandall, H., Dunn, D. M., Ma, Y., Wooten, R. M., Zachary, J. F., Weis, J. H., et al. (2006). Gene expression proﬁling reveals unique pathways associated with differential severity of Lyme arthritis. J. Immunol. 177, 7930–7942. doi: 10.4049/jimmunol.177.11.7930 Crandall, H., Ma, Y., Dunn, D. M., Sundsbak, R. S., Zachary, J. F., Olofsson, P., et al. (2005). Bb2Bb3 regulation of murine Lyme arthritis is distinct from Ncf1 and independent of the phagocyte nicotinamide adenine dinucleotide phosphate oxidase. Am. J. Pathol. 167, 775–785. doi: 10.1016/S0002-9440(10) 62050-0 ACKNOWLEDGMENTS h h CURRENT PROGRESS: MAPPING QUANTITATIVE TRAIT LOCI THAT REGULATE LYME ARTHRITIS SEVERITY IN MICE FIGURE 5 \| Alcian blue staining reveals excess deposition of GAGs within severely arthritic ankle joints, modiﬁed from Bramwell et al. (2014). Top left panel: Ankle joint section from a day 7 K/BxN treated B6 mouse at ×4 magniﬁcation. Remaining panels: Ankle joint section from a day 7 K/BxN treated B6.C3H-Gusbh mouse. Original magniﬁcation ×4, ×20, and ×40. Scale bars: 500, 100, 50 μm, respectively. Boxes on the top right and lower left images indicate the location of the ﬁeld magniﬁed in subsequent images. ACKNOWLEDGMENTS h h Del Rio, R., Noubade, R., Subramanian, M., Saligrama, N., Diehl, S., Rincon, M., et al. (2008). SNPs upstream of the minimal promoter control IL- 2 expression and are candidates for the autoimmune disease-susceptibility locus Aod2/Idd3/Eae3. Genes Immun. 9, 115–121. doi: 10.1038/sj.gene. 6364455 The mouse chromosomal idiograms used in Figure 2 were derived from images produced by Dr. David Adler (University of Washington). This work was supported by 6133/Arthritis Foundation (to Kenneth K.C. Bramwell) and AR-43521/National June 2014 \| Volume 4 \| Article 76 \| 7 Frontiers in Cellular and Infection Microbiology www.frontiersin.org Forward genetics of lyme arthritis Bramwell et al. Drouin, E. E., Seward, R. J., Strle, K., McHugh, G., Katchar, K., Londono, D., et al. (2013). A novel human autoantigen, endothelial cell growth factor, is a target of T and B cell responses in patients with Lyme disease. Arthritis Rheum. 65, 186–196. doi: 10.1002/art.37732 Oosting, M., Ter Hofstede, H., Van De Veerdonk, F. L., Sturm, P., Kullberg, B. J., Van Der Meer, J. W., et al. (2011b). Role of interleukin-23 (IL-23) receptor signaling for IL-17 responses in human Lyme disease. Infect. Immun. 79, 4681–4687. doi: 10.1128/IAI.05242-11 Paterson, A. H., Lander, E. S., Hewitt, J. D., Peterson, S., Lincoln, S. E., and Tanksley, S. D. (1988). Resolution of quantitative traits into Mendelian factors by using a complete linkage map of restriction fragment length polymorphisms. Nature 335, 721–726. doi: 10.1038/335721a0 Eyre, S., Bowes, J., Diogo, D., Lee, A., Barton, A., Martin, P., et al. (2012). High-density genetic mapping identiﬁes new susceptibility loci for rheumatoid arthritis. Nat. Genet. 44, 1336–1340. doi: 10.1038/ng.2462 Feng, S., Barthold, S. W., Bockenstedt, L. K., Zaller, D. M., and Fikrig, E. (1995). Lyme disease in human DR4Dw4-transgenic mice. J. Infect. Dis. 172, 286–289. Radolf, J. D., Caimano, M. J., Stevenson, B., and Hu, L. T. (2012). Of ticks, mice and men: understanding the dual-host lifestyle of Lyme disease spirochaetes. Nat. Rev. Microbiol. 10, 87–99. doi: 10.1038/nrmicro2714 Garcia-Monco, J. C., and Benach, J. L. (2013). A disconnect between the neu- rospirochetoses in humans and rodent models of disease. PLoS Pathog. 9:e1003288. doi: 10.1371/journal.ppat.1003288 Nat. Rev. Microbiol. 10, 87–99. doi: 10.1038/nrmicro2714 Radolf, J. D., Norgard, M. V., Brandt, M. E., Isaacs, R. D., Thompson, P. A., and Beutler, B. (1991). Lipoproteins of Borrelia burgdorferi and Treponema pal- lidum activate cachectin/tumor necrosis factor synthesis. Analysis using a CAT reporter construct. J. Immunol. 147, 1968–1974. Glant, T. T., Adarichev, V. ACKNOWLEDGMENTS h h A., Wong, K., Yalcin, B., et al. (2011). Mouse genomic variation and its effect on phenotypes and gene regulation. Nature 477, 289–294. doi: 10.1038/nature10413 Sellati, T. J., Bouis, D. A., Kitchens, R. L., Darveau, R. P., Pugin, J., Ulevitch, R. J., et al. (1998). Treponema pallidum and Borrelia burgdorferi lipoproteins and synthetic lipopeptides activate monocytic cells via a CD14-dependent pathway distinct from that used by lipopolysaccharide. J. Immunol. 160, 5455–5464. Klempner, M. S., Wormser, G. H., Wade, K., Trevino, R. P., Tang, J., Kaslow, R. A., et al. (2005). A case-control study to examine HLA haplotype associations in patients with posttreatment chronic Lyme disease. J. Infect. Dis. 192, 1010–1013. doi: 10.1086/432733 Spach, K. M., Case, L. K., Noubade, R., Petersen, C. B., McElvany, B., Zalik, N., et al. (2010). Multiple linked quantitative trait loci within the Tmevd2/Eae3 interval control the severity of experimental allergic encephalomyelitis in DBA/2J mice. Genes Immun. 11, 649–659. doi: 10.1038/gene.2010.40 Kuehn, B. M. (2013). CDC estimates 300,000 US cases of Lyme disease annually. JAMA 310, 1110. doi: 10.1001/jama.2013.278331 Kumar, K. R., Li, L., Yan, M., Bhaskarabhatla, M., Mobley, A. B., Nguyen, C., et al. (2006). Regulation of B cell tolerance by the lupus susceptibility gene Ly108. Science 312, 1665–1669. doi: 10.1126/science.1125893 Spach, K. M., Noubade, R., McElvany, B., Hickey, W. F., Blankenhorn, E. P., and Teuscher, C. (2009). A single nucleotide polymorphism in Tyk2 controls suscep- tibility to experimental allergic encephalomyelitis. J. Immunol. 182, 7776–7783. doi: 10.4049/jimmunol.0900142 Lindvall, T., Karlsson, J., Holmdahl, R., and Andersson, A. (2009). Dissection of a locus on mouse chromosome 5 reveals arthritis promoting and inhibitory genes. Arthritis Res. Ther. 11, R10. doi: 10.1186/ar2597 Steere, A. C., and Glickstein, L. (2004). Elucidation of Lyme arthritis. Nat. Rev. Immunol. 4, 143–152. doi: 10.1038/nri1267 Liu, K., Li, Q. Z., Delgado-Vega, A. M., Abelson, A. K., Sanchez, E., Kelly, J. A., et al. (2009). Kallikrein genes are associated with lupus and glomerular basement membrane-speciﬁc antibody-induced nephritis in mice and humans. J. Clin. Invest. 119, 911–923. doi: 10.1172/JCI36728 Steere, A. C., Dwyer, E., and Winchester, R. (1990). Association of chronic Lyme arthritis with HLA-DR4 and HLA-DR2 alleles. N. Engl. J. Med. 323, 219–223. doi: 10.1056/NEJM199007263230402 Ma, R. Z., Gao, J., Meeker, N. D., Fillmore, P. D., Tung, K. S., Watanabe, T., et al. (2002). Identiﬁcation of Bphs, an autoimmune disease locus, as histamine receptor H1. Science 297, 620–623. doi: 10.1126/science.1072810 Steere, A. ACKNOWLEDGMENTS h h A., Nesterovitch, A. B., Szanto, S., Oswald, J. P., Jacobs, J. J., et al. (2004). Disease-associated qualitative and quantitative trait loci in proteoglycan-induced arthritis and collagen-induced arthritis. Am. J. Med. Sci. 327, 188–195. doi: 10.1097/00000441-200404000-00004 Rahman, Z. S., Niu, H., Perry, D., Wakeland, E., Manser, T., and Morel, L. (2007). Expression of the autoimmune Fcgr2b NZW allele fails to be upregulated in germinal center B cells and is associated with increased IgG production. Genes Immun. 8, 604–612. doi: 10.1038/sj.gene.6364423 Gregory, S. G., Sekhon, M., Schein, J., Zhao, S., Osoegawa, K., Scott, C. E., et al. (2002). A physical map of the mouse genome. Nature 418, 743–750. doi: 10.1038/nature00957 Hinks, A., Cobb, J., Marion, M. C., Prahalad, S., Sudman, M., Bowes, J., et al. (2013). Dense genotyping of immune-related disease regions identiﬁes 14 new susceptibility loci for juvenile idiopathic arthritis. Nat. Genet. 45, 664–669. doi: 10.1038/ng.2614 Risch, N., and Merikangas, K. (1996). The future of genetic studies of complex human diseases. Science 273, 1516–1517. Roper, R. J., Weis, J. J., McCracken, B. A., Green, C. B., Ma, Y., Weber, K. S., et al. (2001). Genetic control of susceptibility to experimental Lyme arthri- tis is polygenic and exhibits consistent linkage to multiple loci on chromo- some 5 in four independent mouse crosses. Genes Immun. 2, 388–397. doi: 10.1038/sj.gene.6363801 Hirschfeld, M., Kirschning, C. J., Schwandner, R., Wesche, H., Weis, J. H., Wooten, R. M., et al. (1999). Cutting edge: inﬂammatory signaling by Borrelia burgdorferi lipoproteins is mediated by Toll-like receptor 2. J. Immunol. 163, 2382–2386. Schaible, U. E., Kramer, M. D., Museteanu, C., Zimmer, G., Mossmann, H., and Simon, M. M. (1989). The severe combined immunodeﬁciency (scid) mouse. A laboratory model for the analysis of Lyme arthritis and carditis. J. Exp. Med. 170, 1427–1432. Hirschfeld, M., Ma, Y., Weis, J. H., Vogel, S. N., and Weis, J. J. (2000). Cutting edge: repuriﬁcation of lipopolysaccharide eliminates signaling through both human and murine Toll-like receptor 2. J. Immunol. 165, 618–622. doi: 10.4049/jim- munol.165.2.618 Jackson Laboratory. (2000). Nomenclature for Mouse Strains. [Online]. Available online at: http://jaxmice.jax.org/support/nomenclature/index.html (Accessed April 9, 2014). Schroder, N. W., Diterich, I., Zinke, A., Eckert, J., Draing, C., Von Baehr, V., et al. (2005). Heterozygous Arg753Gln polymorphism of human TLR-2 impairs immune activation by Borrelia burgdorferi and protects from late stage Lyme disease. J. Immunol. 175, 2534–2540. doi: 10.4049/jimmunol.175.4.2534 Keane, T. M., Goodstadt, L., Danecek, P., White, M. ACKNOWLEDGMENTS h h Major histocompatibility complex (I region) linkage and antibody correlates. J. Exp. Med. 154, 688–700. Teuscher, C., Hickey, W. F., Grafer, C. M., and Tung, K. S. (1998). A common immunoregulatory locus controls susceptibility to actively induced experimen- tal allergic encephalomyelitis and experimental allergic orchitis in BALB/c mice. J. Immunol. 160, 2751–2756. Wooten, R. M., Modur, V. R., McIntyre, T. M., and Weis, J. J. (1996). Borrelia burgdorferi outer membrane protein A induces nuclear translocation of nuclear factor-kappa B and inﬂammatory activation in human endothelial cells. J. Immunol. 157, 4584–4590. Teuscher, C., Rhein, D. M., Livingstone, K. D., Paynter, R. A., Doerge, R. W., Nicholson, S. M., et al. (1997). Evidence that Tmevd2 and eae3 may represent either a common locus or members of a gene complex controlling suscepti- bility to immunologically mediated demyelination in mice. J. Immunol. 159, 4930–4934. Wormser, G. P., Dattwyler, R. J., Shapiro, E. D., Halperin, J. J., Steere, A. C., Klempner, M. S., et al. (2006). The clinical assessment, treatment, and preven- tion of Lyme disease, human granulocytic anaplasmosis, and babesiosis: clinical practice guidelines by the Infectious Diseases Society of America. Clin. Infect. Dis. 43, 1089–1134. doi: 10.1086/508667 Yang, L., Ma, Y., Schoenfeld, R., Grifﬁths, M., Eichwald, E., Araneo, B., et al. (1992). Evidence for B-lymphocyte mitogen activity in Borrelia burgdorferi-infected mice. Infect. Immun. 60, 3033–3041. Teuscher, C., Wardell, B. B., Lunceford, J. K., Michael, S. D., and Tung, K. S. (1996). Aod2, the locus controlling development of atrophy in neonatal thymectomy- induced autoimmune ovarian dysgenesis, co-localizes with Il2, Fgfb, and Idd3. J. Exp. Med. 183, 631–637. Conﬂict of Interest Statement: The authors declare that the research was con- ducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. Todd, J. A., Aitman, T. J., Cornall, R. J., Ghosh, S., Hall, J. R., Hearne, C. M., et al. (1991). Genetic analysis of autoimmune type 1 diabetes mellitus in mice. Nature 351, 542–547. doi: 10.1038/351542a0 Tomatsu, S., Montano, A. M., Dung, V. C., Grubb, J. H., and Sly, W. S. (2009). Mutations and polymorphisms in GUSB gene in mucopolysaccharidosis VII (Sly Syndrome). Hum. Mutat. 30, 511–519. doi: 10.1002/humu.20828 Received: 30 January 2014; paper pending published: 28 March 2014; accepted: 20 May 2014; published online: 05 June 2014. Tschirren, B., Andersson, M., Scherman, K., Westerdahl, H., Mittl, P. R., and Raberg, L. (2013). ACKNOWLEDGMENTS h h C., Klitz, W., Drouin, E. E., Falk, B. A., Kwok, W. W., Nepom, G. T., et al. (2006). Antibiotic-refractory Lyme arthritis is associated with HLA-DR molecules that bind a Borrelia burgdorferi peptide. J. Exp. Med. 203, 961–971. doi: 10.1084/jem.20052471 Ma, Y., Miller, J. C., Crandall, H., Larsen, E. T., Dunn, D. M., Weiss, R. B., et al. (2009). Interval-speciﬁc congenic lines reveal quantitative trait Loci with pene- trant Lyme arthritis phenotypes on chromosomes 5, 11, and 12. Infect. Immun. 77, 3302–3311. doi: 10.1128/IAI.00396-09 Strle, K., Shin, J. J., Glickstein, L. J., and Steere, A. C. (2012). Association of a Toll- like receptor 1 polymorphism with heightened Th1 inﬂammatory responses and antibiotic-refractory Lyme arthritis. Arthritis Rheum. 64, 1497–1507. doi: 10.1002/art.34383 Meeker, N. D., Hickey, W. F., Korngold, R., Hansen, W. K., Sudweeks, J. D., Wardell, B. B., et al. (1995). Multiple loci govern the bone marrow-derived immunoreg- ulatory mechanism controlling dominant resistance to autoimmune orchitis. Proc. Natl. Acad. Sci. U.S.A. 92, 5684–5688. Sudweeks, J. D., Todd, J. A., Blankenhorn, E. P., Wardell, B. B., Woodward, S. R., Meeker, N. D., et al. (1993). Locus controlling Bordetella pertussis-induced his- tamine sensitization (Bphs), an autoimmune disease-susceptibility gene, maps distal to T-cell receptor beta-chain gene on mouse chromosome 6. Proc. Natl. Acad. Sci. U.S.A. 90, 3700–3704. Monach, P. A., Mathis, D., and Benoist, C. (2008). The K/BxN arthritis model. Curr. Protoc. Immunol. Chapter 15, Unit 15.22. doi: 10.1002/0471142735.im1522s81 Tchepeleva, S. N., Thurman, J. M., Ruff, K., Perkins, S. J., Morel, L., and Boackle, S. A. (2010). An allelic variant of Crry in the murine Sle1c lupus susceptibil- ity interval is not impaired in its ability to regulate complement activation. J. Immunol. 185, 2331–2339. doi: 10.4049/jimmunol.1000783 Oosting, M., Ter Hofstede, H., Sturm, P., Adema, G. J., Kullberg, B. J., Van Der Meer, J. W., et al. (2011a). TLR1/TLR2 heterodimers play an impor- tant role in the recognition of Borrelia spirochetes. PLoS ONE 6:e25998. doi: 10.1371/journal.pone.0025998 June 2014 \| Volume 4 \| Article 76 \| 8 Frontiers in Cellular and Infection Microbiology www.frontiersin.org Bramwell et al. Forward genetics of lyme arthritis Teuscher, C. (1985). Experimental allergic orchitis in mice. II. Association of disease susceptibility with the locus controlling Bordetella pertussis-induced sensitivity to histamine. Immunogenetics 22, 417–425. Wooley, P. H., Luthra, H. S., Stuart, J. M., and David, C. S. (1981). Type II collagen- induced arthritis in mice. I. June 2014 \| Volume 4 \| Article 76 \| 9 Frontiers in Cellular and Infection Microbiology ACKNOWLEDGMENTS h h Polymorphisms at the innate immune receptor TLR2 are associated with Borrelia infection in a wild rodent population. Proc. Biol. Sci. 280, 20130364. doi: 10.1098/rspb.2013.0364 Citation: Bramwell KK, Teuscher C and Weis JJ (2014) Forward genetic approaches for elucidation of novel regulators of Lyme arthritis severity. Front. Cell. Infect. Microbiol. 4:76. doi: 10.3389/fcimb.2014.00076 This article was submitted to the journal Frontiers in Cellular and Infection Microbiology. Weis, J. J., McCracken, B. A., Ma, Y., Fairbairn, D., Roper, R. J., Morrison, T. B., et al. (1999). Identiﬁcation of quantitative trait loci governing arthritis severity and humoral responses in the murine model of Lyme disease. J. Immunol. 162, 948–956. Copyright © 2014 Bramwell, Teuscher and Weis. This is an open-access article dis- tributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this jour- nal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Wicker, L. S., Clark, J., Fraser, H. I., Garner, V. E., Gonzalez-Munoz, A., Healy, B., et al. (2005). Type 1 diabetes genes and pathways shared by humans and NOD mice. J. Autoimmun. 25 (Suppl.), 29–33. doi: 10.1016/j.jaut.2005.09.009 June 2014 \| Volume 4 \| Article 76 \| 9 Frontiers in Cellular and Infection Microbiology www.frontiersin.org www.frontiersin.org
https://openalex.org/W2161856054	https://www.frontiersin.org/articles/10.3389/fpls.2013.00141/pdf	English	null	Root-microbe systems: the effect and mode of interaction of Stress Protecting Agent (SPA) Stenotrophomonas rhizophila DSM14405T	Frontiers in plant science	2,013	cc-by	8,559	ORIGINAL RESEARCH ARTICLE p blished 14 Ma 2013 ORIGINAL RESEARCH ARTICLE published: 14 May 2013 published: 14 May 2013 doi: 10.3389/fpls.2013.00141 Root-microbe systems: the effect and mode of interaction of Stress Protecting Agent (SPA) Stenotrophomonas rhizophila DSM14405T Peyman Alavi 1, Margaret R. Starcher 1, Christin Zachow 2, Henry Müller 1 and Gabriele Berg 1* 1 Institute of Environmental Biotechnology, Graz University of Technology, Graz, Austria 2 Austrian Centre of Industrial Biotechnology, Graz, Austria Stenotrophomonas rhizophila has great potential for applications in biotechnology and biological control due to its ability to both promote plant growth and protect roots against biotic and a-biotic stresses, yet little is known about the mode of interactions in the root-environment system. We studied mechanisms associated with osmotic stress using transcriptomic and microscopic approaches. In response to salt or root extracts, the transcriptome of S. rhizophila DSM14405T changed drastically. We found a notably similar response for several functional gene groups responsible for general stress protection, energy production, and cell motility. However, unique changes in the transcriptome were also observed: the negative regulation of ﬂagella-coding genes together with the up-regulation of the genes responsible for bioﬁlm formation and alginate biosynthesis were identiﬁed as a single mechanism of S. rhizophila DSM14405T against salt shock. However, production and excretion of glucosylglycerol (GG) were found as a remarkable mechanism for the stress protection of this Stenotrophomonas strain. For S. rhizophila treated with root exudates, the shift from the planktonic lifestyle to a sessile one was measured as expressed in the down-regulation of ﬂagellar-driven motility. These ﬁndings ﬁt well with the observed positive regulation of host colonization genes and microscopic images that show different colonization patterns of oilseed rape roots. Spermidine, described as a plant growth regulator, was also newly identiﬁed as a protector against stress. Overall, we identiﬁed mechanisms of Stenotrophomonas to protect roots against osmotic stress in the environment. In addition to both the changes in life style and energy metabolism, phytohormons, and osmoprotectants were also found to play a key role in stress protection. Correspondence: Correspondence: Gabriele Berg, Institute of Environmental Biotechnology, Graz University of Technology, Petersgasse 12, 8010 Graz, Austria. e-mail: gabriele.berg@tugraz.at Keywords: plant-microbe interaction, oilseed rape, PGPR, SPA, transcriptomics, root exudates, FISH–CLSM Edited by: Edited by: Boris Rewald, University of Natural Resources and Life Sciences (BOKU), Austria Reviewed by: Robert P. Ryan, University College Cork, Ireland Esperanza Martinez-Romero, Universidad Nacional Autonoma de Mexico, Mexico *Correspondence: Gabriele Berg, Institute of Environmental Biotechnology, Graz University of Technology, Petersgasse 12, 8010 Graz, Austria. e-mail: gabriele.berg@tugraz.at Boris Rewald, University of Natural Resources and Life Sciences (BOKU), Austria Reviewed by: Reviewed by: Robert P. Ryan, University College Cork, Ireland Esperanza Martinez-Romero, Universidad Nacional Autonoma de Mexico, Mexico Reviewed by: Robert P. Ryan, University College Cork, Ireland Esperanza Martinez-Romero, Universidad Nacional Autonoma de Mexico, Mexico INTRODUCTION rhizophila is equipped with several genes which may play a role in root colonization, such as those that encode the O-antigen, capsule polysaccha- ride biosynthesis pathways, hemagglutinin, and outer membrane adhesion proteins. However, despite this knowledge, there is still no evidence that these genes are involved in successful root-microbe interactions under salinated conditions. In addi- tion to the high salinity, the role of root exudates for this interaction was pointed out in other studies (González-Pasayo and Martínez-Romero, 2000; rev. in Bais et al., 2006). The ability of cells to respond appropriately to changing environ- mental conditions can be investigated using a transcriptomic approach. This technique offers a new and powerful tool to eval- uate these hypothetical mechanisms in situ, as shown already by van de Mortel et al. (2012) for the Pseudomonas-Arabidopsis and by López-Guerrero et al. (2012) for the Rhizobium-Phaseolus interaction. INTRODUCTION Plant growth promotion by S. rhizophila strain DSM14405T (syn. strain e-p10) was observed under greenhouse conditions (Schmidt et al., 2012) and in the highly salinated soils of Uzbekistan at levels up to 180% (Egamberdieva et al., 2011). Use of classical physiological and biochemical methods unveiled the mechanisms of plant growth promotion and biocontrol against soil-borne pathogens (Berg and Ballin, 1994; Kobayashi et al., 1995; Jacobi et al., 1996; Dunne et al., 2000; Suckstorff and Berg, 2003) as well as the production of high amounts of osmolytes trehalose and GG in response to salt stress (Roder et al., 2005). Next generation sequencing techniques have allowed for new possibilities to study plant-microbe interaction. For exam- ple, genome sequencing has given new insight into the genetic sources that provide beneﬁcial plant-associated bacteria with traits such as plant growth promotion, protection against phy- topathogens, and osmoprotection. In general, the described mode of action could be conﬁrmed due to the presence of genes pos- sibly responsible in the genome of S. rhizophila DSM14405T (Berg et al., 2013). For example, S. rhizophila possesses genes Crop cultivation in salinated soils is one of the major chal- lenges facing agriculture today. Salinated areas are increasing world-wide and plants growing under saline or water-imbalance stress are more vulnerable to diseases caused by soil-borne pathogens (FAO, 2005). Biocontrol using salt-tolerant, plant growth-promoting rhizobacteria (PGPR) to protect plant roots against high salinity and pathogens offers sustainable solutions for plant protection, and Stenotrophomonas rhizophila is a model bacterium for a rhizosphere- and phylloplane-competent, salt- tolerant PGPR (Ryan et al., 2009; Berg et al., 2010, 2013). While the species S. maltophilia has become important as a nosoco- mial human pathogen, no pathogenic potential for humans has ever been observed in the related species S. rhizophila (Wolf et al., 2002). Moreover, both species can be easily distinguished by the production of the osmoprotective substance glucosylglyc- erol (GG) (only present in S. rhizophila) and the occurrence of speciﬁc multidrug-efﬂux pumps (only present in S. maltophilia) (Ribbeck-Busch et al., 2005). May 2013 \| Volume 4 \| Article 141 \| 1 www.frontiersin.org Stress protection by Stenotrophomonas rhizophila Alavi et al. responsible for the synthesis and transport of osmoprotective molecules out of the cell. In addition, it contains a number of genes involved in the biocontrol of soil-borne pathogens and important genes that aid in the competition for nutri- ents and niches as well. Additionally, S. SALT SHOCK S. rhizophila DSM14405T was cultivated in 50 ml CAA under agi- tation at 30◦C for 13 h (per liter: 5.0 g casamino acids, 1.54 g K2HPO4·3H2O, 0.25 g MgSO4·7H2O) until an optical density of OD600 0.8 was reached. A ﬁnal salt (NaCl) concentration of 3% in the medium was reached by using a sterile concentrated sodium chloride stock solution (0.3 g l−1). After 2.7 h cultivation in the medium containing 3% salt, the S. rhizophila DSM14405T culture (OD600 = 0.9) was used for RNA extraction. Two independent replicates were performed as described above. RNA EXTRACTION AND TRANSCRIPTOMIC ANALYSES RNA was extracted using the RNAprotect® Bacteria Reagent (Qiagen, Hilden, Germany). Total rRNA was removed and mRNA was enriched using the MICROBExpress kit, according to the manufacturer’s protocol (Invitrogen, Carlsbad, USA). The mRNA was sequenced using LGC Genomics (Berlin, Germany), and data collection was performed using MicroDiscovery (Berlin, Germany). The data used for assessing the changes in gene tran- scription correspond to normalized values for the number of reads that uniquely mapped to each CDS. Transcription fold change for each CDS was assessed by dividing the corresponding value from the cells that were either treated with root exudates or exposed to salt shock by those from the control group. Of the total genes either up or down-regulated, only those showing fold changes greater than or equal to 1.5 and less than or equal to 0.6 were considered signiﬁcantly impacted. The objective of our study was to investigate the response to changing environmental conditions associated with osmotic stress (1) salt stress and (2) root exudates to understand stress protection against changing osmotic conditions of roots by the endophytic bacterium S. rhizophila DSM14405T in more detail. We hypothesized that there is a general response to changing osmolarities, but also a speciﬁc answer to each other of the two parameters which are important for colonizing the root system of plants. GERMINATION POUCH COLONIZATION ASSAY A batch of 200 oilseed rape seeds were surface-sterilized with 40 ml of 3% NaOCl for 1 min and subsequently washed twice with 40 ml of water for 1 min each time. Surface-sterilized seeds were inoculated with S. rhizophila DSM14405T by incubating in a 2 ml cell suspension containing 107 CFU ml−1. The con- trol included seeds treated with 0.85% NaCl. Twelve seeds per treatment were placed into 2 (6 seeds per pouch) sterile Cyg™ger- mination pouches (Mega International, West St. Paul, MN, USA) wetted with 10 ml of sterilized deionized water or 1.25% NaCl solution. Germination pouches were then placed in sterile, asepti- cally sealed containers and placed in a growing chamber for 9 days with controlled day and night settings (12 h of light at 25◦C and 12 h of dark at 20◦C). After 9 days of growth, roots of 3 seedlings were combined for determination of cell counts resulting in 4 replicates per individual treatment. All root material was cut and transferred to Whirlpak® bags (Carl Roth, Karlsruhe, Germany) containing 2 ml of 0.85% NaCl solution. The roots in the bags were then crushed using a pestle to form a homogenous suspen- sion, which was subsequently serially diluted and drop-streaked onto LB Petri dishes. The plates were then incubated at 30◦C for 24 h. TREATMENT WITH OILSEED RAPE EXUDATES Root exudates were collected from oilseed rape cultivar Californium (Kwizda, Austria) and grown for 14 days in gnoto- biotic systems of 50 ml of sterilized vermiculite packaged in pots and covered with lids (Metro, Austria). Prior to sowing the seeds, about 50 ml of tap water was amended with 1/10 [v/v] of mini- mal medium (Gamborgs B5 basal salt mixture; Duchefa), and the seeds were surface-sterilized in sodium hypochlorite (10% wt/wt) for 10 min and washed successively with sterile water under ster- ile conditions. No seeds were sown in the control system. Plant and control systems were arranged in a replicate randomized block design and maintained at 20◦C under 16-h light and 8-h dark conditions. After 14 days, plants were removed and the root exudates and liquid from the control system were collected in sterile bags and squeezed. To corroborate sterility, both root mate- rial and exudates were plated on nutrient agar. Root exudates were centrifuged (10 min, 5000 × g), and the supernatant was collected, ﬁlter-sterilized (ﬁrst 0.45 µm, second 0.22 µm ﬁlter, Millipore), and stored at −20◦C in the dark until use. S. rhi- zophila DSM14405T was cultivated under agitation in 40 ml CAA (per liter: 5.0 g casamino acids, 1.54 g K2HPO4·3H2O, 0.25 g MgSO4·7H2O) and supplemented with 10 ml of the root exudates and the control liquid, respectively, at 30◦C for 48 h. Cells were harvested using centrifugation at 2500 × g for 1 min for RNA extraction. TRANSCRIPTIONAL RESPONSE OF S. rhizophila DSM14405T TO SALT STRESS Under salt stress of 3% NaCl, a total number of 912 and 1521 genes of S. rhizophila DSM14405T were signiﬁcantly up and down-regulated, respectively. The impact of salt shock on the transcription of S. rhizophila DSM14405T with respect to vari- ous functional gene groups is shown in Figure 1. The majority of functional groups were strongly affected, such as up-regulated genes involved in translation, synthesis of the cell wall, outer or cytoplasm membrane, nucleotide and amino acid transport and metabolism, and the production and conversion of energy. In contrast, genes involved in cell motility, secretion, intracel- lular trafﬁcking, defense mechanisms, and the transport and metabolism of carbohydrates and inorganic ions are down- regulated. Moreover, genes responsible for lipid metabolism and hypothetical genes are somewhat ambiguously affected by salt stress as some are up while others down-regulated. Of those genes with a signiﬁcant transcription fold change discussed above, some code for products with a known phys- iological function and are presented in Table 2. The complete list of S. rhizophila DSM14405T genes with a signiﬁcant tran- scription fold change is presented in Tables S2A, S2B. Cell wall breakdown and cell adherence are early and crucial steps in host- plant colonization. As presented in Table 2, the treatment of S. rhizophila DSM14405T cells with oilseed rape seedling exudates resulted in enhanced expression of cbg-1 and xynB that code for beta-glucosidase and xylanase B, respectively, and are involved in cell wall breakdown. Furthermore, both these genes are con- served among plant-associated Stenotrophomonas strains as they are present in both S. rhizophila DSM14405T and the plant- beneﬁting S. maltophilia R551-3, but absent from the human- pathogenic S. maltophilia K279a. In addition, Sr14405 2818, which is also up-regulated by 2.4 folds, codes for an adhesin pro- tein and is homologous to the haemagglutinin-like protein coding gene from the human-pathogenic S. maltophilia K279a (Table 2). Other up-regulated genes include two adjacent genes, mdtI and mdtJ that both code for spermidine export proteins. Spermidine is a plant growth regulator and has been recently shown to strongly promote the growth of arugula plants (Al-Whaibi et al., 2012). Moreover, several genes that code for multidrug resistance pumps, efﬂux transporters, heavy metal transport systems, and Of the genes that are signiﬁcantly impacted by salt stress in S. rhizophila DSM14405T (Table 1, Tables S1A, S1B), a number of those responsible for general and speciﬁc stress responses are up-regulated. www.frontiersin.org FLUORESCENT in situ HYBRIDIZATION (FISH) To study the oilseed rape colonization ability of S. rhizophila DSM14405T using confocal microscopy, the oilseed rape roots grown in seed germination pouches were ﬁxed with 4% paraformaldehyde/phosphate buffered saline (PBS) (3:1 vol/vol). The control group contained roots without bacterial treatment. May 2013 \| Volume 4 \| Article 141 \| 2 Frontiers in Plant Science \| Functional Plant Ecology Stress protection by Stenotrophomonas rhizophila Alavi et al. The ﬁxed samples were then stored in PBS/ 96% ethanol (1:1) at −20◦C. The FISH probes were purchased from genXpress® (Wiener Neudorf, Austria), and the in-tube FISH was performed as described by Cardinale et al. (2008). The FISH probes used for the hybridization step were labeled with the ﬂuorescent dye Cy3 and included EUB338 (Amman et al., 1990), EUB338 II, and EUB338 III (Daims et al., 1999), all directing eubacteria. An equimolar ratio of the FISH probes was used for the hybridization step to detect S. rhizophila DSM14405T. In this step, 30% for- mamide was added to the samples which were then subsequently incubated in a water bath (43◦C) for 90 min. After hybridiza- tion, the samples were washed at 44◦C for 15 min. Microscopy and image capturing were performed using a Leica TCS SPE con- focal microscope (Leica Microsystems, Wetzlar, Germany) with the Leica ACS APO 63X OIL CS objective (NA: 1.30). A z-step of 0.4–0.8 µm was applied to acquire confocal stacks. and other harmful environmental factors (Monday and Schiller, 1996; Stapper et al., 2004). Furthermore, speciﬁc secretion and transport systems such as those that code for the type VI secre- tion system (TVISS) are strongly up-regulated in S. rhizophila DSM14405T under salt shock, however, it should be noted that closely related plant-associated Stenotrophomonas strains such as S. maltophilia R551-3 lack the TVISS. Moreover, genes involved in the conversion and transport of substances through the cell wall and those responsible for cell division are also up-regulated. Genes responsible for ﬂagellar apparatus and ﬁmbriae- biosynthesis genes are comparatively down-regulated in S. rhi- zophila DSM14405T under salt shock. Similarly, salt shock also negatively impacted the predicted capsule biosynthesis genes. TRANSCRIPTIONAL RESPONSE OF S. rhizophila DSM14405T TO ROOT EXUDATES TRANSCRIPTIONAL RESPONSE OF S. rhizophila DSM14405T TO ROOT EXUDATES TRANSCRIPTIONAL RESPONSE OF S. rhizophila DSM14405T TO ROOT EXUDATES A total of 763 and 246 genes were signiﬁcantly up and down- regulated, respectively, as a result of the addition of oilseed-rape root exudates (Figure 2). In general, the effect of root exudates on the functional groups is indeterminate, as some genes of a partic- ular group are up-regulated while others are transcribed in lesser numbers. However, some functional groups are equally affected by plant root exudates, as the transcription of almost all cor- responding gene sequences is either up or down-regulated. For instance, as shown in Figure 2, root exudates have only a positive effect on the transcription of genes responsible for amino acid, nucleotide, and carbohydrate transport and metabolism, as well as biogenesis of cell membranes, transport of substances through the cell, and the genes responsible for the transport of secondary metabolites and coenzymes. Conversely, genes involved in the secretion, transport, and metabolism of inorganic ions as well as in cell motility are mainly down regulated in response to root exudate stress. TRANSCRIPTIONAL RESPONSE OF S. rhizophila DSM14405T TO SALT STRESS While surA and dnaJ code for general stress chap- erones, ggpS and ycaD build a well-known salt stress response mechanism in S. rhizophila DSM14405T through the synthesis of the osmolyte GG and show a fold change of 8.3, and 7.7, respec- tively (Hagemann et al., 2008). Moreover, two genes responsible for cold shock, deaD and cspA, are also strongly up-regulated under salt stress. In addition, the transcription of ousA that codes for an osmotic stress protein is positively affected as well. Cellular ion exchange mechanisms and some iron uptake genes are also up-regulated in S. rhizophila DSM14405T as the result of salt stress. Although unable to synthesize xanthan, S. rhizophila DSM14405T possesses some of the up-regulated xanthan-coding genes including xanA, xanB, and rmlAC. These genes are involved in bioﬁlm formation in addition to their role in xanthan biosynthesis (Huang et al., 2006). Likewise, the alginate coding gene algJ shows a fold change of 3.2 as a result of salt shock. Alginate is an exopolysaccharide involved in the development and architecture of bioﬁlms that protect bacteria from antibiotics May 2013 \| Volume 4 \| Article 141 \| 3 www.frontiersin.org Stress protection by Stenotrophomonas rhizophila Alavi et al. FIGURE 1 \| The effect of salt shock on the gene expression of various functional gene groups in S. rhizophila DSM14405T. A total of 912 and 1521 genes were signiﬁcantly up and down-regulated, respectively. The impact of salt stress on most functional gene groups is clearly pronounced, as a given functional group shows either an increase or decrease in the transcription of genes belonging to that group. Genes involved in translation, synthesis of the cell wall, outer or cytoplasm membrane, nucleotide and amino acid transport and metabolism, energy production and conversion are up-regulated. In contrast, genes involved in cell motility, secretion, and intracellular trafﬁcking defense mechanisms and transport and metabolism of carbohydrates and inorganic ions are down-regulated. Genes involved in lipid metabolism, and the hypothetical genes are rather ambiguously affected by salt stress, as some of these are up and others down-regulated. The values above each column correspond to the percentage abundance of the corresponding functional group relative to the total count of the up and down-regulated genes. The transcription fold change for each CDS corresponds to the ratio calculated for S. rhizophila under salt shock compared with the control. Data are presented as the mean value of two independent replicates. TRANSCRIPTIONAL RESPONSE OF S. rhizophila DSM14405T TO SALT STRESS The error bar shown on each functional group corresponds to the mean value of errors for all genes belonging to that functional group FIGURE 1 \| The effect of salt shock on the gene expression of various functional gene groups in S. rhizophila DSM14405T. A total of 912 and 1521 genes were signiﬁcantly up and down-regulated, respectively. The impact of salt stress on most functional gene groups is clearly pronounced, as a given functional group shows either an increase or decrease in the transcription of genes belonging to that group. Genes involved in translation, synthesis of the cell wall, outer or cytoplasm membrane, nucleotide and amino acid transport and metabolism, energy production and conversion are up-regulated. In contrast, genes involved in cell motility, secretion, and intracellular trafﬁcking, defense mechanisms, and transport and metabolism of carbohydrates and inorganic ions are down-regulated. Genes involved in lipid metabolism, and the hypothetical genes are rather ambiguously affected by salt stress, as some of these are up and others down-regulated. The values above each column correspond to the percentage abundance of the corresponding functional group relative to the total count of the up and down-regulated genes. The transcription fold change for each CDS corresponds to the ratio calculated for S. rhizophila under salt shock compared with the control. Data are presented as the mean value of two independent replicates. The error bar shown on each functional group corresponds to the mean value of errors for all genes belonging to that functional group. of carbohydrates and inorganic ions are down-regulated. Genes involved in lipid metabolism, and the hypothetical genes are rather ambiguously affected by salt stress, as some of these are up and others down-regulated. The values above each column correspond to the percentage abundance of the corresponding functional group relative to the total count of the up and down-regulated genes. The transcription fold change for each CDS corresponds to the ratio calculated for S. rhizophila under salt shock compared with the control. Data are presented as the mean value of two independent replicates. The error bar shown on each functional group corresponds to the mean value of errors for all genes belonging to that functional group. FIGURE 1 \| The effect of salt shock on the gene expression of various functional gene groups in S. rhizophila DSM14405T. A total of 912 and 1521 genes were signiﬁcantly up and down-regulated, respectively. TRANSCRIPTIONAL RESPONSE OF S. rhizophila DSM14405T TO SALT STRESS The impact of salt stress on most functional gene groups is clearly pronounced, as a given functional group shows either an increase or decrease in the transcription of genes belonging to that group. Genes involved in translation, synthesis of the cell wall, outer or cytoplasm membrane, nucleotide and amino acid transport and metabolism, energy production and conversion are up-regulated. In contrast, genes involved in cell motility, secretion, and intracellular trafﬁcking, defense mechanisms, and transport and metabolism FIGURE 1 \| The effect of salt shock on the gene expression of various functional gene groups in S. rhizophila DSM14405T. A total of 912 and 1521 genes were signiﬁcantly up and down-regulated, respectively. The impact of salt stress on most functional gene groups is clearly pronounced, as a given functional group shows either an increase or decrease in the transcription of genes belonging to that group. Genes involved in translation, synthesis of the cell wall, outer or cytoplasm membrane, nucleotide and amino acid transport and metabolism, energy production and conversion are up-regulated. In contrast, genes involved in cell motility, secretion, and intracellular trafﬁcking, defense mechanisms, and transport and metabolism resistance against antibiotics are positively affected by seedling exudates. by the addition of oilseed rape seedling exudates. The complete list of the ﬂagellar apparatus-coding genes that are down-regulated is not conﬁned to those noted in Table 2, and is presented in Tables S2A, S2B. Likewise, the S. rhizophila DSM14405T contains two ﬂagella-encoding gene blocks that are almost entirely negatively affected May 2013 \| Volume 4 \| Article 141 \| 4 Frontiers in Plant Science \| Functional Plant Ecology Stress protection by Stenotrophomonas rhizophila Alavi et al. Table 1 \| Selected S. TRANSCRIPTIONAL RESPONSE OF S. rhizophila DSM14405T TO SALT STRESS rhizophila DSM14405T genes with known biological roles imp Gene (Putative) Product Transcr ggpS Glucosylglycerol-phosphate synthase 8.3 ycaD MFS-type transporter 7.7 xanA Hosphohexane mutases 3.0 xanB Xanthan biosynthesis protein xanB 2.9 rmlC dTDP-4-dehydrorhamnose 3,5-epimerase 2.3 algJ Alginate biosynthesis protein 3.2 Sr14405 2749 TVISS effector, Hcp1 family protein 2.6 Sr14405 2755 Rhs element Vgr protein 4.2 Sr14405 2761 Rhs element Vgr protein 3.4 icmF TVISS protein 5.0 Sr14405 2781 TVISS-associated protein, ImpA family 2.4 Sr14405 2791 Rhs element Vgr protein 6.8 deaD Cold-shock DEAD box protein A homolog 8.4 cspA Major cold shock protein 4.3 Sr14405 1916 Beta-lactamase L2 protein 6.3 tetA Tetracycline resistance protein 3.8 Sr14405 1293 Bacterioferritin-associated ferredoxin 2.1 bfr Bacterioferritin 5.9 hisl Histidine biosynthesis bifunctional protein 4.6 ousA Osmoprotectant uptake system protein 4.2 surA Chaperone protein 3.8 dnaJ Chaperone 3.1 ompW Outer membrane protein 3.6 oprF Outer membrane protein 2.6 ftsQ Cell division protein 3.4 ftsA Cell division protein 2.3 ftsY Cell division protein 2.6 ftsZ Cell division protein 2.0 lptF Lipopolysaccharide export system permease protein 4.5 IptG Lipopolysaccharide export system permease protein 3.3 Sr14405 2454 Peptidoglycan-associated outer membrane lipoprotein 2.5 mltD Muramidase 3.2 Sr14405 1936 Peptidoglycan-associated lipoprotein 2.8 Sr14405 4324 Cell morphology protein 2.7 clcA H(+)/Cl(−) exchange transporter 2.7 kefA Potassium efﬂux system 2.3 ﬂgA Flagellar basal body P-ring biosynthesis 0.5 ﬂgC Flagellar basal body P-ring biosynthesis 0.3 ﬂgG Flagellar basal body P-ring biosynthesis 0.3 ﬂgF Flagellar basal body P-ring biosynthesis 0.3 ﬂiF Flagellar basal body P-ring biosynthesis 0.3 ﬂhA Flagellar biosynthesis 0.3 ﬂhB Flagellar biosynthesis 0.3 cfaB CFA/I ﬁmbrial subunit B 0.3 cfaC CFA/I ﬁmbrial subunit C 0.4 csoB Fimbrial subunit B 0.2 Sr14405 3215 Capsule polysaccharide biosynthesis protein 0.1 Sr14405 3217 Putative UDP-glucose 4-epimerase 0.2 wzc Tyrosine-protein kinase 0.2 The values for fold changes correspond to the S. rhizophila DSM14405T subjected to the 3% Table 1 \| Selected S. rhizophila DSM14405T genes with known biological roles impacted by salt shock. TRANSCRIPTIONAL RESPONSE OF S. rhizophila DSM14405T TO SALT STRESS Gene (Putative) Product Transcription fold change Biological function ggpS Glucosylglycerol-phosphate synthase 8.3 Salt shock response protein ycaD MFS-type transporter 7.7 Salt shock response protein transporter xanA Hosphohexane mutases 3.0 Xanthan biosynthesis; bioﬁlm formation xanB Xanthan biosynthesis protein xanB 2.9 Xanthan biosynthesis; bioﬁlm formation rmlC dTDP-4-dehydrorhamnose 3,5-epimerase 2.3 Xanthan biosynthesis; bioﬁlm formation algJ Alginate biosynthesis protein 3.2 Alginate biosynthesis Sr14405 2749 TVISS effector, Hcp1 family protein 2.6 Type VI secretion system Sr14405 2755 Rhs element Vgr protein 4.2 Type VI secretion system Sr14405 2761 Rhs element Vgr protein 3.4 Type VI secretion system icmF TVISS protein 5.0 Type VI secretion system Sr14405 2781 TVISS-associated protein, ImpA family 2.4 Type VI secretion system Sr14405 2791 Rhs element Vgr protein 6.8 Type VI secretion system deaD Cold-shock DEAD box protein A homolog 8.4 Cell shock response cspA Major cold shock protein 4.3 Cell shock response Sr14405 1916 Beta-lactamase L2 protein 6.3 Antibiotic resistance tetA Tetracycline resistance protein 3.8 Antibiotic resistance Sr14405 1293 Bacterioferritin-associated ferredoxin 2.1 Iron uptake and transport bfr Bacterioferritin 5.9 Iron uptake and transport hisl Histidine biosynthesis bifunctional protein 4.6 Histidine biosynthesis ousA Osmoprotectant uptake system protein 4.2 Osmotic stress response surA Chaperone protein 3.8 Cellular stress response dnaJ Chaperone 3.1 Stress response ompW Outer membrane protein 3.6 Transport oprF Outer membrane protein 2.6 Transport ftsQ Cell division protein 3.4 Cell division ftsA Cell division protein 2.3 Cell division ftsY Cell division protein 2.6 Cell division ftsZ Cell division protein 2.0 Cell division lptF Lipopolysaccharide export system permease protein 4.5 Cell wall transport IptG Lipopolysaccharide export system permease protein 3.3 Cell wall transport Sr14405 2454 Peptidoglycan-associated outer membrane lipoprotein 2.5 Cell wall protein mltD Muramidase 3.2 Bacterial cell wall biodegradation Sr14405 1936 Peptidoglycan-associated lipoprotein 2.8 Cell wall structure protein Sr14405 4324 Cell morphology protein 2.7 Unknown clcA H(+)/Cl(−) exchange transporter 2.7 Ion regulation kefA Potassium efﬂux system 2.3 Ion regulation ﬂgA Flagellar basal body P-ring biosynthesis 0.5 Flagellar-driven motility ﬂgC Flagellar basal body P-ring biosynthesis 0.3 Flagellar-driven motility ﬂgG Flagellar basal body P-ring biosynthesis 0.3 Flagellar-driven motility ﬂgF Flagellar basal body P-ring biosynthesis 0.3 Flagellar-driven motility ﬂiF Flagellar basal body P-ring biosynthesis 0.3 Flagellar-driven motility ﬂhA Flagellar biosynthesis 0.3 Flagellar-driven motility ﬂhB Flagellar biosynthesis 0.3 Flagellar-driven motility cfaB CFA/I ﬁmbrial subunit B 0.3 Fimbriae synthesis cfaC CFA/I ﬁmbrial subunit C 0.4 Fimbriae synthesis csoB Fimbrial subunit B 0.2 Fimbriae synthesis Table 1 \| Selected S. TRANSCRIPTIONAL RESPONSE OF S. rhizophila DSM14405T TO SALT STRESS While some functional groups are both positively and negatively regulated by root exudates, others show a clear and pronounced alteration, as the majority of the corresponding genes are either up or down-regulated. For example, genes responsible for amino acid, nucleotide, and carbohydrate transport and metabolism, and those TRANSCRIPTIONAL RESPONSE OF S. rhizophila DSM14405T TO SALT STRESS rhizophila DSM14405T genes with known biological roles impacted by salt shock. May 2013 \| Volume 4 \| Article 141 \| 5 www.frontiersin.org Stress protection by Stenotrophomonas rhizophila Alavi et al. FIGURE 2 \| The effect of oilseed rape seedling exudates on gene expression of various functional gene groups in S. rhizophila DSM14405T. A total of 763 and 246 genes were signiﬁcantly up and down-regulated, respectively. While some functional groups are both positively and negatively regulated by root exudates, others show a clear and pronounced alteration, as the majority of the corresponding genes are either up or down-regulated. For example, genes responsible for amino acid, nucleotide, and carbohydrate transport and metabolism, and those involved in cell wall, outer-membrane or cytoplasmic membrane biogenesis and transport as well as genes responsible for the transport of secondary metabolites and coenzymes are mainly up-regulated. In contrast, genes involved in cell motility and secretion, and those responsible for the transport and metabolism of inorganic ions are mainly down-regulated. The value above each column corresponds to percentage abundance of the corresponding functional group in the total count of the up or down-regulated genes. involved in cell wall, outer-membrane or cytoplasmic membrane biogenesis and transport as well as genes responsible for the transport of secondary metabolites and coenzymes are mainly up-regulated. In contrast, genes involved in cell motility and secretion, and those responsible for the transport and metabolism of inorganic ions are mainly down-regulated. The value above each column corresponds to percentage abundance of the corresponding functional group in the total count of the up or down-regulated genes. involved in cell wall, outer-membrane or cytoplasmic membrane biogenesis and transport as well as genes responsible for the transport of secondary metabolites and coenzymes are mainly up-regulated. In contrast, genes involved in cell motility and secretion, and those responsible for the transport and metabolism of inorganic ions are mainly down-regulated. The value above each column corresponds to percentage abundance of the corresponding functional group in the total count of the up or down-regulated genes. FIGURE 2 \| The effect of oilseed rape seedling exudates on gene expression of various functional gene groups in S. rhizophila DSM14405T. A total of 763 and 246 genes were signiﬁcantly up and down-regulated, respectively. DISCUSSION expression. Numerous functional gene groups are up-regulated in response to osmotic stress factors and include those involved in energy production, as well as those involved in the synthesis and transport of cell wall, outer membrane, and cytoplasmic membrane, and those responsible for the metabolism and transport of amino acids, nucelotides, and secondary metabolites (Figure 3). Conversely, genes responsi- ble for cell motility, secretion, intracellular trafﬁcking, and the transport and metabolism of inorganic ions are down- regulated under both salt stress and treatment with root exudates. We studied the response of S. rhizophila DSM14405T to osmotic changes in the form of plant root exudates and salt shock at both the physiological and molecular level. Even though we found a notable similarity in how the cell copes with these stressors, the individual responses included a great deal of speciﬁcity at the gene level thus. The response of S. rhizophila DSM14405T to both oilseed rape root exudates and salt corresponds with several functional gene groups including those responsible for the synthesis and transport of cell wall, outer membrane, and cytoplasmic membrane, the metabolism and transport of amino acids, nucleotide, and secondary metabolites, energy production, cell motility, secretion and intracellular trafﬁcking, and the trans- port and metabolism of inorganic ions. For S. rhizophila treated with root exudates, however, the shift from the planktonic lifestyle to a sessile one as expressed in the down-regulation of ﬂagellar- driven motility is targeted to colonize the plant host, and is well in accordance with the observed positive regulation of host colo- nization genes. In addition to the changes in behavior and lifestyle of the bacterium, several bioactive substances were identiﬁed as key factors in stress protection. The ﬁrst among them is the plant growth regulator, spermidine. Although this substance is known to strongly promote growth, this is the ﬁrst evidence to show its involvement in stress protection of roots. The second group includes osmoprotective substances which were both produced SIMILARITIES IN THE TRANSCRIPTIONAL RESPONSE OF S. rhizophila DSM14405T TO SALT AND ROOT EXUDATES expression of the genes responsible for ﬁmbriae-driven cell motility, such as cfaB and csoB is negatively impacted by seedling exudates. Moreover, genes involved in the uptake, transport, and bioavailability of iron are also down-regulated. In response to both oilseed rape root exudates and salt shock, S. rhizophila DSM14405T copes with osmotic stress in a surprisingly similar way through the alteration of gene Frontiers in Plant Science \| Functional Plant Ecology Frontiers in Plant Science \| Functional Plant Ecology May 2013 \| Volume 4 \| Article 141 \| 6 Stress protection by Stenotrophomonas rhizophila Alavi et al. Table 2 \| Selected S. rhizophila DSM14405T genes with known biological roles impacted by plant root exudates. Gene (Putative) Product Transcription fold change Biological function mdtI Spermidine export protein 6.3 Export of the plant growth regulator spermidine mdtJ Spermidine export proteins 7.6 Export of the plant growth regulator spermidine Sr14405 2818 Adhesin 2.4 Host cell surface attachment/colonization cbg-1 Beta-glucosidase 1.7 Plant cell wall biodegradation/colonization xynB Xylanase B 1.6 Plant cell wall biodegradation/colonization Sr14405 4324 Cell morphology protein 2.2 Unknown Sr14405 1672 Generally characterized MFS-type transporter 3.0 Antibiotic resistance Sr14405 1673 Multidrug synthesis protein 8.8 Antibiotic resistance Sr14405 2718 Multidrug synthesis protein 3.5 Antibiotic resistance tetX Tetracycline resistance protein 3.5 Antibiotic resistance Sr14405 4658 Acriﬂavin resistance protein 1.6 Antibiotic resistance Sr14405 2827 Heavy metal transport and detoxciﬁcation protein 2.0 Heavy metal efﬂux system Sr14405 1538 Efﬂux transporter 1.5 Efﬂux of unknown target ﬂgA Flagellar basal body P-ring biosynthesis 0.5 Flagellar-driven motility ﬂgC Flagellar basal body P-ring biosynthesis 0.5 Flagellar-driven motility ﬂgG Flagellar basal body P-ring biosynthesis 0.5 Flagellar-driven motility ﬂgF Flagellar basal body P-ring biosynthesis 0.5 Flagellar-driven motility ﬂiF Flagellar basal body P-ring biosynthesis 0.5 Flagellar-driven motility ﬂhA Flagellar biosynthesis 0.6 Flagellar-driven motility ﬂhB Flagellar biosynthesis 0.5 Flagellar-driven motility cfaB CFA/I ﬁmbrial subunit B 0.5 Fimbriae synthesis csoB Fimbrial subunit B 0.4 Fimbriae synthesis Sr14405 1293 Bacterioferritin-associated ferredoxin 0.6 Iron uptake and transport Sr14405 1746 Heme oxygenase 0.4 Iron bioavailability fpvA Ferripyoverdine 0.5 Iron uptake and transport Sr14405 4245 Outer-membrane hemin receptor 0.4 Iron uptake and transport The values for fold changes correspond to the S. rhizophila DSM14405T treated with plant root exudates compared to the control. le 2 \| Selected S. rhizophila DSM14405T genes with known biological roles impacted by plant root exudates. COLONIZATION PATTERNS OF S. rhizophila DSM14405T ON ROOTS UNDER STRESS S. rhizophila DSM14405T intensely colonizes oilseed rape plants, as revealed by the cell count of log10 9.47 CFU g−1 root fresh weight (±0.08) for seeds treated with deionized water. The treatment of seeds with 1.25% NaCl, however, decreased the colonization ability by nearly half resulting in a cell count of log10 9.09 CFU g−1 root fresh weight (±0.18). Furthermore, microscopic images captured using FISH combined with confocal laser scanning microscopy (CLSM) also revealed a signiﬁcant decrease in the colo- nization of oilseed rape roots treated with 1.25% NaCl (Figure 4). May 2013 \| Volume 4 \| Article 141 \| 7 www.frontiersin.org Stress protection by Stenotrophomonas rhizophila Alavi et al. FIGURE 3 \| Model showing the response of S. rhizophila DSM14405T to osmotic stress: salt shock and root exudates. Functional gene groups shared in the response to oilseed rape root exudates and salt shock are presented. Several functional gene groups are up-regulated as a result of both oilseed rape root exudates and salt shock including those responsible for the synthesis and transport of cell wall, outer membrane, and cytoplasmic membrane, the metabolism and transport of amino acids, nucleotide, and secondary metabolites, and energy production. In contrast, genes responsible for cell motility, secretion and intracellular trafﬁcking, and the transport and metabolism of inorganic ions are down-regulated. d t d i hi h l d ib d li (R d t l for the synthesis and transport of cell wall, outer membrane, and cytoplasmic membrane, the metabolism and transport of amino acids, nucleotide, and secondary metabolites, and energy production. In contrast, genes responsible for cell motility, secretion and intracellular trafﬁcking, and the transport and metabolism of inorganic ions are down-regulated. FIGURE 3 \| Model showing the response of S. rhizophila DSM14405T to osmotic stress: salt shock and root exudates. Functional gene groups shared in the response to oilseed rape root exudates and salt shock are presented. Several functional gene groups are up-regulated as a result of both oilseed rape root exudates and salt shock including those responsible and excreted in high volumes as described earlier (Roder et al., 2005). FIGURE 4 \| The impact of salt stress on the capability of S. rhizophila DSM14405T to colonize the oilseed rape rhizosphere visualized using FISH-CLSM. S. rhizophila DSM14405T intensely colonizes the oilseed rape rhizosphere (left) while the treatment of seeds with 1.25% NaCl (right) severely decreases the colonization capability. An equimolar ratio of the FISH probes EUB338, EUB338 II, and EUB338 III labeled with the ﬂuorescent dye Cy3 was used in the hybridization step. Microscopic images were captured using a Leica TCS SPE confocal microscope. The Leica ACS APO 63X OIL CS objective (NA: 1.30) was used to acquire confocal stacks by applying a z-step of 0.4–0.8 µm. Spermidine is a well-known plant growth regulator and has been revealed to play a critical role in plant embryo develop- ment (Imai et al., 2004). Moreover, it has been recently shown to strongly promote the growth of arugula plants (Al-Whaibi et al., 2012). In addition, spermidine affects bioﬁlm formation in various bacterial species via multiple pathways that involve both transport and signaling networks (McGinnis et al., 2009). As a result, enhanced bioﬁlm formation or possible plant growth regulation resulting from the up-regulation of S. rhizophila sper- midine export genes would well-serve the lifestyle shift that ultimately leads to efﬁcient colonization of the plant host in the presence of oilseed rape exudates. Spermidine was found to pro- long the life span of several eukaryotic model organisms including yeasts, nematodes, ﬂies, and plants as well as signiﬁcantly reduce age-related oxidative protein damage in mice which could indi- cate a potential universal anti-aging drug for eukaryotes (Imai et al., 2004; Eisenberg et al., 2009). FIGURE 4 \| The impact of salt stress on the capability of S. REFERENCES with plants and other organ- isms. Annu. Rev. Plant Biol. 57, 233–266. with plants and other organ- isms. Annu. Rev. Plant Biol. 57, 233–266. Dunne, C., Moenne-Loccoz, Y., de Bruijn, F. J., and O’Gara, F. (2000). Overproduction of an inducible extracellular serine protease improves biological control of Pythium ultimum by Stenotrophomonas maltophilia strain W81. Microbiology 146, 2069–2078. London; New York: Springer), 445–460. Al-Whaibi, M. H., Siddiqui, M. H., Al-Munqadhi, B. M. A., Sakran, A. M., Ali, H. M., and Basalah, M. O. (2012). Inﬂuence of plant growth regulators on growth performance and photosyn- thetic pigments status of Eruca sativa Mill. J. Med. Plants Res. 6, 1948–1954. Bingle, L. E. H., Bailey, C. M., and Pallen, M. J. (2008). Type VI secre- tion: a beginner’s guide. Curr. Opin. Microbiol. 11, 3–8. Berg, G., Alavi, M., Schmidt, C. S., Egamberdieva, D., Kamilova, F., and Lugtenberg, B. (2013). “Biocontrol and osmoprotection for plants under saline conditions,” in Molecular Microbial Ecology of the Rhizosphere, ed F. J. de Bruijn (John Wiley and Sons, Inc.). Cardinale, M., De Castro, J. V. Jr., Müller, H., Berg, G., and Grube, M. (2008). In situ analysis of the bacterial community associated with the reindeer lichen Cladonia arbuscula reveals predominance of Alphaproteobacteria. FEMS Microbiol. Ecol. 66, 1–9. strain W81. Microbiology 146, 2069–2078. Egamberdieva, D., Kucharova, Z., Davranov, K., Berg, G., Makarova, N., Azarova, T., et al. (2011). Bacteria able to control foot and root rot and to promote growth of cucumber in salinated soils. Biol. Fertil. Soils 47, 197–205. Amman, R. I., Binder, B. J., Olson, R. J., Chisholm, S. W., Devereux, R., and Stahl, D. A. (1990). Combination of 16S rRNA-targeted oligonucleotide probes with ﬂow cytometry for analyzing mixed microbial popula- tions. Appl. Environ. Microbiol. 56, 1919–1925. Berg, G., and Ballin, G. (1994). Bacterial antagonists to Verticillium dahliae. J. Phytopathol. 141, 99–110. Berg, G., Egamberdieva, D., Lugtenberg, B., and Hagemann, M. (2010). “Symbiotic plant-microbe interactions: stress protection, plant growth promotion and biocontrol by Stenotrophomonas,” in Symbiosis and Stress, eds J. M. G. Seckbach and M. Grube (Dordrecht; Heidelberg; Daims, H., Brühl, A., Amann, R., Schleifer, K. H., and Wagner, M. (1999). The domain-speciﬁc probe EUB338 is insufﬁcient for the detec- tion of all bacteria: development and evaluation of a more com- prehensive probe set. Syst. Appl. Microbiol. 22, 434–444. Eisenberg, T., Knauer, H., Schauer, A., Büttner, S., and Ruckenstuhl, C. (2009). Induction of autophagy by spermidine promotes longevity. rhizophila DSM14405T to colonize the oilseed rape rhizosphere visualized using FISH-CLSM. S. rhizophila DSM14405T intensely FIGURE 4 \| The impact of salt stress on the capability of S. rhizophila DSM14405T to colonize the oilseed rape rhizosphere visualized using FISH-CLSM. S. rhizophila DSM14405T intensely colonizes the oilseed rape rhizosphere (left) while the treatment of seeds with 1.25% NaCl (right) severely decreases the colonization capability. An equimolar ratio of the FISH probes EUB338, EUB338 II, and EUB338 III labeled with the ﬂuorescent dye Cy3 was used in the hybridization step. Microscopic images were captured using a Leica TCS SPE confocal microscope. The Leica ACS APO 63X OIL CS objective (NA: 1.30) was used to acquire confocal stacks by applying a z-step of 0.4–0.8 µm. FIGURE 4 \| The impact of salt stress on the capability of S. rhizophila DSM14405T to colonize the oilseed rape rhizosphere visualized using FISH-CLSM. S. rhizophila DSM14405T intensely colonizes the oilseed rape rhizosphere (left) while the treatment of seeds with 1.25% NaCl (right) severely decreases the colonization capability. An equimolar ratio of the FISH probes EUB338, EUB338 II, and EUB338 III labeled with the ﬂuorescent dye Cy3 was used in the hybridization step. Microscopic images were captured using a Leica TCS SPE confocal microscope. The Leica ACS APO 63X OIL CS objective (NA: 1.30) was used to acquire confocal stacks by applying a z-step of 0.4–0.8 µm. GG and trehalose are well-studied general osmoprotective substances that protect cells from high salt May 2013 \| Volume 4 \| Article 141 \| 8 Frontiers in Plant Science \| Functional Plant Ecology Stress protection by Stenotrophomonas rhizophila Alavi et al. concentrations (Ferjani et al., 2003; Hincha and Hagemann, 2004). While both species produce trehalose, GG is synthesized exclusively in S. rhizophila thus distinguishing itself from the pathogenic S. maltophilia (Ribbeck-Busch et al., 2005; Roder et al., 2005). In S. rhizophila DSM14405T, ggpS and ycaD are both strongly up-regulated under 3% salt and are essential for the syn- thesis and transport of GG. This ﬁnding corresponds completely with both the general role of GG as a cell protector and previ- ous ﬁndings that the amount of GG excreted into the medium increases substantially in comparison with intracellular GG con- tent resulting from a shift of lower (less than 2%) to higher salt concentrations (Roder et al., 2005). Thus, GG production is the speciﬁc mechanism of S. SUPPLEMENTARY MATERIAL The Supplementary Material for this article can be found online at: http://www.frontiersin.org/Functional_Plant_Ecology/ 10.3389/fpls.2013.00141/abstract The Supplementary Material for this article can be found online at: http://www.frontiersin.org/Functional_Plant_Ecology/ 10.3389/fpls.2013.00141/abstract Table S1A \| Signiﬁcantly up-regulated genes in S. rhizophila DSM14405T under salt shock. The treatment of S. rhizophila DSM14405T with oilseed rape exudates resulted in the down-regulation of iron uptake and transport genes. This change is possibly due to the fact that once treated with oilseed rape exudates, S. rhizophila is provided with biologically available iron bound with plant siderophores result- ing in less demand for the synthesis of bacterial siderophores to bind and uptake biologically unavailable iron ions present in the Table S1B \| Signiﬁcantly down-regulated genes in S. rhizophila DSM14405T under salt shock. Table S2A \| Signiﬁcantly up-regulated genes in S. rhizophila DSM14405T under treatment with root exudates. Table S2B \| Signiﬁcantly down-regulated genes in S. rhizophila DSM14405T under treatment with root exudates. rhizophila DSM14405T to cope with salt stress. medium. Moreover, treatment with root exudates resulted in the up-regulation of several multidrug resistance pumps thus demon- strating that the role of the multidrug pumps is not conﬁned to the export of antibiotics out of the cell, but also includes a more general function with the transport of other substances once in a hyperosmotic environment. Several questions still remain, such as the reason for posi- tive regulation of several genes responsible for iron uptake and transport, as well as the reason for cell division under salt shock or the role of other remaining genes that are signiﬁcantly up or down-regulated by osmotic stress factors. However, this work has shed light on the so far unknown mode of action of S. rhizophila DSM14405T to a-biotic changes by unveiling the mechanisms that are harnessed to establish in highly salinated plant root ecosystems. TVISS genes represent a novel key virulence system used by many important pathogenic bacteria in eukaryotic host infection (Bingle et al., 2008; Pieper et al., 2009) and are intensely up- regulated under salt shock. In addition, plant growth promotion increased up to 180% in the highly salinated soils of Uzbekistan in the presence of S. rhizophila DSM14405T (Egamberdieva et al., 2011). Similarly, Schmidt et al. (2012) reported that this plant growth promotion effect was more pronounced in soil than under gnotobiotic conditions suggesting it is due to the control of diseases and deleterious microorganisms. Together with the absence of TVISS genes from the other known plant- beneﬁcial Stenotrophomonas strains with no plant growth pro- moting effect under saline conditions, these ﬁndings imply that the salt-stimulated S. rhizophila DSM14405T TVISS is indirectly harnessed to promote plant growth by eliminating harmful and deleterious microorganisms in soil. ACKNOWLEDGMENTS This study was supported by the Austrian Science Foundation FWF (P 20542-B16) by a grant to Gabriele Berg. The genome sequence was funded by a project in the Austrian Centre of Industrial Biotechnology, which has been supported by the Austrian BMWFJ, BMVIT, SFG, Standortagentur Tirol, and ZIT through the Austrian FFG-COMET-Funding Program. REFERENCES Conﬂict of Interest Statement: The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. González-Pasayo, R., and Martínez- Romero, E. (2000). Multiresistance genes of Rhizobium etli CFN42. Mol. Plant Microbe Interact. 13, 572–577. Ryan, R. P., Monchy, S., Cardinale, M., Taghavi, S., Crossman, L., Avison, M. B., et al. (2009). Versatility and adaptation of bacteria from the genus Stenotrophomonas. Nat. Microbiol. Rev. 7, 514–525. López-Guerrero, M. G., Ormeño- Orrillo, E., Acosta, J. L., Mendoza-Vargas, A., Rogel, M. A., Ramírez, M. A., et al. (2012). Rhizobial extrachromosomal replicon variability, stability and expression in natural niches. Plasmid 68, 149–158. Received: 28 March 2013; paper pend- ing published: 12 April 2013; accepted: 25 April 2013; published online: 14 May 2013. Hagemann, M., Ribbeck-Busch, K., Klähn, S., Hasse, D., Steinbruch, R., and Berg, G. (2008). The plant-associated bacterium Stenotrophomonas rhizophila expresses a new enzyme for the synthesis of the compatible solute glucosylglycerol. J. Bacteriol. 190, 5898–5906. Schmidt, C. S., Alavi, M., Cardinale, M., Müller, H., and Berg, G. (2012). Stenotrophomonas rhi- zophila DSM14405T promotes plant growth probably by altering fungal communities in the rhizosphere. Biol. Fertil. Soils 48, 947–960. Citation: Alavi P, Starcher MR, Zachow C, Müller H and Berg G (2013) Root-microbe systems: the effect and mode of interaction of Stress Protecting Agent (SPA) Stenotrophomonas rhi- zophila DSM14405T. Front. Plant Sci. 4:141. doi: 10.3389/fpls.2013.00141 McGinnis, M. W., Parker, Z. M., Walter, N. E., Rutkovsky, A. C., Cartaya-Marin, C., and Karatan, E. (2009). Spermidine regulates Vibrio cholerae bioﬁlm formation via transport and signaling path- ways. FEMS Microbiol. Lett. 299, 166–174. Hincha, D. K., and Hagemann, M. (2004). Stabilization of model membranes during drying by compatible solutes involved in the stress tolerance of plants and microorganisms. Biochem. J. 383, 277–283. Stapper, A. P., Narasimhan, G., Ohman, D. E., Barakat, J., Hentzer, M., Molin, S., et al. (2004). Alginate production affects Pseudomonas aeruginosa bioﬁlm development and architecture, but is not essential for bioﬁlm formation. J. Med. Microbiol. 53, 679–690. This article was submitted to Frontiers in Functional Plant Ecology, a specialty of Frontiers in Plant Science. Monday, S. R., and Schiller, N. L. (1996). Alginate synthesis in Pseudomonas aeruginosa: the role of algL (alginate lyase) and algX. J. Bacteriol. 178, 625–632. Copyright © 2013 Alavi, Starcher, Zachow, Müller and Berg. REFERENCES Nat. Cell Biol. 11, 1305–1314. Bais, H. P., Weir, T. L., Perry, L. G., Gilroy, S., and Vivanco, J. M. (2006). The role of root exu- dates in rhizosphere interactions longevity. Nat. Cell Biol. 11, 1305–1314. May 2013 \| Volume 4 \| Article 141 \| 9 www.frontiersin.org Stress protection by Stenotrophomonas rhizophila Alavi et al. the expression of a type VI secre- tion system in Yersinia pestis. Microbiology 155, 498–512. van de Mortel, J. E., de Vos, R. C., Dekkers, E., Pineda, A., Guillod, L., Bouwmeester, K., et al. (2012). Metabolic and transcriptomic changes induced in Arabidopsis by the rhizobacterium Pseudomonas ﬂuorescens SS101. Plant Physiol. 160, 2173–8218. FAO, Rome (Italy), Land, and Water Development Div.; International Programme for Technology and Research in Irrigation and Drainage, Rome (Italy). Secretariat; Centre Virtuel de l’Eau Agricole et ses Usages, Rome (Italy). Centre d’Information sur l’Eau Agricole et ses Usage. (2005). Management of Irrigation-Induced Salt-Affected Soils. Rome: CISEAU/FAO/IPTRID. Imai, A., Matsuyama, T., Hanzawa, Y., Akiyama, T., Tamaoki, M., and Saji, H. (2004). Spermidine syn- thase genes are essential for survival of Arabidopsis. Plant Physiol. 135, 1565–1573. Ribbeck-Busch, K., Roder, A., Hasse, D., de Boer, W., Martínez, J. L., Hagemann, M., et al. (2005). A molecular biological proto- col to distinguish potentially human-pathogenic strains of Stenotrophomonas maltophilia from non-pathogenic S. rhizophila strains. Environ. Microbiol. 7, 1853–1858. Jacobi, M., Kaiser, D., Berg, G., Jung, G., Winkelmann, G., and Bahl, H. (1996). Maltophilin – a new antifungal compound produced by Stenotrophomomas maltophilia R3089. J. Antib. 49, 1101–1104. Wolf, A., Fritze, A., Hagemann, M., and Berg, G. (2002). Stenotrophomonas rhizophila sp. nov., a novel plant-associated bacterium with antifungal properties. Int. J. Syst. Evol. Microbiol. 52, 1937–1944. Ferjani, A., Mustardy, L., Sulpice, R., Marin, K., Suzuki, I., Hagemann, M., et al. (2003). Glucosylglycerol, a compatible solute, sustains cell divi- sion under salt stress. Plant Physiol. 131, 1628–1637. Kobayashi, D. Y., Gugliemoni, M., and Clarke, B. B. (1995). Isolation of chitinolytic bacteria Xanthomonas maltophilia and Serratia marcescens as biological control agents for summer patch disease of turf grass. Soil Biol. Biochem. 27, 1479–1487. Roder, A., Hoffmann, E., Hagemann, M., and Berg, G. (2005). Synthesis of the compatible solutes glucosylglyc- erol and trehalose by salt-stressed cells of Stenotrophomonas strains. FEMS Microbiol. Lett. 243, 219–226. May 2013 \| Volume 4 \| Article 141 \| 10 REFERENCES This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and subject to any copyright notices concerning any third- party graphics etc. Huang, T. P., Somers, E. B., and Wong, A. C. (2006). Differential bioﬁlm formation and motility associated with lipopolysaccharide/ exopolysaccharide-coupled biosyn- thetic genes in Stenotrophomonas maltophilia. J. Bacteriol. 188, 3116–3120. Suckstorff, I., and Berg, G. (2003). Evidence for dose-dependent effects on plant growth by Stenotrophomonas strains from different origins. J. Appl. Microbiol. 95, 656–663. Pieper, R., Huang, S. T., Robinson, J. M., Clark, D. J., Alami, H., Parmar, P. P., et al. (2009). Temperature and growth phase inﬂuence the outer-membrane proteome and May 2013 \| Volume 4 \| Article 141 \| 10 Frontiers in Plant Science \| Functional Plant Ecology
https://openalex.org/W4365516362	https://digibug.ugr.es/bitstream/10481/83603/1/s0218202523500392.pdf	English	null	Singular patterns in Keller–Segel-type models	Mathematical models and methods in applied sciences/Mathematical models & methods in applied sciences	2,023	cc-by	13,588	OPEN ACCESS Mathematical Models and Methods in Applied Sciences Vol. 33, No. 8 (2023) 1693–1719 © The Author(s) DOI: 10.1142/S0218202523500392 OPEN ACCESS Mathematical Models and Methods in Applied Sciences Vol. 33, No. 8 (2023) 1693–1719 © The Author(s) DOI: 10.1142/S0218202523500392 OPEN ACCESS Mathematical Models and Methods in Applied Sciences Vol. 33, No. 8 (2023) 1693–1719 © The Author(s) DOI: 10.1142/S0218202523500392 Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. The classical KS model consists in a reaction–diﬀusion system of two coupled parabolic equations ∂tU = divx(DU∇xU −χU∇xQ) + H(U, Q), x ∈RN, t > 0, τ∂tQ = DQ∆Q + K(U, Q), x ∈RN, t > 0. (1.1) (1.1) In the biological context of cell dynamics, U represents the cell-density and Q the chemoattractant concentration. The positive deﬁnite terms DQ and DU are, respec- tively, the diﬀusivity of the chemoattractant and of the cells, χ ≥0 is the chemotac- tic sensitivity and the functions H(U, Q) and K(U, Q) in (1.1) model the interaction (production and degradation) between the cell density and the chemical substance. In most simpliﬁed models and in the original KS system, these terms are modeled as K(U, Q) = U −Q and H(U, Q) = 0. The hyperbolic limit (high-ﬁeld limit) and some special parabolic limit (low-ﬁeld limit) have been derived from kinetic equa- tions describing the run and tumble process for bacterial motion.7–11, 28, 29, 38, 44 In the biological context of cell dynamics, U represents the cell-density and Q the chemoattractant concentration. The positive deﬁnite terms DQ and DU are, respec- tively, the diﬀusivity of the chemoattractant and of the cells, χ ≥0 is the chemotac- tic sensitivity and the functions H(U, Q) and K(U, Q) in (1.1) model the interaction (production and degradation) between the cell density and the chemical substance. In most simpliﬁed models and in the original KS system, these terms are modeled as K(U, Q) = U −Q and H(U, Q) = 0. The hyperbolic limit (high-ﬁeld limit) and some special parabolic limit (low-ﬁeld limit) have been derived from kinetic equa- tions describing the run and tumble process for bacterial motion.7–11, 28, 29, 38, 44 The parameter τ ≥0 is introduced to distinguish if there is an adjustment of the chemo-attacker during the evolution of the process, it is standard that it only takes values only 0 or 1 giving rise to diﬀerent models that may not be dynamically equivalents. Singular patterns in Keller–Segel-type models Singular patterns in Keller–Segel-type models Juan Campos∗, Carlos Pulido† and Juan Soler‡ Departamento de Aatem´atica Aplicada and Research Unit “Modeling Nature” (MNat), Facultad de Ciancias. Universidad de Granada, 18071-Granada, Spain ∗campos@ugr.es †cpulidog@ugr.es ‡jsoler@ugr.es Mario Veruete Applied Mathematics Division, EPF, Montpellier, France mario.veruete@epf.fr Mario Veruete Applied Mathematics Division, EPF, Montpellier, France mario.veruete@epf.fr Received 12 December 2022 Revised 19 March 2023 Accepted 19 March 2023 Published 29 May 2023 Communicated by N. Bellomo Received 12 December 2022 Revised 19 March 2023 Accepted 19 March 2023 Published 29 May 2023 Communicated by N. Bellomo The aim of this paper is to elucidate the existence of patterns for Keller–Segel-type models that are solutions of the traveling pulse form. The idea is to search for transport mechanisms that describe this type of waves with compact support, which we ﬁnd in the so-called nonlinear diﬀusion through saturated ﬂux mechanisms for the movement cell. At the same time, we analyze various transport operators for the chemoattractant. The techniques used combine the analysis of the phase diagram in dynamic systems together with its counterpart in the system of partial diﬀerential equations through the concept of entropic solution and the admissible jump conditions of the Rankine–Hugoniot type. We found traveling pulse waves of two types that correspond to those found experimentally. Keywords: Flux-saturated; Keller–Segel; traveling waves; patterns; block solution; cross- diﬀusion; soliton. AMS Subject Classiﬁcation: 35K57, 35A15, 35C07 ‡Corresponding author. This is an Open Access article published by World Scientiﬁc Publishing Company. It is distributed under the terms of the Creative Commons Attribution 4.0 (CC BY) License which permits use, distribution and reproduction in any medium, provided the original work is properly cited. 1693 1694 J. Campos et al. 1. Introduction The collective behavior of species and how dynamic patterns emerge (defense, inva- sion, resilience, . . .) is one of the most important topics in this research that requires a multidisciplinary approach to be addressed. In addition to the intrinsic value of studying the dynamics of a population of birds, ﬁsh, ants or sheep, these models could provide foundations for understanding other, more microscopic problems such as morphogen-cell interaction or the evolution of tumors. However, the impressive evolution in microscopy and in antibody concentration morphogenesis cell signaling allowed the study of collective behavior at the subcellular and cellular level to be analyzed and modeled directly.1 This provides a new impetus in which the models initially developed by Keller and Segel (KS)39, 40, 45 for chemotaxis processes (the movement of biological entities in response to chemical gradients) take on a new dimension. In addition, in recent years, various applications have been developed in exotic contexts11 beyond cell signaling mechanisms that have provided more ﬂexible and diverse variants of KS-type models. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. c = lim s→∞Φ(s), and it is ﬁnite. Also Φ ∈C1(R) in order to have uniqueness of the initial value problems. The value µ = Φ′(0) is the kinematic viscosity for small velocities and near ux = 0 the ﬂow means and it is ﬁnite. Also Φ ∈C1(R) in order to have uniqueness of the initial value problems. The value µ = Φ′(0) is the kinematic viscosity for small velocities and near ux = 0 the ﬂow means U mΦ U −1 ∂U ∂x ∼µU m−1Ux, U mΦ U −1 ∂U ∂x ∼µU m−1Ux, being m ≥1 a parameter that measures the porosity of the medium. In some sense, we have a ﬂux-saturated combined with a porous media operator.16 Diﬀerent proposals to ours to use ﬂux-saturated operators as an alternative to linear diﬀu- sion for the KS model use hyperbolic, fractional diﬀusion or porous medium type approximations, see, for example, Refs. 15, 27 and 30 and the references therein. One of the main objectives of our study is to consider the so-called relativistic heat case Φ(s) = µ s q 1 + µ2 c2 s2 that leads to that leads to ∂U ∂t = µ ∂ ∂x   U m ∂U ∂x q U 2 + µ2 c2 ∂U ∂x 2 −aU ∂Q ∂x  . Other examples of great interest are Φ(s) = µ s 1+ µ c \|s\| usually referred as Wilson operator,43 the Larson operator16 Φ(s) = µ s pq 1+ µp cp sp that include the relativistic case, and Φ(s) = c tanh( µs c ) usually referred as the hyperbolic tangent operator, see, for instance, Ref. 42. If m = 1, which is the case of the relativistic heat equation, then c is the speed at which the solution support moves.3 In the general choice of Φ, c represents the maximum speed at which the solution support can move.16 Therefore, c is a parameter that can be taken from the biological experimental data.47 Note that for any ﬂux-saturated Φ, if c tends to inﬁnity the heat equation or the porous media equation are recovered for the diﬀerent values of m ≥1.26 The aim of this paper is to ﬁnd, in the context of KS models, soliton-type patterns with compact support, which represent collective models of cell invasion, propagation or behavior in which the interface with the medium is singular. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. In this paper, we will consider two variants of the ﬂux-saturated Keller–Segel (FSKS) model: ∂U ∂t = ∂ ∂x U mΦ U −1 ∂U ∂x −aU ∂Q ∂x , (1.2) (1.2) T (Q) = U, T (Q) = U, where T = T (Q) is one of the following linear diﬀerential operators: where T = T (Q) is one of the following linear diﬀerential operators: T (Q) = ∂2Q ∂t∂x −ν ∂2Q ∂x2 , (1.3) (1.3) Singular patterns in Keller–Segel-type models 1695 1695 or T (Q) = τ ∂Q ∂t + α∂Q ∂x −ν ∂2Q ∂x2 . (1.4) (1.4) The parameters α ≥0 and ν ≥0 stand for the transport and diﬀusion coeﬃcients, respectively. The ﬂux function Φ = Φ(s) is a bounded, regular, increasing and odd function. The value c > 0 is deﬁned as The parameters α ≥0 and ν ≥0 stand for the transport and diﬀusion coeﬃcients, respectively. The ﬂux function Φ = Φ(s) is a bounded, regular, increasing and odd function. The value c > 0 is deﬁned as Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. This type of patterns usually appears with diverse geometry in the experimental data, 1696 J. Campos et al. 1696 J. Campos et al. (a) (b) Fig. 1. (Color online) Pattern prototypes with compact support associated with ﬂux-saturated operators. Pattern prototypes with compact support associated with saturated-ﬂux operators. In blue we represent the cell concentration and in black the proﬁle of the chemoattractant. (a) (a) (b) (b) Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. els Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com RANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articl (b) (a) Fig. 1. (Color online) Pattern prototypes with compact support associated with ﬂux-saturated operators. Pattern prototypes with compact support associated with saturated-ﬂux operators. In blue we represent the cell concentration and in black the proﬁle of the chemoattractant. and cannot be captured with linear diﬀusion terms in the classical KS model. In Fig. 1, we provide various conﬁgurations of some of the results that we obtain here, coming from the analysis of the nonlinear variants of the KS models that we will study throughout the paper. Experimental data show that the movement of cells aﬀected by a chemoat- tractant occurs through a pulse or soliton-type solution.33, 47 Moreover, from the point of view of modeling, the KS model combines a system of partial diﬀerential equations that represents the evolution of the cell density and the chemoattrac- tant concentration. However, the classical KS system, although it admits regular traveling waves with a birth term of either a Fisher–KPP term-type, does not seem to admit soliton-type solutions. The modiﬁcation of the transport terms, especially preventing free diﬀusion, allows to build solutions that better reﬂect the experimental results. This fact was rigorously proved in the case of a ﬂux- saturated as an alternative to linear diﬀusion in cell density in Ref. 6. A great eﬀort has been devoted in recent years to study the properties of the evolution by ﬂux-saturated mechanisms, in particular the existence of traveling waves, see, for instance, Refs. 3, 4, 12, 13, 17–25, 31, 32, 36, 41 and 46 and the references therein. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly In the case where the time evolution of the chemoattractant is negligible, the resulting model produces a self-generated potential in terms of cell density. This has been the most studied approach in the context of the KS models.14, 34, 35, 37, 38 The main reason to modify the linear diﬀusion by a nonlinear one is that it reproduces more faithfully the experimental data. In this context, the FSKS is a macroscopic model describing cell motion by chemotaxis, in which saturation of the velocity is taken into account. The FSKS model also has the advantage that trav- eling pulse or soliton-type solutions with compact support emerging as a prototype of pattern under this system.6 The existence of this type of solutions is relevant for biological applications since, from a modeling perspective, the compactly supported property is well suited. The paper is structured as follows. Section 2 is devoted to deﬁning the solution concept and the soliton-type geometric structure of the solutions we seek. In this sense, we deﬁne the block-type solution, which will be the object of study in this paper. Section 3 deals with the case in which the chemoattractant gradient is transported without diﬀusion, proving that any maximal solution of the associated Singular patterns in Keller–Segel-type models 169 dynamic system is a block-type solution. Section 4 deals with the case where transport and diﬀusion terms are combined in the chemoattractant. Conditions are given for the existence and non-existence of block-type solutions. In the case of dif- fusion without transport of the chemoattractant, a complete analysis of all types of traveling waves is carried out (particularly those cases in which there are block-type solutions), classifying all types of solutions according to the system parameters. 2. Block Solitons Moving at a Constant Speed Let us consider the general system of cell evolution U together with the chemoat- tractant Q ∂U ∂t = ∂ ∂x U mΦ U −1 ∂U ∂x −aU ∂Q ∂x τ1 ∂Q ∂t + τ2 ∂ ∂t ∂Q ∂x + α∂Q ∂x −ν ∂2Q ∂x2 = U. (2.1) (2.1) In this model, the coeﬃcient ν ≥0 is the viscosity of the chemoattractant. Addi- tionally, we assume that a > 0, α > 0 and m ≥1. The parameters τ1 and τ2 are taken greater than or equal to zero, in fact in the models considered τ1 and τ2 take values {0, 1}. We are going to study the existence of biological blocks that move at a constant speed. Mathematically, the concept of block solutions is associated to that of trav- eling waves type solutions of the previous problem (2.1) for cell dynamics whose mass is concentrated in a bounded region. If σ > 0 denotes a speed of propagation, we look for solutions of the kind U(t, x) = u(x −σt), Q(t, x) = q(x −σt), where u, q : R →R are scalar functions and u has the mass concentrated in a compact interval. Formally, the resulting system for u, q veriﬁes −σu′(ξ) = um(ξ)Φ u′(ξ) u(z) −au(ξ)q′(ξ) ′ , (α −στ1)q′(ξ) −(στ2 + ν)q′′(ξ) = u(ξ), (2.2) (2.2) where ξ := x −σt. A ﬁrst question to consider is the concept of solution for (2.1). The appropri- ate framework for our analysis is that of solutions of bounded variation. However, the theory of existence in the context of bounded variation solutions for KS-type systems is not suﬃciently fully established, and this is not the aim of our paper. To avoid entering the theory of bounded variation functions of several variables, we are going to focus on our study on Eq. (2.2) directly. The cell structure that gives rise to the u component is going to be assumed to be much larger and heavier than the molecular structure of the chemoattractant given by the q component, therefore a singularization of the component u can be expected. This appreciation is supported by the presence of a ﬂux-saturated as the basis of the movement of u in the ﬁrst equation. On the other hand, we expect a 1698 J. Campos et al. 1698 milder behavior of the chemoattractant q. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Z R um(ξ)Φ u′(ξ) u(ξ) −au(ξ)q′(ξ) + σu(ξ) ψ′(ξ)dξ = 0, (α −στ1) Z R q(ξ)ψ′(ξ)dξ −(στ2 + ν) Z R q′(ξ)ψ′(ξ)dξ = − Z R u(ξ)ψ(ξ)dξ, (2.4) (2.4) holds, for each ψ ∈D(R). In this expression, the role of u′(ξ) has to be clariﬁed. When a function u ∈BV (R) its derivative in the sense of the distributions Du decomposes as an absolutely continuous part u′(ξ) and a singular part Dsu that is orthogonal to the Lebesgue measure. The singular part of the measure is not easy to absorb, see Ref. 2. The set S = supp{Dsu} is called the set of singularities of u. It is common in this type of operators that S is a ﬁnite set and also u ∈C1(R\S). holds, for each ψ ∈D(R). In this expression, the role of u′(ξ) has to be clariﬁed. When a function u ∈BV (R) its derivative in the sense of the distributions Du decomposes as an absolutely continuous part u′(ξ) and a singular part Dsu that is orthogonal to the Lebesgue measure. The singular part of the measure is not easy to absorb, see Ref. 2. The set S = supp{Dsu} is called the set of singularities of u. It is common in this type of operators that S is a ﬁnite set and also u ∈C1(R\S). A block structure is going to be requested on u. This concept of block solution materializes in the existence of a compact interval [ξ1, ξ2], not reduced to a point, such that u(ξ) > 0, for a.e. ξ ∈[ξ1, ξ2], and u(ξ) = 0, otherwise. For q no restrictions will be imposed on its support. One last assumption is that the singularities of u have been formed by the saturation of the cell ﬂux. If ¯ξ ∈S is a singular point, then the lateral limit values are always deﬁned. The point ¯ξ is said of saturation to the left if lim ξ→¯ξ u′(ξ) = −∞, lim ξ→¯ξ−u(ξ) ≥lim ξ→¯ξ+ u(ξ), while it will be saturation to the right if lim ξ→¯ξ u′(ξ) = +∞, lim ξ→¯ξ−u(ξ) ≤lim ξ→¯ξ+ u(ξ). 2. Block Solitons Moving at a Constant Speed The formation of discontinuities in q is not expected if στ2 + ν > 0. (2.3) (2.3) Even in the degenerate case τ2 = 0 = ν the existence of fronts is not apparent because in q, even in that case, we have a linear transport equation. These reasons make us assume that q is of class 1 in R while u is only going to be a bounded variation function. We are in a position to consider distributional solutions imposing that Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. verify verify um−1 ± c −aq′(¯ξ) + σ = 0, um−1 ± c −aq′(¯ξ) + σ = 0, um−1 ± c −aq′(¯ξ) + σ = 0, where it has been used that we have a saturation on the left. Then, both one-sided limits are equal. That implies the continuity of u in ¯ξ. Using the regularity and the (2.3) condition in the second equation of (2.4) we get that q′′ is deﬁned in ¯ξ. In particular, the function ξ →cum−1(ξ) −q′(ξ) + σ, has inﬁnite derivative at ¯ξ, and it is an increasing function in a neighborhood of that point. This would give us ξ values such that ξ →cum−1(ξ) −q′(ξ) + σ < 0, which is contradictory to (2.5) since Φ(s) < c, for all s ∈R. has inﬁnite derivative at ¯ξ, and it is an increasing function in a neighborhood of that point. This would give us ξ values such that ξ →cum−1(ξ) −q′(ξ) + σ < 0, which is contradictory to (2.5) since Φ(s) < c, for all s ∈R. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Down by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is In conclusion, assuming (2.3) we can deﬁne a block-type solution as follows. Deﬁnition 2.1. Given an interval [ξ1, ξ2], we will say that a pair of function u ∈C0[ξ1, ξ2] ∩C1(ξ1, ξ2) and q ∈C0[ξ1, ξ2] ∩C2(ξ1, ξ2) constitute a block-type solution as long as u ∈C0[ξ1, ξ2] ∩C1(ξ1, ξ2) and q ∈C0[ξ1, ξ2] ∩C2(ξ1, ξ2) constitute a block-type solution as long as constitute a block-type solution as long as ξ) > 0, for each ξ ∈[ξ1, ξ2] and both verify • u(ξ) > 0, for each ξ ∈[ξ1, ξ2] and both verify um−1(ξ)Φ u′(ξ) u(ξ) −aq′(ξ) + σ = 0 (α −στ1)q′(ξ) −(στ2 + ν)q′′(ξ) = u(ξ). (2.6) (2.6) • The singular points are S = {ξ1, ξ2}, and both are lateral saturation points for u, that is • The singular points are S = {ξ1, ξ2}, and both are lateral saturation points for u, that is • The singular points are S = {ξ1, ξ2}, and both are lateral saturation points for u, that is lim ξ→ξ+ 1 u′(ξ) = ∞, lim ξ→ξ− 2 u′(ξ) = −∞. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. If ¯ξ is a boundary point of the support, then the saturation condition is only one-sided, that is, if ¯ξ = ξ1, then u(ξ) = 0, for ξ < ξ1, which means If ¯ξ is a boundary point of the support, then the saturation condition is only one-sided, that is, if ¯ξ = ξ1, then u(ξ) = 0, for ξ < ξ1, which means lim ξ→ξ+ 1 u′(ξ) = ∞, and a symmetric condition on ξ2. Lemma 2.1. Assume that (2.3) holds, then there are no saturation points inside the support. Lemma 2.1. Assume that (2.3) holds, then there are no saturation points inside the support. Proof. It follows from (2.4) that the function um(ξ)Φ u′(ξ) u(ξ) −au(ξ)q′(ξ) + σu(ξ) has zero weak derivative. By Stampacchia’s Lemma um(ξ)Φ u′(ξ) u(ξ) −au(ξ)q′(ξ) + σu(ξ) = K, for some constant K. We can assume that this constant is going to be Singular patterns in Keller–Segel-type models 1699 zero when considering ξ outside the support of u. Therefore, we have zero when considering ξ outside the support of u. Therefore, we have um−1(ξ)Φ u′(ξ) u(ξ) −aq′(ξ) + σ = 0, a.e. ξ ∈[ξ1, ξ2]. (2.5) (2.5) Whence, both values lim ξ→¯ξ± u(ξ) = u±, lim ξ→¯ξ± u(ξ) = u±, lim ξ→¯ξ± u(ξ) = u±, Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. lim ξ→ξ+ 1 u′(ξ) = ∞, lim ξ→ξ− 2 u′(ξ) = −∞. Under these conditions, we will discuss throughout the paper the existence of a cell block moving at speed σ, and this will be obtained by extending by zero the cell density u outside the interval [ξ1, ξ2], q extends to a function from class 1 to R through two straight lines. Taking g = Φ−1, in the sense of the composition of applications, then g : (−c, c) →R is a C1 function deﬁned as g(y) = s ↔y = Φ(s). 1700 J. Campos et al. 1700 J. Campos et al. 1700 If, in addition, we deﬁne r(ξ) = g′(ξ), we get If, in addition, we deﬁne r(ξ) = g′(ξ), we get u′ = ug ar −σ um−1 . Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Finally, we can rebuild from (2.6) the following system: Finally, we can rebuild from (2.6) the following system: Finally, we can rebuild from (2.6) the following system: This implies ar(ξn) −σ um−1(ξn) →0. ar(ξn) −σ um−1(ξn) →0. Since u+ is bounded, this implies that r(ξn) →σ a, but this is not possible since r+ ∈γ+. Therefore ξ+ < +∞. The reasoning for ξ−is analogous. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. u′(ξn) →0. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. u′ = ug ar −σ um−1 , u′ = ug ar −σ um−1 , (2.7) u′ = ug ar −σ um−1 , r′ = (α −σδ1)r −u δ + . (2.7) (2.7) r′ = (α −σδ1)r −u σδ2 + ν . (2.7) Note that under the hypothesis on g in the previous section, this system is deﬁned only for values of (u, r) belonging to the domain Γ deﬁned by Note that under the hypothesis on g in the previous section, this system is deﬁned only for values of (u, r) belonging to the domain Γ deﬁned by Γ := {(u, r) : u > 0, \|(ar −σ)u1−m\| < c}. (2.8) (2.8) such that Γ = Γ−∪Γ0 ∪Γ+, where such that Γ = Γ−∪Γ0 ∪Γ+, where Γ0 = n (u, r) ∈Γ : u > 0, r = σ a o , Γ−= n (u, r) ∈Γ : r > σ a o Γ0 = n (u, r) ∈Γ : u > 0, r = σ a o , Γ−= n (u, r) ∈Γ : r > σ a o and and Γ+ = n (u, r) ∈Γ : r < σ a o . We denote by γ = ∂Γ = γ+ ∪γ−∪ 0, σ a , where γ± := (u, r) ∈(0, ∞) × R : ar −σ um−1 = ∓c . (2.9) (2.9) A block-type solution corresponds to a maximal solution of (2.7) such that A block-type solution corresponds to a maximal solution of (2.7) such that lim ξ→¯ξ± u(ξ) = u±, lim ξ→¯ξ± r(ξ) = r±, (2.10) (2.10) where (ξ−, ξ+) is the maximum interval of deﬁnition and (u±, r±) ∈γ±, see Fig. 2. where (ξ−, ξ+) is the maximum interval of deﬁnition and (u±, r±) ∈γ±, see Fig. 2. (a) (b) Fig. 2. (a) Representation of the ﬁeld of tangent vectors of (2.7) and (b) representation of a block-type solution. (a) (b) (b) (a) Fig. 2. (a) Representation of the ﬁeld of tangent vectors of (2.7) and (b) representation of a block-type solution. Singular patterns in Keller–Segel-type models 1701 Proposition 2.1. Under these conditions, the spatial support of the solution is necessarily bounded. Proposition 2.1. Under these conditions, the spatial support of the solution is necessarily bounded. Proof. Let us prove it by reductio ad absurdum. Assume that ξ+ = +∞. Then, there exists a sequence ξn →+∞such that 3. Transport in the Gradient of Q This section is devoted to analyze the case where the gradient of the chemoattrac- tant (∂xQ) is solution of the non-homogeneous linear transport equation: ∂U ∂t = ∂ ∂x U mΦ U −1 ∂U ∂x −aU ∂Q ∂x , ∂ ∂t ∂Q ∂x −ν ∂ ∂x ∂Q ∂x = U. (3.1) (3.1) In this model, we assume that ν > 0 so we are in a non-degenerate situation (remember that a > 0). The case m = 1 can be analyzed following the guidelines of the case m > 1, therefore we also assume that m > 1. As we will see, this model can be seen as a particular case of the one analyzed in the following section where the values would have another expression, but we have considered studying this case ﬁrst for clarity in the exposition. Hence, we will focus on study the existence of block solutions, deﬁned in the pre- vious section, which correspond with the search of orbits of the diﬀerential equation u′ = ug ar −σ um−1 , r′ = − u σ + ν . (3.2) (3.2) r′ = − u σ + ν . ( ) This orbits connect γ−with γ+, where γ± were deﬁned in (2.9). Therefore, for any initial condition in Γ deﬁned in (2.8), we will be able to ﬁnd a block solution, see Fig. 3. This orbits connect γ−with γ+, where γ± were deﬁned in (2.9). Therefore, for any initial condition in Γ deﬁned in (2.8), we will be able to ﬁnd a block solution, see Fig. 3. Theorem 3.1. Every maximal solution of (3.2) is a block solution. A key ingredient to prove this theorem are the following results, in which we will show that there exist an orbit connecting γ−with γ+, for every initial condition in the line s = σ a. 1702 J. Campos et al. (a) 1 < m < 2 (b) m = 2 (c) m ≥2 Fig. 3. Representation of the tangent vector ﬁeld associated to (3.2). (a) 1 < m < 2 Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. a Assume that r0 > σ a and r(ξ) > σ a, for all ξ ∈(ξ−, ξ+) and hence, u′(ξ) > 0 and there exist the limξ→ξ+ u(ξ). If this limit is ﬁnite, then ξ+ = +∞and we will have a critical point, but this is not possible. Therefore, we have lim ξ→ξ+ u(ξ) = +∞. (3.4) (3.4) Since u(ξ) ≥u0 and r(ξ) ≤r0, then we obtain Since u(ξ) ≥u0 and r(ξ) ≤r0, then we obtain g ar(ξ) −σ um−1(ξ) ≤g ar0 −σ um−1 0 , for ξ ∈(ξ0, ξ+). Using that r → ar−σ um−1 is increasing in r, then we deduce that u →ar−σ um−1 is decreasing in u and y →g(y) is decreasing. Deﬁning M as the value obtained above, we have that for ξ ∈(ξ0, ξ+). Using that r → ar−σ um−1 is increasing in r, then we deduce that u →ar−σ um−1 is decreasing in u and y →g(y) is decreasing. Deﬁning M as the value obtained above, we have that u′(ξ) ≤Mu(ξ). Then, combining the Gronwall Lemma together with (3.4) we deduce ξ+ = +∞. However, this cannot be possible, because r′(ξ) = − 1 σ+µu(ξ) and u(ξ) ≥u0, for ξ ∈(ξ0, ξ+), and we obtain r′(ξ) ≤− 1 σ + µu0 < 0, ∀ξ ∈(ξ0, ξ+), diction. which is a contradiction. which is a contradiction. Therefore, thanks to these two results and the fact that the solutions are invari- ant under time translation, the proof of Proposition 3.1 follows. Lemma 3.1. Every maximal solution of (3.2) intersects the curve Γ0. Proof. Let u, r : (ξ−, ξ+) →R a solution of (3.2), and assume that for some value ξ0 ∈(ξ−, ξ+) we have (u0, r0) := (u(ξ0), r(ξ0)) ∈Γ. Let us prove that if r0 > σ a, then there exits a ξ1 ∈(ξ−, ξ+), such that r(ξ1) = σ a. Similarly, if r0 < σ a, we can ﬁnd a value ξ2 such that r(ξ2) = σ a, which will conclude the proof. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. (b) m = 2 Fig. 3. Representation of the tangent vector ﬁeld associated to (3.2). Proposition 3.1. Consider the initial conditions r(0) = σ a and u(0) = u0 > 0 associated to (3.2). Then, there exists a block solution (u, r) corresponding to these initial data. Proof. Let us prove that the maximal solution of the previous problem gives rise to a block solution, by seeing that it connect a point in γ−with a point in γ+, according to the deﬁnition of the previous section. We can check that u(ξ) has uni-modal shape with a unique maximum at ξ = 0, and r(ξ) is a strictly decreasing function. Therefore, we can prove the existence of the ﬁnite limits lim ξ→ξ±(u(ξ), r(ξ)) =: (u±, r±). (3.3) (3.3) Due to the decrease of r we obtain r+ < σ a < r−, and taking limits in the inequality ar −σ um−1 < c, we deduce that u± are necessarily strictly positive. we deduce that u± are necessarily strictly positive. Let us see that r+ = σ −cum−1 + a . If r+ ̸= σ−cum−1 + a , then (u+, r+) is not in the boundary of Γ, and by a prolongation argument we get ξ+ = +∞. Therefore (u+, r+) will be a critical point. However, there are no critical points in the problem, so ξ+ < +∞. Using again a prolonga- bility argument we obtain that (u+, r+) is in the boundary of Γ. Also, we can prove that Also, we can prove that r−= σ + cum−1 − a , r−= σ + cum−1 − a , by using similar arguments. by using similar arguments. Once demonstrated the existence of solutions for initial conditions in the vertical isocline, we will proceed to prove that for every initial condition in Γ, the associated solutions always reach the vertical isocline. Singular patterns in Keller–Segel-type models 1703 Lemma 3.1. Every maximal solution of (3.2) intersects the curve Γ0. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. The analysis of the existence of such orbits will be studied in terms of the parameter m and the relation between α and σ. Remark 4.1. If σ = α, the diﬀerential equation (4.2) is similar to (3.2). Therefore, we will have existence of block solutions for σ > 0, thanks to Theorem 3.1. The main result describing the existence of block solution in this context is the following one. Theorem 4.1. Block solutions exist if one of the following conditions holds true: Theorem 4.1. Block solutions exist if one of the following conditions holds true: If 1 < < 2 d > 0 • If 1 < m < 2 and σ > 0. • If 1 < m < 2 and σ > 0. • If 1 < m < 2 and σ > 0. • If m ≥2, σ > 0 and α ≤α∗, for α∗> 0, where • If m ≥2, σ > 0 and α ≤α∗, for α∗> 0, where α∗= a m −1 m−1 2m−3 1 c3−2m (m −2) m−2 2m−3 . (4.3) (4.3) • If m ≥2, σ > σ∗(α) > 0 and α > α∗> 0, where • If m ≥2, σ > σ∗(α) > 0 and α > α∗> 0, where σ∗(α) = α − a m −1c 1 1−m m −2 α m−2 m−1 . (4.4) (4.4) Remark 4.2. In the case m = 2, we have that σ∗(α) = α −a c , which is the limit when m →2 of (4.4), and therefore solving σ∗(α) = 0, we obtain the value α∗= a c . In Figure 4 we can see a qualitative representation, for m ≥2, of the region of σ values for which a solution exists as a function of the parameter α. Remark 4.2. In the case m = 2, we have that σ∗(α) = α −a c , which is the limit when m →2 of (4.4), and therefore solving σ∗(α) = 0, we obtain the value α∗= a c . In Figure 4 we can see a qualitative representation, for m ≥2, of the region of σ values for which a solution exists as a function of the parameter α. To carry out the proof of Theorem 4.1, we are going to introduce a series of previous results. 4. Transport and Diﬀusion in the Chemoattractant In this section, we consider that the chemoattractant concentration is solution of a linear transport-diﬀusion equation. ∂U ∂t = ∂ ∂x U mΦ U −1 ∂U ∂x −aU ∂Q ∂x , ∂Q ∂t + α∂Q ∂x −ν ∂2Q ∂x2 = U, (4.1) ∂U ∂t = ∂ ∂x U mΦ U −1 ∂U ∂x −aU ∂Q ∂x , (4 1) (4.1) ∂Q ∂t + α∂Q ∂x −ν ∂2Q ∂x2 = U, (4.1) ∂Q ∂t + α∂Q ∂x −ν ∂2Q ∂x2 = U, where α is the transport speed coeﬃcient of the chemoattractant density and ν stands for its diﬀusion coeﬃcient. Our goal is to study the existence of orbits that where α is the transport speed coeﬃcient of the chemoattractant density and ν stands for its diﬀusion coeﬃcient. Our goal is to study the existence of orbits that 1704 J. Campos et al. 1704 connect γ−with γ+, deﬁned in (2.9), of the diﬀerential equation: u′ = ug ar −σ um−1 , (4.2) (4.2) r′ = 1 ν ((α −σ)r −u). ( ) Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Fig. 5. Scheme of the proof of Lemma 4.1. Fig. 5. Scheme of the proof of Lemma 4.1. Lemma 4.1. Let 1 < m ≤2. If η has positive slope and cuts γ−, then there exist (¯u, ¯r) ∈Γ−such as Lemma 4.1. Let 1 < m ≤2. If η has positive slope and cuts γ−, then there exist (¯u, ¯r) ∈Γ−such as r′ < 0, u′ > 0, ∀(u, r) ∈B, where B := B(¯u,¯r) = {(u, r) ∈Γ−: u ≤¯u, r ≥¯r}. Proof. Since 1 < m ≤2, then γ−is the graph of a concave function or a straight line. It is easy to see that an increasing line below u = 0 will intersect γ−at a single point. u∗, see Fig. 5. Therefore, it allows us to ﬁnd a curved triangular region, denoted by B, just build taking as vertex of B any points (¯u, ¯r) ∈Γ−such that ¯u > u∗and ¯r > γ−(u∗). Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Let us denote by η the horizontal isocline, whose equation is r = 1 α−σu and represents the points (u, r) ∈Γ, where r′ = 0. First, we will analyze the case in which η has positive slope, and we will focus on ﬁnding some initial values (¯u, ¯r) from which we can construct the solution. Fig. 4. Representation of the region of existence obtained in Theorem 4.1 for m ≥2. Fig. 4. Representation of the region of existence obtained in Theorem 4.1 for m ≥2. Singular patterns in Keller–Segel-type models 1705 Fig. 5. Scheme of the proof of Lemma 4.1. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictl The points (¯u, ¯r), deﬁned in Lemma 4.1, will allow us to construct the desired orbits of (4.2) when these are taken as initial data. Proposition 4.1. If η has positive slope and cuts γ−, then there exists a block solution of (4.2). Proof. In the case 1 < m ≤2 (see Fig. 6), let us take the points (¯u, ¯r), previously deﬁned in Lemma 4.1, as the initial condition of the problem (4.2). The solution of the initial value problem (¯u, ¯r) remains in B as long as it is deﬁned. Bendixson’s theorem assures us that this solution has to touch γ−, as ξ →ξ−, since there is no equilibrium points in the set B. Moreover, this solution will always intersect the line r = σ a, for some ξ1 ∈(ξ−, ξ+). This is because r′ < 0 and u′ > 0, for r > σ a. An argument similar to that used in Lemma 3.1 allows us to prove the existence of a value ξ1 at which the solution intersects the straight line r = σ a. a We cannot know how the solutions (u1, r1), connecting γ−with r = σ a, will behave once they go through the line r = σ a. Let us deﬁne u∗= u1(ξ1). Observe that the solution of (4.2), with initial con- dition u(0) = ˜u, r(0) = σ a, will touch γ−when ξ →ξ−, for any value ˜u ≥u∗. This is due to the fact that the orbits of the autonomous systems cannot intersect. Therefore, to ﬁnish the proof it remains to ﬁnd a value ˜u such that the solution of the initial value problem (˜u, σ a) reaches γ+. 1706 J. Campos et al. (b) m = 2 (a) m < 2 (b) m = 2 (c) m > 2 Fig. 6. Representation of the phase diagram as a function of m, when η intersects γ+. (a) m < 2 Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. (c) m > 2 (c) m > 2 Fig. 6. Representation of the phase diagram as a function of m, when η intersects γ+. Let (ˆu, ˆr) be the intersection of the straight line η with the curve γ+. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. (a) 1 < m < 2 (b) m = 2 (c) m > 2 Fig. 7. Representation of the phase diagram as a function of m, when η is decreasing. Proposition 4.2. If η has negative slope, then there exists a block solution of (4.2) Proposition 4.2. If η has negative slope, then there exists a block solution of (4.2). Proof. The directions of the vector ﬁeld, represented in Fig. 7, show that the solution with initial data (ˆu, σ a) will touch γ−and γ+, for any point (ˆu, σ a) in the straight line r = σ a which is to the left of the intersection point of the straight line η with γ+. The proof argument is similar to the one made in the proof of Proposition 2.1. These results inform us under which conditions we can ﬁnd block solutions. To ﬁnish the proof of Theorem 4.1, it is necessary to see what relationships between the parameters allow us to obtain these solutions. Proof of Theorem 4.1. The equation of the line η is 1 α−σu. If σ > α, η has negative slope, and by Proposition 4.2 we have existence of a block solution. If σ = α, we have also existence of solution by arguing as in Remark 4.1. On the other hand, if σ < α, we have to study the relative position between γ− and η. Proposition 4.1 establishes the existence of solution when η has positive slope. Therefore, we have to analyze the possibilities of intersection between η and γ−. and η. Proposition 4.1 establishes the existence of solution when η has positive slope. Therefore, we have to analyze the possibilities of intersection between η and γ−. If 1 < m < 2, η will always intersect γ−, when 0 < σ < α. If 1 < m < 2, η will always intersect γ−, when 0 < σ < α. If m = 2, the slope of η must be greater than the slope of γ−, which is a line in this case. This is fulﬁlled when c a < 1 α−σ, i.e. σ > α −a c . (4.5) (4.5) Finally, for m > 2, the intersection points are given by the roots of the following equation: cum−1 a + σ a − u α −σ = 0. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictl Then, the solution of the initial value problem (ˆr, u∗) touches γ+ as ξ →ξ+, due to the fact that r′ < 0 for r ≤ˆr. In addition, this solution will intersect the straight line r = σ a at a point (˜u, σ a), by a symmetric argument to the one made in Lemma 3.1. a In the case m > 2, the proof is carried out in a similar way. Indeed, if η intersects γ−, this intersection can be made at two points or at one tangent point (see Fig. 6). In both cases, we can deﬁne u = max u ∈(0, +∞) : 1 α −σ u = γ−(u) . Therefore, the set A = {(u, r) ∈Γ−: u > u, r ≥η(u)} is negatively invariant, since u′ > 0 and r′ ≥0, for all (u, r) ∈A. Therefore, following the same ideas as in the previous case, we can show that the solution connects γ−to r = σ a, reaching the line r = σ a, for some ξ1 ∈(ξ−, ξ+). So, as in the previous case, we are able to ﬁnd an initial condition (˜u, σ a) whose solution connects γ−with γ+. Let us now study the case in which η has negative slope. Singular patterns in Keller–Segel-type models 1707 1707 (a) 1 < m < 2 (b) m = 2 (c) m > 2 Fig. 7. Representation of the phase diagram as a function of m, when η is decreasing. (a) 1 < m < 2 (b) m = 2 (c) m > 2 Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. The existence of roots of this equation is equivalent to prove that the minimum takes negative values, this holds true for σ > α − a m −1c 1 1−m m −2 α m−2 m−1 = f(α). Note that if m = 2 this equation coincides with (4.5), as m →2. 1708 J. Campos et al. It can be seen that f(α) ≤0, if α ≤α∗, where α∗= a m −1 m−1 2m−3 1 c3−2m (m −2) m−2 2m−3 . If α > α∗, then the expression σ > f(α) is equivalent to σ > σ∗(α) which is given by (4.4). If α > α∗, then the expression σ > f(α) is equivalent to σ > σ∗(α) which is given by (4.4). 4.1. Non-existence of block solution Once we have analyzed the existence of solution in the previous section, let us see under what conditions we can prove the non-existence of block solutions. For this purpose, we will consider the function θ : (−c, c) →R deﬁned as θ(y) = (α −σ)y ν −(m −1)yg(y). Observe that the function satisﬁes that θ(−c) = θ(c) = −∞, therefore there exists the value θ0 = maxy∈(−c,c) θ(y). On the other hand, let us consider the function ω : (0, +∞) →R deﬁned as ω(u) = au −σ(α −σ) νum−1 . Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly Arguing as before, let us consider the value ω0 = maxu∈(0,+∞) ω(u), whose expres- sion is Arguing as before, let us consider the value ω0 = maxu∈(0,+∞) ω(u), whose expres- sion is ω0 = 1 v σ(α −σ) m −2 2−m a (m −1) m−1 . With these two constants we obtain the following non-existence result. Theorem 4.2. If θ0 > ω0, then there is no block-type solution. rem 4.2. If θ0 > ω0, then there is no block-type solution. Proof. Let us take y(ξ) = ar(ξ)−σ um−1(ξ), which satisﬁes the diﬀerential equation Proof. Let us take y(ξ) = ar(ξ)−σ um−1(ξ), which satisﬁes the diﬀerential equation . Let us take y(ξ) = ar(ξ)−σ um−1(ξ), which satisﬁes the diﬀerential equation u′ = ug(y), y′ = a v[(α −σ)[ um−1y+σ a ] −u] −(m −1)um−1yg(y) um−1 . (4.6) (4.6) Math. Models M by UNIVERSIDAD DE GRANA Observe that a block-type solution is now a connection between y = c and y = −c. Taking ¯y such that θ(¯y) > maxu∈(0,+∞) ω(u) then the expression of the second equation of (4.6) provides y′ > 0, for all u ∈(0, +∞). Therefore, it would not be possible to connect y = c with y = −c. Observe that a block-type solution is now a connection between y = c and y = −c. Taking ¯y such that θ(¯y) > maxu∈(0,+∞) ω(u) then the expression of the second equation of (4.6) provides y′ > 0, for all u ∈(0, +∞). Therefore, it would not be possible to connect y = c with y = −c. Remark 4.3. For example, in the Wilson operator g(u) is deﬁned by g(u) = 1 µ u 1−\|u\| c . Then, we can calculate explicitly the values of θ0 and ω0. Those values are Remark 4.3. For example, in the Wilson operator g(u) is deﬁned by g(u) = 1 µ u 1−\|u\| c . Then, we can calculate explicitly the values of θ0 and ω0. Those values are θ0 = c sα −σ ν + (m −1) c µ − r (m −1) c µ !2 , ω0 = 1 ν σ(α −σ) m −2 2−m a m −1 m−1 . Singular patterns in Keller–Segel-type models 1709 (a) m = 2 (b) m = 3 Fig. 8. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. (b) m = 3 (b) m = 3 (a) m = 2 Fig. 8. Representation of the region of existence (gray region) and non-existence (pattern region) of solution in Wilson’s model, for c = µ = a = ν = 1. Fig. 8. Representation of the region of existence (gray region) and non-existence (pattern region) of solution in Wilson’s model, for c = µ = a = ν = 1. Fig. 8. Representation of the region of existence (gray region) and non-existence (pattern region) of solution in Wilson’s model, for c = µ = a = ν = 1. Combining and approximating them we obtain the following inequality: Combining and approximating them we obtain the following inequality: σ m 2 −1(α −σ) m 2 ≥Θ sα ν + (m −1) c µ + r (m −1) c µ ! , (4.7) σ m 2 −1(α −σ) m 2 ≥Θ sα ν + (m −1) c µ + r (m −1) c µ ! , (4.7) Θ = rν c a m −1 m−1 2 1 (m −2) 2−m 2 . (4.7) where Θ = rν c a m −1 m−1 2 1 (m −2) 2−m 2 . Using Theorems 4.1 and 4.2, we can establish the region of existence and non- existence of solution for a given parametric conﬁguration, (see Fig. 8). Observe that the left-hand side of the inequality (4.7) has uni-modal shape and, therefore, we obtain an interval of σ-values for which there is no solution, with all parameters ﬁxed. We can see this behavior in Fig. 8. Note that in the limit case m = 2, the region of non-existence is bounded by a straight line. Moreover, it has been numerically observed, it is possible to ﬁnd a block-type solution under certain parameter settings for m > 2 in the region between the non-existence zone and σ = 0. 4.2. Case m = 1 Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly Representation of the region of existence (gray region) and non-existence (pattern region) of solution in Wilson’s model, for c = µ = a = ν = 1. (a) m = 2 (b) m = 3 Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly Fig. 9. Representation of the phase diagram of (4.8), as a function of σ, m and α. the curves γ+ and γ−, we have more diﬃculties in ﬁnding block solutions of the equation, i.e. solutions that touch these curves. This fact is due to the behavior of the function g(u) as u →±c. In fact, we can show that we may not ﬁnd a block solution, for a certain behavior of the function g, as the following result shows. Proposition 4.3. Assume that Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. 4.2. Case m = 1 In the previous sections, we have always considered the case m > 1, for which we have shown the existence and non-existence of solution under certain conﬁgurations of the parameters. But what happens in the case m = 1? Taking m = 1, the system can be written as follows: Taking m = 1, the system can be written as follows: u′ = ug(ar −σ), r′ = 1 ν ((α −σ)r −u). (4.8) (4.8) Here, γ+ and γ−are horizontal straight lines and it is possible that the point (u, r) = (0, 0) belongs to Γ. If we remove the σ = α and σ = c cases, an isocline analysis reveals the situations given by Fig. 9. In this case, due to the shape of 1710 J. Campos et al. Fig. 9. Representation of the phase diagram of (4.8), as a function of σ, m and α. 2023.33:1693-1719. Downloaded from www.worldscientific.com Re-use and distribution is strictly not permitted, except for Open Access articles. On the other hand, we can also establish conditions for the existence of solution. Proposition 4.4. Assume that Proposition 4.3. Assume that Proposition 4.3. Assume that 1 g(u) = O(c −u), as u →c, 1 g(u) = O(c −u), as u →c, 1 g(u) = O(c −u), as u →c, then there is no block-type solution. Remark 4.4. A type of ﬂux-saturated function that satisﬁes this condition is the Wilson operator. Proof. We are going to use a reductio ad absurdum argument to prove the lemma. Suppose that there is a block solution, then there would be a connection between γ+ and γ−. This means we can ﬁnd a solution branch (¯u, ¯r) in the interval (ξ−, ξ−+ ϵ] that starts over points of γ−. On the other hand, we can consider the problem hand, we can consider the problem r′(u) = 1 ν (α −σ)r −u ug(ar −σ) , r(u0) = r0, (4.9) (4.9) Singular patterns in Keller–Segel-type models 1711 where (u0, r0) ∈γ−. The Picard–Lindelof theorem can be applied to (4.9) extended by zero, prove uniqueness of solution that takes the form r(u) = σ+c a . However, we had assumed that there was a solution branch (¯u, ¯r) that in the form ¯r(¯u) would be the solution of the problem (4.9), which is not possible due to the previous uniqueness argument. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly Lemma 4.2. Consider the equation Proposition 4.4. Assume that 1 g(u) = O((c −u)1/p), as u →c, then there exists a block-type solution. then there exists a block-type solution. Remark 4.5. We can ﬁnd some ﬂux-saturated functions satisfying these conditions on g, such as the hyperbolic tangent operator, or the Larson operator to which it is associated g(u) = 1 µ u pq 1−( \|u\| c )p . The non-uniqueness of the problem (4.9), under the hypothesis of Proposi- tion 4.4, will allow us to prove the existence of a solution for this model. To do this, we will consider the following result. 4.3. Diﬀusion without transport Formally, letting α = 0 in the model (4.1) leads to the following system: Formally, letting α = 0 in the model (4.1) leads to the following system: ∂U ∂t = ∂ ∂x U mΦ U −1 ∂U ∂x −aU ∂Q ∂x , ∂Q ∂t −ν ∂2Q ∂x2 = U, (4.11) ∂U ∂t = ∂ ∂x U mΦ U −1 ∂U ∂x −aU ∂Q ∂x , (4 11) (4.11) ∂Q ∂t −ν ∂2Q ∂x2 = U, (4.11) corresponding to a process where diﬀusion of the chemoattractant dominates the dynamics. As we discussed in the previous section, using the jump condition we can derive a diﬀerential system for the description of entropy solutions. In this case, the equations can be obtained by letting α = 0 in Eq. (4.2) corresponding to a process where diﬀusion of the chemoattractant dominates the dynamics. As we discussed in the previous section, using the jump condition we can derive a diﬀerential system for the description of entropy solutions. In this case, the equations can be obtained by letting α = 0 in Eq. (4.2) u′ = ug ar −σ um−1 , r′ = −1 ν (u + σr). (4.12) (4.12) Therefore, we can analyze the existence of block-type solutions of this problem as a particular case of (4.2). From Theorem 4.1, we have the following result. Corollary 4.1. There is a block-type solution of (4.12), for all σ > 0. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Let us denote by (u−, r−) one of the solution launched from γ−and (u+, r+) one of the solution launched from γ+. Those solutions touch the curve r = σ a at some value u− 1 and u+ 1 , respectively. Assume u− 1 < u+ 1 , the opposite case can be treated similarly. Then, we have that the solution (u+, r+), once it crosses the line r = σ a, it will always touch the curve γ′, since u′ > 0, and it cannot touch the orbit of the solution (u−, r−). Therefore, the solution (u+, r+) connects γ− with γ+. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Lemma 4.2. Consider the equation x′ = x 1 p a(t, x), x > 0, p > 1, (4.10) (4.10) where the function a admits a continuous extension in a neighborhood of (0, 0) and a(0, 0) > 0. Then, the initial value problem (4.10) with x(0) = 0 has a solution x(t) > 0 in a neighborhood of the right-hand side of t = 0. where the function a admits a continuous extension in a neighborhood of (0, 0) and a(0, 0) > 0. Then, the initial value problem (4.10) with x(0) = 0 has a solution x(t) > 0 in a neighborhood of the right-hand side of t = 0. Proof. Let us make the change of variable y = x p−1 p . We have the following initial value problem: y′ = p −1 p a t, y p p−1 , y(0) = 0. y′ = p −1 p a t, y p p−1 , y(0) = 0. By using the Picard theorem, the problem has a solution y(t) with y′(0) = p−1 p a(0, 0) > 0. Therefore, we have that y(t) > 0 on positive values in the neigh- borhood of t = 0. Remark 4.6. If a(0, 0) < 0, then the neighborhood is on the left-hand side of t = 0. We can now proceed to prove Proposition 4.4. We can now proceed to prove Proposition 4.4. Proof. We will use the same constructive scheme developed in the proof of Proposition 4.1. 1712 J. Campos et al. 1712 J. Campos et al. 1712 J. Campos et al. First, we will consider ˜u suﬃciently large, such that r′ < 0 and u′ has an uni- modal shape, under the conditions u > ˜u and r ∈ σ+c a , σ−c a . a a Using Lemma 4.2 and system (4.9), we are be able to launch solutions from the curves γ+ and γ−. Taking the initial conditions such that u0 > ˜u, these solutions will always touch the curve r = σ a (see the proof of Lemma 3.1). Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Using this in the ﬁrst expression of (2.4), as in Sec. 2, we obtain the existence of a value K such as 4.4. Transport without diﬀusion in Q In this last section, we consider that the chemoattractant concentration is solution of a linear transport-diﬀusion equation, which corresponds to the case τ1 = 1 and τ2 = 0 = ν in Eq. (2.1), namely ∂U ∂t = ∂ ∂x U mΦ U −1 ∂U ∂x −aU ∂Q ∂x , ∂Q ∂t + α∂Q ∂x = U. (4.13) (4.13) Singular patterns in Keller–Segel-type models 1713 With a similar argument to the one in Sec. 2, we would obtain (2.4). However, in this case, the expression obtained for q′ does not need to be regular since (2.3) is not satisﬁed. However, if σ ̸= α we can expect that q ∈H1 loc(R) and (α −σ)q′(ξ) = u(ξ), a.e. ξ ∈R. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Fig. 10. Behavior of H in Case 1.1 and the proﬁle of the solution. Fig. 10. Behavior of H in Case 1.1 and the proﬁle of the solution. Fig. 11. Behavior of H in Case 1.2 and the proﬁle of the solution. Note that for 2 ○there is solution because the diﬀerential equation is not deﬁned. Fig. 11. Behavior of H in Case 1.2 and the proﬁle of the solution. Note that for 2 ○there is no solution because the diﬀerential equation is not deﬁned. Fig. 11. Behavior of H in Case 1.2 and the proﬁle of the solution. Note that for 2 ○there is no solution because the diﬀerential equation is not deﬁned. Fig. 11. Behavior of H in Case 1.2 and the proﬁle of the solution. Note that for 2 ○there is no solution because the diﬀerential equation is not deﬁned. Case 1.2. When m = 2, H is again increasing, negative but H(+∞) = − a σ−α < 0. This case opens two possibilities see Fig. 11. 1 ○The ﬁrst alternative is σ > α + a c , which is similar to Case 1.1, see Fig. 11. ○ c , , g 1.2. 2 ○The second case corresponds with σ ≤α + a c . In this situation there are no solutions because there are no points in which the diﬀerential equation is deﬁned. 1.2. 2 ○The second case corresponds with σ ≤α + a c . In this situation there are no solutions because there are no points in which the diﬀerential equation is deﬁned. Case 1.3. In the case 1 < m < 2 the function H satisﬁes H(∞) = −∞and there is only a critical point at u∗with a maximum value H∗, which are given by H∗:= H(u∗) < 0, u∗= σ(α −σ)(1 −m) a(2 −m) . (4.17) (4.17) According to the relative position of H∗with respect to c, we can distinguish the following cases, see Fig. 12: According to the relative position of H∗with respect to c, we can distinguish the following cases, see Fig. 12: 1.3. 1 ○If H∗≤−c, there are no solutions since there are no points for which the diﬀerential equation is deﬁned. 1.3. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. um(ξ)Φ u′(ξ) u(ξ) − a α −σ u2(ξ) + σu(ξ) = K, a.e. ξ ∈R, where u′ is the Radon–Nikodym derivative. Since um(ξ)Φ u′(ξ) u(ξ) is assumed to be 0 if u(ξ) = 0, it follows that if u has compact support, then outside this support u = 0, and thus K = 0. Therefore, solutions of (4.13) will satisfy the equation u′ = ug a α−σu −σ um−1 , q′ = u α −σ , (4.14) (4.14) at the points of its support. at the points of its support. at the points of its support. Remark 4.7. Since condition (2.3) is not veriﬁed, there is no clear description of the block-type solutions. Therefore, we will describe the maximal branches solutions of (4.16) in order to describe a possible connection between them. Because of the large casuistry, the description of the chemoattractant will not be discussed here. In this section, it will be assumed that σ ̸= α, since the solution u = cte is obtained for σ = α. Setting H(u) = a α−σu −σ um−1 (4.15) the ﬁrst equation of (4.14) can be written as ﬁrst equation of (4.14) can be written as u′ = ug(H(u)), (4.16) (4.16) and the expression of the chemoattractant follows after integrate the second equa- tion of (4.14). and the expression of the chemoattractant follows after integrate the second equa- tion of (4.14). Since the function g is only deﬁned in (−c, c), it is important to know the values of u for which H(u) ∈(−c, c). We will distinguish several cases based on the values of α, σ, m. Case 1. If σ > α, H is always negative because H(0) = −∞and there is no root of H. Case 1.1. If m > 2, H is increasing and H(∞) = 0. Solutions live in the interval (u+, ∞) and are decreasing. It can be extended to −∞, and u(−∞) = +∞. For a ﬁnite value u(ξ+) = u+ and u′(ξ+) = −∞. This can be observed in Fig. 10. 1714 J. Campos et al. Fig. 10. Behavior of H in Case 1.1 and the proﬁle of the solution. Fig. 10. Behavior of H in Case 1.1 and the proﬁle of the solution. Fig. 11. Behavior of H in Case 1.2 and the proﬁle of the solution. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Note that for 2 ○there is n solution because the diﬀerential equation is not deﬁned. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. 2 ○If −c < H∗, the diﬀerential equation is only deﬁned for u ∈(u+, u−). Therefore, the solutions are deﬁned in a bounded interval (ξ−, ξ+), where u(ξ−) = u−, u(ξ+) = u+, and u′(ξ−) = u′(ξ+) = −∞. Case 2. If 0 < σ < α, then H(0) = −∞, but H changes sign in ˆu, which is given by 2. If 0 < σ < α, then H(0) = −∞, but H changes sign in ˆu, which is given by ˆu = σ(α −σ) a . (4.18) (4.18) a Singular patterns in Keller–Segel-type models 1715 Fig. 12. Behavior of H in Case 1.3 and the proﬁle of the solution. Note that for 1 ○there is no solution because the diﬀerential equation is not deﬁned. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Fig. 12. Behavior of H in Case 1.3 and the proﬁle of the solution. Note that for 1 ○there is no solution because the diﬀerential equation is not deﬁned. Fig. 13. Behavior of H in the case 1 < m < 2 and 0 < σ < α. Note that both traveling waves are continued by zero. Fig. 13. Behavior of H in the case 1 < m < 2 and 0 < σ < α. Note that both traveling waves are continued by zero. Case 2.1. In the case 1 < m < 2, H has no critical points and H(+∞) = +∞. Therefore, there are two types of traveling waves, one increasing and one decreasing that are represented in Fig. 13. Case 2.2. If m = 2 and 0 < α < σ, H has no critical points, but there is a ﬁnite asymptotic value H(+∞) = a α −σ > 0. The position of this asymptotic value with respect to c gives us three diﬀerent situations: 2.2. 1 ○If a α−σ > c. This is a scenario similar to Case 2.1, see the graph in Fig. 14. Fig. 14. Behavior of H in the case m = 2 and 0 < α < σ. In this case, the traveling waves that have a ﬁnite height are continued by zero. The traveling wave of type 2 ○and 3 ○are not bounded, and, therefore, is conditioned by a more general theory of the initial value problem. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. 3.33:1693-1719. Downloaded from www.worldscientific.com use and distribution is strictly not permitted, except for Open Access articles. Fig. 15. Behavior of H in the case m > 2 and 0 < α < σ. 2.2. 2 ○If a α−σ = c. The modiﬁcation in this case is the non-existence of a point r+ that cuts the graph of H to the c level (see Fig. 14), and now the solutions end at inﬁnity. They can reach inﬁnity in ﬁnite or inﬁnite time depending on the properties of Φ. 2.2. 2 ○If a α−σ = c. The modiﬁcation in this case is the non-existence of a point r+ that cuts the graph of H to the c level (see Fig. 14), and now the solutions end at inﬁnity. They can reach inﬁnity in ﬁnite or inﬁnite time depending on the properties of Φ. 2.2. 3 ○If 0 < a α−σ < c. This is a situation similar to the previous one, but inﬁnite is reached in inﬁnite time, see Fig. 14. 2.2. 3 ○If 0 < a α−σ < c. This is a situation similar to the previous one, but inﬁnite is reached in inﬁnite time, see Fig. 14. Case 2.3. In the case m > 2, H(+∞) = 0, and the function H reaches a maximum level H∗> 0. Then, it is necessary to compare this number with c and we can deﬁne three diﬀerent scenarios and the values of σ for which the diﬀerent traveling waves are deﬁned, see Fig. 15. Case 2.3. In the case m > 2, H(+∞) = 0, and the function H reaches a maximum level H∗> 0. Then, it is necessary to compare this number with c and we can deﬁne three diﬀerent scenarios and the values of σ for which the diﬀerent traveling waves are deﬁned, see Fig. 15. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Fig. 14. Behavior of H in the case m = 2 and 0 < α < σ. In this case, the traveling waves that have a ﬁnite height are continued by zero. The traveling wave of type 2 ○and 3 ○are not bounded, and, therefore, is conditioned by a more general theory of the initial value problem. 1716 J. Campos et al. p Fig. 15. Behavior of H in the case m > 2 and 0 < α < σ. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. Acknowledgments This work is partially supported by the RTI2018-098850-B-I00 from the MICINN- Feder (Spain), B-FQM-580-UGR20 & PY18-RT-2422 from the Junta de Andaluc´ıa (Spain). 4.4.1. Conclusion and summary The idea of this last section is to determine under which conditions we can ﬁnd block-type solutions. To do this we have analyzed all the diﬀerent solution proﬁles satisfying Eq. (4.14). We have basically found two types of proﬁles that we can denominate, according to their character, as increasing or decreasing proﬁles. Both are separated by the level u∗, deﬁned in (4.18), which deﬁnes the point of possible sign change of the H function. However, the only type of compact support solution we have found has a decreasing proﬁle, see Fig. 16. This compact support solution exists for σ > α, Fig. 16. Representation of the only block solution of (4.13). Fig. 16. Representation of the only block solution of (4.13). Singular patterns in Keller–Segel-type models 1717 1 < m < 2 and H∗> −c. The last condition can be expressed, after several standard calculations, as 1 < m < 2 and H∗> −c. The last condition can be expressed, after several standard calculations, as σ(2−m) (σ −α)m−1 > c 2 −m m −1 a(2 −m) m−1 . References 1. A. Aguirre-Tamaral, M. Camb´on, D. Poyato, J. Soler and I. Guerrero, A guidance model predicting the role of Ihog–glypicans interaction in cytoneme orientation for Hedgehog signaling, Nature Commun. 13 (2022) 5647. 2. L. Ambrosio and N. Fusco, Functions of Bounded Variation and Free Discontinuity Problems, Oxford Mathematical Monographs (Oxford Univ. Press, 2000). 3. F. Andreu, V. Caselles and J. Maz´on, The Cauchy problem for a strongly degenerate quasilinear equation, J. Eur. Math. Soc. 7 (2005) 361–393. 4. F. Andreu, V. Caselles, J. Maz´on, J. Soler and M. Verbeni, Radially symmetric solu- tions of a tempered diﬀusion equation. A porous media, ﬂux-limited case, SIAM J. Math. Anal. 44 (2012) 1019–1049. ( ) 5. G. Anzellotti, Pairings between measures and bounded functions and compensated compactness, Ann. Mat. Pura. Appl. 135 (1983) 293–318. 6. M. Arias, J. Campos and J. Soler, Cross-diﬀusion and traveling waves in porous- media ﬂux-saturated Keller–Segel models, Math. Models Methods Appl. Sci. 28 (2018) 2103–2129. 7. N. Bellomo, A. Bellouquid, J. Nieto and J. Soler, Multicellular growing systems: Hyperbolic limits towards macroscopic description, Math. Models Methods Appl. Sci. 17 (2007) 1675–1693. 8. N. Bellomo, A. Bellouquid, J. Nieto and J. Soler, Complexity and mathematical tools toward the modelling of multicellular growing systems, Math. Comput. Model. 51 (2010) 441–451. 9. N. Bellomo, A. Bellouquid, J. Nieto and J. Soler, Multiscale biological tissue models and ﬂux-limited chemotaxis from binary mixtures of multicellular growing systems, Math. Models Methods Appl. Sci. 10 (2010) 1179–1207. ( ) 10. N. Bellomo, A. Bellouquid, Y. Tao and M. Winkler, Toward a mathematical theory of Keller–Segel models of pattern formation in biological tissues, Math. Models Methods Appl. Sci. 25 (2015) 1663–1763. 11. N. Bellomo, N. Outada, J. Soler, Y. Tao and M. Winkler, Chemotaxis and cross- diﬀusion models in complex environments: Models and analytic problems toward a multiscale vision, Math. Models Methods Appl. Sci. 32 (2022) 713–792. 12. N. Bellomo and M. Winkler, Finite-time blow-up in a degenerate chemotaxis system with ﬂux limitation, Trans. Amer. Math. Soc. Ser. B 4 (2017) 31–67. 13. N. Bellomo and M. Winkler, A degenerate chemotaxis system with ﬂux limitation. Maximally extended solutions and absence of gradient blow-up, Commun. Partial Diﬀerential Equations 42 (2017) 436–473. 1718 J. Campos et al. 14. A. Blanchet, J. Dolbeault and B. Perthame, Two-dimensional Keller–Segel model: Optimal critical mass and qualitative properties of the solutions, Electron. J. Diﬀer- ential Equations 32 (2006) 44. 15. N. References Bournaveas and V. Calvez, The one-dimensional Keller–Segel model with fractional diﬀusion of cells, Nonlinearity 23 (2010) 923. 16. J. Calvo, J. Campos, V. Caselles, O. S´anchez and J. Soler, Flux–saturated porous media equation and applications, EMS Surv. Math. Sci. 2 (2015) 131–218. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded from www.worldscientific.com by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly not permitted, except for Open Access articles. 17. J. Calvo, J. Campos, V. Caselles, O. S´anchez and J. Soler, Pattern formation in a ﬂux limited reaction–diﬀusion equation of porous media type, Invent. Math. 206 (2016) 57–108. 18. J. Calvo, J. Campos, V. Caselles, O. S´anchez and J. Soler, Qualitative behaviour for ﬂux-saturated mechanisms: Travelling waves, waiting time and smoothing eﬀects, J. Eur. Math. Soc. 19 (2017) 441–472. ( ) 19. J. Calvo, J. Maz´on, J. Soler and M. Verbeni, Qualitative properties of the solutions of a nonlinear ﬂux-limited equations arising in the transport of morphogens, Math. Models Methods Appl. Sci. 21 (2011) 893–937. 20. J. Campos, P. Guerrero, O. S´anchez and J. Soler, On the analysis of traveling waves to a nonlinear ﬂux limited reaction–diﬀusion equation, Ann. Inst. H. Poincare (C) Non Linear Anal. 30 (2013) 141–155. 21. J. Campos and J. Soler, Qualitative behavior and traveling waves for ﬂux-saturated porous media equations arising in optimal mass transportation, Nonlinear Anal. TMA 137 (2016) 266–290. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly 22. J. A. Carrillo, V. Caselles and S. Moll, On the relativistic heat equation in one space dimension, Proc. Lond. Math. Soc. 107 (2013) 1395–1423. ( ) 23. V. Caselles, An existence and uniqueness result for ﬂux limited diﬀusion equations, Discrete Contin. Dyn. Syst. 31 (2011) 1151–1195. 24. V. Caselles, On the entropy conditions for some ﬂux limited diﬀusion equations, J. Diﬀerential Equations 250 (2011) 3311–3348. 25. V. Caselles, Flux limited generalized porous media diﬀusion equations, Publ. Mat. 57 (2013) 144–217. 26. V. Caselles, Convergence of ﬂux limited porous media diﬀusion equations to its clas- sical counterpart, Ann. Sc. Norm. Super. Pisa Cl. Sci. XIV (2015) 481–505. 27. F. Cerreti, B. Perthame, Ch. Schmeiser, M. Tang and N. Vauchelet, Waves for an hyperbolic Keller–Segel model and branching instabilities, Math. Models Methods Appl. Sci. 21 (2011) 825–842. ( ) 28. F. A. Chalub, Y. Dolak-Struss, P. Markowich, D. Oeltz, C. Schmeiser and A. Soref, Model hierarchies for cell aggregation by chemotaxis, Math. Models Methods Appl. Sci. 16 (2006) 1173–1198. 29. F. A. Chalub, P. Markovich, B. Perthame and C. Schmeiser, Kinetic models for chemo- taxis and their drift–diﬀusion limits, Monatsh. Math. 142 (2004) 123–141. 30. P. H. Chavanis, Nonlinear mean ﬁeld Fokker–Planck equations. Application to the chemotaxis of biological populations, Eur. Phys. J. B 62 (2008) 179–208. 31. A. Chertock, A. Kurganov and P. Rosenau, Formation of discontinuities in ﬂux-saturated degenerate parabolic equations, Nonlinearity 16 (2003) 1875– 1898. 32. A. Chertock, A. Kurganov, X. Wang and Y. Wu, On a chemotaxis model with satu- rated chemotactic ﬂux, Kinet. Related Models 5 (2012) 51–95. ( ) 33. E. Dessaud, L. L. Yang, K. Hill, B. Cox, F. Ulloa, A. Ribeiro, A. Mynett, B. G. Novitch and J. Briscoe, Interpretation of the sonic hedgehog morphogen gradient by a temporal adaptation mechanism, Nature 450 (2007) 717–720. Singular patterns in Keller–Segel-type models 1719 34. J. Dolbeault and B. Perthame, Optimal critical mass in the two–dimensional Keller– Segel model in R2, C. R. Math. Acad. Sci. Paris 339 (2004) 611–616. 35. J. Dolbeault and C. Schmeiser, The two–dimensional Keller–Segel model after blow- up, Discrete Contin. Dyn. Syst. 25 (2009) 109–121. 36. L. Giacomelli, Finite speed of propagation and waiting-time phenomena for degenerate parabolic equations with linear growth Lagrangian, SIAM J. Math. Anal. 47 (2015) 2426–2441. 37. T. Hillen and A. Potapov, The one-dimensional chemotaxis model: Global existence and asymptotic proﬁle, Math. Methods Appl. Sci. Math. Models Methods Appl. Sci. 2023.33:1693-1719. Downloaded by UNIVERSIDAD DE GRANADA on 07/10/23. Re-use and distribution is strictly 27 (2004) 1783–1801. 38. T. Hillen and K. J. Painter, A user’s guide to PDE models for chemotaxis, J. Math. Biol. 58 (2009) 183–217 39. E. F. Keller, Assessing the Keller–Segel model: how has it fared? in Proc. Conf. Biological Growth and Spread (Springer, 1980), pp. 379–387. 40. E. F. Keller and L. A. Segel, Traveling bands of chemotactic bacteria: A theoretical analysis, J. Theor. Biol. 30 (1971) 235–248. 41. A. Kurganov and P. Rosenau, On reaction processes with saturating diﬀusion, Non- linearity 19 (2006) 171–193 . 42. C. D. Levermore and G. C. Pomraning, A ﬂux-limited diﬀusion theory, Astrophys. J. 248 (1981) 321–334. 43. D. Mihalas and B. Mihalas, Foundations of Radiation Hydrodynamics (Oxford Univ. Press, 1984). 44. H. Othmer and T. Hillen, The diﬀusion limit of transport equations II: Chemotaxis equations, SIAM J. Appl. Math. 62 (2002) 1222–1250. 45. C. S. Patlak, Random walk with persistent and external bias, Bull. Math. Biophys. 15 (1953) 311–338. 46. P. Rosenau, Tempered diﬀusion: A transport process with propagating front and inertial delay, Phys. Rev. A 46 (1992) 7371–7374. 47. M. Verbeni, O. S´anchez, E. Mollica, I. Siegl-Cachedenier, A. Carleton, I. Guerrero, A. Ruiz i Altaba and J. Soler, Modeling morphogenetic action through ﬂux-limited spreading, Phys. Life Rev. 10 (2013) 457–475.
https://openalex.org/W4315786008	https://link.springer.com/content/pdf/10.1007/s10854-022-09564-0.pdf	English	null	High dielectric-energy storage and ferromagnetic-superparamagnetic properties: tetra-doping CuO nanocompositions	Journal of materials science. Materials in electronics	2,023	cc-by	12,027	High dielectric-energy storage and ferromagnetic- superparamagnetic properties: tetra-doping CuO nanocompositions A. M. Youssef1 and S. M. Yakout1,* 1Inorganic Chemistry Department, National Research Centre (NRC), El Buhouth St., Dokki 12622, Cairo, Egypt J Mater Sci: Mater Electron (2023) 34:105 (0123456789().,-volV)(0123456789().,-volV) High dielectric-energy storage and ferromagnetic- superparamagnetic properties: tetra-doping CuO nanocompositions A. M. Youssef1 and S. M. Yakout1,* 1Inorganic Chemistry Department, National Research Centre (NRC), El Buhouth St., Dokki 12622, Cairo, Egypt J Mater Sci: Mater Electron (2023) 34:105 (0123456789().,-volV)(0123456789().,-volV) High dielectric-energy storage and ferromagnetic- superparamagnetic properties: tetra-doping CuO nanocompositions A. M. Youssef1 and S. M. Yakout1,* 1Inorganic Chemistry Department, National Research Centre (NRC), El Buhouth St., Dokki 12622, Cairo, Egypt J Mater Sci: Mater Electron (2023) 34:105 (0123456789().,-volV)(0123456789().,-volV) J Mater Sci: Mater Electron (2023) 34:105 ABSTRACT Tetra-doping by (Mn, Fe, Co, Ni) ions strongly boosted the room temperature dielectric constant and the ferromagnetic-superparamagnetic characteristics of monoclinic CuO structure. In this study, undoped CuO, Cu0.98Mn0.005Fe0.005- Co0.005Ni0.005O, Cu0.96Mn0.01Fe0.01Co0.01Ni0.01O and Cu0.94Mn0.015Fe0.015Co0.015- Ni0.015O nanocompositions were synthesized through coprecipitation technique. The crystal structure analysis veriﬁed that all samples have a pure single phase, corresponding to monoclinic CuO structure. The substitution of Cu2?-sites into CuO lattice by Mn2?, Fe2?/3?, Co2? and Ni2? ions has been deduced from the expansions of lattice constant, shifts of XRD diffraction peaks and band gap energy alteration. The additions of (Mn, Fe, Co, Ni) ions lead to the formation of homogenous distributed very ﬁne spherical nanoparticles, especially at large concentrations of dopants (Cu0.94Mn0.015Fe0.015Co0.015Ni0.015O sample). The tetra-doping by (Mn, Fe, Co, Ni) ions reduced the intensity of the diffuse reﬂectance alongside red shifted the absorption edge and the band gap energy of monoclinic CuO structure. Cu0.98Mn0.005Fe0.005Co0.005Ni0.005O exhibits a high relative permittivity value of 6096 at low frequency of 42 Hz with small dielectric loss tangent (tan d) compared to pure one. The tetra-doping by (Mn, Fe, Co, Ni) dopants induced excellent intrinsic ferromagnetic and superpara- magnetic hysteresis loops into monoclinic CuO structure with full saturation loops shape and variable coercivity values. [1–3]. The modern renewable resources such as solar power and wind enable the electrical energy to being produced in a mass amount [4]. For economic and environmental beneﬁt, storing the electrical energy to be consumed efﬁciently later is an important issue [4]. For these reasons, the quality of the electronic High dielectric-energy storage and ferromagnetic- superparamagnetic properties: tetra-doping CuO nanocompositions A. M. Youssef1 and S. M. Yakout1,* A. M. Youssef1 and S. M. Yakout1,* 1Inorganic Chemistry Department, National Research Centre (NRC), El Buhouth St., Dokki 12622, Cairo, Egypt Inorganic Chemistry Department, National Research Centre (NRC), El Buhouth St., Dokki 12622, Cairo, Egyp Received: 22 August 2022 Accepted: 8 November 2022 Published online: 12 January 2023 The Author(s) 2023 Address correspondence to E-mail: s_mabrok2002@yahoo.com 1 Introduction In recent years, the p- and n-types metal oxide semiconductors have gained more attention owing to their applicable dependent electrical, optoelectronic, magnetic and dielectric energy storage properties https://doi.org/10.1007/s10854-022-09564-0 J Mater Sci: Mater Electron (2023) 34:105 105 Page 2 of 19 105 resonance imaging and drug delivery [29–33]. Cop- per oxide (CuO) structure is an interesting p-type semiconductor material with remarkable electrical, magnetic, optical and thermal characteristics as well as inexpensive and nontoxic [34]. CuO semiconduc- tor is technologically well-known material having multifunctional properties with promising applica- tions in magnetic storage media [35], gas sensor [36], optical devices [37], catalysts [38], lithium-ion bat- teries [39], p–n diode [40], solar energy [41] and superconductors [42]. The physical electrical, dielec- tric, magnetic and optical properties of CuO can be controlled or enriched through incorporation of appropriate dopants into its lattice. Previous studies on the inﬂuences of the transition metals doping and codoping on the optical, electrical and magnetic properties of CuO were carried out [43–49]. As far we known, no available study on the competition effect of the four dopants including Mn, Fe, Co and Ni ions on the optical, electrical, dielectric, ferromagnetic and superparamagnetic characteristics of CuO was detected. The combination between variable dopants can induce and enhances many different features. Compared to single-doping, the multi-doping can help in improving the concentration and stability of the defects, tuning the dopants populations as well as the durable bonding between dopants considerably decreases the formation energy. The ionic radii of Mn, Fe, Co and Ni ions are analogous to that of Cu2?- sites; hence the replacement can take place without causing much lattice distortion. Furthermore, 3d- electrons of Mn, Fe, Co and Ni ions can interact into CuO and enhance the optical, electrical and magnetic properties. In this study, Mn2?, Fe2?/3?, Co2? and Ni2? as multi-substituted ions were used to tune the optical, dielectric and magnetic properties of mono- clinic CuO semiconductor for multi-functional applications. In this work, equal concentration was used for each element equal to 0.5%, 1 and 1.5 wt% to achieve the following compositions Cu0.98Mn0.005- Fe0.005Co0.005Ni0.005O, Cu0.96Mn0.01Fe0.01Co0.01Ni0.01O and Cu0.94Mn0.015Fe0.015Co0.015Ni0.015O. A future separate study on the inﬂuence of varying the con- centrations of these dopants (in unequal proportions) will be carried out. systems that can store energy plays a leading role in the storage of the electrical energy [5]. J Mater Sci: Mater Electron (2023) 34:105 J Mater Sci: Mater Electron (2023) 34:105 105 2 Synthesis, characterization and measurements 3 mm thickness. The room temperature magnetic behavior (VSM loop) of 0%TC, 2%TC, 4%TC and 6%TC samples was found through using a vibrating sample magnetometer (VSM, LakeShore Model 7410). Pure CuO, Cu0.98Mn0.005Fe0.005Co0.005Ni0.005O, Cu0.96- Mn0.01Fe0.01Co0.01Ni0.01O and Cu0.94Mn0.015Fe0.015- Co0.015Ni0.015O nanocompositions were synthesized through the coprecipitation process. For simplicity and ease, the obtained samples are coded as 0%TC (pure CuO), 2%TC (Cu0.98Mn0.005Fe0.005Co0.005Ni0.005- O), 4%TC (Cu0.96Mn0.01Fe0.01Co0.01Ni0.01O) and 6%TC (Cu0.94Mn0.015Fe0.015Co0.015Ni0.015O). Copper chloride dihydrate (CuCl22H2O) was used as a source of copper ions while MnCl24H2O, Fe(NO3)39H2O, CoCl26H2O and Ni(NO3)26H2O were used as sour- ces for dopants ions. Desired weights of the starting materials were dissolved in 200 ml deionized water under continuous stirring for 2 h. Ammonia solution (30% NH4OH) was added dropwise to form the hydroxide precipitates at pH = 8, then the total solution left for 2 h. For many times, the obtained precipitates were washed by using deionized water until removing the undesirable dissolved ions such as chloride ions, the veriﬁcation was achieved by using silver nitrate test for chloride ions in residual solu- tion. After complete removing of the chloride ions, the obtained precipitates were also washed for two times, dried and calcined at 500 C. The phase structure of the synthesized 0%TC, 2%TC, 4%TC and 6%TC samples was investigated by X-ray diffraction (XRD) (PANalytical X-ray diffraction equipment model X0Pert PRO) with step size [2h] = 0.0260. The XRD results were analyzed by Rietveld reﬁnement to estimate the lattice constants. The microstrain present in the lattice and the crystallite size of the synthesized compositions were calculated by simpliﬁed integral breadth technique of Williamson–Hall. The diffuse reﬂectance (R), absorption edge and energy of the band gap of 0%TC, 2%TC, 4%TC and 6%TC samples were studied by double beam spectrophotometer- JASCO (model V-570 UV–Vis-NIR). The morpholog- ical structure of 0%TC, 2%TC, 4%TC and 6%TC samples was investigated by scanning electron microscope (SEM, model Quanta 250 FEG). The room temperature relative permittivity (dielectric constant, e0), dielectric loss tangent (tan d) and the ac electrical conductivity characteristics of the fabricated 0%TC, 2%TC, 4%TC and 6%TC pellets were carried out by using LCR meter (Hitester, model Hioki 3532-50, Japan). The synthesized powders were compressed to 3 Results and discussion 1 Introduction The dielectric capacitors as storage devices have many advantages including high power, fast charging and long life time as well as can deliver the electrical energy almost instantly [6, 7]. These remarks make the dielectric capacitors a perfect choice for great energy storage. Also, the fast development of the micro- electronic devices for aerospace engineering and electrical vehicles make immediately need for dielectric materials with high energy storage perfor- mance [8–10]. The dielectric properties (dielectric constant and dielectric loss tangent, tan d) of the fabricated materials are very important for its appli- cability in the electronics industry [8]. To investigate the permittivity or the dielectric characteristics of any structure, the dielectric constant (e0) is a fundamental aspect that determines the capacity or amplitude of the material for charge storage [11–14]. Colossal rel- ative permittivity materials are utilized as memory cells, gate dielectrics in metal oxide semiconductor (MOS) transistors, capacitors and supercapacitors [15–18]. In the same context, low relative permittivity materials have advanced applications in high-speed integrated circuits and electrical devices [19]. The metal oxides semiconductors (MOS) as dielectric substances are main components for varied thin-ﬁlm electronic chips due to their remarkable mechanical and dielectric properties [20]. On the other hand, the metal oxides semiconductors are looked upon as promising materials in diluted magnetic semicon- ductor (DMS) systems for spin-electronics and magnetic-medical applications [21–23]. Reinforce the ferromagnetic and superparamagnetic characteristics of the metal oxides semiconductors alongside the dielectric properties can inspire their multi-func- tionality in spin-electronics and medical technology [24, 25]. Spin-electronics is a novel future technology that utilizes the electron spin together with its charge to upsurge the storage capacity, speed and efﬁciency [26–28]. For spin-electronics technology, one of the essential requirements is the synthesis of single phase structure which holds both the ferromagnetic-semi- conducting characteristics at room temperature [26–28]. The metal oxides semiconductors with room J Mater Sci: Mater Electron (2023) 34:105 Page 3 of 19 105 Page 3 of 19 105 3.1 XRD: phase, crystallite size and microstrain The lattice constant (a, b c) and unit cell volume (V) of 0%TC sample are Fig. 1 XRD patterns of 0%TC, 2%TC, 4%TC and 6%TC samples synthesized by coprecipitation method Fig. 2 Williamson–Hall plots of 0%TC, 2%TC, 4%TC and 6%TC samples, calculation of crystallite size and microstrain 105 Page 4 of 19 J Mater Sci: Mater Electron (2023) 34:105 J Mater Sci: Mater Electron (2023) 34:105 105 Page 4 of 19 d d by Fig. 1 XRD patterns of 0%TC, 2%TC, 4%TC and 6%TC samples synthesized by coprecipitation method respectively. These increases in the microstrain can be ascribed to the mismatch between the ionic radius of Cu2?-sites as parent cation and the ionic radii of Mn, Fe, Co and Ni as guest ions. The insertion of Mn, Fe, Co and Ni ions alongside the defects such as Cu and O vacancies is somewhat inﬂuenced on the microstrain of the pure CuO sample. Like micros- train, the crystallite size of the 0%TC sample was slightly increased from 50 nm to 61, 56 and 60 nm after tetra-doping by 2 wt%, 4 wt% and 6 wt% (Mn, Fe, Co, Ni) ions, respectively. The variations in the lattice constant, unit cell volume and position of XRD peaks were used to evaluate the actual replacement of Cu2? by Mn, Fe, Co, Ni ions alongside deduce their most effective oxidation states. The lattice constant (a, b, c) and unit cell volume (V) of 0%TC sample are found to be a = 4.6876 A˚ , b = 3.4262 A˚ , c = 5.1311 A˚ and V = 81.7233 A˚ 3. Upon tetra-doping by Mn, Fe, Co and Ni ions, both the lattice constant and the unit cell volume of CuO were increased as illustrated in Fig. 3 and Table 1. The growths in the unit cell vol- ume of the pure CuO indicated that large ionic radii Fig. 2 Williamson–Hall plots of 0%TC, 2%TC, 4%TC and 6%TC samples, calculation of crystallite size and microstrain Fig. 2 Williamson–Hall plots of 0%TC, 2%TC, 4%TC and 6%TC samples, calculation of crystallite size and microstrain respectively. These increases in the microstrain can be ascribed to the mismatch between the ionic radius of Cu2?-sites as parent cation and the ionic radii of Mn, Fe, Co and Ni as guest ions. 3.1 XRD: phase, crystallite size and microstrain Figure 1 depicts the X-ray powder diffraction (XRPD) patterns of the synthesized 0%TC, 2%TC, 4%TC and 6%TC samples within 2h range from 20 to 80. For 0%TC sample, thirteen diffraction peaks located at 2h = 32.562, 35.561, 38.756, 46.298, 48.818, 53.499, 58.258, 61.588, 65.86, 66.312, 68.053, 72.40 and 75.15 were detected, corresponding to (110) (- 111), (111), (112), (- 202), (020), (202), (- 113), (311), (022), (220), (- 312) and (203) crystal- lographic planes of monoclinic CuO (JCPDS card ﬁle No. 48-1548, space group: C2/c), respectively. Like- wise, thirteen high intensity crystallographic X-ray diffraction peaks were found for 2%TC, 4%TC and 6%TC samples. The nonappearance of any other diffraction peaks veriﬁed the high purity and exclude the occurrence of any impurities. For each composi- tion, the diffraction peaks have high intensities, sig- nifying a well crystallinity. The microstrain and crystallite size of the 0%TC, 2%TC, 4%TC and 6%TC samples were estimated via simpliﬁed integral breadth technique of Williamson–Hall. The Wil- liamson–Hall plot formula can be represented as follow [50]: bhkl cos hhkl ¼ ðKk=DÞ þ 4e sin hhkl ð1Þ ð1Þ bhkl cos hhkl ¼ ðKk=DÞ þ 4e sin hhkl where b, h, K, k, D, e and hkl symbolize the width at half maximum (radian), position of the peaks (ra- dian), K is the Scherrer constant, wavelength of inci- dent X-ray (0.15406 nm), crystallite size, microstrain and Miller indices, respectively. The relation between the 4sinhhkl (X-axis) and bcoshhkl (Y-axis) provides the microstrain by estimating the slope of the ﬁtting line and the size of crystallite (D) was calculated using the intercept value on bcoshhkl axis (D = Kk/intercept) as illustrated in Fig. 2. The microstrain of the 0%TC sample was found to be 0.000043, Fig. 3. The incor- poration of 2 wt%, 4 wt% and 6 wt% (Mn, Fe, Co, Ni) ions inside CuO lattice increases the microstrain of CuO progressively to 0.00013, 0.00025 and 0.00033, respectively. These increases in the microstrain can be ascribed to the mismatch between the ionic radius of Cu2?-sites as parent cation and the ionic radii of Mn, Fe, Co and Ni as guest ions. The insertion of Mn, Fe Co and Ni ions alongside the defects such as Cu lattice constant, unit cell volume and position of XRD peaks were used to evaluate the actual replacement of Cu2? by Mn, Fe, Co, Ni ions alongside deduce their most effective oxidation states. 3.1 XRD: phase, crystallite size and microstrain The insertion of Mn, Fe, Co and Ni ions alongside the defects such as Cu and O vacancies is somewhat inﬂuenced on the microstrain of the pure CuO sample. Like micros- train, the crystallite size of the 0%TC sample was slightly increased from 50 nm to 61, 56 and 60 nm after tetra-doping by 2 wt%, 4 wt% and 6 wt% (Mn, Fe, Co, Ni) ions, respectively. The variations in the lattice constant, unit cell volume and position of XRD peaks were used to evaluate the actual replacement of Cu2? by Mn, Fe, Co, Ni ions alongside deduce their most effective oxidation states. The lattice constant (a, b, c) and unit cell volume (V) of 0%TC sample are found to be a = 4.6876 A˚ , b = 3.4262 A˚ , c = 5.1311 A˚ and V = 81.7233 A˚ 3. Upon tetra-doping by Mn, Fe, Co and Ni ions, both the lattice constant and the unit cell volume of CuO were increased as illustrated in Fig. 3 and Table 1. The growths in the unit cell vol- ume of the pure CuO indicated that large ionic radii Page 5 of 19 105 J Mater Sci: Mater Electron (2023) 34:105 0%TC 2%TC 4%TC 6%TC 0.0000 0.0001 0.0002 0.0003 0.0004 0.0005 Microstrain 0.000043 0.00013 0.00025 0.00033 0%TC 2%TC 4%TC 6%TC 0 30 60 90 nm Crystallite size (A) (A)3 O 50 61 56 60 0%TC 2%TC 4%TC 6%TC 5.125 5.130 5.135 5.140 5.145 5.150 Lattice constant c O 5.1311 5.1410 5.1420 5.1411 0%TC 2%TC 4%TC 6%TC 81 82 Unit cell volume 81.2975 81.7233 81.8128 81.7664 Fig. 3 Microstrain, crystallite size, lattice constant c (A˚ ) and unit cell volume (A˚ )3 of 0%TC, 2%TC, 4%TC and 6%TC samples Table 1 Lattice parameters (a, b, c), unit cell volume (V) and relative permittivity (e0) values at 42, 100, 200, 400 and 1000 Hz for 0%TC Fig. 3 Microstrain, crystallite size, lattice constant c (A˚ ) and unit cell volume (A˚ )3 of 0%TC, 2%TC, 4%TC Fig. 3 Microstrain, crystallite size, lattice constant c (A˚ ) and unit cell volume (A˚ )3 of 0%TC, 2%TC, 4%TC and 6%TC samples Table 1 Lattice parameters (a, b, c), unit cell volume (V) and relative permittivity (e0) values at 42, 100, 200, 400 and 1000 Hz for 0%TC Fig. 3 Microstrain, crystallite size, lattice constant c (A˚ ) and unit cell volume (A˚ )3 of 0%TC, 2%TC, 4%TC and 6%TC samples Table 1 Lattice parameters (a, b, c), unit cell volume (V) and relative permittivity (e0) values at 42, 100, 200, 400 and 1000 Hz for 0%TC, 2%TC, 4%TC and 6%TC samples Samples a (A˚ ) b (A˚ ) c (A˚ ) V (A˚ 3) e0 42 Hz e0 100 Hz e0 200 Hz e0 400 Hz e0 1000 Hz 0%TC 4.6876 ± 0.0003 3.4262 ± 0.0003 5.1311 ± 0.0004 81.2975 850 351 305 270 236 2%TC 4.6957 ± 0.0006 3.4316 ± 0.0005 5.1410 ± 0.0001 81.7233 6096 2936 1706 1053 608 4%TC 4.6980 ± 0.0001 3.4324 ± 0.0006 5.1420 ± 0.0007 81.8128 3014 1156 857 659 462 6%TC 4.6971 ± 0.0012 3.4317 ± 0.001 5.1411 ± 0.0012 81.7664 2328 1232 759 485 289 Table 1 Lattice parameters (a, b, c), unit cell volume (V) and relative permittivity (e0) values at 42, 100, 200 2%TC, 4%TC and 6%TC samples ions were substituted the Cu2?-sites into CuO lattice. The transition metals including Mn, Fe, Co and Ni ions have variable oxidation states and each oxida- tion state has its particular ionic radius. In sixfold coordination (VI), Mn2? has two ionic radii of 0.67 A˚ (low spin state, VI) or 0.83 A˚ (high spin state, VI) while Mn3? has ionic radius of 0.58 A˚ (low spin state, VI) or 0.645 A˚ (high spin state, VI). The ionic radius of Fe2? in sixfold coordination is 0.61 A˚ (low spin ions were substituted the Cu2?-sites into CuO lattice. The transition metals including Mn, Fe, Co and Ni ions have variable oxidation states and each oxida- tion state has its particular ionic radius. In sixfold coordination (VI), Mn2? has two ionic radii of 0.67 A˚ (low spin state, VI) or 0.83 A˚ (high spin state, VI) while Mn3? has ionic radius of 0.58 A˚ (low spin state, VI) or 0.645 A˚ (high spin state, VI). The ionic radius of Fe2? in sixfold coordination is 0.61 A˚ (low spin state, VI) or 0.78 A˚ (high spin state, VI) while Fe3? possesses ionic radius of 0.55 (low spin state, VI) or 0.645 A˚ (high spin state, VI). The ionic radius of Co2? ions is 0.65 A˚ (low spin state, VI) or 0.745 A˚ (high spin state, VI) and that of Ni2? is 0.69 A˚ (VI). The measured increases in the unit cell volume give a clear sign that Mn, Fe, Co and Ni ions have been mostly inserted as Mn2? (0.83 A˚ , high spin state), Fe2? (0.78 A˚ , high spin state), Co2? (0.745 A˚ , high J Mater Sci: Mater Electron (2023) 34:105 Page 6 of 19 105 spin state) ions of larger ionic radii plus Ni2? (0.69 A˚ ) ions. The changes of the 2-theta position of the (- 111) and (111) crystallographic planes support the above conclusions as illustrated in Fig. 4. The sub- stitution of Mn, Fe, Co and Ni ions for Cu2? sites in CuO lattice appears successful, because the main XRD peaks, (- 111) or (111), of the 2%TC, 4%TC and 6%TC samples have been shifted towards lower 2h angle as well as the peaks intensity was somewhat reduced relative to 0%TC sample. It can be noticed that with increasing the concentration of (Mn, Fe, Co, Ni) ions from 2 wt% to 4 wt% or 6 wt%, the expan- sion in the unit cell volume and the shift of the (- 111) and (111) crystallographic planes aren’t changed. This means that at high concentrations of (Mn, Fe, Co, Ni), mixed oxidation states may be incorporated as Fe2? and Fe3? ions and the integrated effects of these ions lead to the observed shifts and expansions. synthesized 0%TC powder, Fig. 5a, demonstrates the formation of asymmetrical grains having variable sizes with appearance of some agglomeration. The SEM image of 2%TC sample has shown very ﬁne grains (major) with occurrence of some rod grains (minor) as illustrated in Fig. 5b. The SEM image of 4%TC powder clearly shows the presence of some asymmetrical large grains besides very ﬁne (major) and rods particles, Fig. 5c. In case of 6%TC sample, homogenous distributed very ﬁne grains are formed, Fig. 5d. These results show that Mn2?, Fe2?/3?, Co2? or Ni2? dopants have remarkable roles in improving the homogenous shape of the grains besides reducing their sizes through acting as restriction ions for the growth of the grains. Figure 6 displays the size dis- tribution histograms of 0%TC, 2%TC, 4%TC and 6%TC samples. It can be noticed that 6%TC sample exhibits the lowest particles size distribution. For photocatalytic, optoelectronics and other applica- tions, the roughness value of the surface is an important factor. The measured average values of the surface roughness were 4.84, 3.99, 2.87, and 3.48 for 0%TC, 2%TC, 4%TC and 6%TC powders, respectively. 3.2 Scanning electron microscope (SEM) Figure 5 reveals the scanning electron microscopy (SEM) and the 3D pictures of 0%TC, 2%TC, 4%TC and 6%TC powders. The micrograph of the Fig. 4 Impact of (Mn, Fe, Co, Ni) ions tetra-doping on 2h- position of (-111) and (111) crystallographic planes of 0%TC sample Fig. 4 Impact of (Mn, Fe, Co, Ni) ions tetra-doping on 2h- position of (-111) and (111) crystallographic planes of 0%TC sample Fig. 4 Impact of (Mn, Fe, Co, Ni) ions tetra-doping on 2h- position of (-111) and (111) crystallographic planes of 0%TC sample J Mater Sci: Mater Electron (2023) 34:105 Page 7 of 19 105 Page 7 of 19 105 J Mater Sci: Mater Electron (2023) 34:105 Page 7 of 19 105 J Mater Sci: Mater Electron (2023) 34:105 Fig. 5 SEM images as well as 3D views of a 0%TC, b 2%TC, c 4%TC and d 6%TC powders Fig. 5 SEM images as well as 3D views of a 0%TC, b 2%TC, c 4%TC and d 6%TC powders 105 Page 8 of 19 J Mater Sci: Mater Electron (2023) 34:105 105 Page 8 of 19 J Mater Sci: Mater Electron (2023) 34:105 Fig. 6 Size distribution histograms of 0%TC, 2%TC, 4%TC and 6%TC samples Fig. 6 Size distribution histograms of 0%TC, 2%TC, 4%TC and 6%TC samples display a marked difference in terms of infrared absorption between pure and tetra-doped CuO samples, in relation with the amount of Mn, Fe, Co and Ni ions introduced. To calculate the band gap energy of 0%TC, 2%TC, 4%TC and 6%TC samples from reﬂectance data, the Kubelka–Munk and Tauc’s equations were used. The band gap energy is esti- mated from the plot between (F(R) ht)2 and ht [52]. The Kubelka–Munk function and the Tauc’s equa- tions can be writing as shown below [52–55]: 3.3 Optical properties: diffuse reﬂectance and band gap energy The relative permittivity and dielectric loss tangent of the syn- thesized compositions are related to the occurrence of space charge polarization alongside on structural defects, temperature, morphology, frequency and chemical composition [59, 60]. Figure 8a illustrates the dependence of the relative permittivity on fre- quency for 0%TC, 2%TC, 4%TC and 6%TC pellets at room temperature. The values of the relative per- mittivity were reduced with growing the applied frequency. Herein, the curves of the dielectric con- stant show three regions with frequency; from 42 Hz to 1 kHz the samples reveal strong decreases, from 1 to 100 kHz the relative permittivity slowly decreased and from 1 kHz to 5 MHz the relative permittivity is nearly not affected by frequency. At 42 Hz, 0%TC sample has a relative permittivity value of 850 while 2%TC, 4%TC and 6%TC samples exhibit high relative permittivity values of 6096, 3014 and 2328 respec- tively. The relative permittivity values of the 0%TC, Fig. 7d. The shrinkages in the band gap can be ascribed to the occurrence of oxygen vacancies and impurity states within the conduction band and the valance band, initiating by the (Mn, Fe, Co, Ni) tetra- doping [56, 57]. The impurity bands originate by Mn2? (3d5), Fe2? (3d6), Co2? (3d7) and Ni2? (3d8) electrons mix with the conduction band and decrease the width of the band gap of CuO. Reengineering of the band gap structure of CuO veriﬁes that Mn2?, Fe2?/3?, Co2? and Ni2? ions have been replaced the Cu2?-sites into CuO structure. Fig. 7d. The shrinkages in the band gap can be ascribed to the occurrence of oxygen vacancies and impurity states within the conduction band and the valance band, initiating by the (Mn, Fe, Co, Ni) tetra- doping [56, 57]. The impurity bands originate by Mn2? (3d5), Fe2? (3d6), Co2? (3d7) and Ni2? (3d8) electrons mix with the conduction band and decrease the width of the band gap of CuO. Reengineering of the band gap structure of CuO veriﬁes that Mn2?, Fe2?/3?, Co2? and Ni2? ions have been replaced the Cu2?-sites into CuO structure. 3.3 Optical properties: diffuse reﬂectance and band gap energy Figure 7 illustrates the room temperature diffuse reﬂectance, Kubelka–Munk function vs. wavelength, [F(R) hm]2 against hm (eV) and band gap values of 0%TC, 2%TC, 4%TC and 6%TC powders. The diffuse reﬂectance spectra for the 0%TC, 2%TC, 4%TC and 6%TC nanopowders were measured and presented in the spectral range of 200–2400 nm. Upon incorpora- tion of Mn, Fe, Co and Ni dopants, the reﬂectance (%) of the 0%TC powder was signiﬁcantly reduced in the wavelength of 700–2400 nm and this effect is more pronounced with increasing the (Mn, Fe, Co, Ni) concentration as shown in Fig. 7a. Also, the absorp- tion edge was noticeably changed and red shifts for higher wavelengths were observed. The reﬂectance (%) was converted to absorptivity through applying the Kubelka–Munk function (F (R) = (1-R)2/2R = a/ S) [51]. The normalized diffuse reﬂectance spectra (Kubelka–Munk function vs. wavelength) are shown in Fig. 7b. The normalized diffuse reﬂectance spectra FðRÞ ¼ ð1 RÞ2=2R ¼ a=S ð2Þ ahv ¼ Aðhv EgÞn ð3Þ ð2Þ ð2Þ ð3Þ ð3Þ R is the reﬂectance, h is known as a plank’s con- stant, ht is the photon energy in eV, Eg is the band gap and n = 2. From the (F(R) hm)2 and hm plot, the band gap was found through the intercept of the linear part of the curve with X-axis as illustrated in Fig. 7c. The measured band gap for 0%TC sample was 1.445 eV and those for 2%TC, 4%TC and 6%TC samples were 1.39, 1.35 and 1.37 eV, respectively, Page 9 of 19 105 J Mater Sci: Mater Electron (2023) 34:105 Fig. 7 Illustrates a: Diffuse reﬂectance, b: Kubelka–Munk function vs. wavelength, c: [F(R)hm]2 against hm (eV) plots for optical band gap energy determination and d Band gap energy values of 0%TC, 2%TC, 4%TC and 6%TC samples Fig. 7 Illustrates a: Diffuse reﬂectance, b: Kubelka–Munk function vs. wavelength, c: [F(R)hm]2 against hm (eV) plots for optical band gap energy determination and d Band gap energy values of 0%TC, 2%TC, 4%TC and 6%TC samples Fig. 7 Illustrates a: Diffuse reﬂectance, b: Kubelka–Munk function vs. wavelength, c: [F(R)hm]2 against energy determination and d Band gap energy values of 0%TC, 2%TC, 4%TC and 6%TC samples polarization and ionic conduction [58]. 3.4 Dielectric properties The relative permittivity (dielectric constant, e0) and the dielectric loss tangent (tan d) are two fundamental electrical parameters that give information concern- ing the conduction behavior in CuO structure. The real part of relative permittivity symbolizes the polarization of the substance while the dielectric loss tangent (tan d) signiﬁes the energy loss due to the J Mater Sci: Mater Electron (2023) 34:105 105 Page 10 of 19 2%TC, 4%TC and 6%TC samples at 42, 100, 200, 400 and 1000 Hz are presented in Table 1. Figure 9 demonstrates the inﬂuence of (Mn, Fe, Co, Ni) con- centration on the values of the dielectric constant of CuO at 42 and 100 Hz. The order of the relative permittivity value following the sequence of 2%TC [ 4%TC [ 6%TC [ 0%TC. The results verify that the 2 wt% (Mn, Fe, Co, Ni) tetra-doping pos- sesses a high afﬁrmative impact on the relative per- mittivity of CuO material. At low concentration, it Fig. 8 Shows a Dependence of room temperature dielectric constant on frequency and b Dependence of room temperature dielectric loss tangent (tan d) on frequency for 0%TC, 2%TC, 4%TC and 6%TC samples 0%TC 2%TC 4%TC 6%TC 0 1000 2000 3000 4000 5000 6000 7000 Dielectric constant 42 Hz 100 Hz 850 6096 3014 2328 351 2936 1156 1232 Fig. 9 Room temperature dielectric constant of 0%TC, 2%TC, 4%TC and 6%TC samples at 42 Hz and 100 Hz Fig. 8 Shows a Dependence of room temperature dielectric constant on frequency and b Dependence of room temperature dielectric loss tangent (tan d) on frequency for 0%TC, 2%TC, 4%TC and 6%TC samples Fig. 9 Room temperature dielectric constant of 0%TC, 2%TC, 4%TC and 6%TC samples at 42 Hz and 100 Hz Fig. 9 Room temperature dielectric constant of 0%TC, 2%TC, 4%TC and 6%TC samples at 42 Hz and 100 Hz 0%TC 2%TC 4%TC 6%TC 0 1000 2000 3000 4000 5000 6000 7000 Dielectric constant 42 Hz 100 Hz 850 6096 3014 2328 351 2936 1156 1232 Fig. 9 Room temperature dielectric constant of 0%TC, 2%TC, 4%TC and 6%TC samples at 42 Hz and 100 Hz 0%TC 2%TC 4%TC 6%TC 0 1000 2000 3000 4000 5000 6000 7000 Dielectric constant 42 Hz 100 Hz 850 6096 3014 2328 351 2936 1156 1232 2%TC, 4%TC and 6%TC samples at 42, 100, 200, 400 and 1000 Hz are presented in Table 1. 3.4 Dielectric properties Figure 9 demonstrates the inﬂuence of (Mn, Fe, Co, Ni) con- centration on the values of the dielectric constant of CuO at 42 and 100 Hz. The order of the relative permittivity value following the sequence of 2%TC [ 4%TC [ 6%TC [ 0%TC. The results verify that the 2 wt% (Mn, Fe, Co, Ni) tetra-doping pos- sesses a high afﬁrmative impact on the relative per- mittivity of CuO material. At low concentration, it permittivity value following the sequence of 2%TC [ 4%TC [ 6%TC [ 0%TC. The results verify that the 2 wt% (Mn, Fe, Co, Ni) tetra-doping pos- sesses a high afﬁrmative impact on the relative per- mittivity of CuO material. At low concentration, it 2%TC, 4%TC and 6%TC samples at 42, 100, 200, 400 and 1000 Hz are presented in Table 1. Figure 9 demonstrates the inﬂuence of (Mn, Fe, Co, Ni) con- centration on the values of the dielectric constant of CuO at 42 and 100 Hz. The order of the relative 3 5 El t i l d l l i seems that the hopping mechanism between Mn2?, Fe2?, Co2?, Ni2? and Cu2? is predominant and leads to higher values of relative permittivity while at greater concentration, the dopants ions existing in the CuO lattice may behave as deep donor. Thus, it may inhibit the conduction mechanism and does not contribute to the conduction process but provides the impedance to it and hence decreases the relative permittivity values. The measured relative permit- tivity behavior can be discussed by Maxwell–Wagner effect [61, 62]. The Maxwell–Wagner view considers the dielectric substance as composed of highly con- ducting grains disconnected by the weak conducting grain boundaries. Under ac electric ﬁeld, by hopping mechanism the electrons pass from the grains to the grain boundaries and crowd owing to the large resistance which yield polarization. As a result, the relative permittivity values are very high at the lower frequencies. When the frequency growth, the hop- ping of the electric charge cannot track the electric ﬁeld, consequently the electrons have weak chance to reach the grain boundary, leading to less polarization (low relative permittivity value). Figure 8b displays the dependence of dielectric loss tangent (tan d, energy dissipation) on frequency for 0%TC, 2%TC, 4%TC and 6%TC samples. At low frequency, the dielectric loss tangent (tan d) has high value espe- cially for 0%TC sample owing to interfacial polar- ization and with increasing the frequency the loss tangent decreases until reach a constant value. Interestingly, at lower frequencies the (Mn, Fe, Co, Ni) tetra-doping strongly reduces the dielectric loss tangent of pure CuO to large extent owing to the inﬂuence of the space charge polarization. This means that the 2%TC, 4%TC and 6%TC samples can store more energy in the low frequency region. The domain wall resonance can be used to explain the reduction in the dielectric loss tangent with fre- quency [59]. At low frequency, additional energy is needed due to the high grain boundary resistance while at high frequency the resistance decreased and the electrons required a small energy to transfer. The high e0 and the low tan d values make the 2%TC composition suitable for energy storage applications. M ¼ M0 þ M00 M0 ¼ e0=½ðe0Þ2 þ ðe00Þ2 M00 ¼ e0=½ðe0Þ2 þ ðe00Þ2 Figure 10 demonstrates the log frequency-depen- dent the real part (M’) and the imaginary part (M’’) of 0%TC, 2%TC, 4%TC and 6%TC samples. The real part (M0) has nearly zero value in the low frequency region, indicating a negligible inﬂuence from elec- trodes [65]. At higher frequency, the real part (M0) of the complex modulus increases throughout the fre- quency range used. In the range of 100 kHz to 5 MHz region, both (M’) and (M’’) show relaxation peaks at certain frequency. The observed peaks point out to that the relaxation phenomenon in CuO samples is due to the dielectric relaxation. At higher frequency, (M’) and (M’’) have higher values for 4%TC and 6%TC samples compared to 0%TC and 2%TC samples. 3.7 Magnetic properties analysis 3.7 Magnetic properties analysis Page 11 of 19 105 105 J Mater Sci: Mater Electron (2023) 34:105 and M’’) is correlated to the complex permittivity e * (e0 and e00) by the next equations [64]: seems that the hopping mechanism between Mn2?, Fe2?, Co2?, Ni2? and Cu2? is predominant and leads to higher values of relative permittivity while at greater concentration, the dopants ions existing in the CuO lattice may behave as deep donor. Thus, it may inhibit the conduction mechanism and does not contribute to the conduction process but provides the impedance to it and hence decreases the relative permittivity values. The measured relative permit- tivity behavior can be discussed by Maxwell–Wagner effect [61, 62]. The Maxwell–Wagner view considers the dielectric substance as composed of highly con- ducting grains disconnected by the weak conducting grain boundaries. Under ac electric ﬁeld, by hopping mechanism the electrons pass from the grains to the grain boundaries and crowd owing to the large resistance which yield polarization. As a result, the relative permittivity values are very high at the lower frequencies. When the frequency growth, the hop- ping of the electric charge cannot track the electric ﬁeld, consequently the electrons have weak chance to reach the grain boundary, leading to less polarization (low relative permittivity value). Figure 8b displays the dependence of dielectric loss tangent (tan d, energy dissipation) on frequency for 0%TC, 2%TC, 4%TC and 6%TC samples. At low frequency, the dielectric loss tangent (tan d) has high value espe- cially for 0%TC sample owing to interfacial polar- ization and with increasing the frequency the loss tangent decreases until reach a constant value. Interestingly, at lower frequencies the (Mn, Fe, Co, Ni) tetra-doping strongly reduces the dielectric loss tangent of pure CuO to large extent owing to the inﬂuence of the space charge polarization. This means that the 2%TC, 4%TC and 6%TC samples can store more energy in the low frequency region. The domain wall resonance can be used to explain the reduction in the dielectric loss tangent with fre- quency [59]. At low frequency, additional energy is needed due to the high grain boundary resistance while at high frequency the resistance decreased and the electrons required a small energy to transfer. The high e0 and the low tan d values make the 2%TC composition suitable for energy storage applications. 3.6 Electrical conductivity Figure 11 displays the room temperature ac electrical conductivity (rac) with log frequency (F) for 0%TC, 2%TC, 4%TC and 6%TC samples, within range from 42 Hz to 5 MHz. Approximately, the rac independent on frequency up to 10 kHz followed by steady or gradual increases up to 100 kHz, and after that the rac reveals exponential increases. The variation of the rac with frequency displays two regions from fre- quency-independent (region of the low frequency) to frequency-dependent (region of the high frequency), indicating relaxation phenomenon of rac. The grad- ual increases of rac with frequency can be discussed based on interfacial Maxwell–Wagner inﬂuence [59, 60]. The weak conducting grain boundaries are more effective at lower frequency, which consecu- tively decreases the rac of the CuO samples. When frequency growth, the electron hopping is improved, consequently the rac was increased. 3.5 Electrical modulus analysis Figure 12 demonstrates magnetic characteristics (VSM hysteresis loops, M-H relation) of 0%TC, 2%TC, 4%TC and 6%TC nanopowders within ± 20 KOe at room temperature. On the whole, all the To explore the nature of the relaxation, the real and imaginary components of the complex modulus have been investigated [63]. The electrical modulus M* (M’ J Mater Sci: Mater Electron (2023) 34:105 synthesized compositions exhibit S-hysteresis shape with different saturation magnetization and coerciv- ity values, Table 2. The M-H relation of 0%TC pow- der illustrates full ferromagnetic performance with saturation magnetization of 0 045 emu/g and coercivity of 150 Oe, attributed to the uncompensated spins on the CuO surface [66, 67]. Upon incorporation of 2 wt% (Mn, Fe, Co, Ni) into CuO structure (2%TC), the ferromagnetic performance converts to nearly superparamagnetic with coercivity value close to Fig. 10 Displays a Dependence of the real part of the complex electric modulus, M’, on frequency and b Dependence of the imaginary part of the complex electric modulus, M’’, on frequency for 0%TC, 2%TC, 4%TC and 6%TC samples Fig. 11 Dependence of room temperature ac electrical conductivity on frequency for 0%TC, 2%TC, 4%TC and 6%TC samples Fig. 10 Displays a Dependence of the real part of the complex electric modulus, M’, on frequency and b Dependence of the imaginary part of the complex electric modulus, M’’, on frequency for 0%TC, 2%TC, 4%TC and 6%TC samples ys of the real part electric on frequency ence of the of the complex us, M’’, on 0%TC, 2%TC, TC samples synthesized compositions exhibit S-hysteresis shape with different saturation magnetization and coerciv- ity values, Table 2. The M-H relation of 0%TC pow- der illustrates full ferromagnetic performance with saturation magnetization of 0.045 emu/g and coercivity of 150 Oe, attributed to the uncompensated spins on the CuO surface [66, 67]. Upon incorporation of 2 wt% (Mn, Fe, Co, Ni) into CuO structure (2%TC), the ferromagnetic performance converts to nearly superparamagnetic with coercivity value close to Fig. 11 Dependence of room temperature ac electrical conductivity on frequency for 0%TC, 2%TC, 4%TC and 6%TC samples synthesized compositions exhibit S-hysteresis shape with different saturation magnetization and coerciv- ity values Table 2 The M H relation of 0%TC pow coercivity of 150 Oe, attributed to the uncom spins on the CuO surface [66, 67]. Upon incor of 2 wt% (Mn Fe Co Ni) into CuO structure Fig. 3.5 Electrical modulus analysis The crystal structure analysis, optical prop- erties and band gap reengineering of 0%TC, 2%TC, 4%TC and 6%TC samples recommended the effective replacement of Cu2?-sites by Mn2?, Fe2?/3?, Co2? and Ni2? ions. The expanded XRD pattern of 2%TC, 4%TC and 6%TC compositions demonstrates notable 2h angle shifts for the (- 111) and (111) crystallographic planes compared to 0%TC sample. As results, these shifts and the changes of lattice constant agree with Vegard’s law and point to the substitutional tetra-doping of Mn2?, Fe2?, Co2? and Ni2? ions at the Cu2?-sites [68]. Based on XRD peaks, no secondary phases associated with any impurities oxides were detected. Linked to these comments, we result, the synthesized compositions exhibits two distinctive kinds of magnetic characteristic based on the coercivity values. The 0%TC and 4%TC compo- sitions display nearly ferromagnetism at room tem- perature, whereas 2%TC and 6%TC compositions reveal approximately room temperature superpara- magnetism with coercivity value close to zero (14 or 15 Oe). The crystal structure analysis, optical prop- erties and band gap reengineering of 0%TC, 2%TC, 4%TC and 6%TC samples recommended the effective replacement of Cu2?-sites by Mn2?, Fe2?/3?, Co2? and Ni2? ions. The expanded XRD pattern of 2%TC, 4%TC and 6%TC compositions demonstrates notable 2h angle shifts for the (- 111) and (111) crystallographic planes compared to 0%TC sample. As results, these shifts and the changes of lattice constant agree with Vegard’s law and point to the substitutional tetra-doping of Mn2?, Fe2?, Co2? and Ni2? ions at the Cu2?-sites [68]. Based on XRD peaks, no secondary phases associated with any impurities oxides were detected. Linked to these comments, we Table 2 Magnetic parameters including saturation magnetization (Ms), coercivity (Hc), retentivity (Mr) and squareness (Sq) of 0%TC, 2%TC, 4%TC and 6%TC samples ( ) y ( ) y ( ) q ( q) 0%TC, 2%TC, 4%TC and 6%TC samples Samples Ms (emu/g) Hc (Oe) Mr (emu/g) Sq 0%TC 0.045 150 0.007 0.15 2%TC 059 14 0.033 0.056 4%TC 0.88 29 0.054 0.061 6%TC 1.04 15 0.047 0.045 zero (14 Oe). As well, the saturation magnetization value is greatly boosted to 0.59 emu/g. With increasing (Mn, Fe, Co, Ni) concentration to 4 wt% (4%TC), the magnetization increased to 0.88 emu/g and the coercivity growth to 29 Oe (close the ferro- magnetic performance). 3.5 Electrical modulus analysis 11 Dependence of room temperature ac electrical conductivity on frequency for 0%TC, 2%TC, 4%TC and 6%TC samples endence of room c electrical on frequency for C, 4%TC and es synthesized compositions exhibit S-hysteresis shape with different saturation magnetization and coerciv- ity values, Table 2. The M-H relation of 0%TC pow- der illustrates full ferromagnetic performance with saturation magnetization of 0.045 emu/g and coercivity of 150 Oe, attributed to the uncompensated spins on the CuO surface [66, 67]. Upon incorporation of 2 wt% (Mn, Fe, Co, Ni) into CuO structure (2%TC), the ferromagnetic performance converts to nearly superparamagnetic with coercivity value close to synthesized compositions exhibit S-hysteresis shape with different saturation magnetization and coerciv- ity values, Table 2. The M-H relation of 0%TC pow- der illustrates full ferromagnetic performance with saturation magnetization of 0.045 emu/g and coercivity of 150 Oe, attributed to the uncompensated spins on the CuO surface [66, 67]. Upon incorporation of 2 wt% (Mn, Fe, Co, Ni) into CuO structure (2%TC), the ferromagnetic performance converts to nearly superparamagnetic with coercivity value close to Page 13 of 19 105 J Mater Sci: Mater Electron (2023) 34:105 -20k -10k 0 10k 20k -0.08 -0.06 -0.04 -0.02 0.00 0.02 0.04 0.06 0.08 -1000 -500 0 500 1000 Ms (emu/g) Ms (emu/g) (a) Hc = 150 Oe Ms = 0.045 emu/g -20k -10k 0 10k 20k -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 -1000 -500 0 500 1000 (b) Hc = 14 Oe Ms = 0.59 emu/g -20k -10k 0 10k 20k -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2 (c) Hc = 15 Oe Hc = 29 Oe Ms = 0.88 emu/g -20k -10k 0 10k 20k -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 H (KOe) H (KOe) (d) Ms = 1.04 emu/g Fig. 12 Room temperature VSM patterns of a 0%TC, b 2%TC, c 4%TC and d 6%TC powders (d) Fig. 12 Room temperature VSM patterns of a 0%TC, b 2%TC, c 4%TC and d 6%TC powders result, the synthesized compositions exhibits two distinctive kinds of magnetic characteristic based on the coercivity values. The 0%TC and 4%TC compo- sitions display nearly ferromagnetism at room tem- perature, whereas 2%TC and 6%TC compositions reveal approximately room temperature superpara- magnetism with coercivity value close to zero (14 or 15 Oe). 3.5 Electrical modulus analysis SEM micrographs show the presence of some large grains in the powder of 4%TC sample while 2%TC and 6%TC samples con- tain only major particles of very ﬁne shape. According to RKKY concept [80], the superparam- agnetic characteristics at room temperature can be assigned to the interactions of the spin-polarized localized electrons with the conductive one. Spin polarization of the conductive electrons take place due to these exchange interactions. As a result, the spin-polarized conductive electrons can make exchange interactions with the spin-polarized local- ized electrons of the Mn2?, Fe2?, Co2? and Ni2? ions. Owing to the prolonged area of the exchange inter- actions, approximately all Mn2?, Fe2?/3?, Co2? and Ni2? ions can exhibit an analogous spin direction. In this case, the spin-polarized conductive electrons may be considered as a media to join all Mn2?, Fe2?/ 3?, Co2? and Ni2? ions. Subsequently, the 2%TC and 6%TC compositions display superparamagnetic behavior at room temperature. As an alternative, the room temperature superparamagnetism of 2%TC and 6%TC compositions can be also discussed based on the existence of two kinds of magnetic characteristics [81]. The ﬁrst kind is the ferromagnetic characteristic, which induced by defects like oxygen vacancies while the second kind is the antiferromagnetic per- formance, which produced by the interaction of Mn2?, Fe2?/3?, Co2? and Ni2? ions (Mn?–Mn? interaction). Under this situation, it looks that the antiferromagnetic interaction at room temperature obstructs the coupling of the ferromagnetism, pro- ducing a quick decreases in the magnetic moment. The immediate occurrence of room temperature fer- romagnetism and antiferromagnetism characteristics induces a disordered state of superparamagnetism. Based on SEM micrographs, the difference in the coercivity values between 4%TC (29 Oe, close to fer- romagnetism), 2%TC (14 Oe, close to superparam- agnetism) and 6%TC (15 Oe, close to superparamagnetism) may be attributed to the dif- ference in particles shape. SEM micrographs show the presence of some large grains in the powder of 4%TC sample while 2%TC and 6%TC samples con- tain only major particles of very ﬁne shape. 3.5 Electrical modulus analysis There is an additional mechanism termed F-center (FC, oxy- gen vacancies mediated ferromagnetism) that it is a sub-mechanism or an equivalent path to the BMPs mechanism to describe the ferromagnetic character- istics at room temperature in undoped or transition (Mn, Fe, Co, Ni) or rare earth (Gd, Ce, Sm) metals lightly doped oxides semiconductor [76–79]. In the F-center mechanism, the electrons trapped by the oxygen vacancies can act as coupling F-centers, over which the doped magnetic ions like Mn, Fe, Co, Ni, Gd and Sm align in ferromagnetic order. The F-center expects that Mn?–VO– Mn? states (Mn? = doped magnetic ions and VO = oxygen vacancies) will be According to RKKY concept [80], the superparam- agnetic characteristics at room temperature can be assigned to the interactions of the spin-polarized localized electrons with the conductive one. Spin polarization of the conductive electrons take place due to these exchange interactions. As a result, the spin-polarized conductive electrons can make exchange interactions with the spin-polarized local- ized electrons of the Mn2?, Fe2?, Co2? and Ni2? ions. Owing to the prolonged area of the exchange inter- actions, approximately all Mn2?, Fe2?/3?, Co2? and Ni2? ions can exhibit an analogous spin direction. In this case, the spin-polarized conductive electrons may be considered as a media to join all Mn2?, Fe2?/ 3?, Co2? and Ni2? ions. Subsequently, the 2%TC and 6%TC compositions display superparamagnetic behavior at room temperature. As an alternative, the room temperature superparamagnetism of 2%TC and 6%TC compositions can be also discussed based on the existence of two kinds of magnetic characteristics [81]. The ﬁrst kind is the ferromagnetic characteristic, which induced by defects like oxygen vacancies while the second kind is the antiferromagnetic per- formance, which produced by the interaction of Mn2?, Fe2?/3?, Co2? and Ni2? ions (Mn?–Mn? interaction). Under this situation, it looks that the antiferromagnetic interaction at room temperature obstructs the coupling of the ferromagnetism, pro- ducing a quick decreases in the magnetic moment. The immediate occurrence of room temperature fer- romagnetism and antiferromagnetism characteristics induces a disordered state of superparamagnetism. Based on SEM micrographs, the difference in the coercivity values between 4%TC (29 Oe, close to fer- romagnetism), 2%TC (14 Oe, close to superparam- agnetism) and 6%TC (15 Oe, close to superparamagnetism) may be attributed to the dif- ference in particles shape. 3.5 Electrical modulus analysis At 6 wt% (Mn, Fe, Co, Ni) concentration (6%TC), the S-hysteresis shape reveals a magnetization of 1.06 emu/g and a coercivity value of 15 Oe, close to the superparamagnetism. As a zero (14 Oe). As well, the saturation magnetization value is greatly boosted to 0.59 emu/g. With increasing (Mn, Fe, Co, Ni) concentration to 4 wt% (4%TC), the magnetization increased to 0.88 emu/g and the coercivity growth to 29 Oe (close the ferro- magnetic performance). At 6 wt% (Mn, Fe, Co, Ni) concentration (6%TC), the S-hysteresis shape reveals a magnetization of 1.06 emu/g and a coercivity value of 15 Oe, close to the superparamagnetism. As a 105 Page 14 of 19 J Mater Sci: Mater Electron (2023) 34:105 105 can deduce that the room temperature ferromagnetic- superparamagnetic characteristics of 0%TC, 2%TC, 4%TC and 6%TC compositions have intrinsic causes and not associated with any impurities. The accurate cause of the ferromagnetic and superparamagnetic characteristics in the undoped or low doped oxides semiconductor (diluted magnetic semiconductor) is still in uncertainty [69]. Many scientists proposed diverse theories or models to illuminate the main cause of the ferromagnetism or superparamagnetism in undoped or low-doped oxides semiconductor including direct interactions like super-exchange and double-exchange interactions [70], Ruderman–Kittel– Kasuya–Yosida (RKKY) exchange mechanism [71], p-d Zener mechanism [72], F-center-based oxygen vacancies interaction and the bound magnetic polar- ons (BMPs) theory [73]. In this context, Coey et al. [74] reported that the observed ferromagnetism at room temperature in lightly doped or multi-doped materials by transition or rare earth metals like Mn, Fe, Co, Ni and Gd dopants could be interpreted via BMP model. The BMP model was proposed for oxi- des semiconductor and it is fundamentally based on the formation of a spin polarized cloud nearby a region holding both magnetic impurities (Mn, Fe, Co, Ni) and donor defects (vacancies). When the size or the number of these spin polarized clouds increases, the BMPs can join which produces magnetically ordered areas. Sequentially, these magnetically ordered areas can powerfully overlap and extent over the total structure which producing a condition comparable to a ferromagnetic state and this type can continue to room temperature or above [75]. 3.5 Electrical modulus analysis There is an additional mechanism termed F-center (FC, oxy- gen vacancies mediated ferromagnetism) that it is a sub-mechanism or an equivalent path to the BMPs mechanism to describe the ferromagnetic character- istics at room temperature in undoped or transition (Mn, Fe, Co, Ni) or rare earth (Gd, Ce, Sm) metals lightly doped oxides semiconductor [76–79]. In the F-center mechanism, the electrons trapped by the oxygen vacancies can act as coupling F-centers, over which the doped magnetic ions like Mn, Fe, Co, Ni, Gd and Sm align in ferromagnetic order. The F-center expects that Mn?–VO– Mn? states (Mn? = doped magnetic ions and VO = oxygen vacancies) will be popular in the composition and the electrons trapped in the oxygen vacancies produce F-centers; where the electron occupies an orbital which overlaps the d orbitals of both magnetic ions neighbors can deduce that the room temperature ferromagnetic- superparamagnetic characteristics of 0%TC, 2%TC, 4%TC and 6%TC compositions have intrinsic causes and not associated with any impurities. The accurate cause of the ferromagnetic and superparamagnetic characteristics in the undoped or low doped oxides semiconductor (diluted magnetic semiconductor) is still in uncertainty [69]. Many scientists proposed diverse theories or models to illuminate the main cause of the ferromagnetism or superparamagnetism in undoped or low-doped oxides semiconductor including direct interactions like super-exchange and double-exchange interactions [70], Ruderman–Kittel– Kasuya–Yosida (RKKY) exchange mechanism [71], p-d Zener mechanism [72], F-center-based oxygen vacancies interaction and the bound magnetic polar- ons (BMPs) theory [73]. In this context, Coey et al. [74] reported that the observed ferromagnetism at room temperature in lightly doped or multi-doped materials by transition or rare earth metals like Mn, Fe, Co, Ni and Gd dopants could be interpreted via BMP model. The BMP model was proposed for oxi- des semiconductor and it is fundamentally based on the formation of a spin polarized cloud nearby a region holding both magnetic impurities (Mn, Fe, Co, Ni) and donor defects (vacancies). When the size or the number of these spin polarized clouds increases, the BMPs can join which produces magnetically ordered areas. Sequentially, these magnetically ordered areas can powerfully overlap and extent over the total structure which producing a condition comparable to a ferromagnetic state and this type can continue to room temperature or above [75]. 4 Conclusions In the study, single phase compositions composed of CuO, Cu0.98Mn0.005Fe0.005Co0.005Ni0.005O, Cu0.96- Mn0.01Fe0.01Co0.01Ni0.01O and Cu0.94Mn0.015Fe0.015- Co0.015Ni0.015O were synthesized for energy storage, medical and spin-electronics applications. For all Page 15 of 19 105 J Mater Sci: Mater Electron (2023) 34:105 compositions, single phase with monoclinic structure was detected via XRD analysis. The expansions of lattice constants, shifts of (- 111) or (111) XRD crys- tallographic planes and the reengineering of the band gap structure veriﬁed the effective replacement of Cu2?-sites by Mn2?, Fe2?/3?, Co2? and Ni2? ions. The SEM results demonstrated that Mn2?, Fe2?/3?, Co2? or Ni2? dopants have remarkable roles in improving the homogenous shape of the grains besides reducing their sizes through acting as restriction ions for the growth of the grains. Obvious red shifts for the absorption edge of CuO were detected due to inser- tion of Mn2?, Fe2?/3?, Co2? and Ni2? ions into its lattice. The 0%TC, 2%TC, 4%TC and 6%TC compo- sitions revealed a band gap energy of 1.445, 1.39, 1.35 and 1.37 eV, respectively. Remarkably, 2%TC sample exhibits a high relative permittivity of 6096 at low frequency of 42 Hz with small loss tangent values compared to 0%TC sample. The synthesized com- positions exhibit two distinctive kinds of magnetic characteristics based on the coercivity values. The 0%TC and 4%TC compositions display full ferro- magnetic behaviour at room temperature with satu- ration magnetization of 0.045 and 0.88 emu/g, whereas 2%TC and 6%TC compositions reveal approximately room temperature superparamag- netism with coercivity value close to zero (14 or 15 Oe). Declarations Conﬂict of interest The authors declare that they have no known competing ﬁnancial interests or per- sonal relationships that could have appeared to inﬂuence the work reported in this paper. Open Access This article is licensed under a Crea- tive Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Crea- tive Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licen ses/by/4.0/. Data availability compositions, single phase with monoclinic structure was detected via XRD analysis. The expansions of lattice constants, shifts of (- 111) or (111) XRD crys- tallographic planes and the reengineering of the band gap structure veriﬁed the effective replacement of Cu2?-sites by Mn2?, Fe2?/3?, Co2? and Ni2? ions. The SEM results demonstrated that Mn2?, Fe2?/3?, Co2? or Ni2? dopants have remarkable roles in improving the homogenous shape of the grains besides reducing their sizes through acting as restriction ions for the growth of the grains. Obvious red shifts for the absorption edge of CuO were detected due to inser- tion of Mn2?, Fe2?/3?, Co2? and Ni2? ions into its lattice. The 0%TC, 2%TC, 4%TC and 6%TC compo- sitions revealed a band gap energy of 1.445, 1.39, 1.35 and 1.37 eV, respectively. Remarkably, 2%TC sample exhibits a high relative permittivity of 6096 at low frequency of 42 Hz with small loss tangent values compared to 0%TC sample. The synthesized com- positions exhibit two distinctive kinds of magnetic characteristics based on the coercivity values. The 0%TC and 4%TC compositions display full ferro- magnetic behaviour at room temperature with satu- ration magnetization of 0.045 and 0.88 emu/g, whereas 2%TC and 6%TC compositions reveal approximately room temperature superparamag- netism with coercivity value close to zero (14 or 15 Oe). The datasets generated during and/or analyzed during the current study are available from the cor- responding author on reasonable request. References The submission of the manuscript has been approved by all coauthors. All authors have equal contributions in preparing the manuscript including experimental, writing, discussion and revision. 1. X. Yu, T.J. Marks, A. Facchetti, Metal oxides for optoelec- tronic applications. Nat. Mater. 15, 383–396 (2016) 2. K. Jeyasubramanian, R.V. William, P. Thiruramanathan, G.S. Hikku, M.V. Kumar, B. Ashima, P. Veluswamy, H. Ikeda, Dielectric and magnetic properties of nanoporous nickel doped zinc oxide for spintronic applications. J. Magn. Magn. Mater. 485, 27–35 (2019) Funding Open access funding provided by The Science, Technology & Innovation Funding Authority (STDF) in cooperation with The Egyptian Knowledge Bank (EKB). 3. G.K. Dalapati, H. Sharma, A. Guchhait, N. Chakrabarty, P. Bamola, Q. Liu, G. Saianand, A.M.S. Krishna, S. Mukhopadhyay, A. Dey, T.K.S. Wong, S. Zhuk, S. Ghosh, S. Chakrabortty, C. Mahata, S. Biring, A. Kumar, C.S. Ribeiro, S. Ramakrishna, A.K. Chakraborty, S. Krishnamurthy, P. Sonar, M. Sharma, Tin oxide for optoelectronic, photovoltaic and energy storage devices: a review. J. Mater. Chem. A 9, 16621–16684 (2021) 105 Page 16 of 19 J Mater Sci: Mater Electron (2023) 34:105 microelectronics. Superlattices Microstruct. 100, 876–885 (2016) 4. P. Duffy, C. Fitzpatrick, T. Conway, R.P. Lynch, Energy sources and supply grids—the growing need for storage, in Energy Storage Options and Their Environmental Impact. ed. by R.E. Hester, R.M. Harrison (Royal Society of Chem- istry; 1st edition, London, 2018), pp.1–41 18. J. Varghese, S. Jasimudeen, K.T. Varghese, Study of the dielectric properties of graphene/CuS/ZnO hybrid nanocom- posites for high performance supercapacitor applications. RSC Adv. 5, 107142–107149 (2015) 5. D. Kaur, T. Sharma, C. Madhu, Dielectric investigations of pristine and modiﬁed ZnO nanoparticles for energy storage devices. J. Mater. Sci.: Mater. Electron. 33, 9905–9917 (2022) 19. R. Singh, R.K. Ulrich, High and low dielectric constant materials. Electrochem. Soc. Interface 8, 26–30 (1999) 6. M. Zhai, A. Li, J. Hu, CuO nanorods grown vertically on graphene nanosheets as a battery-type material for high per- formance supercapacitor electrodes. RSC Adv. 10, 36554–36561 (2020) 20. S. Park, C. Kim, W. Lee, S. Sung, M. Yoon, Sol gel metal oxide dielectrics for all-solution-processed electronics. Mater. Sci. Eng. R. Rep. 114, 1–22 (2017) 21. V.R. Akshay, B. Arun, G. Mandal, M. Vasundhara, Structural, optical and magnetic behavior of sol–gel derived Ni-doped dilute magnetic semiconductor TiO2 nanocrystals for advanced functional applications. Phys. Chem. Chem. Phys. 21, 2519–2532 (2019) 7. H. Xie, X. Liu, R. Wu, J. Liu, J. Wu, L. Li, High-performance supercapacitor with faster energy storage and long cyclic life based on CuO@MnO2 nano-core-shell array on carbon ﬁber surface. ACS Appl. Energy Mater. 3, 7325–7334 (2020) 8. D. Kaur, A. Bharti, T. Sharma, C. Madhu, Dielectric prop- erties of ZnO-based nanocomposites and their potential applications. Int. J. Optics 2021, 9950202 (2021) 22. B. Poornaprakash, S. Ramu, K. Subramanyam, Y.L. Kim, M. Kumar, M.S.P. Reddy, Robust ferromagnetism of ZnO:(- Ni?Er) diluted magnetic semiconductor nanoparticles for spintronic applications. Ceram. Int. Funding 47, 18557–18564 (2021) 9. X. Liu, M. Zheng, G. Chen, Z. Dang, J. Zha, High-temper- ature polyimide dielectric materials for energy storage: theory, design, preparation and properties. Energy Environ. Sci. 15, 56–81 (2022) 23. S. Palanisamy, Y. Wang, Superparamagnetic iron oxide nanoparticulate system: synthesis, targeting, drug delivery and therapy in cancer. Dalton Trans. 48, 9490–9515 (2019) 10. H. Pan, A. Kursumovic, Y. Lin, C. Nan, J.L. MacManus- Driscoll, Dielectric ﬁlms for high performance capacitive energy storage: multiscale engineering. Nanoscale 12, 9582–19591 (2020) 24. J. Kazmi, P.C. Ooi, B.T. Goh, M.K. Lee, M.F.M.R. Wee, S.S.A. Karim, S.R.A. Raza, M.A. Mohamed, Bi-doping improves the magnetic properties of zinc oxide nanowires. RSC Adv. 10, 23297–23311 (2020) 11. W.R. Tinga, S.O. Nelson, Dielectric properties of materials for microwave processing—tabulated. J. Microw. Power 8, 23–65 (1973) 25. M.E. Grigore, E.R. Biscu, A.M. Holban, M.C. Gestal, A.M. Grumezescu, Methods of synthesis, properties and biomedical applications of CuO nanoparticles. Pharmaceuticals 9, 75 (2016) 12. Y. Wang, W. Jie, C. Yang, X. Wei, J. Hao, Colossal permit- tivity materials as superior dielectrics for diverse applications. Adv. Funct. Mater. 29, 1808118 (2019) 26. S.B. Ogale, Dilute doping, defects, and ferromagnetism in metal oxide systems. Adv. Mater. 22, 3125–3155 (2010) 27. Z. Yang, A perspective of recent progress in ZnO diluted magnetic semiconductors. Appl. Phys. A 112, 241–254 (2013) 13. M. Tse, X. Wei, J. Hao, High-performance colossal permit- tivity materials of (Nb ? Er) co-doped TiO2 for large capacitors and high-energy-density storage devices. Phys. Chem. Chem. Phys. 18, 24270–24277 (2016) 28. S. Ning, P. Zhan, Q. Xie, W. Wang, Z. Zhang, Defects-driven ferromagnetism in undoped dilute magnetic oxides: a review. J. Mater. Sci. Technol. 31, 969–978 (2015) 14. H.V.S. Pessoni, P. Banerjee, A. Franco, Colossal dielectric permittivity in Co-doped ZnO ceramics prepared by a pres- sure-less sintering method. Phys. Chem. Chem. Phys. 20, 28712–28719 (2018) 29. J. Lee, B. Kim, Y. Kim, S. Kim, Ultra-high rate of tempera- ture increment from superparamagnetic nanoparticles for highly efﬁcient hyperthermia. Sci. Rep. 11, 4969 (2021) 15. X. Huang, P. Jiang, Core-shell structured high-k polymer nanocomposites for energy storage and dielectric applica- tions. Adv. Mater. 27, 546–554 (2015) 30. P.L. Rathi, B. Ponraj, S. Deepa, Structural, magnetic and electrical properties of electroactive-superparamagnetic PVDF-Sn0.2Fe2.8O4 nanocomposite ﬁlms. Ceram. Int. 47, 9727–9735 (2021) 16. J. Robertson, R.M. Wallace, High-K materials and metal gates for CMOS applications. Mater. Sci. Eng. R. Rep. 88, 1–41 (2015) 31. Y. Funding Wang, C. Xu, Y. Chang, L. Zhao, K. Zhang, Y. Zhao, F. Gao, X. Gao, Ultrasmall super-paramagnetic iron oxide nanoparticle for T2-weighted magnetic resonance imaging. ACS Appl. Mater. Interfaces 9, 28959–28966 (2017) 17. S. Gyanan, A. Mondal, Kumar, Tunable dielectric properties of TiO2 thin ﬁlm based MOS systems for application in Page 17 of 19 105 Page 17 of 19 105 J Mater Sci: Mater Electron (2023) 34:105 32. M. Musielak, I. Piotrowski, W.M. Suchorska, Superparam- agnetic iron oxide nanoparticles (SPIONs) as a multifunc- tional tool in various cancer therapies. Rep. Pract. Oncol. Radiother. 24, 307–314 (2019) 44. Y. Li, M. Xu, L. Pan, Y. Zhang, Z. Guo, C. Bi, Structural and room-temperature ferromagnetic properties of Fe-doped CuO nanocrystals. J. Appl. Phys. 107, 113908 (2010) 45. S.P. Kamble, V.D. Mote, Optical and room-temperature fer- romagnetic properties of Ni-doped CuO nanocrystals pre- pared via autocombustion method. J. Mater. Sci.: Mater. Electron 32, 5309–5315 (2021) 33. M. Nikolaou, K. Avraam, A. Kolokithas-Ntoukas, A. Bakandritsos, F. Lizal, O. Misik, M. Maly, J. Jedelsky, I. Savva, F. Balanean, T. Krasia-Christoforou, Superparamag- netic electrospun microrods for magnetically-guided pul- monary drug delivery with magnetic heating. Mater. Sci. Eng., C 126, 112117 (2021) 46. N.M. Basith, J.J. Vijay, L.J. Kennedy, M. Bououdina, Struc- tural, optical and room-temperature ferromagnetic properties of Fe-doped CuO nanostructures. Phys. E 53, 193–199 (2013) 47. Z. Jing, X. Qinglin, L. Jinmin, Ferromagnetism in Fe-doped CuO nanopowder. J. Semicond. 33, 013001 (2012) 34. A.S. Zoolfakar, R.A. Rani, A.J. Morfa, A.P. O’Mullaned, K. Kalantar-zadeh, Nanostructured copper oxide semiconduc- tors: a perspective on materials, synthesis methods and applications. J. Mater. Chem. C 2, 5247–5270 (2014) 48. S.M. Yakout, A.M. El-Sayed, Structural, morphological and ferromagnetic properties of pure and (Mn, Co) codoped CuO nanostructures. J. Supercond. Nov. Magn. 29, 2961–2968 (2016) 35. M.M. Rashad, D.A. Rayan, A.A. Ramadan, Optical and magnetic properties of CuO/CuFe2O4 nanocomposites. J. Mater. Sci.: Mater. Electron. 24, 2742–2749 (2013) 49. S.M. Yakout, Robust ferromagnetism and active visible-near infrared photocatalytic properties: Fe based Mn Co and Ni codoped CuO nanostructures. Opt. Mater. 112, 110769 (2021) 36. F. Wang, H. Li, Z. Yuan, Y. Sun, F. Chang, H. Deng, L. Xie, H. Li, A highly sensitive gas sensor based on CuO nanoparticles synthetized via a sol–gel method. RSC Adv. 6, 79343–79349 (2016) 50. F.I.H. Rhouma, F. Belkhiria, E. Bouzaiene, M. Daoudi, K. Taibi, J. Dhahri, R. Funding Chtourou, The structure and photolumi- nescence of a ZnO phosphor synthesized by the sol gel method under praseodymium doping. RSC Adv. 9, 5206–5217 (2019) 37. K.H. Al-Mayalee, N. Saadi, E. Badradeen, F. Watanabe, T. Karabacak, Optical and photoconductive response of CuO nanostructures grown by a simple hot water treatment method. J. Phys. Chem. C 122, 23312–23320 (2018) 51. M. Fracchia, P. Ghigna, A. Minguzzi, A. Vertova, F. Turco, G. Cerrato, D. Meroni, Role of synthetic parameters on the structural and optical properties of N, Sn-copromoted nanostructured TiO2: A combined Ti K-edge and Sn L2,3- edges X-ray absorption investigation. Nanomaterials 10, 1224 (2020) 38. O.J.S. Junior, A.F.F. Monteiro, J.B.L. Oliveira, A.M.U. Ara- u´jo, D.G. Silva, J. Kulesza, B.S. Barros, Coordination poly- mer-derived CuO catalysts for oxidative degradation of methylene blue. Mater. Chem. Phys. 235, 121737 (2019) 39. P. Wang, X. Gou, S. Xin, F. Cao, Facile synthesis of CuO nanochains as high-rate anode materials for lithium-ion bat- teries. New J. Chem. 43, 6535–6539 (2019) 52. S.M. Yakout, H.A. Mousa, H.T. Handal, W. Sharmoukh, Role of non-magnetic dopants on the room temperature ferro- magnetism and optical properties of BaSnO3 perovskite. J. Solid State Chem. 281, 121028 (2020) 40. R. Lo´pez, G. Villa-Sa´nchez, I.V. de la Cruz, C. Encarnacio´n- Go´mez, V.H. Castrejo´n-Sa´nchez, A. Coyopol, J.E. Mastache, C. Leyva-Porras, Cupric oxide (CuO)/zinc oxide (ZnO) heterojunction diode with low turn-on voltage. Results Phys. 22, 103891 (2021) 53. M.R. Islam, Md. Saiduzzaman, S.S. Nishat, A. Kabir, S.F.U. Farhad, Synthesis, characterization and visible light-respon- sive photocatalysis properties of Ce doped CuO nanoparti- cles: a combined experimental and DFT?U study. Colloids Surf., A 617, 126386 (2021) 41. T. Jiang, M. Bujoli-Doeuff, Y. Farre´, Y. Pellegrin, E. Gautron, M. Boujtita, L. Cario, S. Jobic, F. Odobel, CuO nanomaterials for p-type dye-sensitized solar cells. RSC Adv. 6, 112765–112770 (2016) 54. Md. Rakibuddin, R. Ananthakrishnan, A novel Ag deposited nanocoordination polymer derived porous SnO2/NiO hetero- nanostructure for the enhanced photocatalytic reduction of Cr(VI) under visible light. New J. Chem. 40, 3385–3394 (2016) 42. J. Li, X. Zeng, A. Mo, Z. Xu, Fabrication of cuprate super- conducting La1.85Sr0.15CuO4 nanoﬁbers by electrospinning and subsequent calcination in oxygen. CrystEngComm 13, 6964–6967 (2011) 55. S. Velliyan, V. Rajendran, Study on the effect of Ce3? doping on structural, morphological and optical properties of CuO nanoparticles synthesized via combustion technique. Phys. B: Phys. Condens. Matter 613, 413015 (2021) 43. S. Manna, S.K. Funding De, Room temperature ferromagnetism in Fe doped CuO nanorods. J. Magn. Magn. Mater. 322, 2749–2753 (2010) 56. S.J. Singh, P. Chinnamuthu, Highly efﬁcient natural-sunlight- driven photodegradation of organic dyes with combustion 105 Page 18 of 19 J Mater Sci: Mater Electron (2023) 34:105 105 derived Ce-doped CuO nanoparticles. Colloids Surf., A 625, 126864 (2021) 68. A. Nakrela, N. Benramdane, A. Bouzidi, Z. Kebbab, M. Medles, C. Mathieu, Site location of Al-dopant in ZnO lattice by exploiting the structural and optical characterisation of ZnO: Al thin ﬁlms. Results Phys. 6, 133–138 (2016) 57. H. Siddiqui, M. Shrivastava, M.R. Parra, P. Pandey, S. Ayaz, M.S. Qureshi, The effect of La3? ion doping on the crystal- lographic, optical and electronic properties of CuO nanorods. Mater. Lett. 229, 225–228 (2018) 69. A. Zunger, S. Lany, H. Raebiger, The quest for dilute ferro- magnetism in semiconductors: guides and misguides by the- ory. Physics 3, 53 (2010) 58. A. Bandyopadhyay, N. Bhakta, S. Sutradhar, B.J. Sarkar, A.K. Deb, S. Kobayashi, K. Yoshimura, P.K. Chakrabarti, Microstructure investigation, optical properties and magnetic phase transition of Tm3? substituted nanocrystalline ZnO (Zn0.95Tm0.05O). RSC Adv. 6, 101818–101826 (2016) 70. K. Sato, H.K. Yoshida, Material design for transparent fer- romagnets with ZnO-based magnetic semiconductors. Jpn. J. Appl. Phys. 39, L555 (2000) 71. T. Story, P.J.T. Eggenkamp, C.H.W. Swu¨ste, H.J.M. Swagten, W.J.M. de Jonge, L.F. Lemmens, Ruderman-Kittel-Kasuya- Yosida exchange interaction in many-valley IV-VI semi- magnetic semiconductors. Phys. Rev. B 45, 1660 (1992) 59. H. Saadi, F.I.H. Rhouma, Z. Benzarti, Z. Bougrioua, S. Guermazi, K. Khirouni, Electrical conductivity improvement of Fe doped ZnO nanopowders. Mater. Res. Bull. 129, 110884 (2020) 72. T. Dietl, H. Ohno, F. Matsukura, J. Cibert, D. Ferrand, Zener model description of ferromagnetism in zinc-blende magnetic semiconductors. Science 287, 1019 (2000) 60. D. Savitha, H.K.E. Latha, H.S. Lalithamba, S. Mala, Y.V. Jeppu, Structural, optical and electrical properties of undoped and doped (Al, Al ? Mn) ZnO nanoparticles synthesised by green combustion method using terminalia catappa seed extract. Mater. Today: Proc. 60, 988–997 (2022) 73. H.T. Handal, H.A. Mousa, S.M. Yakout, W. Sharmoukh, V. Thangadurai, Effect of Mn and Ni-doping on structure, pho- toluminescence and magnetic properties of perovskite-type BaSn0.99Gd0.01O3. J. Magn. Magn. Mater. 498, 165946 (2020) 61. A.M. Youssef, S.M. Yakout, Colossal permittivity, electrical conductivity and ferromagnetic properties of pure TiO2: mono and binary doping. Materialia 21, 101277 (2022) 74. J.M.D. Coey, M. Venkatesan, C.B. Funding Fitzgerald, Donor impu- rity band exchange in dilute ferromagnetic oxides. Nat. Mater. 4, 173–179 (2005) 62. A.M. Youssef, S.M. Yakout, Colossal dielectric constant, electric modulus and electrical conductivity of nanocrystalline SnO2: role of Zr/Mn, Fe or Co dopants. J. Solid State Chem. 308, 122902 (2022) 75. E. Bellingeri, S. Rusponi, A. Lehnert, H. Brune, F. Nolting, A. Leveratto, A. Plaza, D. Marre´, Inﬂuence of free charge carrier density on the magnetic behavior of (Zn, Co)O thin ﬁlm studied by ﬁeld effect modulation of magnetotransport. Sci. Rep. 9, 149 (2019) 63. H.M. Ahsan, K. Lal, M. Saleem, G.M. Mustafa, M.A. Khan, A.S. Haidyrah, S. Atiq, Tuning the dielectric behavior and energy storage properties of Mn/Co co-doped ZnO. Mater. Sci. Semicond. Process. 134, 105977 (2021) 76. P. Joshi, S. Tiwari, K. Punia, S. Kumar, Defect mediated mechanism in greenly synthesized undoped, Al?3, Cu?2 and Zn?2 doped TiO2 nanoparticles for tailoring bandgap, luminescence, magnetic and electrical properties. Opt. Mater. 132, 112778 (2022) 64. M. Cos¸ kun, O. Polat, F.M. Cos¸ kun, Z. Durmus¸ , M. C¸ ag˘lar, A. Tu¨ru¨t, The electrical modulus and other dielectric properties by the impedance spectroscopy of LaCrO3 and LaCr0.90- Ir0.10O3 perovskites. RSC Adv. 8, 4634–4648 (2018) 65. N. Kanabar, K. Gadani, V.G. Shrimali, K. Sagapariya, K.N. Rathod, B. Udeshi, J. Joseph, D.D. Pandya, P.S. Solanki, N.A. Shah, Structural and electrical properties of sol–gel grown nanostructured ZnO and LaMnO3 particle-based nanocomposites. Appl. Phys. A 127, 122 (2021) 77. H. Bhoi, S. Tiwari, G. Lal, K.K. Jani, S.K. Modi, P. Seal, V. Saharan, K.B. Modi, J.P. Borah, K. Punia, S. Kumar, Green synthesis and characterization of Mg0.93Na0.07O nanoparticles for antimicrobial activity, cytotoxicity and magnetic hyper- thermia. Ceram. Int. 48, 28355–28373 (2022) 66. D. Gao, G. Yang, J. Li, J. Zhang, J. Zhang, D. Xue, Room- temperature ferromagnetism of ﬂowerlike CuO nanostruc- tures. J. Phys. Chem. C 114, 18347–18351 (2010) 78. K. Punia, G. Lal, S. Dalela, S.N. Dolia, P.A. Alvi, S.K. Barbar, K.B. Modi, S. Kumar, A comprehensive study on the impact of Gd substitution on structural, optical and magnetic properties of ZnO nanocrystals. J. Alloys Compd. 868, 159142 (2021) 67. M. Chandrasekar, M. Subash, S. Logambal, G. Udhayakumar, R. Uthrakumar, C. Inmozhi, W.A. Al-Onazi, A.M. Al- Mohaimeed, T. Chen, K. Kanimozhi, Synthesis and charac- terization studies of pure and Ni doped CuO nanoparticles by hydrothermal method. J. King Saud University Science 34, 101831 (2022) 79. K. Punia, G. Funding Lal, S.K. Barbar, S.N. Dolia, P.A. Alvi, S. Dalela, S. Kumar, Oxygen vacancies mediated cooperative magnetism in ZnO nanocrystals: a d0 ferromagnetic case study. Vacuum 184, 109921 (2021) Page 19 of 19 105 J Mater Sci: Mater Electron (2023) 34:105 105 Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. 80. J. Wang, J. Wan, K. Chen, Facile synthesis of superparam- agnetic Fe-doped ZnO nanoparticles in liquid polyols. Mater. Lett. 64, 2373–2375 (2010) 81. J. Mohapatra, D.K. Mishra, S.K. Singh, Superparamagnetic behavior in chemically synthesized nanocrystalline Zn0.99- Ni0.01O powders. Mater. Lett. 75, 91–94 (2012)
https://openalex.org/W3201848506	https://figshare.com/articles/journal_contribution/Developmental_role_of_PHD2_in_the_pathogenesis_of_pseudohypoxic_pheochromocytoma_/16895350/1/files/31243090.pdf	English	null	Developmental role of PHD2 in the pathogenesis of pseudohypoxic pheochromocytoma	Endocrine-related cancer	2,021	cc-by	10,831	Abstract Despite a general role for the HIF hydroxylase system in cellular oxygen sensing and tumour hypoxia, cancer-associated mutations of genes in this pathway, including PHD2, PHD1, EPAS1 (encoding HIF-2α) are highly tissue-restricted, being observed in pseudohypoxic pheochromocytoma and paraganglioma (PPGL) but rarely, if ever, n other tumours. In an effort to understand that paradox and gain insights into the pathogenesis of pseudohypoxic PPGL, we constructed mice in which the principal HIF prolyl hydroxylase, Phd2, is inactivated in the adrenal medulla using TH-restricted Cre recombinase. Investigation of these animals revealed a gene expression pattern closely mimicking that of pseudohypoxic PPGL. Spatially resolved analyses demonstrated a binary distribution of two contrasting patterns of gene expression among adrenal medullary cells. Phd2 inactivation resulted in a marked shift in this distribution towards a Pnmt−/Hif-2α+/Rgs5+ population. This was associated with morphological abnormalities of adrenal development, including ectopic TH+ cells within the adrenal cortex and external to the adrenal gland. These changes were ablated by combined inactivation of Phd2 with Hif-2α, but not Hif-1α. However, they could not be reproduced by inactivation of Phd2 in adult life, suggesting that they arise from dysregulation of this pathway during adrenal development. Together with the clinical observation that pseudohypoxic PPGL manifests remarkably high heritability, our findings suggest that this type of tumour likely arises from dysregulation of a tissue-restricted action of the PHD2/HIF-2α pathway affecting adrenal development in early life and provides a model for the study of the relevant processes. Endocrine-Related Cancer (2021) 28, 757–772 Key Words f f PHD f f HIF f f hypoxia f f adrenal medulla f f pheochromocytoma Despite a general role for the HIF hydroxylase system in cellular oxygen sensing and tumour hypoxia, cancer-associated mutations of genes in this pathway, including PHD2, PHD1, EPAS1 (encoding HIF-2α) are highly tissue-restricted, being observed in pseudohypoxic pheochromocytoma and paraganglioma (PPGL) but rarely, if ever, in other tumours. In an effort to understand that paradox and gain insights into the pathogenesis of pseudohypoxic PPGL, we constructed mice in which the principal HIF prolyl hydroxylase, Phd2, is inactivated in the adrenal medulla using TH-restricted Cre recombinase. Investigation of these animals revealed a gene expression pattern closely mimicking that of pseudohypoxic PPGL. Spatially resolved analyses demonstrated a binary distribution of two contrasting patterns of gene expression among adrenal medullary cells. Phd2 inactivation resulted in a marked shift in this distribution towards a Pnmt−/Hif-2α+/Rgs5+ population. orrespondence should be addressed to P J Ratcliffe or T Bishop: peter.ratcliffe@ndm.ox.ac.uk or tammie.bishop@ (L Eckardt and M Prange-Barczynska contributed equally to this work) 28:12 757–772 28:12 757–772 Endocrine-Related Cancer Endocrine-Related Cancer 757–772 The role of PHD2 in pseudohypoxic PCCs L Eckardt, M Prange- Barczynska et al. 28:12 Developmental role of PHD2 in the pathogenesis of pseudohypoxic pheochromocytoma Luise Eckardt1,2,, Maria Prange-Barczynska1,3,, Emma J Hodson4,5, James W Fielding1,3, Xiaotong Cheng1,3, Joanna D C C Lima1, Samvid Kurlekar1, Gillian Douglas6, Peter J Ratcliffe1,3,4 and Tammie Bishop 1 Luise Eckardt1,2,, Maria Prange-Barczynska1,3,, Emma J Hodson4,5, James W Fielding1,3, Xiaotong Cheng1,3, Joanna D C C Lima1, Samvid Kurlekar1, Gillian Douglas6, Peter J Ratcliffe1,3,4 and Tammie Bishop 1 , , 5The Department of Experimental Medicine and Immunotherapeutics, University of Cambridge, Cambridge, UKff 5The Department of Experimental Medicine and Immunotherapeutics, University of Cambridge, Cambridge, UK 6BHF Centre of Research Excellence Division of Cardiovascular Medicine Radcliffe Department of Medicine John R 5The Department of Experimental Medicine and Immunotherapeutics, University of Cambridge, Cambridge, UK 6BHF Centre of Research Excellence, Division of Cardiovascular Medicine, Radcliffe Department of Medicine, John Radcliffe p p p y g g BHF Centre of Research Excellence, Division of Cardiovascular Medicine, Radcliffe Department of Medicine, John Radcliffe H xford, UK Endocrine-Related Cancer (2021) 28, 757–772 Introduction Pheochromocytoma and paraganglioma (PPGL) are tumours of the autonomic paraganglia that arise in diverse anatomical locations from the skull base to the pelvis. Those found within the adrenal glands (AGs) are known as pheochromocytoma (PCC) and those in extra-adrenal structures including the carotid body are commonly termed paraganglioma (PGL). Molecular analysis of these tumours has revealed a number of subtypes or clusters, with distinct patterns of gene expression within the tumour being associated with different groups of tumour- associated mutations. Genetic profiling has revealed four such subtypes: kinase signalling, Wnt-altered, cortical admixture and pseudohypoxia (Crona et al. 2017, Fishbein & Wilkerson 2018). The pseudohypoxia subtype (or Cluster I) is associated with mutations affecting transcriptional pathways induced by hypoxia. However, there are a number of puzzling features in these associations. First, PPGLs in general manifest an unusually high ratio of inherited to sporadic forms. For instance, up to 40% of PPGLs are associated with germline or post-zygotic but very early somatic mutation, as assessed by family history or distribution of mutant cells (Buffet et al. 2020). Secondly, the spectrum of gene dysregulation in pseudohypoxic PPGLs does not align exactly with that of dynamically regulated HIF transcriptional targets (Dahia et al. 2005, Favier et al. 2009, 2012, Waldmann et al. 2010, Burnichon et al. 2011, Lorenzo et al. 2013, Toledo et al. 2013, Welander et al. 2014, Yang et al. 2015, Fishbein et al. 2017). Thirdly, human VHL mutations associated with PCCs have a complex relationship to dysregulation of HIF: type 1 VHL mutations, which show complete dysregulation of HIF, are not associated with pheochromocytoma whereas type 2A, B and C VHL mutations, which are associated with pheochromocytoma, show either less severe or no dysregulation of HIF, at least when assayed in vitro or in heterologous cell types (Kaelin 2008). Fourthly, Vhl inactivation in the adrenal medulla (AM) and carotid body of the mouse results in tissue atrophy rather than tumour formation (Macias et al. 2014). The transcriptional response to hypoxia is mediated by hypoxia-inducible factor (HIF) (reviewed in Bishop & Ratcliffe 2014), a heterodimer consisting of an oxygen- regulated α and a constitutively expressed β subunit. HIF-α is regulated by a series of 2-oxoglutarate-dependent dioxygenases that generate an oxygen-dependent signal by hydroxylation of specific prolyl residues in HIF-α subunits which are targeted for proteasomal degradation via the E3 ubiquitin ligase von Hippel-Lindau (VHL) protein. Endocrine-Related Cancer Endocrine-Related Cancer 758 2005, Favier et al. 2009, 2012, Waldmann et al. 2010, Burnichon et al. 2011, Lorenzo et al. 2013, Toledo et al. 2013, Welander et al. 2014, Yang et al. 2015, Fishbein et al. 2017). They are associated with loss-of-function mutations in VHL, PHD2 and PHD1 and gain-of-function mutations in HIF-2α (otherwise known as EPAS1) as well as mutations in genes encoding the tricarboxylic acid cycle enzymes succinate dehydrogenase (SDHB/D/C/A or SDHx) and fumarate hydratase (FH) (Dahia et al. 2005, Ladroue et al. 2008, Zhuang et al. 2012, Yang et al. 2015). Impaired function of the latter two enzymes leads to the accumulation of succinate and fumarate, respectively, which are able to inhibit 2-oxoglutarate-dependent dioxygenases including the HIF prolyl hydroxylases. All these mutations therefore have the potential to activate HIF, suggesting that in this setting it is HIF that provides the oncogenic drive. This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access This work is licensed under a Creative Commons Attribution 4.0 International License. Abstract This was associated with morphological abnormalities of adrenal development, including ectopic TH+ cells within the adrenal cortex and external to the adrenal gland. These changes were ablated by combined inactivation of Phd2 with Hif-2α, but not Hif-1α. However, they could not be reproduced by inactivation of Phd2 in adult life, suggesting that they arise from dysregulation of this pathway during adrenal development. Together with the clinical observation that pseudohypoxic PPGL manifests remarkably high heritability, our findings suggest that this type of tumour likely arises from dysregulation of a tissue-restricted action of the PHD2/HIF-2α pathway affecting adrenal development in early life and provides a model for the study of the relevant processes. Key Words f f PHD f f HIF f f hypoxia f f adrenal medulla f f pheochromocytoma Endocrine-Related Cancer (2021) 28, 757–772 This work is licensed under a Creative Commons Attribution 4.0 International License. Printed in Great Britain Published by Bioscientifica Ltd. © 2021 The authors https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com Downloaded from Bioscientifica.com at 10/28/2021 08 via fre Printed in Great Britain Published by Bioscientifica Ltd. © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. © 2021 The authors This work is licensed under a Creative Commons Attribution 4.0 International License. This work is licensed under a Creative Commons Attribution 4.0 International License. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Endocrine-Related Cancer Drug administration For adult-onset Phd2 inactivation, ~6 week-old mice were dosed once daily with 2 mg tamoxifen orally (20 mg/mL in corn oil containing 10% ethanol, Sigma) for 5 consecutive days. Phd2f/f;RosaCreER and Phd2f/f;THCreER mice were sacrificed 17 days or ~3 months, respectively, after the start of treatment. Immunohistochemistry Sections were immunostained for TH using an EnVision+ kit (Dako Denmark A/S) with a polyclonal rabbit anti-TH antibody (1:5000, NB300-109, Novus Biologicals, Cambridge, UK) (Bishop et al. 2013). In situ hybridisation mRNA was detected in sections using the manual RNAScope 2.5 HD BROWN assay or, for dual in situ hybridisation, the RNAScope 2.5 HD Duplex assay (Advanced Cell Diagnostics, Newark, US). RNAScope probes: Mm-Pnmt (426421 or 426421-C2 for dual RNAScope), Mm-Epas1 (314371), Mm-Rgs5 (430181), Mm-Vegfa-OI (43961). Imaging was performed with a Leica DM 1000 LED microscope (Leica Biosystems). Ethical approval and animals Blood was centrifuged at 800 g for 5 min and plasma was separated and stored at −80°C. Adrenaline and noradrenaline were detected in diluted plasma samples using the Epinephrine/Norepinephrine ELISA Kit (KA1877, Abnova, Taoyuan City, Taiwan). Signal absorbance was read at 450 nm using a FLUOstar Omega microplate reader (BMG Labtech, Aylesbury, UK). Catecholamine concentrations were calculated using a linear standard curve. Animal experimental protocols were approved by the University of Oxford Medical Science Division Ethical Review Committee and are compliant with the UK Home Office Animals (Scientific Procedures) Act 1986. Experiments were performed on ~3 month-old mice and sex-matched controls, unless stated otherwise. Mice were kept in individually ventilated cages with free access to water and food. Phd2f/f, Hif-1αf/f and Hif-2αf/f alleles are as described (Cramer et al. 2003, Gruber et al. 2007, Mazzone et al. 2009). Note that Phd2 is equivalent to Egln1 and Hif-2α is equivalent to Epas1; we have used the Phd2/Hif-2α terminology to simplify mechanistic interpretation for the reader. TH+ cell-specific inactivation was achieved using the constitutively expressed TH-IRES- Cre (THCre; Lindeberg et al. 2004) or inducible TH-IRES- CreER (THCreER; Rotolo et al. 2008) and ubiquitous inactivation using the inducible Rosa26CreERT2 (RosaCreER; Vooijs et al. 2001). Each mouse line was backcrossed with C57BL/6 for at least five generations. Proximity ligation in situ hybridisation (PLISH) Multiplex fluorescence in situ hybridisation was performed using PLISH (Nagendran et al. 2018). Sections were de-paraffinised, boiled in a 10 mM citrate buffer (pH 6.0) with 0.05% lithium dodecyl sulfate (Sigma) and processed in sealed hybridisation chambers (Grace Biolabs, Bend, US). Tissues were treated with Endocrine-Related Cancer Endocrine-Related Cancer 759 heparinised needles. AGs for RT-qPCR were dissected into ice-cold phosphate buffered saline (PBS, in mM: 137 NaCl, 2.7 KCl, 4 Na2H2PO4.7H2O, 1.5 KH2PO4) in diethyl pyrocarbonate-treated water (0.1%, v/v, Sigma). For histology, mice were perfused-fixed with 5 mL PBS followed by 5 mL 4% paraformaldehyde (PFA)/PBS (w/v) (Sigma). Dissected AGs were fixed in 4% PFA/PBS overnight, transferred into 70% (v/v) ethanol then dehydrated in an ascending ethanol series ending in histoclear (National Diagnostics, Atlanta, US), embedded in 60°C paraffin and sectioned to 4 μm thickness with a Microm HM 355S microtome (Thermo Fisher Scientific). gene expression associated with the presence or absence of phenylethanolamine N-methyltransferase (PNMT). Changes in gene expression were accompanied by morphological abnormalities including ectopic TH+ cells within the adrenal cortex and in peri-adrenal structures. These findings, together with marked differences between constitutive and inducible inactivation of Phd2 in adult life, suggest that the pathological activation of the PHD2/ HIF-2 pathway during adrenal development is critical for its tumourigenic action. Introduction Mammalian species have multiple HIF-α isoforms, of which HIF-1α and HIF-2α are the most abundant and best studied. They also express three closely related isoforms of the HIF prolyl hydroxylase enzymes (PHD1, PHD2 and PHD3, otherwise known as EGLN2, EGLN1 and EGLN3), of which PHD2 is the most abundant and important regulator of HIF. HIF is commonly activated in cancer and its role in oncogenesis has attracted widespread interest, particularly in view of the recent development of drugs with the potential to activate or inactivate components of the pathway therapeutically (reviewed in Semenza 2019, Choueiri & Kaelin 2020). Nevertheless, this relationship of HIF activation to oncogenesis has proved more complex than anticipated. For instance, although HIF is commonly upregulated in cancer, direct genetic activation by mutation of any of the key components of the pathway is rare in most forms of cancer. In an attempt to shed light on these paradoxical findings and better understand the role of activation of HIF pathways in the AM, we have examined the effects of inactivation of the principal HIF prolyl hydroxylase Phd2 in the AM, using Cre recombinase restricted by the tyrosine hydroxylase (TH) promoter, in the mouse. We report that TH-restricted constitutive inactivation of Phd2 in the AM results in a ‘pseudohypoxic pattern’ of gene expression in which dynamic activation of HIF transcription is superimposed on a developmental shift in populations of AM cells manifesting specific patterns of The pseudohypoxic subtype of PPGLs is an important exception. These tumours manifest a ‘pseudohypoxic’ pattern of gene expression encompassing upregulation of certain HIF target genes, together with alterations in the expression of differentiation markers (Dahia et al. This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Results 0.1 mg/mL Pepsin (Roche-10108057001, Sigma-Aldrich) in 0.1 M HCl, followed by (at 37°C): hybridisation of barcoded gene probes and bridge sequences, DNA ligation (10 CEU/mL T4 DNA ligase, M0202T, New England Biolabs, Ipswich, US) and extension by rolling circle amplification (1 U/mL Nxgen phi29 polymerase, Lucigen, Middleton, US). Samples were incubated with fluorophore-conjugated oligonucleotides specific to the barcode for the targeted gene probe (Supplementary Methods, see section on supplementary materials given at the end of this article), washed then treated with TruVIEW autofluoresence quenching kit (Vector Laboratories, Burlingame, US), stained with DAPI and mounted. Imaging was performed with a Zeiss Axio Imager M1 microscope (Jena, Germany) and analysed with Qupath software (Bankhead et al. 2017). Since pseudohypoxic PPGLs have a characteristic gene expression profile, we first sought to test whether and to what extent this was mimicked by Phd2 inactivation in the AM. To this end, we intercrossed mice bearing a conditionally inactivated (Phd2f/f) allele with a transgenic line expressing Cre recombinase restricted by the TH promoter (THCre) to generate Phd2f/f;THCre mice and measured effects on the expression of a panel of 14 genes that are frequently dysregulated in pseudohypoxic PPGLs (Supplementary Table 1). This comprised several classes of genes, including HIF target genes Vegfa, Slc2a1 and Ldha; atypical mitochondrial subunits Ndufa4l2 and Cox4i2; G-protein signalling pathway components Rgs4, Rgs5 and Adora2a; Pnmt, the terminal enzyme in catecholamine synthesis; genes with oncogenic potential including Stc1. These experiments revealed that many, though not all, genes identified as dysregulated in pseudohypoxic PPGLs were also dysregulated in a similar way in the AM of Phd2f/f;THCre mice (Fig. 1A), suggesting that their expression is directly or indirectly altered as a consequence of dysregulation of the PHD2/HIF system in this setting. Interestingly, not all these genes have been identified as being dynamically regulated by HIF itself in the manner observed. In particular, Pnmt has been reported to be upregulated by HIF-1 in rat pheochromocytoma PC12 cells (Tai et al. 2009), whereas we observed striking downregulation of Pnmt (mRNA and protein) with Phd2 inactivation/HIF activation (Fig. 1A and Supplementary Fig. 1). PNMT is the terminal enzyme in the catecholamine synthesis pathway which converts noradrenaline into adrenaline. In keeping with the loss of Pnmt, we observed a shift in plasma catecholamines: an increase in noradrenaline together with a reduction in adrenaline (Fig. 1B). Results This predominantly noradrenergic secretory profile, together with a loss of Pnmt expression, is similar to clinical observations in patients with pseudohypoxic PCCs (Eisenhofer et al. 2001). RT-qPCR AGs were collected from five mice per genotype and AMs sub-dissected under an SMZ-745 stereo microscope (Nikon) and stored in RNAProtect (Qiagen) on ice. Pooled tissues were homogenised in RLT+ buffer (Qiagen) using a ProScientific PRO200 Homogenizer (Cole-Parmer, Eaton Socon, UK). RNA was isolated using the RNeasy Plus Micro Kit (Qiagen), cDNA prepared using the QuantiTect RT Kit (Qiagen) and RT-qPCR performed with the TaqMan Fast Advanced Master Mix Kit (Thermo Fisher Scientific) in the StepOnePlus Real-Time PCR System (Applied Biosystems). Three technical replicates were used in each biological repeat with Actb serving as a reference gene (see Supplementary methods for TaqMan probes). Fold change in gene expression was reported as 2−ΔΔCT, where ΔΔCT = Phd2f/f;THCre (CTtarget – CTreference) – Phd2f/f (CTtarget – CTreference). We therefore sought to determine whether Phd2f/f; THCre mice have morphological features of PCC, in particular evidence of intra-adrenal pathology. However, no frank tumours, nodules, cortical compression or loss of nest-like/formation of sheet-like chromaffin cell clusters (Smith-Hicks et al. 2000, Park et al. 2015) were observed, and there was no change in overall AM volume or proliferation (Supplementary Table 2). Tissue collection Animals were killed by an overdose of isoflurane (Piramal Critical Care, West Drayton, UK) and exsanguination from the inferior vena cava. Blood was collected using https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Morphometric analysis AM volume was modelled by measuring the TH+ area of one in eight consecutive slides of the AG using ImageJ software (NIH) (Bishop et al. 2008, Fielding et al. 2018). In situ hybridisation signal was quantified using trainable Weka segmentation plugin in Fiji ImageJ 1.53c software (NIH) (Cheng et al. 2020). This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Figure 1f Figure 1 Effect of TH-restricted Phd2 inactivation on gene expression, plasma catecholamines and AM morphology. (A) Gene expression profile of AMs in Phd2f/f;THCre mice (relative to Phd2f/f controls), analysed by RT-qPCR on RNA extracted from ten AMs per biological replicate. Data for individual genes were analysed by unpaired two-tailed Student’s t-tests. Gene expression pattern in Phd2f/f;THCre AMs resembles that of pseudohypoxic PPGLs. (B) Plasma catecholamine levels in Phd2f/f;THCre (vs Phd2f/f control) mice. Data were analysed by a two-way ANOVA, interaction (Phd2 inactivation vs noradrenaline/ adrenaline ratio) P < 0.0001 followed by Sidak’s multiple comparisons test. Graph shows a reduction in adrenaline and an increase in noradrenaline in the plasma of Phd2f/f;THCre mice. (C) Representative images of Phd2f/f;THCre (compared with Phd2f/f control) AGs. The AM is encircled with a black dashed line in this and other figures. TH antibody (brown); Harris haematoxylin counterstain (blue). Enlarged images show ectopic TH+ cells in the adrenal cortex (red frame) and in a peri-adrenal structure (black frame). Scale bars: 0.5 mm (far and middle left panels); 0.1 mm (red frame); 0.05 mm (black frame). Data show mean ± s.e.m. P < 0.05, ***P < 0.0001. adrenal cortex that disrupt the outer cortical boundary, adjacent to a TH+ extra-adrenal ganglion-like structure (Fig. 1C). Altogether, 11 out of 14 Phd2f/f;THCre mice and 0 out of 15 littermate controls manifest these abnormalities. AG is known to express Pnmt in a restricted set of mature chromaffin cells that produce adrenaline (Coupland & Hopwood 1966). Analysis of the spatial distribution of Pnmt expression in WT mice confirmed this, with the majority (~75%) of chromaffin cells expressing Pnmt (Fig. 2A). We next considered whether this expression pattern extended to other genes which were dysregulated in AMs with loss of Phd2. The following genes were selected for analysis: Hif-2α, since its upregulation is characteristic of VHL-associated neoplasia including PCCs (Toledo 2017); Vegfa and Rgs5, since these are reported HIF target genes (Jin et al. 2009, Fielding et al. 2018) and the latter is also a proposed regulator of chromaffin cell differentiation (Chan et al. 2019, Hanemaaijer et al. 2021). These experiments revealed a striking inverse pattern of gene expression, with Hif-2α and Rgs5 mRNA being expressed TH+ chromaffin cells migrate through the adrenal cortex to reach their final destination in the AM (Furlan et al. 2017, Hanemaaijer et al. Statistical analysis Data are shown as mean ± s.e.m. Statistical analyses were performed using unpaired Student’s t-tests, unless otherwise stated, and using GraphPad Prism 9.0 Software. However, we noted striking abnormalities of adrenal morphology, in particular the abnormal location of TH+ cells. These comprised clusters of ectopic cells in the https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. 761 L Eckardt, M Prange- Barczynska et al. The role of PHD2 in pseudohypoxic PCCs 28:12 Endocrine-Related Cancer Figure 1 Effect of TH-restricted Phd2 inactivation on gene expression, plasma catecholamines and AM morphology. (A) Gene expression profile of AMs in Phd2f/f;THCre mice (relative to Phd2f/f controls), analysed by RT-qPCR on RNA extracted from ten AMs per biological replicate. Data for individual genes were analysed by unpaired two-tailed Student’s t-tests. Gene expression pattern in Phd2f/f;THCre AMs resembles that of pseudohypoxic PPGLs. (B) Plasma catecholamine levels in Phd2f/f;THCre (vs Phd2f/f control) mice. Data were analysed by a two-way ANOVA, interaction (Phd2 inactivation vs noradrenaline/ adrenaline ratio) P < 0.0001 followed by Sidak’s multiple comparisons test. Graph shows a reduction in adrenaline and an increase in noradrenaline in the plasma of Phd2f/f;THCre mice. (C) Representative images of Phd2f/f;THCre (compared with Phd2f/f control) AGs. The AM is encircled with a black dashed line in this and other figures. TH antibody (brown); Harris haematoxylin counterstain (blue). Enlarged images show ectopic TH+ cells in the adrenal cortex (red frame) and in a peri-adrenal structure (black frame). Scale bars: 0.5 mm (far and middle left panels); 0.1 mm (red frame); 0.05 mm (black frame). Data show mean ± s.e.m. P < 0.05, ***P < 0.0001. 761 L Eckardt, M Prange- Barczynska et al. The role of PHD2 in pseudohypoxic PCCs 28:12 Endocrine-Related Cancer Endocrine-Related Cancer This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Figure 1f 2021), during which time they acquire Pnmt expression in the final stages of maturation to become adrenergic (Verhofstad et al. 1979). We therefore considered whether the abnormally located TH+ cells (both within and directly adjacent to the AG) represent abnormal migration of chromaffin cells and that the reduction in Pnmt might also reflect failure to acquire Pnmt due to dysregulated/ arrested development with Phd2 inactivation. To address this, we proceeded to analyse the spatial distribution of gene expression within the AM. The normal https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. 762 L Eckardt, M Prange- Barczynska et al. The role of PHD2 in pseudohypoxic PCCs 28:12 Endocrine-Related Cancer 762 L Eckardt, M Prange- Barczynska et al. The role of PHD2 in pseudohypoxic PCCs 28:12 Endocrine-Related Cancer Cancer gure 2f ect of Phd2 inactivation on the spatial expression of genes in the AM. (A) In situ hybridisation for Pnmt, Hif-2α, Rgs5 and Vegfa mRNA (brown) in adj ctions of a WT AG showing Hif-2α and Rgs5 mRNA expression in Pnmt− cells. Gill’s haematoxylin counterstain (grey-blue). Scale bars: 0.2 mm. (B) mparison of Pnmt, Hif-2α, Rgs5 and Vegfa mRNA expression in Phd2f/f vs Phd2f/f;THCre AMs. Red dashed lines outline Pnmt−/Hif-2α+/Rgs5+ cell popul this and other figures. Images show a switch from predominantly Pnmt+/Hif-2α−/Rgs5− to Pnmt−/Hif-2α+/Rgs5+ cells following Phd2 inactivation. Har ematoxylin counterstain (blue). Scale bars: 0.1 mm. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com Figure 2f Figure 2 g Effect of Phd2 inactivation on the spatial expression of genes in the AM. (A) In situ hybridisation for Pnmt, Hif-2α, Rgs5 and Vegfa mRNA (brown) in adjacent sections of a WT AG showing Hif-2α and Rgs5 mRNA expression in Pnmt− cells. Gill’s haematoxylin counterstain (grey-blue). Scale bars: 0.2 mm. (B) Comparison of Pnmt, Hif-2α, Rgs5 and Vegfa mRNA expression in Phd2f/f vs Phd2f/f;THCre AMs. Red dashed lines outline Pnmt−/Hif-2α+/Rgs5+ cell populations in this and other figures. Images show a switch from predominantly Pnmt+/Hif-2α−/Rgs5− to Pnmt−/Hif-2α+/Rgs5+ cells following Phd2 inactivation. Harris haematoxylin counterstain (blue). Scale bars: 0.1 mm. © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. This work is licensed under a Creative Commons Attribution 4.0 International License. This work is licensed under a Creative Commons Attribution 4.0 International License. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Endocrine-Related Cancer Endocrine-Related Cancer Pnmt expression in cells inversely correlates with Hif-2α and Rgs5 in both genotypes but Pnmt+/Hif-2α−/Rgs5− cell populations are dominant in Phd2f/f whereas this switches to Pnmt−/Hif-2α+/Rgs5+ cells in Phd2f/f;THCre AMs, which additionally show induction of Vegfa. Endocrine-Related Cancer Endocrine-Related Cancer 763 a major switch of cell populations in AMs following Phd2 inactivation. The same pattern of Pnmt−/Hif-2α+/Rgs5+ was maintained, but cells manifesting this pattern now became the dominating cell population distributed across most of the AM with the exception of small areas around the periphery (Fig. 2B). Within the dominating population of Pnmt−/Hif-2α+/Rgs5+ expressing cells, there was also a mild induction of the HIF target gene Vegfa (Fig. 2B). This inverse expression of Pnmt and Hif-2α/Rgs5/Vegfa in Phd2f/f and Phd2f/f;THCre AMs was confirmed by dual in situ hybridisation and PLISH, which allow multiplexing of different mRNA probes to measure overlapping patterns of expression (Fig. 3 and Supplementary Fig. 2). only in the normal AM cells that do not express Pnmt (Fig. 2A). Vegfa, on the other hand, could not be detected, except in the adrenal cortex where it was strongly expressed (Fig. 2A). This suggested the possibility that a binary, cell- specific pattern of expression exists within AM cells, with low Pnmt associating with high Hif-2α and upregulation of at least some of its transcriptional targets such as Rgs5. The findings therefore raised the interesting question as to whether inactivation of Phd2 might affect the expression of these genes (Pnmt, Hif-2α, Rgs5 and Vegfa) within a specific population of cells or whether the prevalence of the two populations manifesting the observed patterns of gene expression might change after Phd2 inactivation. To address this, we performed further in situ mRNA hybridisation studies. These studies revealed We next went on to examine the ectopic TH+ cells within the adrenal cortex. Interestingly, essentially Figure 3 Spatial co-localisation of genes in the AM. (A) Dual in situ hybridisation or (B) proximity ligation in situ hybridisation (PLISH) for Pnmt, Hif-2α, Rgs5 and Vegfa mRNA in Phd2f/f and Phd2f/f;THCre AMs. Pnmt expression in cells inversely correlates with Hif-2α and Rgs5 in both genotypes but Pnmt+/Hif-2α−/Rgs5− cell populations are dominant in Phd2f/f whereas this switches to Pnmt−/Hif-2α+/Rgs5+ cells in Phd2f/f;THCre AMs, which additionally show induction of Vegfa. For dual in situ hybridisation, Gill’s haematoxylin counterstain (grey-blue). Scale bars for dual in situ hybridisation: 0.1 mm, for PLISH: 0.05 mm. Figure 3 Spatial co-localisation of genes in the AM. (A) Dual in situ hybridisation or (B) proximity ligation in situ hybridisation (PLISH) for Pnmt, Hif-2α, Rgs5 and Vegfa mRNA in Phd2f/f and Phd2f/f;THCre AMs. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com Endocrine-Related Cancer Endocrine-Related Cancer L Eckardt, M Prange- Barczynska et al. 764 all these cells were Pnmt−/Hif-2α+/Rgs5+/Vegfa+ (Fig. 4) (and negative for the adrenal cortical cell marker Cyp11a1, Supplementary Fig. 3), again suggestive of arrested migration of an immature Pnmt− population of chromaffin cells within the adrenal cortex during development. Taken together, these experiments suggest that TH-restricted Phd2 inactivation results in a pattern of gene expression similar to that of pseudohypoxic PPGLs. Morphological abnormalities suggestive of an effect on adrenal development were coupled to a major switch in an apparently binary pattern of gene expression observed in populations of cells within the AM. were studied somewhat earlier, approximately 17 days after tamoxifen dosing. In striking contrast to Phd2f/f;THCre mice, inactivation of Phd2 in adult mice by either protocol did not result in morphological abnormalities or a change in the spatial distribution of Pnmt− and Pnmt+ cells (Fig. 6). Additionally, no change in AM volume or proliferation was noted (Supplementary Table 2). In contrast, induction of Rgs5 and Vegfa mRNA was observed in both models of adult- onset Phd2 inactivation and was apparently confined to the population of cells that did not express Pnmt and were Hif-2α+ (Figs 6, 7 and Supplementary Fig. 4). Together, this suggests that two distinct effects of HIF-2 activation contribute to the pseudohypoxic phenotype observed with TH-restricted Phd2 inactivation: first, as a regulator of chromaffin cell differentiation during development; secondly, as a dynamic regulator of HIF target gene expression within Hif-2α expressing cells. To further understand this process, we intercrossed animals to generate Phd2f/f;Hif-1αf/f;THCre and Phd2f/f; Hif-2αf/f;THCre mice, which were examined with respect to the above phenotypes. No reversion of the morphological phenotype was observed with concomitant Hif-1α inactivation (Fig. 5). In striking contrast, Hif-2α inactivation (Phd2f/f;Hif-2αf/f;THCre mice) completely reversed all the morphological abnormalities associated with Phd2 inactivation, such that AMs were similar to those of control (Phd2f/f) mice (Fig. 5). Similar results were obtained with analysis of Pnmt, Rgs5 and Vegfa gene expression. The inverse expression pattern of Pnmt and Rgs5 was invariant, but the proportion of cells of each type was strikingly different, with Phd2f/f;Hif-1αf/f;THCre mice retaining a dominant population of Pnmt−/Rgs5+ cells while Phd2f/f;Hif-2αf/f;THCre mice apparently reverted to a phenotype indistinguishable from controls, including loss of Vegfa mRNA (Fig. 5). Together, this indicates that Hif-2α, not Hif-1α, is necessary for the abnormal phenotype resembling pseudohypoxic PCCs. Figure 3 Figure 3 Spatial co-localisation of genes in the AM. (A) Dual in situ hybridisation or (B) proximity ligation in situ hybridisation (PLISH) for Pnmt, Hif-2α, Rgs5 and Vegfa mRNA in Phd2f/f and Phd2f/f;THCre AMs. Pnmt expression in cells inversely correlates with Hif-2α and Rgs5 in both genotypes but Pnmt+/Hif-2α−/Rgs5− cell populations are dominant in Phd2f/f whereas this switches to Pnmt−/Hif-2α+/Rgs5+ cells in Phd2f/f;THCre AMs, which additionally show induction of Vegfa. For dual in situ hybridisation, Gill’s haematoxylin counterstain (grey-blue). Scale bars for dual in situ hybridisation: 0.1 mm, for PLISH: 0.05 mm. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free acces © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. This work is licensed under a Creative Commons Attribution 4.0 International License. This work is licensed under a Creative Commons Attribution 4.0 International License. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Endocrine-Related Cancer This work is licensed under a Creative Commons Attribution 4.0 International License. This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Discussion 765 L Eckardt, M Prange- Barczynska et al. The role of PHD2 in pseudohypoxic PCCs 28:12 Endocrine-Related Cancer Endocrine-Related Cancer Endocrine-Related C 765 Barczynska et al. pseudohypoxic PCCs r sion profile of the ectopic TH+ cell tracks in adrenal glands from Phd2f/f;THCre mice. Representative images of ectopic TH+ cells wo different adrenal glands (left vs middle and right hand columns) from Phd2f/f;THCre mice showing TH protein and Pnmt, Hif- TH+ cells in these ectopic tracks are essentially all Pnmt−/Hif-2α+/Rgs5+/Vegfa+. Harris haematoxylin counterstain (blue). Scale ba 0.5 mm (middle column), 0.2 mm (right column). g/10 1530/ERC 21 0211 oscientifica.com © 2021 The authors Published by Bioscientifica Ltd This work is licensed under a Creativ Attribution 4 0 International License Discussion Our findings demonstrate that TH-restricted inactivation of Phd2 results in a pattern of gene expression within the AM that resembles pseudohypoxic PPGLs. Importantly, several lines of evidence reveal that in addition to dynamic activation of the HIF transcriptional response, this alteration in gene expression reflects developmental consequences of Phd2 inactivation on the AM. First, the alteration in gene expression involves spatial changes in cell-specific patterns of gene expression that reflect lack of terminal differentiation to Pnmt+ cells. Secondly, several of the genes involved in the altered pattern of expression, including Pnmt and Hif-2α itself, are not dynamic HIF transcriptional targets. Thirdly, altered patterns of gene expression were associated with morphological abnormalities, including ectopic TH+ cell populations with a Pnmt− pattern of gene expression. Finally, neither the spatial change in gene expression within the AM nor the morphological abnormalities in the position of TH+ cells could be induced by Phd2 inactivation in adult life. We therefore hypothesised that inactivation of Phd2 might have at least two distinct effects in order to generate this pseudohypoxic phenotype: first, a dynamic induction of HIF transcriptional target genes including Vegfa; secondly, a switch to an immature noradrenergic cellular phenotype within the AM. Since we also observed morphological abnormalities suggestive of interrupted differentiation and/or migration, we sought to determine whether these components of the pseudohypoxic phenotype might reflect an action of the PHD2/HIF-2 axis during differentiation. To test this, we proceeded to compare the effects with those in animals where inactivation of Phd2 was restricted to adult life, using two different models. We analysed adult Phd2f/f;THCreER at 3 months after inducing recombination with tamoxifen. To assess any effect of more extensive recombination, adult Phd2f/f;RosaCreER mice were also studied. Since these develop systemic abnormalities (Hodson et al. 2016), they These findings are of particular interest when considered alongside several unusual observations on the clinical genetics of PPGL. Mutations that directly affect components of the PHD-HIF system are frequently observed in the uncommon syndrome of pseudohypoxic PPGL, but rarely, if ever, seen in other much more common forms of cancer. This is surprising since dynamic regulation of the transcriptional response This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com Figure 4 Figure 4 Gene expression profile of the ectopic TH+ cell tracks in adrenal glands from Phd2f/f;THCre mice. Representative images of ectopic TH+ cells in the adrenal cortex from two different adrenal glands (left vs middle and right hand columns) from Phd2f/f;THCre mice showing TH protein and Pnmt, Hif-2α, Rgs5 and Vegfa mRNA. TH+ cells in these ectopic tracks are essentially all Pnmt−/Hif-2α+/Rgs5+/Vegfa+. Harris haematoxylin counterstain (blue). Scale bars: 0.05 mm (left column), 0.5 mm (middle column), 0.2 mm (right column). Figure 4 Gene expression profile of the ectopic TH+ cell tracks in adrenal glands from Phd2f/f;THCre mice. Representative images of ectopic TH+ cells in the adrenal cortex from two different adrenal glands (left vs middle and right hand columns) from Phd2f/f;THCre mice showing TH protein and Pnmt, Hif-2α, Rgs5 and Vegfa mRNA. TH+ cells in these ectopic tracks are essentially all Pnmt−/Hif-2α+/Rgs5+/Vegfa+. Harris haematoxylin counterstain (blue). Scale bars: 0.05 mm (left column), 0.5 mm (middle column), 0.2 mm (right column). © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. This work is licensed under a Creative Commons Attribution 4.0 International License. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access 766 L Eckardt, M Prange- Barczynska et al. The role of PHD2 in pseudohypoxic PCCs 28:12 Endocrine-Related Cancer Figure 5 Effect of combined Hif-α and Phd2 inactivation on morphological abnormalities and gene expression in the AM. Representative images of TH immunohistochemistry and in situ hybridisation for Pnmt, Rgs5 and Vegfa mRNA in Phd2f/f, Phd2f/f;THCre, Phd2f/f;Hif-1αf/f;THCre and Phd2f/f;Hif-2αf/f;THCre AMs. Concomitant inactivation of Hif-2α and Phd2 (but not Hif-1α and Phd2) reverses the morphological and gene expression changes observed in Phd2f/f;THCre mice such that these AMs resemble those of control (Phd2f/f) mice. Harris haematoxylin counterstain (blue). Scale bars: 0.5 mm (top row), 0 1 mm (lower panels) Endocrine-Related Cancer Figure 5f Effect of combined Hif-α and Phd2 inactivation on morphological abnormalities and gene expression in the AM. Representative images of TH immunohistochemistry and in situ hybridisation for Pnmt, Rgs5 and Vegfa mRNA in Phd2f/f, Phd2f/f;THCre, Phd2f/f;Hif-1αf/f;THCre and Phd2f/f;Hif-2αf/f;THCre AMs. Concomitant inactivation of Hif-2α and Phd2 (but not Hif-1α and Phd2) reverses the morphological and gene expression changes observed in Phd2f/f;THCre mice such that these AMs resemble those of control (Phd2f/f) mice. Harris haematoxylin counterstain (blue). Scale bars: 0.5 mm (top row), 0.1 mm (lower panels). This work is licensed under a Creative Commons Attribution 4.0 International License. Figure 4 Our findings suggest a mechanism by which these apparently paradoxical findings could be explained. Specifically, the observation that dysregulation of the PHD- HIF system has a tissue-specific action on AM development suggests that these tumours have origins in early life, with to hypoxia by the PHD-HIF system is a general function observed in all cells. In addition, pseudohypoxic PPGL is associated with a very much higher incidence of germline (or post-zygotic but early somatic) mutations vs sporadic mutations, as compared to other forms of neoplasia. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access 767 L Eckardt, M Prange- Barczynska et al. The role of PHD2 in pseudohypoxic PCCs 28:12 Endocrine-Related Cancer Endocrine-Related Cancer g Barczynska et al. pseudohypoxic PCCs Cancer Figure 6 Effect of adult-onset Phd2 inactivation on morphology and gene expression in the AM. (A) Representative images of TH protein and Pnmt and Rgs5 mRNA in adrenal glands from Phd2f/f;THCreER, Phd2f/f;RosaCreER (and their respective Phd2f/f controls) killed ~3 months or 17 days post tamoxifen treatment, respectively. Harris haematoxylin counterstain (blue). Scale bars: 0.5 mm (top row) or 0.2 mm (bottom two rows). (B) Quantification of the percentage of Pnmt+ and Rgs5+ AM area; AMs from Phd2f/f;THCre, Phd2f/f;THCreER and Phd2f/f;RosaCreER mice (filled bars and rhombi) and their respective controls (open bars and rhombi). Bars show mean ± s.e.m. Data within individual genotype groups were compared by unpaired two-tailed Student’s t-tests: P < 0.05, P < 0.01, *P < 0.0001. Adult-onset Phd2 inactivation did not phenocopy morphological abnormalities or Pnmt loss in the AM observed with early-onset Phd2 inactivation, although a small but significant induction in Rgs5+ was noted within the Pnmt− cells of Phd2f/f;THCreER AMs. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:2 via free Figure 6 Figure 6f g Effect of adult-onset Phd2 inactivation on morphology and gene expression in the AM. (A) Representative images of TH protein and Pnmt and Rgs5 mRNA in adrenal glands from Phd2f/f;THCreER, Phd2f/f;RosaCreER (and their respective Phd2f/f controls) killed ~3 months or 17 days post tamoxifen treatment, respectively. Harris haematoxylin counterstain (blue). Scale bars: 0.5 mm (top row) or 0.2 mm (bottom two rows). (B) Quantification of the percentage of Pnmt+ and Rgs5+ AM area; AMs from Phd2f/f;THCre, Phd2f/f;THCreER and Phd2f/f;RosaCreER mice (filled bars and rhombi) and their respective controls (open bars and rhombi). Bars show mean ± s.e.m. Data within individual genotype groups were compared by unpaired two-tailed Student’s t-tests: P < 0.05, P < 0.01, **P < 0.0001. Adult-onset Phd2 inactivation did not phenocopy morphological abnormalities or Pnmt loss in the AM observed with early-onset Phd2 inactivation, although a small but significant induction in Rgs5+ was noted within the Pnmt− cells of Phd2f/f;THCreER AMs. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. This work is licensed under a Creative Commons Attribution 4.0 International License. © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Endocrine-Related Cancer 768 L Eckardt, M Prange- Barczynska et al. The role of PHD2 in pseudohypoxic PCCs 28:12 Endocrine-Related Cancer Barczynska et al. pseudohypoxic PCCs Cancer Figure 7 Spatial distribution of gene expression in the AM with constitutive or adult-onset Phd2 inactivation. Representative images of in situ hybridisation for Pnmt, Hif-2α, Rgs5 and Vegfa mRNA in Phd2f/f;THCre (constitutive) or Phd2f/f;THCreER (adult-onset) mice and their respective controls. Pnmt expression in cells inversely correlates with Hif-2α and Rgs5 across all genotypes but the Pnmt+/Hif-2α−/Rgs5− cell population is dominant in Phd2f/f controls whereas this switches to Pnmt−/Hif-2α+/Rgs5+ cells in Phd2f/f;THCre AMs. This contrasts to Phd2f/f;THCreER AMs, where Pnmt+/Hif-2α−/Rgs5− cells remain the dominant population; additionally, Rgs5 and Vegfa are induced within the minority population of Pnmt−/Hif-2α+/Rgs5+ cells compared to control mice. Harris haematoxylin counterstain (blue). Scale bars: 0.05 mm. Endocrine-Related Cancer Endocrine-Related Cancer 769 HIF-1α, are associated with human PPGL (Buffet et al. 2020). The finding of extra-adrenal tissue reported in our study might also be relevant to the not infrequent occurrence of chromaffin cell tumours at extra-adrenal sites (Tischler 2008). Although a correlation between the distribution of chromaffin tissue and paraganglioma has been reported (Coupland 1965), our data raises the possibility that these extra-adrenal PGLs may arise due to impaired migration and differentiation of sympathetic precursors that would normally populate the adrenal primordia to acquire features of mature, adrenergic chromaffin cells during development (Furlan et al. 2017, Hanemaaijer et al. 2021). Díaz-Castro et al. 2012, Macias et al. 2014, Lepoutre- Lussey et al. 2016, Al Khazal et al. 2021), to assess whether a switch from Pnmt+/Hif-2α−/Rgs5− to Pnmt−/ Hif-2α+/Rgs5+ cell populations also occurs in these settings. At the clinical level, our findings suggest that attempts at prevention or treatment should rationally include a focus on early life. Interestingly, patients with congenital cyanotic heart disease (a condition associated with life-long hypoxaemia beginning perinatally) have been reported to be susceptible to PCCs, many of which harbour sporadic HIF-2α mutations (Opotowsky et al. 2015, Vaidya et al. 2018); our data suggests that it is the early-onset hypoxia in these patients which predisposes to subsequent PCC formation. HIF-1α, are associated with human PPGL (Buffet et al. 2020). The finding of extra-adrenal tissue reported in our study might also be relevant to the not infrequent occurrence of chromaffin cell tumours at extra-adrenal sites (Tischler 2008). Although a correlation between the distribution of chromaffin tissue and paraganglioma has been reported (Coupland 1965), our data raises the possibility that these extra-adrenal PGLs may arise due to impaired migration and differentiation of sympathetic precursors that would normally populate the adrenal primordia to acquire features of mature, adrenergic chromaffin cells during development (Furlan et al. 2017, Hanemaaijer et al. 2021). The precise action of HIF-2 in promoting abnormal adrenal development will require further investigation, including the dissection of effects on differentiation, migration and interaction with other processes including innervation (Vollmer 1996). Notably, inactivation of Hif-2α in the setting of Phd2 inactivation (Supplementary Fig. 5) did not obliterate Pnmt−/Rgs5+ cells in Phd2f/f;Hif-2αf/f; THCre mice. Furthermore, in the normal AM, Pnmt− cells that show increased Hif-2α and Rgs5 mRNA levels did not express detectable levels of HIF-2α protein (data not shown). Endocrine-Related Cancer Thus, in this cell population, HIF-2 does not appear necessary to generate a Pnmt−/Hif-2α+/Rgs5+ or noradrenergic phenotype; nor is HIF-2 necessary for PNMT acquisition in adrenergic cells, as evidenced from the normal AMs in Hif-2αf/f;THCre mice (Macias et al. 2018) as well as in Phd2f/f;Hif-2αf/f;THCre mice reported here. Rather, excess stabilisation of HIF-2α acts in some way to interrupt a developmental programme that ordinarily generates Pnmt+ in other AM cells. Interestingly, the HIF-2-dependent effects of Phd2 inactivation, including paraganglioma-like carotid bodies (Fielding et al. 2018), are strikingly different from those of Vhl inactivation in catecholaminergic tissues using the same THCre promoter as in this study, which results in atrophy of multiple organs of the sympathetic nervous system, including the AM and the carotid body (Macias et al. 2014). There is a complex association between mutations in VHL disease and the tumour phenotypes, with type 2 VHL mutations that result in PCCs having only modest (or minimal) effects on HIF dysregulation, while type 1 mutations do not develop PCCs and result in greater HIF stabilisation (Kaelin 2008). Although levels of HIF activation have not been compared directly, our findings support the hypothesis that more moderate HIF activation associated with inactivation of a single PHD (as opposed to another function of VHL distinct from an action on HIF) is the most likely explanation for this paradox. Although Phd2 inactivation resulted in changes characteristic of pseudohypoxic PCCs, in this model we did not detect frank PCCs in ~3 month-old mice. Since PCC development might have a longer latency, we also analysed older cohorts of animals but again did not observe PCC development in n = 7 Phd2f/f;THCre and littermate control mice analysed aged ~18 months. However, we did observe three TH+, chromogranin A+ nodules of chromaffin cells amongst a group of n = 11 Phd2f/f;Hif-1αf/f;THCre mice aged in parallel (Supplementary Fig. 6). The significance of this is unclear, but HIF-1α has been reported to act as a tumour suppressor in VHL-associated renal clear cell carcinoma (Shen et al. 2011), and it may be that HIF-1 also has a tumour suppressive role in this context. Several other aspects of the dysregulated gene expression pattern merit comment. In particular, several genes that were upregulated in Phd2-inactivated AMs (including Rgs5, Cox4i2 and Adora2a) are very highly and specifically expressed in normal carotid body type I cells (Zhou et al. Figure 7 g Spatial distribution of gene expression in the AM with constitutive or adult-onset Phd2 inactivation. Representative images of in situ hybridisation for Pnmt, Hif-2α, Rgs5 and Vegfa mRNA in Phd2f/f;THCre (constitutive) or Phd2f/f;THCreER (adult-onset) mice and their respective controls. Pnmt expression in cells inversely correlates with Hif-2α and Rgs5 across all genotypes but the Pnmt+/Hif-2α−/Rgs5− cell population is dominant in Phd2f/f controls whereas this switches to Pnmt−/Hif-2α+/Rgs5+ cells in Phd2f/f;THCre AMs. This contrasts to Phd2f/f;THCreER AMs, where Pnmt+/Hif-2α−/Rgs5− cells remain the dominant population; additionally, Rgs5 and Vegfa are induced within the minority population of Pnmt−/Hif-2α+/Rgs5+ cells compared to control mice. Harris haematoxylin counterstain (blue). Scale bars: 0.05 mm. Bishop et al. 2008). Consistent with the relevance of our findings to the human syndrome, we found that changes in adrenal morphology and gene expression were ablated by inactivation of Hif-2α, but not Hif-1α, in line with the observation that activating mutations in HIF-2α, but not PHD2-dependent cell differentiation changes predisposing to subsequent tumourigenesis. Interestingly, inactivation of the related HIF prolyl hydroxylase isoform PHD3 prevents developmental culling of neurons, and this has also been proposed to pre-dispose to tumours (Lee et al. 2005, This work is licensed under a Creative Commons Attribution 4.0 International License. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access This work is licensed under a Creative Commons Attribution 4.0 International License. doi.org/10.1016/s0092-8674(03)00154-5) The authors thank Douglas Dos Santos Passos and Jade Harris for their help in various forms. Crona J, Taïeb D & Pacak K 2017 New perspectives on pheochromocytoma and paraganglioma: toward a molecular classification. Endocrine Reviews 38 489–515. (https://doi.org/10.1210/er.2017-00062) Dahia PL, Ross KN, Wright ME, Hayashida CY, Santagata S, Barontini M, Kung AL, Sanso G, Powers JF, Tischler AS, et al. 2005 A HIF1alpha regulatory loop links hypoxia and mitochondrial signals in pheochromocytomas. PLoS Genetics 1 72–80. (https://doi.org/10.1371/ journal.pgen.0010008) Endocrine-Related Cancer The role of PHD2 in pseudohypoxic PCCs 770 Bishop T, Gallagher D, Pascual A, Lygate CA, De Bono JP, Nicholls LG, Ortega-Saenz P, Oster H, Wijeyekoon B, Sutherland AI, et al. 2008 Abnormal sympathoadrenal development and systemic hypotension in PHD3-/- mice. Molecular and Cellular Biology 28 3386–3400. (https:// doi.org/10.1128/MCB.02041-07) This is linked to the online version of the paper at https://doi.org/10.1530/ ERC-21-0211. Bishop T, Talbot NP, Turner PJ, Nicholls LG, Pascual A, Hodson EJ, Douglas G, Fielding JW, Smith TG, Demetriades M, et al. 2013 Carotid body hyperplasia and enhanced ventilatory responses to hypoxia in mice with heterozygous deficiency of PHD2. Journal of Physiology 591 3565–3577. (https://doi.org/10.1113/jphysiol.2012.247254) ps://doi.org/10.1172/JCI133194) Choueiri TK & Kaelin Jr WG 2020 Targeting the HIF2-VEGF axis in renal cell carcinoma. Nature Medicine 26 1519–1530. (https://doi. org/10.1038/s41591-020-1093-z) org/10.1038/209590a0) Cramer T, Yamanishi Y, Clausen BE, Forster I, Pawlinski R, Mackman N, Haase VH & Jaenisch R, Corr M, Nizet V 2003 HIF-1alpha is essential for myeloid cell-mediated inflammation. Cell 112 645–657. (https:// doi.org/10.1016/s0092-8674(03)00154-5) Declaration of interesti P J R is a scientific co-founder of, and holds equity in, ReOx Ltd, a University spin-out company that seeks to develop therapeutic HIF hydroxylases inhibitors and a non-executive director of Immunocore Holdings PLC. E J H is employed under the Cambridge Experimental Medicine Initiative, partly funded by AstraZeneca, although they have not been involved in this project. The other authors declare no financial interests. Buffet A, Burnichon N, Favier J & Gimenez-Roqueplo AP 2020 An overview of 20 years of genetic studies in pheochromocytoma and paraganglioma. Best Practice and Research: Clinical Endocrinology and Metabolism 34 101416. (https://doi.org/10.1016/j.beem.2020.101416) Burnichon N, Vescovo L, Amar L, Libe R, De Reynies A, Venisse A, Jouanno E, Laurendeau I, Parfait B, Bertherat J, et al. 2011 Integrative genomic analysis reveals somatic mutations in pheochromocytoma and paraganglioma. Human Molecular Genetics 20 3974–3985. (https:// doi.org/10.1093/hmg/ddr324) Author contribution statement Experiments were designed by L E, M P B, P J R and T B. Data were collected and analysed by all authors. Manuscript was prepared by L E, M P B, P J R and T B and reviewed by all authors. Figures were prepared and statistical analyses performed by L E and M P B with input from other authors. P J R and T B conceived the study and managed the project. P J R and T B are co-senior authors. Coupland RE 1965 The Natural History of Chromaffin Cell. London, UK: Longmans. Coupland RE & Hopwood D 1966 Mechanism of a histochemical reaction differentiating between adrenaline- and noradrenaline-storing cells in the electron microscope. Nature 209 590–591. (https://doi. Endocrine-Related Cancer 2016), a cell type that responds to low oxygen with the rapid release of neurotransmitters to mediate hypoxic ventilatory control in what is termed acute oxygen sensing (Gao et al. 2019). Chromaffin cells are also reportedly acutely oxygen-sensitive during development, but this is lost in adulthood (Thompson et al. 1997). The retention of an immature phenotype in AMs with Phd2 inactivation may extend beyond gene expression to include retention of acute oxygen sensitivity. In future studies, it will be of interest to determine the extent to which Phd2 inactivation in the AM recreates oxygen sensitivity in the adult. Our findings have relevance for both experimental and clinical research into PCC. They suggest that it may also be useful to revisit mouse models of pseudohypoxic PCC including models of inactivation of Sdhx that have not resulted in PCC (Piruat et al. 2004, Bayley et al. 2009, https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. Funding Funding for the work was received from the Oxford Branch of the Ludwig Institute for Cancer Research, the Wellcome Trust (106241/Z/14/Z) and the Paradifference Foundation. This work was also supported by the Francis Crick Institute, which receives its core funding from Cancer Research UK (FC001501), the UK Medical Research Council (FC001501), and the Wellcome Trust (FC001501). L E was sponsored by an MD fellowship from Boehringer Ingelheim Fonds; J D C C L by a FAPESP fellowship (2018/20083-1); S K by a Christopher Welch Scholarship and the Clarendon Fund. Chan WH, Komada M, Fukushima T, Southard-Smith EM, Anderson CR & Wakefield MJ 2019 RNA-seq of isolated chromaffin cells highlights the role of sex-linked and imprinted genes in adrenal medulla development. Scientific Reports 9 3929. (https://doi.org/10.1038/ s41598-019-40501-0) Cheng X, Prange-Barczynska M, Fielding JW, Zhang M, Burrell AL, Lima JD, Eckardt L, Argles I, Pugh CW, Buckler KJ, et al. 2020 Marked and rapid effects of pharmacological HIF-2α antagonism on hypoxic ventilatory control. Journal of Clinical Investigation 130 2237–2251. (https://doi.org/10.1172/JCI133194) emmm.201404153) Fishbein L & Wilkerson MD 2018 Chromaffin cell biology: inferences from the Cancer Genome Atlas. Cell and Tissue Research 372 339–346. (https://doi.org/10.1007/s00441-018-2795-0) Macias D, Cowburn AS, Torres-Torrelo H, Ortega-Saenz P, Lopez-Barneo J & Johnson RS 2018 HIF-2alpha is essential for carotid body development and function. eLife 7 e34681. (https://doi.org/10.7554/ eLife.34681) Fishbein L, Leshchiner I, Walter V, Danilova L, Robertson AG, Johnson AR, Lichtenberg TM, Murray BA, Ghayee HK, Else T, et al. 2017 Comprehensive molecular characterization of pheochromocytoma and paraganglioma. Cancer Cell 31 181–193. (https://doi.org/10.1016/j. ccell.2017.01.001) Mazzone M, Dettori D, De Oliveira RL, Loges S, Schmidt T, Jonckx B, Tian YM, Lanahan AA, Pollard P, De Almodovar CR, et al. 2009 Heterozygous deficiency of PHD2 restores tumor oxygenation and inhibits metastasis via endothelial normalization. Cell 136 839–851. (https://doi.org/10.1016/j.cell.2009.01.020) Furlan A, Dyachuk V, Kastriti ME, Calvo-Enrique L, Abdo H, Hadjab S, Chontorotzea T, Akkuratova N, Usoskin D, Kamenev D, et al. 2017 Multipotent peripheral glial cells generate neuroendocrine cells of the adrenal medulla. Science 357 eaal3753. (https://doi.org/10.1126/ science.aal3753) Nagendran M, Riordan DP, Harbury PB & Desai TJ 2018 Automated cell- type classification in intact tissues by single-cell molecular profiling. eLife 7 e30510. (https://doi.org/10.7554/eLife.30510) Opotowsky AR, Moko LE, Ginns J, Rosenbaum M, Greutmann M, Aboulhosn J, Hageman A, Kim Y, Deng LX, Grewal J, et al. 2015 Pheochromocytoma and paraganglioma in cyanotic congenital heart disease. Journal of Clinical Endocrinology and Metabolism 100 1325–1334. (https://doi.org/10.1210/jc.2014-3863) Gao L, Ortega-Sáenz P & López-Barneo J 2019 Acute oxygen sensing-Role of metabolic specifications in peripheral chemoreceptor cells. Respiratory Physiology and Neurobiology 265 100–111. (https://doi. org/10.1016/j.resp.2018.08.007) Gruber M, Hu CJ, Johnson RS, Brown EJ, Keith B & Simon MC 2007 Acute postnatal ablation of Hif-2alpha results in anemia. PNAS 104 2301–2306. (https://doi.org/10.1073/pnas.0608382104) Park WJ, Brenner O, Kogot-Levin A, Saada A, Merrill Jr AH, Pewzner-Jung Y & Futerman AH 2015 Development of pheochromocytoma in ceramide synthase 2 null mice. Endocrine-Related Cancer 22 623–632. (https://doi.org/10.1530/ERC-15-0058) Hanemaaijer ES, Margaritis T, Sanders K, Bos FL, Candelli T, Al-Saati H, Van Noesel MM, Meyer-Wentrup FAG, Van De Wetering M, Holstege FCP, et al. 2021 Single-cell atlas of developing murine adrenal gland reveals relation of Schwann cell precursor signature to neuroblastoma phenotype. PNAS 118 e2022350118. (https://doi. (https://doi.org/10.1113/JP271050) Semenza GL 2019 Pharmacologic targeting of hypoxia-inducible factors. Annual Review of Pharmacology and Toxicology 59 379–403. (https://doi. org/10.1146/annurev-pharmtox-010818-021637) Jin Y, An X, Ye Z, Cully B, Wu J & Li J 2009 RGS5, a hypoxia-inducible apoptotic stimulator in endothelial cells. Journal of Biological Chemistry Semenza GL 2019 Pharmacologic targeting of hypoxia-inducible factors. Annual Review of Pharmacology and Toxicology 59 379–403. (https://doi. org/10.1146/annurev-pharmtox-010818-021637) 284 23436–23443. (https://doi.org/10.1074/jbc.M109.032664) Kaelin Jr WG 2008 The von Hippel-Lindau tumour suppressor protein: O2 sensing and cancer. Nature Reviews: Cancer 8 865–873. (https://doi. org/10.1038/nrc2502) Shen C, Beroukhim R, Schumacher SE, Zhou J, Chang M, Signoretti S & Kaelin Jr WG 2011 Genetic and functional studies implicate HIF1α as a 14q kidney cancer suppressor gene. Cancer Discovery 1 222–235. (https://doi.org/10.1158/2159-8290.CD-11-0098) (https://doi.org/10.1158/2159-8290.CD-11-0098) Ladroue C, Carcenac R, Leporrier M, Gad S, Le Hello C, Galateau-Salle F, Feunteun J, Pouysségur J, Richard S & Gardie B 2008 PHD2 mutation and congenital erythrocytosis with paraganglioma. New England Journal of Medicine 359 2685–2692. (https://doi.org/10.1056/ NEJMoa0806277) Smith-Hicks CL, Sizer KC, Powers JF, Tischler AS & Costantini F 2000 C-cell hyperplasia, pheochromocytoma and sympathoadrenal malformation in a mouse model of multiple endocrine neoplasia type 2B. EMBO Journal 19 612–622. (https://doi.org/10.1093/ emboj/19.4.612) Lee S, Nakamura E, Yang H, Wei W, Linggi MS, Sajan MP, Farese RV, Freeman RS, Carter BD, Kaelin Jr WG, et al. 2005 Neuronal apoptosis linked to EglN3 prolyl hydroxylase and familial pheochromocytoma genes: developmental culling and cancer. Cancer Cell 8 155–167. (https://doi.org/10.1016/j.ccr.2005.06.015) Tai TC, Wong-Faull DC, Claycomb R & Wong DL 2009 Hypoxic stress- induced changes in adrenergic function: role of HIF1 alpha. Journal of Neurochemistry 109 513–524. (https://doi. org/10.1111/j.1471-4159.2009.05978.x) Lepoutre-Lussey C, Thibault C, Buffet A, Morin A, Badoual C, Bénit P, Rustin P, Ottolenghi C, Janin M, Castro-Vega LJ, et al. 2016 From Nf1 to Sdhb knockout: successes and failures in the quest for animal models of pheochromocytoma. Molecular and Cellular Endocrinology 421 40–48. (https://doi.org/10.1016/j.mce.2015.06.027) Thompson RJ, Jackson A & Nurse CA 1997 Developmental loss of hypoxic chemosensitivity in rat adrenomedullary chromaffin cells. Journal of Physiology 498 503–510. (https://doi.org/10.1113/jphysiol.1997.sp021876) Tischler AS 2008 Pheochromocytoma and extra-adrenal paraganglioma: updates. Archives of Pathology and Laboratory Medicine 132 1272–1284. (https://doi.org/10.5858/2008-132-1272-PAEPU) Lindeberg J, Usoskin D, Bengtsson H, Gustafsson A, Kylberg A, Söderström S & Ebendal T 2004 Transgenic expression of Cre recombinase from the tyrosine hydroxylase locus. Genesis 40 67–73. (https://doi.org/10.1002/gene.20065) Toledo RA 2017 New HIF2α inhibitors: potential implications as therapeutics for advanced pheochromocytomas and paragangliomas. References Al Khazal F, Kang S, Nelson Holte M, Choi DS, Singh R, Ortega-Sáenz P, López-Barneo J & Maher 3rd LJ 2021 Unexpected obesity, rather than tumorigenesis, in a conditional mouse model of mitochondrial complex II deficiency. FASEB Journal 35 e21227. (https://doi. org/10.1096/fj.202002100R) Díaz-Castro B, Pintado CO, García-Flores P, López-Barneo J & Piruat JI 2012 Differential impairment of catecholaminergic cell maturation and survival by genetic mitochondrial complex II dysfunction. Molecular and Cellular Biology 32 3347–3357. (https://doi.org/10.1128/ MCB.00128-12) Bankhead P, Loughrey MB, Fernández JA, Dombrowski Y, McArt DG, Dunne PD, Mcquaid S, Gray RT, Murray LJ, Coleman HG, et al. 2017 QuPath: open source software for digital pathology image analysis. Scientific Reports 7 16878. (https://doi.org/10.1038/s41598-017-17204-5) Eisenhofer G, Walther MM, Huynh TT, Li ST, Bornstein SR, Vortmeyer A, Mannelli M, Goldstein DS, Linehan WM, Lenders JW, et al. 2001 Pheochromocytomas in von Hippel-Lindau syndrome and multiple endocrine neoplasia type 2 display distinct biochemical and clinical phenotypes. Journal of Clinical Endocrinology and Metabolism 86 1999–2008. (https://doi.org/10.1210/jcem.86.5.7496) Bayley JP, Van Minderhout I, Hogendoorn PC, Cornelisse CJ, Van Der Wal A, Prins FA, Teppema L, Dahan A, Devilee P & Taschner PE 2009 Sdhd and SDHD/H19 knockout mice do not develop paraganglioma or pheochromocytoma. PLoS ONE 4 e7987. (https://doi.org/10.1371/ journal.pone.0007987) Favier J, Brière JJ, Burnichon N, Rivière J, Vescovo L, Benit P, Giscos- Douriez I, De Reyniès A, Bertherat J, Badoual C, et al. 2009 The Warburg effect is genetically determined in inherited pheochromocytomas. PLoS ONE 4 e7094. (https://doi.org/10.1371/ journal.pone.0007094) Bishop T & Ratcliffe PJ 2014 Signaling hypoxia by hypoxia-inducible factor protein hydroxylases: a historical overview and future perspectives. Hypoxia 2 197–213. (https://doi.org/10.2147/HP.S47598) Bishop T & Ratcliffe PJ 2014 Signaling hypoxia by hypoxia-inducible factor protein hydroxylases: a historical overview and future perspectives. Hypoxia 2 197–213. (https://doi.org/10.2147/HP.S47598) This work is licensed under a Creative Commons Attribution 4.0 International License. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access The role of PHD2 in pseudohypoxic PCCs The role of PHD2 in pseudohypoxic PCCs L Eckardt, M Prange- Barczynska et al. 771 Lorenzo FR, Yang C, Ng Tang Fui M, Vankayalapati H, Zhuang Z, Huynh T, Grossmann M, Pacak K & Prchal JT 2013 A novel EPAS1/HIF2A germline mutation in a congenital polycythemia with paraganglioma. Journal of Molecular Medicine 91 507–512. (https://doi.org/10.1007/ s00109-012-0967-z) Favier J, Buffet A & Gimenez-Roqueplo AP 2012 HIF2A mutations in paraganglioma with polycythemia. New England Journal of Medicine 367 2161; author reply 2161–2162. (https://doi.org/10.1056/ NEJMc1211953) Fielding JW, Hodson EJ, Cheng X, Ferguson DJP, Eckardt L, Adam J, Lip P, Maton-Howarth M, Ratnayaka I, Pugh CW, et al. 2018 PHD2 inactivation in Type I cells drives HIF-2alpha-dependent multilineage hyperplasia and the formation of paraganglioma-like carotid bodies. Journal of Physiology 596 4393–4412. (https://doi.org/10.1113/ JP275996) Macias D, Fernandez-Aguera MC, Bonilla-Henao V & Lopez-Barneo J 2014 Deletion of the von Hippel-Lindau gene causes sympathoadrenal cell death and impairs chemoreceptor-mediated adaptation to hypoxia. EMBO Molecular Medicine 6 1577–1592. (https://doi.org/10.15252/ emmm 201404153) emmm.201404153) org/10.1073/pnas.2022350118) Piruat JI, Pintado CO, Ortega-Sáenz P, Roche M & López-Barneo J 2004 The mitochondrial SDHD gene is required for early embryogenesis, and its partial deficiency results in persistent carotid body glomus cell activation with full responsiveness to hypoxia. Molecular and Cellular Biology 24 10933–10940. (https://doi.org/10.1128/MCB.24.24.10933- 10940.2004) Hodson EJ, Nicholls LG, Turner PJ, Llyr R, Fielding JW, Douglas G, Ratnayaka I, Robbins PA, Pugh CW, Buckler KJ, et al. 2016 Regulation of ventilatory sensitivity and carotid body proliferation in hypoxia by the PHD2/HIF-2 pathway. Journal of Physiology 594 1179–1195. (https://doi org/10 1113/JP271050) Rotolo T, Smallwood PM, Williams J & Nathans J 2008 Genetically- directed, cell type-specific sparse labeling for the analysis of neuronal morphology. PLoS ONE 3 e4099. (https://doi.org/10.1371/journal. pone.0004099) (https://doi.org/10.1113/JP271050) Endocrine- Related Cancer 24 C9–C19. (https://doi.org/10.1530/ERC-16-0479) This work is licensed under a Creative Commons Attribution 4.0 International License. https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com https://doi.org/10.1530/ERC-21-0211 https://erc.bioscientifica.com © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access L Eckardt, M Prange- Barczynska et al. dependent. EMBO Reports 2 292–297. (https://doi.org/10.1093/embo- reports/kve064) Toledo RA, Qin Y, Srikantan S, Morales NP, Li Q, Deng Y, Kim SW, Pereira MA, Toledo SP, Su X, et al. 2013 In vivo and in vitro oncogenic effects of HIF2A mutations in pheochromocytomas and paragangliomas. Endocrine-Related Cancer 20 349–359. (https://doi. org/10.1530/ERC-13-0101) Waldmann J, Fendrich V, Holler J, Buchholz M, Heinmöller E, Langer P, Ramaswamy A, Samans B, Walz MK, Rothmund M, et al. 2010 Microarray analysis reveals differential expression of benign and malignant pheochromocytoma. Endocrine-Related Cancer 17 743–756. (https://doi.org/10.1677/ERC-09-0118) Vaidya A, Flores SK, Cheng ZM, Nicolas M, Deng Y, Opotowsky AR, Lourenço Jr DM, Barletta JA, Rana HQ, Pereira MA, et al. 2018 EPAS1 mutations and paragangliomas in cyanotic congenital heart disease. New England Journal of Medicine 378 1259–1261. (https://doi. Welander J, Andreasson A, Brauckhoff M, Bäckdahl M, Larsson C, Gimm O & Söderkvist P 2014 Frequent EPAS1/HIF2α exons 9 and 12 mutations in non-familial pheochromocytoma. Endocrine-Related Cancer 21 495–504. (https://doi.org/10.1530/ERC-13-0384) Verhofstad AA, Hökfelt T, Goldstein M, Steinbusch HW & Joosten HW 1979 Appearance of tyrosine hydroxylase, aromatic amino-acid decarboxylase, dopamine beta-hydroxylase and phenylethanolamine N-methyltransferase during the ontogenesis of the adrenal medulla: an immunohistochemical study in the rat. Cell and Tissue Research 200 1–13. (https://doi.org/10.1007/ BF00236882) Verhofstad AA, Hökfelt T, Goldstein M, Steinbusch HW & Joosten HW 1979 Appearance of tyrosine hydroxylase, aromatic amino-acid decarboxylase, dopamine beta-hydroxylase and phenylethanolamine N-methyltransferase during the ontogenesis of the adrenal medulla: an immunohistochemical study in the rat. Cell and Tissue Research 200 1–13. (https://doi.org/10.1007/ BF00236882) Yang C, Zhuang Z, Fliedner SM, Shankavaram U, Sun MG, Bullova P, Zhu R, Elkahloun AG, Kourlas PJ, Merino M, et al. 2015 Germ-line PHD1 and PHD2 mutations detected in patients with pheochromocytoma/ paraganglioma-polycythemia. Journal of Molecular Medicine 93 93–104. (https://doi.org/10.1007/s00109-014-1205-7) Zhou T, Chien MS, Kaleem S & Matsunami H 2016 Single cell transcriptome analysis of mouse carotid body glomus cells. Journal of Physiology 594 4225–4251. (https://doi.org/10.1113/JP271936) Vollmer RR 1996 Selective neural regulation of epinephrine and norepinephrine cells in the adrenal medulla – cardiovascular implications. Clinical and Experimental Hypertension 18 731–751. (https://doi.org/10.3109/10641969609081778) Vollmer RR 1996 Selective neural regulation of epinephrine and norepinephrine cells in the adrenal medulla – cardiovascular implications. Clinical and Experimental Hypertension 18 731–751. (https://doi.org/10.3109/10641969609081778) Zhuang Z, Yang C, Lorenzo F, Merino M, Fojo T, Kebebew E, Popovic V, Stratakis CA, Prchal JT & Pacak K 2012 Somatic HIF2A gain-of-function mutations in paraganglioma with polycythemia. Endocrine-Related Cancer 772 Received in final form 1 September 2021 Accepted 20 September 2021 Accepted Manuscript published online 20 September 2021 L Eckardt, M Prange- Barczynska et al. New England Journal of Medicine 367 922–930. (https://doi.org/10.1056/NEJMoa1205119) p g Vooijs M, Jonkers J & Berns A 2001 A highly efficient ligand-regulated Cre recombinase mouse line shows that LoxP recombination is position Received in final form 1 September 2021 Accepted 20 September 2021 Accepted Manuscript published online 20 September 2021 © 2021 The authors Printed in Great Britain Published by Bioscientifica Ltd. This work is licensed under a Creative Commons Attribution 4.0 International License. Downloaded from Bioscientifica.com at 10/28/2021 08:22:43AM via free access
https://openalex.org/W2024241323	https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0096473&type=printable	English	null	Diabetes Attitude Scale: Validation in Type-2 Diabetes Patients in Multiple Centers in China	PloS one	2,014	cc-by	6,989	Abstract Competing Interests: The funding to support this study was provided by Novo Nordisk (China) Pharmaceutical Co., Ltd (NNCP). The in development or marketed products to declare. This does not alter the authors’ adherence to PLOS ONE policies on sharing data * E-mail: guoxh@medmail.com.cn * E-mail: guoxh@medmail.com.cn ¤ Current address: Beijing Tsinghua Hospital, Beijing, China Empowerment Scale [11,12]. All of these established scales have been used in different studies [13,14,15,16,17,18,19,20,21,22]. Furthermore, they have been translated into Chinese versions [23,24,25,26,27,28], and used in China [29,30,31]. Attitude towards diabetes is very important. According to the attitude behavior model, a patient’s intention to behave in a certain way has two major determinants, one of which is the patients’ attitudes towards the behavior [32]. And it has been shown that attitudes can affect health care behavior [33,34,35], diabetes control and patient outcomes [36,37]. Indeed, it has been proposed that changing patients’ attitudes by monitoring the psychosocial impact of diabetes may provide a cost-effective way to improve disease control outcomes [38]. Unlike many other diseases, diabetes requires ongoing self-management of care, even when patients are Diabetes Attitude Scale: Validation in Type-2 Diabetes Patients in Multiple Centers in China Qingqing Lou1,2, Yufeng Chen3, Xiaohui Guo4*¤, Li Yuan5, Tao Chen5, Chun Wang5, Li Shen6¤, Zilin Sun7, Fang Zhao8, Xia Dai9, Jin Huang10, Huiying Yang11, on behalf of Chinese Diabetes Education Status Survey study group 1 Jiangsu Province Hospital on Integration of Chinese and Western Medicine, Nanjing, Jiangsu Province, China, 2 Sir Run Run Shaw Hospital, Zhejiang University Medical School, Hangzhou, Zhejiang Province, China, 3 School of Nursing, Nanjing University of Traditional Chinese Medicine, Nanjing, Jiangsu Province, China, 4 Department of Endocrinology, Peking University First Hospital, Beijing, China, 5 Department of Endocrinology, West China Medical School, West China Hospital, Sichuan University, Chengdu, Sichuan Province, China, 6 Peking University First Hospital, Beijing, China, 7 Institute of Diabetes, Zhongda Hospital, Medical School, Southeast University, Nanjing, Jiangsu Province, China, 8 Department of Endocrinology, China-Japan Friendship Hospital, Beijing, China, 9 First Affiliated Hospital, Guangxi Medical University, Nanning, Jiangsu Province, China, 10 The Second Xiangya Hospital of Central South University, Changsha, Hunan Province, China, 11 The First Affiliated Hospital of Kunming Medical College, Kunming, Yunnan Province, China Abstract Objective: The aim of the paper is to report the development and psychometric testing of Diabetes Att Method: A prospective study was performed. The cultural equivalency and content validity of the Diabetes Attitude Scale were determined by panels of endocrinologists, physiologists, nurses and dieticians. An accurate and usable translation was obtained for each of five subscales examining attitudes on need for special training, the seriousness of type-2 diabetes, the need for controlling the condition, its psychosocial impact and the degree of autonomy given to patients in decision making. The validation was derived from 5961 patients with type-2 diabetes, recruited from 50 centers in 29 provinces throughout China between March 1st and September 30th, 2010. Results: The modified Diabetes Attitude Scale showed an acceptable level of internal consistency. The strength of the inter- correlations among the domains of five subscales suggests that the instrument measures related but separate domains of patients’ attitudes toward diabetes. Moreover, the test-retest intraclass correlation coefficients were high enough to support the stability of the Chinese version of the third version of the scale. Conclusions: The psychometric properties of the Chinese version of Diabetes Attitude Scale demonstrated satisfactory validity and reliability and appeared to effectively evaluate attitudes toward diabetes in patients with type-2 diabetes. Citation: Lou Q, Chen Y, Guo X, Yuan L, Chen T, et al. (2014) Diabetes Attitude Scale: Validation in Type-2 Diabetes Patients in Multiple Centers in China. PLoS ONE 9(5): e96473. doi:10.1371/journal.pone.0096473 Editor: Antony Bayer, Cardiff University, United Kingdom Received November 7, 2013; Accepted April 9, 2014; Published May 6, 2014 Received November 7, 2013; Accepted April 9, 2014; Published May 6, 2014 Received November 7, 2013; Accepted April 9, 2014; Published May 6, 2014 Copyright: 2014 Lou et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: Funding to support this study was provided by Novo Nordisk (China) Pharmaceutical Co., Ltd (NNCP). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. he funding to support this study was provided by Novo Nordisk (China) Pharmaceutical Co., Ltd (NNCP). There are no patents, products eted products to declare. This does not alter the authors’ adherence to PLOS ONE policies on sharing data and materials. Methodology The study incorporated a two-phase design (Fig. 1) that had previously been used by Shiu et al [27] to validate the Diabetes Empowerment Scale. This approach enabled both qualitative and quantitative assessments of the psychometric properties of the C-DAS-3 to be evaluated. Phase I. Phase I included translation of the English language version of the DAS-3 into Chinese and examination of the Chinese version of DAS-3 (C-DAS-3) for cultural equivalency and content validity. The Chinese translation was guided by the Brislin’s translation model [46]. A bilingual translator who was a medical doctor translated the DAS-3 into Chinese. It was then translated back into English by another physician who was also a bilingual translator. Both the original and back-translated versions were compared to determine the accuracy of the translation. The translated version (C-DAS-3) was then examined by a panel of experts comprising an endocrinologist, three diabetes nurse educators and a dietitian. The panel tested whether the translated parameters were equivalent to the original parameters and assessed whether the translated version could be readily under- stood and tested in a sample population of Chinese patients with type-2 diabetes. The translated C-DAS-3 was presented to a second panel of experts for content validity assessment. This panel comprised two endocrinologists, a psychologist, five diabetes nurse specialists, and a dietician. Content validity was assessed by asking the members to rate each item as a valid measure of the construct using a three- point scale where: 1 = disagree, 2 = neutral and 3 = agree. The content validity was calculated in this manner for each item and for the overall C-DAS-3. The panel also assessed each translated item individually for accuracy, clarity and for the cultural relevance of the translation. Following minor revisions, a panel- modified version was developed and was pilot-tested in 20 patients with type-2 diabetes to check the data collection procedure and ease of understanding. A second pilot study involving 106 patients recruited from a single center (three hospitals) in China, was undertaken to establish the psychometric properties [47] and internal consistency of the translation prior to the large scale assessment of the C-DAS-3. In this second pilot study the overall Cronbach’s alpha coefficient for internal consistency was satisfactory (0.771), but the coefficient for the second dimension consistency was only 0.390. The translation was rechecked under the guidance of the original author of the DAS-3 (Dr. Study Population The sample for Phase II was selected from among 6043 patients (. = 16 years old), who had been diagnosed with type-2 diabetes for at least 1 year. The patients were selected from 50 centers in 29 provinces in China between March 1st and September 30th, 2010. Demographic characteristics, HbA1c, fasting and 2 h postprandial blood glucose levels were obtained from all patients. To avoid bias a single nurse from each center was trained to administer the C-DAS-3 using a defined protocol. The patients completed the Methodology Anderson) and changes were made to the translation of item 21: ‘Type-2 diabetes is a very serious disease’. Content validity was repeated by the same expert panel and a final version was developed. Phase II. In the second phase of the study we established the test-retest reliability, internal consistency, construct validity, and criterion validity of the final version of the C-DAS-3 in a large number of patients with type-2 diabetes. obtained from each patient and the primary caregivers of the minors enrolled. This version collected data on 33 items within five discrete subscales and was based on the original versions of the DAS [45]. The five subscales examine patient attitudes on need for specialist healthcare training, the seriousness of type-2 diabetes, the need for controlling the condition, its psychosocial impact and the degree of autonomy given to patients in decision making. These items were chosen on the basis that they were important beliefs that were likely to predict the behavior of patients with type-2 diabetes. The items cover a range of issues relevant to the effects of diabetes on everyday life and patients’ well being. However, the vase cultural differences between China and western countries make applying DAS in Chinese patients a complex task, requiring cultural consideration rather than simply translation to accurately assess psychological attitudes towards diabetes in Chinese patients. obtained from each patient and the primary caregivers of the minors enrolled. asymptomatic. In contemporary research, compliance with self- management programs during the asymptomatic period has been reported to be very low, especially in key areas of diet and exercise [39]. In a wide cross-section of international type 2 diabetes patients, patient attitudes, wishes, and needs have been shown to be the foundation for successful care [40]. In Western research, several studies have produced classification systems that allow for clinical stratification of type 2 diabetes patients based on opinions and attitudes that can influence self-care behaviors [41]. It has been demonstrated, however, that diabetes patients’ attitudes and opinions are highly dependent on specific cultural factors, including patients’ social networks, knowledge and opinions of family and friends, and concern about the disease [33]. The rate of diabetic compliance with self-management, including insulin therapy thus varies highly by country [34], potentially as a result of variant patient opinions and attitudes towards their diabetes care. In particular, insulin adherence, a behavior known to detrimentally affect many non-compliant patients, was reported to impact patient financial situation, family and social life, and emotional well-being by the Global Attitudes of Patients and Physicians in Insulin Therapy study of 1530 insulin-treated patients, including 1350 Type 2 diabetes patients in China, France, Japan, Germany, Spain, Turkey, the UK, and the USA [4]. In developing countries, the patient population’s knowledge, attitude, and practice (KAP) of diabetes is generally much worse than those in developed countries, partially due to the lack of training programs for care providers and education programs for patients [42]. Thus, many cultural factors play a role in successful diabetes care by influencing patient attitudes and opinions, and attitudes vary widely between different countries. In fact, increasing evidence suggests that improving patient educations is the most effective way to lessen the complications and costs associated with diabetes and its management [43]; however, targeted and culturally sensitive patient education programs are not possible without improved understanding of diabetic patient attitudes in these developing regions. To assess patient attitudes, we need a specific tool. The original version of the Diabetes Attitude Scale (DAS) was developed by Anderson et al in 1989. The scale which, included 31-items arranged in eight subscales, was designed to measure the attitudes of health-care professionals (HCPs) concerning important issues that affect diabetes control [44]. The third version of the scale (DAS-3) published in 1998 was designed to obtain information from patients as well as from HCPs [41]. Introduction Successful control of diabetes greatly depends on patients being able to manage their disease [1]. The patients can, therefore, be regarded as core team members who administer their treatment on a daily basis [2]. Studies show that diabetes patients experience various types of psychosocial and emotional problems [3], therefore, psychological factors are important to diabetes man- agement. Many studies focusing on the effects of psychological problems on diabetes reveal that psychological factors have an impact on diabetes control [4,5,6]. And researchers have also developed different questionnaires to provide valuable insights in patients’ view on their disease e.g. the Diabetes Distress Scale [7,8], the Problem Areas in Diabetes Questionnaire [9,10] or the May 2014 \| Volume 9 \| Issue 5 \| e96473 PLOS ONE \| www.plosone.org 1 May 2014 \| Volume 9 \| Issue 5 \| e96473 The Chinese Version of Diabetes Attitude Scale (C-DAS-3) The Chinese Version of Diabetes Attitude Scale (C-DAS-3) asymptomatic. In contemporary research, compliance with self- management programs during the asymptomatic period has been reported to be very low, especially in key areas of diet and exercise [39]. In a wide cross-section of international type 2 diabetes patients, patient attitudes, wishes, and needs have been shown to be the foundation for successful care [40]. In Western research, several studies have produced classification systems that allow for clinical stratification of type 2 diabetes patients based on opinions and attitudes that can influence self-care behaviors [41]. It has been demonstrated, however, that diabetes patients’ attitudes and opinions are highly dependent on specific cultural factors, including patients’ social networks, knowledge and opinions of family and friends, and concern about the disease [33]. The rate of diabetic compliance with self-management, including insulin therapy thus varies highly by country [34], potentially as a result of variant patient opinions and attitudes towards their diabetes care. In particular, insulin adherence, a behavior known to detrimentally affect many non-compliant patients, was reported to impact patient financial situation, family and social life, and emotional well-being by the Global Attitudes of Patients and Physicians in Insulin Therapy study of 1530 insulin-treated patients, including 1350 Type 2 diabetes patients in China, France, Japan, Germany, Spain, Turkey, the UK, and the USA [4]. In developing countries, the patient population’s knowledge, attitude, and practice (KAP) of diabetes is generally much worse than those in developed countries, partially due to the lack of training programs for care providers and education programs for patients [42]. Thus, many cultural factors play a role in successful diabetes care by influencing patient attitudes and opinions, and attitudes vary widely between different countries. In fact, increasing evidence suggests that improving patient educations is the most effective way to lessen the complications and costs associated with diabetes and its management [43]; however, targeted and culturally sensitive patient education programs are not possible without improved understanding of diabetic patient attitudes in these developing regions. To assess patient attitudes, we need a specific tool. The original version of the Diabetes Attitude Scale (DAS) was developed by Anderson et al in 1989. The scale which, included 31-items arranged in eight subscales, was designed to measure the attitudes of health-care professionals (HCPs) concerning important issues that affect diabetes control [44]. The third version of the scale (DAS-3) published in 1998 was designed to obtain information from patients as well as from HCPs [41]. The Chinese Version of Diabetes Attitude Scale (C-DAS-3) This version collected data on 33 items within five discrete subscales and was based on the original versions of the DAS [45]. The five subscales examine patient attitudes on need for specialist healthcare training, the seriousness of type-2 diabetes, the need for controlling the condition, its psychosocial impact and the degree of autonomy given to patients in decision making. These items were chosen on the basis that they were important beliefs that were likely to predict the behavior of patients with type-2 diabetes. The items cover a range of issues relevant to the effects of diabetes on everyday life and patients’ well being. However, the vase cultural differences between China and western countries make applying DAS in Chinese patients a complex task, requiring cultural consideration rather than simply translation to accurately assess obtained from each patient and the primary caregivers of the minors enrolled. Figure 1. Flow diagram of the study. doi:10.1371/journal.pone.0096473.g001 Figure 1. Flow diagram of the study. doi:10.1371/journal.pone.0096473.g001 questionnaires in a comfortable and quiet room. For illiterate participants the C-DAS-3 questions were read by a trained nurse and their answers were recorded. was considered to be long enough for the respondents not to recall their initial answers but not too long for their attitudes to change [48]. Instrument SPSS version 15(IBM, USA) was used to analyze results from psychometric tests and scale analyses. In all analyses, values of P,0.05 were considered statistically significant. The test–retest reliability was evaluated using the paired t-test and the Pearson correlation coefficient. The internal consistency of the scale was examined by measuring the item reliability index, performed by calculating the Cronbach’s alpha values for assessing the relatedness of each domain in the questionnaire and of each item in every domain. A principal component analysis with Varimax rotation was used for factor analysis to determine the extent of change in the questionnaire that resulted from the translation. Factors with eigenvalues $1.0 were included in the model. The reliability of the C-DAS-3 was evaluated by estimating internal consistency using Cronbach’s alpha statistic for the overall score and for each subscale. Correlations between the C-DAS-3 subscales and the criterion validity of the total C-DAS-3 score in comparison with HbA1c were examined using Pearson’s correla- tion statistics, and test-retest consistency was evaluated using intraclass correlation coefficients. The final Chinese version included translations of all 33 items in the original DAS-3. Five statements evaluated beliefs about the need for special training of healthcare staff, seven items each examined attitudes about the need for tight control of diabetes and seriousness of the disease, six items reviewed the psychosocial impact of diabetes and eight items examined what patients believed about the degree of autonomy they had in decision making. Patients indicated their agreement with each of the 33 statement as: strongly agree ( = 5), agree ( = 4), neutral ( = 3), disagree ( = 2) or strongly disagree ( = 1). Ethics Statements The study protocol was approved by the Hospital Ethical Committee of West China Hospital, Medical School Sichuan University (approval ID: 2010(10)). Written informed consent was May 2014 \| Volume 9 \| Issue 5 \| e96473 PLOS ONE \| www.plosone.org May 2014 \| Volume 9 \| Issue 5 \| e96473 2 The Chinese Version of Diabetes Attitude Scale (C-DAS-3) Psychometric Tests and Scale Statistics After factor analysis, 6 factors with minimum eigenvalue of 1.0 were extracted, accounted for 59.67% of the total variance. The subscales of the original DAS could not be confirmed in the Chinese population. The correlation coefficient between the total Chinese version of Diabetes Attitude Scale score and HbA1c was 20.040 (p = 0.018), and weak correlations between 4 subscales (subscale 1,subscale 4) and A1c were also found. The correlation coefficients were 20.033 (p = 0.042), 20.047 (p = 0.006), 20.077 (p = 0.000), 20.066 (p = 0.000), and 20.024 (p = 0.153), respectively. Descriptive statistics for the five C-DAS-3 subscales are presented in Table 2. The mean scores ranged from 3.5760.51 for subscale 3 (‘value of tight control’) to 4.360.49 for subscale 1 (‘need for special training’). The reliability (internal consistency) of the subscales based on Cronbach’s alpha statistic ranged from 0.654 for subscale 5 (‘patient autonomy’) to 0.848 for subscale 4 (‘psychosocial impact of diabetes’). The comparison of Cronbach’s alpha of each subscale with the original DAS-3 was also presented in Table 2. Subscale inter-correlations ranged from 0.836 for subscale 1 (‘need for special training’) versus subscale 5 (‘patient autonomy’) to 0.497 for subscale 1 (‘need for special training’) versus subscale 2 (‘seriousness of type-2 diabetes’). This high correlation between the subscales maybe partially due to the fact that the items of the subscales did not load in different factors. There were correlations between the C-DAS score and age, education level, duration of diabetes, presence of complications, and accepted diabetes education, but there was no association between the Chinese DAS score and gender, insulin treatment, and BMI. (Table 3) Demographic and Clinical Data Eighty two of the initial 6043 questionnaires had missing data, and questionnaires from the remaining 5961 patients (98.6%, male 3233 and female 2728) were included in the analysis. The mean age was 59.5612.4 years, the mean BMI was 24.4964.10 kg/m2 Test-retest reliability was assessed after evaluating internal consistency, in 60 patients who were willing to complete the questionnaire on two occasions 2 to 4 weeks apart. This interval May 2014 \| Volume 9 \| Issue 5 \| e96473 PLOS ONE \| www.plosone.org 3 The Chinese Version of Diabetes Attitude Scale (C-DAS-3) Table 1. Demographic data and baseline characteristics of participants. Character Sex (female), n 2728 (45.8%) Age (mean 6 SD), years 59.50612.48 Level of education No formal education 365 Primary school 737 Middle school 1509 High school 1611 College level or above 1739 Diabetes duration (mean 6 SD), years 8.7966.85 Treatment, n Oral medication only 2069 (34.73%) Insulin only 1399 (23.48%) Oral medication + insulin 2269 (38.08%) Neither oral medication nor insulin 221 (3.71%) At least one diabetic complication, n 4238 (71.2%) BMI (mean 6 SD), kg/m2 24.4964.10 HbA1c (mean 6 SD; n = 3480) 8.2762.23 ,7.0 1117 (32.10%) 7.0–8.5 1086 (31.21%) $8.5 1277 (36.69%) Total C-DAS-3 score (mean 6 SD) 3.7660.30 Data are expressed as means6SD or as numbers and % (n = 5961). Table 1. Demographic data and baseline characteristics of participants. diabetes’) (Table 2). Paired t-test was also applied to evaluate the test-retest reliability which shows no statistical difference between the test and re-test (t = 0.18, p = 0.857). and the mean duration of diabetes was 8.7566.78 years. The majority of patients (71.2%) had at least one complication of chronic diabetes. HbA1c values were obtained from 3480 patients (Table 1). The norms of the C-DAS-3 in the diabetes patients whose HbA1c, fasting and 2 h postprandial blood glucose levels were successfully controlled were 3.7960.29, 3.7960.30, and 3.7960.29 respectively. Psychosocial Impact, Training, and Patient Well-Being in China The mean nurse evaluated subscale scores for the C-DAS-3 were all somewhat lower than that in the original DAS-3 (subscale 1: 4.23 versus 4.67; subscale 2: 3.62 versus 4.58; subscale 3: 3.57 versus 4.43; subscale 4: 3.79 versus 4.39; subscale 5: 3.68 versus 4.33) [45]. The reliabilities (‘internal consistency’) of the subscale 1, subscale 3 and subscale 4 of the C-DAS-3 were higher than that in the original DAS-3 (subscale 1: 0.74 versus 0.67; subscale 3: 0.820 versus 0.72; subscale 4: 0.848 versus 0.65) [45] but others of the C-DAS-3 were lower than that in the original DAS-3 (subscale 2: 0.706 versus 0.80; subscale 5: 0.654 versus 0.76) [45]. The results were comparable. Cultural difference is the possible reason for different reliability outcomes in the Chinese and in the western populations. For example, the reliabilities of the subscale 3 (value of tight control) in the C-DAS-3 was higher than that in the original DAS-3. Compared with western patients, Chinese patients are more stressed on value of tight glycemic control, therefore patients tended to give a more homogeneous response pattern in the Chinese version of the DAS tight control than in a western population. In America, diabetes education has evolved from primarily didactic presentations to more theoretically based empowerment models [51] emphasizing on patient autonomy. Therefore Americans tended to give a more homogeneous response pattern in the original version of the DAS patient autonomy. That is why the reliabilities of the subscale 5 (patient autonomy) in the C-DAS-3 was lower than that in the original DAS-3. Both the original and Chinese versions of DAS-3 have validated by the experts, but in the factor analysis, the subscales of the original DAS could not be confirmed in the Chinese population, this may be due to the culture difference, diabetes education qualities, as well as the education levels of the participants between two countries. We found that the highest inter-correlations were for the relationships between the psychosocial impact of diabetes (subscale 4; 0.768) and the need for special training (subscale 1) and between patient autonomy (subscale 5; 0.836) and the need for special training (subscale 1). These findings suggest that patients’ sense of wellbeing may be related to the degree of specialist training received by medical professionals. However, in the original DAS-3 the highest inter-correlation (0.63) was between the seriousness of diabetes and the need for tight control. Theoretical Framework Based on the results presented here, the C-DAS-3 is a validated tool that can be used in the Chinese population with type-2 diabetes. Our findings provide support for the construct validity and test-retest reliability of the C-DAS-3. The internal consistency of the scale was 0.813, which was validated by previously proposed theoretical frameworks [49,50]. The strength of the inter- correlations among the domains of five subscales suggests that the instrument measures related but separate domains of patients’ The C-DAS-3 test-retest reliability using intraclass correlation coefficients in a sample of 48 patients who completed the questionnaire on two occasions 2 to 4 weeks apart, was 0.82 (95% CI: 0.68–0.89). The coefficients between the two repeat tests among the same subscales ranged from 0.66 for subscale 5 (‘patient autonomy’) to 0.87 for subscale 4 (‘psychosocial impact of May 2014 \| Volume 9 \| Issue 5 \| e96473 PLOS ONE \| www.plosone.org May 2014 \| Volume 9 \| Issue 5 \| e96473 4 The Chinese Version of Diabetes Attitude Scale (C-DAS-3) Table 2. Descriptive statistics, comparison with the original DAS-3 and test-retest intraclass correlation coefficients for C-DAS-3. Scale name Number of items (N = 5961) Mean ± SD (N = 5961) Range (N = 5961) Cronbach’s alpha of C-DAS-3 (N = 5961) Cronbach’s alpha of DAS-3 (N = 1843) Test-retest Test (N = 48) Retest (N = 48) Intraclass Correlation (P-value) Subscale 1 5 4.3260.49 1.80,5.00 0.740 0.67 4.4360.45 4.3760.37 0.740 (,0.0001) Subscale 2 7 3.6260.46 1.43,5.00 0.706 0.80 2.8860.44 2.6860.35 0.657 (,0.0001) Subscale 3 7 3.5760.51 1.71,5.00 0.820 0.72 2.9360.54 2.8960.47 0.749 (,0.0001) Subscale 4 6 3.7960.53 1.67,5.00 0.848 0.65 3.3060.52 3.1960.59 0.865 (,0.0001) Subscale 5 8 3.6860.49 1.63,5.00 0.654 0.76 3.8060.47 3.7560.43 0.654 (,0.0001) Total scale 33 3.7660.30 2.45,4.76 0.813 3.4760.29 3.3760.26 0.706 (,0.0001) DAS 3 Th thi d i f di b t ttit d l PLOS ONE \| l Table 2. Descriptive statistics, comparison with the original DAS-3 and test-retest intraclass correlation coefficients for C-DAS-3. Psychosocial Impact, Training, and Patient Well-Being in Chi Psychosocial Impact, Training, and Patient Well-Being in China Theoretical Framework Scale name Number of items (N = 5961) Mean ± SD (N = 5961) Range (N = 5961) Cronbach’s alpha of C-DAS-3 (N = 5961) Cronbach’s alpha of DAS-3 (N = 1843) Test-retest Test (N = 48) Retest (N = 48) Intraclass Correlation (P-value) Subscale 1 5 4.3260.49 1.80,5.00 0.740 0.67 4.4360.45 4.3760.37 0.740 (,0.0001) Subscale 2 7 3.6260.46 1.43,5.00 0.706 0.80 2.8860.44 2.6860.35 0.657 (,0.0001) Subscale 3 7 3.5760.51 1.71,5.00 0.820 0.72 2.9360.54 2.8960.47 0.749 (,0.0001) Subscale 4 6 3.7960.53 1.67,5.00 0.848 0.65 3.3060.52 3.1960.59 0.865 (,0.0001) Subscale 5 8 3.6860.49 1.63,5.00 0.654 0.76 3.8060.47 3.7560.43 0.654 (,0.0001) Total scale 33 3.7660.30 2.45,4.76 0.813 3.4760.29 3.3760.26 0.706 (,0.0001) PLOS ONE \| www.plosone.org attitudes toward diabetes. Moreover, the test-retest intraclass correlation coefficients were high enough to support the stability of the C-DAS-3. Psychosocial Impact, Training, and Patient Well-Being in China In our study, we found correlations between the C-DAS score and age, education level, duration of diabetes, presence of complications, and accepted diabetes education. Patients with higher education level and those who received diabetes education have more serious attitude on diabetes. On the other hand, younger patients (usually with higher education degrees), and patients with complications and longer diabetes duration take diabetes more seriously. In China, attitudes towards diabetes have rarely been explored, though limited reports of attitudes towards diabetes complications have been reported. In a study of diabetic glaucoma in rural Chinese patients, Yan et al. [52] reported that misconceptions about the nature of the disease commonly results in poor adherence to routine examination schedules during asymptomatic periods. Furthermore, it was proposed that patient education by trained nurses should be implemented through home contact to improve patient compliance [52]. Similar results were May 2014 \| Volume 9 \| Issue 5 \| e96473 5 The Chinese Version of Diabetes Attitude Scale (C-DAS-3) Table 3. Correlations between C-DAS-3 and demographic/medical variables. Table 3. Correlations between C-DAS-3 and demographic/medical variables. Total scale of C-DAS-3 Pearson’s r Spearman Mann-Whitney U P-value Gender no number 0.101 Age 20.164 0.000 Diabetes duration 0.274 0.000 BMI 0.003 0.862 A1c 0.040 0.018 Education level 0.352 0.000 Complications 0.387 0.000 Diabetes Education 0.226 0.000 Insulin treatment 21.213 0.225 doi:10.1371/journal.pone.0096473.t003 doi:10.1371/journal.pone.0096473.t003 disease [54]. The attitudes of Chinese patients may be similarly improved through education programs. Before education pro- grams can be designed, however, there is an urgent need for better assessment of attitudes of type-2 diabetes patients, as provided in the current study. Thus, this research provides an essential first step to improving care for diabetes patients in China and potentially implementing education programs in the future. demonstrated in a wide cross-section of international diabetic patients by the Diabetes Attitudes, Wishes, and Needs (DAWN) study, which indicated that patient compliance with self-manage- ment behaviors, particular diet and exercise, was as low as 2.9% in Type 2 diabetic patients [53]. The findings of the current research combined with these previous indications highlights the need for both improved nursing training that includes specialization in diabetes care and patient education programs, particularly in under-served and rural regions of China. In conclusion, psychometric properties of a translated version of the C-DAS-3 demonstrated satisfactory validity and reliability and provided an effective measure for evaluating attitudes toward diabetes in a Chinese population with type-2 diabetes. Prior Presentation Data from this paper were presented, in part, at the 71st Scientific Sessions of the American Diabetes Association, 24–28 June 2011, San Diego, CA, USA. Only a low level of statistical evidence for a negative relationship between overall C-DAS-3 scores and HbA1c levels was found (r = 20.04). It is theoretically possible that the statistical signifi- cance of this correlation may have been due to the large sample size without signifying any clinical significance or importance. However, a previous study undertaken in Argentina study provided evidence to suggest that changing the attitudes of diabetic patients through reeducation contributed to improved care and quality of life and decreases the financial burden of the Limitations The study is to some degree limited by the relatively small populations used to pilot the questionnaire and identify areas that needed changing in the final version, and also by the relatively small population (48 out of 60 patients) available to evaluate the test-retest reliability, it means only 80% of the patients completed the second questionnaire, therefore, there is a selection bias. It would also be beneficial to compare the validity of this version of the C-DAS-3 in other provinces and regions of China. Psychosocial Impact, Training, and Patient Well-Being in China This assessment requires further validation in other populations of Chinese patents with diabetes. Author Contributions Conceived and designed the experiments: QL XG. Performed the experiments: QL XG LY TC CW LS ZS FZ XD JH HY YC. Analyzed the data: QL XG LY TC CW LS ZS FZ XD JH HY YC. Wrote the paper: QL. Acknowledgments We would like to express our gratitude to the participants for their participation. References 7. Polonsky WH, Fisher L, Earles J, Dudl RJ, Lees J, et al. (2005) Assessing psychosocial distress in diabetes: development of the diabetes distress scale. Diabetes Care 28: 626–631. 1. Peyrot M, Rubin RR, Lauritzen T, Skovlund SE, Snoek FJ, et al. (2006) Patient and provider perceptions of care for diabetes: results of the cross-national DAWN Study. Diabtologia 49: 279–288. y g 2. Donnelly MB, Anderson RM (1990) The role related attitudes of physicians, nurses, and dieticians in the treatment of diabetes. Medical Care 28: 175–179. 8. Fisher L, Glasgow RE, Mullan JT, Skaff MM, Polonsky WH (2008) Development of brief diabetes distress screening instrument. Ann Fam Med 6: 246–252. 3. Dziemidok P, Makara-Studzin´ska M, Jarosz MJ (2011) Diabetes and depression: a combination of civilization and life-style diseases is more than simple problem adding - literature review. Ann Agric Environ Med 18: 318–322. 9. Polonsky WH, Anderson BJ, Lohrer PA, Welch G, Jacobson AM, et al. (1995) Assessment of diabetes-related distress. Diabetes Care 18: 754–760. 10. Welch GW, Jacobson AM, Polonsky WH (1997) The Problem Areas in Diabetes scale: an evaluation of its clinical utility. Diabetes Care 20: 760–766. g g 4. Ismail K, Winkley K, Rabe-Hesketh S (2004) Systematic review and meta- analysis of randomised controlled trials of psychological interventions to improve glycaemic control in patients with type 2 diabetes. Lancet 363: 1589–1597. scale: an evaluation of its clinical utility. Diabetes Care 20: 760–766. 11. Anderson RM, Funnell MM, Fitzgerald JT, Marrero DG (2000) The Diabetes Empowerment Scale: a measure of psychosocial self-efficacy. Diabetes Care 23: 739–743. 5. Berk KA, Buijks H, Ozcan B, Van’t Spijker A, Busschbach JJ, et al. (2012) The prevention of weight regain in diabetes type 2 (POWER) study: the effectiveness of adding a combined psychological intervention to a very low calorie diet, design and pilot data of a randomized controlled trial. BMC Public Health 12: 1026–1039. 12. Anderson RM, Fitzgerald JT, Gruppen LD, Funnell MM, Oh MS (2003) The diabetes empowerment scale-short form (DES-SF). Diabetes Care 26: 1641– 1642. 13. Pandit AU, Bailey SC, Curtis LM, Seligman HK, Davis TC, et al. (2014) Disease-related distress, self-care and clinical outcomes among low-income patients with diabetes. J Epidemiol Community Health. In press. doi:10.1136/ jech-2013-203063. 6. Eccleston C, Palermo TM, Fisher E, Law E (2012) Psychological interventions for parents of children and adolescents with chronic illness. Cochrane Database Syst Rev doi:10.1002/14651858. 6. Eccleston C, Palermo TM, Fisher E, Law E (2012) Psychological interventions for parents of children and adolescents with chronic illness. Cochrane Database Syst Rev doi:10.1002/14651858. 13. Pandit AU, Bailey SC, Curtis LM, Seligman HK, Davis TC, et al. (2014) Disease-related distress, self-care and clinical outcomes among low-income patients with diabetes. J Epidemiol Community Health. In press. doi:10.1136/ jech-2013-203063. 11. Anderson RM, Funnell MM, Fitzgerald JT, Marrero DG (2000) The Diabetes Empowerment Scale: a measure of psychosocial self-efficacy. Diabetes Care 23: 739–743. 1. Peyrot M, Rubin RR, Lauritzen T, Skovlund SE, Snoek FJ, et al. (2006) Patient and provider perceptions of care for diabetes: results of the cross-national DAWN Study. Diabtologia 49: 279–288. 2. Donnelly MB, Anderson RM (1990) The role related attitudes of physicians, nurses, and dieticians in the treatment of diabetes. Medical Care 28: 175–179. The Chinese Version of Diabetes Attitude Scale (C-DAS-3) The Chinese Version of Diabetes Attitude Scale (C-DAS-3) 14. Baradaran HR, Mirghorbani SM, Javanbakht A, Yadollahi Z, Khamseh ME (2013) Diabetes distress and its association with depression in patients with type 2 diabetes in iran. Int J Prev Med 4: 580–584. 33. Mosnier-Pudar H, Hochberg G, Eschwege E, Halimi S, Virally ML, et al. (2010) How patients’ attitudes and opinions influence self-care behaviours in type 2 diabetes. Insights from the French DIABASIS Survey. Diabetes and Metab 36: 476–83. J 15. Fisher L, Glasgow RE, Strycker LA (2010) The relationship between diabetes distress and clinical depression with glycemic control among patients with type 2 diabetes. Diabetes Care 33: 1034–1036. 34. Peyrot M, Barnett AH, Meneghini LF, Schumm-Draeger PM (2012) Insulin adherence behaviours and barriers in the multinational Global Attitudes of Patients and Physicians in Insulin Therapy study. Diabet Med 29: 682–689. 16. Hendrieckx C, Halliday JA, Bowden JP, Colman PG, Cohen N, et al. (2014) Severe hypoglycaemia and its association with psychological well-being in Australian adults with type 1 diabetes attending specialist tertiary clinics. Diabetes Res Clin Prac. In press. doi:10.1016/j.diabres.12.005. 35. Anderson RM, Fitzgerald JT, Oh MS (1993) The relationship between diabetes- related attitudes and patients’ self-reported adherence. Diabetes Educator 19: 287–292. Diabetes Res Clin Prac. In press. doi:10.1016/j.diabres.12.005. 17. Reddy J, Wilhelm K, Campbell L (2013) Putting PAID to diabetes-related distress: the potential utility of the problem areas in diabetes (PAID) scale in patients with diabetes. Psychosomatics 54: 44–51. 36. Polly RK (1992) Diabetes health beliefs, self-care behaviors, and glycemic control among older adults with non-insulin-dependent diabetes mellitus. Diabetes Educator 18: 321–327. 18. Hayashino Y, Okamura S, Matsunaga S, Tsujii S, Ishii H, et al. (2012) The association between problem areas in diabetes scale scores and glycemic control is modified by types of diabetes therapy: diabetes distress and care registry in Tenri (DDCRT 2). Diabetes Res Clin Pract 97: 405–410. 37. Chen A, Huang Z, Wan X, Deng W, Wu J, et al. (2012) Attitudes toward diabetes affect maintenance of drug-free remission in patients with newly diagnosed type 2 diabetes after short-term continuous subcutaneous insulin infusion treatment. Diabetes Care 35: 474–481. 19. Leyva B, Zagarins SE, Allen NA, Welch G (2011) The relative impact of diabetes distress vs depression on glycemic control in Hispanic patients following a diabetes self-management education intervention. Ethn Dis 21: 322–327. 38. The Chinese Version of Diabetes Attitude Scale (C-DAS-3) Mahjouri MY, Arzaghi SM, Qorbani M, Nasli-Esfahani E, Larijani B (2011) Evaluation of psychometric properties of the third version of the Iranian Diabetes Attitude Scale (IR-DAS-3). Iranian Journal of Diabetes and Lipid Disorders 10: 1–6. 20. Anderson RM, Funnell MM, Aikens JE, Krein SL, Fitzgerald JT, et al. (2009) Evaluating the Efficacy of an Empowerment-Based Self-Management Consul- tant Intervention: Results of a Two-Year Randomized Controlled Trial. Ther Patient Educ 1: 3–11. 39. Saddinne JB, Cadwell B, Gregg EB, Engelgau MM, Vinicor F, et al. (2006) Improvements in diabetes processes of care and intermediate outcomes: United States, 1988-2002. Ann Intern Med 144: 465–474. , 40. Alberti G (2002) The DAWN (Diabetes Attitudes, Wishes, and Needs) study. Pract Diabetes Int 19: 22–24. 21. Tol A, Baghbanian A, Mohebbi B, Shojaeizadeh D, Azam K, et al. (2013) Empowerment assessment and influential factors among patients with type 2 diabetes. Diabetes Metab Disord 12: 6. Available: http://www.jdmdonline. com/content/12/1/6. Accessed 2014 March 11. 41. Mani N, Caiola E, Fortuna RJ (2011) The influence of social networks on patients’ attitudes toward type II diabetes. J Community Health 36: 728–732. 22. Samoocha D, Bruinvels DJ, Elbers NA, Anema JR, van der Beek AJ (2010) Effectiveness of web-based interventions on patient empowerment: a systematic review and meta-analysis. Med Internet Res 12: e23. Available: http://www. jmir.org/2010/2/e23/. Accessed 2014 March 11. 42. Shah VN, Kamdar PK, Shah N (2009) Assessing the knowledge, attitudes and practice of type 2 diabetes among patients of Saurashtra region, Gujarat. Int J Diabetes Dev Ctries 29: 118–122. J 43. Mazzuca SA, Moorman NH, Wheeler ML (1986) The diabetes education study: a controlled trial of the effects of diabetes education. Diabetes Care 9:1–10. j g 23. Yang Q, Liu XQ (2010) The reliability and validity of Chinese version of Diabetes Distress Scale. Journal of Nursing (China) 17: 8–10. 44. Anderson RM, Donnelly MB, Gressard CP, Dedrick RF (1989) Development of diabetes attitude scale for health-care professionals. Diabetes Care 12: 120–127. 24. Ting RZ, Nan H, Yu MW, Kong AP, Ma RC, et al. (2011) Diabetes-related distress and physical and psychological health in Chinese type 2 diabetic patients. Diabetes Care 34: 1094–1096. 45. Anderson RM, Fitzgerald JT, Funnell MM, Gruppen LD (1998) The third version of the Diabetes Attitude Scale. Diabetes Care 21: 1403–1407. p 25. Hsu HC, Chang YH, Lee PJ, Chen SY, Hsieh CH, et al. The Chinese Version of Diabetes Attitude Scale (C-DAS-3) (2013) Developing and psychometric testing of a short-form problem areas in diabetes scale in Chinese patients. J Nurs Res 21: 212–218. 46. Brislin R (1986) The wording and translation of research instruments. In: Loner W, Berry J, editors. Field Methods in Cross-Cultural Research Newbury Park, CA, pp. 137–164. 26. Hu BB, Lou QQ, Tian Y, Zhang QW, Zhu JY (2011) Empowerment and its influencing factors of diabetes inpatients. Journal of Nursing (China) 46: 225– 228. 47. Zhou X, Lou Q, Zhang X, Zhu W (2011) Survey on attitude and behavior of patients with diabetes in self-management education. Nursing Rehab 10: 947– 949. 27. Shiu AT, Wong RY, Thompson DR (2003) Development of a reliable and valid Chinese version of the diabetes empowerment scale. Diabetes Care 26: 2817– 2821. 48. Jackson CJ, Furnham A (2000) Designing and analysing questionnaires and surveys: a manual for health professionals and administrators, Whurr Publishers, London. 28. Shiu AT, Choi KC, Wong RY (2012) The Chinese version of the Diabetes Empowerment Scale-short form. Patient Educ Couns 87: 258–260. 49. Indrayan A, Sarmukaddam A (2001) Medical statistics, Marcel Dekker, New York. 50. MacDougall JD, Wenger HA, Green HJ, CASS (1991) Physiological testing the high-performance athlete, Human Kinetics Books, Champaing, IL. 29. Kong L, Cai Y, Mei G, Gu R, Zhang X, et al. (2013) Psychological status and diabetes-related distress of Chinese type 1 diabetes patients in Jiangsu province, China. J Biomed Res 27: 380–385. 51. Funnell MM, Brown TL, Childs BP, Haas LB, Hosey GM, et al. (2008) National standards for diabetes self-management education. Diabetes Care 31: S97–S104. J 30. Zhang J, Xu CP, Wu HX, Xue XJ, Xu ZJ, et al. (2013) Comparative study of the influence of diabetes distress and depression on treatment adherence in Chinese patients with type 2 diabetes: a cross-sectional survey in the People’s Republic of China. Neuropsychiatr Dis Treat 9: 1289–1294. 52. Yan X, Liu T, Gruber L, He M, Congdon N (2012) Attitudes of physicians, patients, and village health workers toward glaucoma and diabetic retinopathy in rural China: a focus group study. Arch Ophthalmol 30: 761–770. p yp y Republic of China. Neuropsychiatr Dis Treat 9: 1289–1294. rural China: a focus group study. Arch Ophthalmol 30: 761–770. p p y 31. Chen B, Zhang X, Xu X, Lv X, Yao L, et al. (2013) Diabetes education improves depressive state in newly diagnosed patients with type 2 diabetes. References May 2014 \| Volume 9 \| Issue 5 \| e96473 PLOS ONE \| www.plosone.org 6 The Chinese Version of Diabetes Attitude Scale (C-DAS-3) Pak J Med Sci 29: 1147–1152. g p 53. Funnell MM (2006) The Diabe Clinical Diabetes 24: 154–155. 53. Funnell MM (2006) The Diabetes Attitudes, Wishes, and Needs (DAWN) Study. Clinical Diabetes 24: 154–155. 54. Gagliardino JJ, Gonzalez C, Caporale JE (2007) The diabetes-related attitudes of health care professionals and persons with diabetes in Argentina. Revista Panamericana de Salud Publica 22: 304–307. J 32. Ajzen I, Fishbein M (1980) Understanding Attitudes and Predicting Social Behavior. Englewood Cliffs, NJ: Prentice Hall. May 2014 \| Volume 9 \| Issue 5 \| e96473 PLOS ONE \| www.plosone.org 7
https://openalex.org/W4386750643	https://pubs.rsc.org/en/content/articlepdf/2023/cc/d3cc90305b	English	null	Inside front cover	Chemical communications	2,023	cc-by	53	ChemComm Chemical Communications rsc.li/chemcomm Volume 59 Number 74 21 September 2023 Pages 11007–11150 Volume 59 Number 74 21 September 2023 Pages 11007–11150 Chemical Communications rsc.li/chemcomm Chemical Communications rsc.li/chemcomm ISSN 1359-7345 ChemComm Chemical Communications rsc.li/chemcomm COMMUNICATION Yasushi Sekine et al . Equilibrium unconstrained low-temperature CO 2 conversion on doped gallium oxides by chemical looping
https://openalex.org/W4327593645	https://journals.e-palli.com/home/index.php/ajec/article/download/1243/573	English	null	Artificial Neural Network for Forecasting Monsoon Rainfall of South-West Region in Bangladesh	American journal of environment and climate	2,023	cc-by	7,885	Page 1 Page 1 mm/year (Ahsan et al., 2010). Weather forecasting is the dispensation of scientific knowledge for the practical purpose of predicting the condition of the atmosphere for a given location and time. Among them, rainfall is a significant and complicated weather phenomenon whose prediction boosts the development of water resources, mainly in Bangladesh’s climate change regions. Since climate change affects the rainfall pattern, it causes floods, droughts, etc. Moreover, rainfall prediction with a good model is inevitable to extrapolate the brunt. In Bangladesh, the highest temperature is seen in the South-West, and the lowest is in the northeast. The average temperature in the cool season varies from 13oto 26o C; in the hot season, it varies from 25ο to 31ο C (Climate Change Profile: Bangladesh, 2018). Rainfall in Bangladesh also differs with location and season, where the central west receives less than 1400mm per year, and the northeast and South-East receive over 3000mm per year. Moreover, Bangladesh is a developing country where most people earn their livelihood from agriculture or shrimp farming, which makes much foreign currency. Moreover, Sundarban and Kuakata Sea Beach are also Among them, rainfall is a significant and complicated weather phenomenon whose prediction boosts the development of water resources, mainly in Bangladesh’s climate change regions. Since climate change affects the rainfall pattern, it causes floods, droughts, etc. Moreover, rainfall prediction with a good model is inevitable to extrapolate the brunt. As Bangladesh is in the climatology of the Asian monsoon system, hence the climate of this regime is differentiated by a seasonal variation of surface wind and a remarkable seasonality of rainfall and stays from June to mid-October (Ahmed and Kim, 2003; Shahid, 2010). The previous record shows that most of the natural calamities and rainfall occurred during this period. In the South-East part of the country, the rain is shown in a changing pattern, and monsoon rainfall is not suggested based on the overall evidence (Rahman et al., 1997). The next highest rainfall occurs in the southeastern region, and the following heights are in the northeastern part of Bangladesh (Ahsan et al., 2010). A unimodal pattern has been seen in the mean monthly rainfall in Bangladesh, with high rainfall in the monsoon season, with the highest in July, and low rainfall between December-February with the weakest in January (Ahsan et al., 2010). 1 Mathematics Discipline, Khulna University, Khulna-9208, Bangladesh 2 Bangladesh Meteorological Department, Ministry of Defence, Bangladesh * Corresponding author’s e-mail: munnujahan@math.ku.ac.bd Keywords Artificial Neural Network, Monsoon Season, Prediction, South-West Region Artificial Neural Network, Monsoon Season, Prediction, South-West Region ABSTRACT Article Information Received: January 27, 2023 Accepted: March 02, 2023 Published: March 16, 2023 Article Information Received: January 27, 2023 Accepted: March 02, 2023 Published: March 16, 2023 Changing patterns of climate factors have become a point of discussion in recent times worldwide. Several aspects of an individual’s prosperity, like communal, financial, and ecological increment, were impacted directly or circuitously by climate change. Moreover, the Bangladeshi people’s life is extremely affected by heavy rainfall because of its geographical structure, especially in the South-West region. Hence, this paper has experimented with the monthly average monsoon data of average temperature, wind speed, relative humidity, mean sea-level pressure, cloud cover, and rainfall from 1981-2018 and predicted the precipitation of 9 meteorological stations from 2019-2028 of the South-West part of Bangladesh. The monthly average monsoon rainfall strongly correlated with relative humidity, mean sea-level pressure, and cloud cover among all the mentioned weather variables. An artificial neural network (ANN) model was formulated with a gradient descent algorithm to predict the rainfall. R2 value was also measured to see the accuracy of the model. Thereafter, the nine stations of the given region have the following order of average monsoon rainfall:Khepupara(15.22mm)>Potuakhali(14.01mm)>Bhola(11.36mm)>Barishal(10.68mm)> Mongla(10.25mm)>Khulna(9.33mm)>Satkhira(9.00mm)>Faridpur(8.67mm)>Jashore(8.64mm). The predicted and real rainfall patterns showed the same escalating or plummeting trends for each station, which justified the ANN model for predicting the monthly average monsoon rainfall of the South-West region in Bangladesh. Such a rainfall prediction can assist people of this region to be more equipped for adverse heavy rain, saving lives and decreasing infrastructure loss during the monsoon season. Page 1 American Journal of Environment and Climate (AJEC) Volume 2 Issue 1, Year 2023 ISSN: 2832-403X (Online) DOI: https://doi.org/10.54536/ajec.v2i1.1243 https://journals.e-palli.com/home/index.php/ajec Volume 2 Issue 1, Year 2023 ISSN: 2832-403X (Online) American Journal of Environment and Climate (AJEC) mm/year (Ahsan et al., 2010). Bangladesh’s mean summer monsoon rainfall is1769.14mm, and the country’s average monsoon rainfall is decreasing by -0.53 Figure 1: The location of the studied weather stations Figure 1: The location of the studied weather stations Am. J. Environ. Clim. 2(1) 11-23, 2023 situated here. Between 1991-2000, about 93 disasters occurred here, caused in 200000 deaths and USD 5.90billion in damages to agriculture and infrastructure (Shaibur et al., 2017). Among them, most of the damages occurred in the South-West region. In figure 1, the topographic areas of the nine meteorological stations of the South-West region have been represented. monsoon and post-monsoon seasons, it has increased all over the country. Ahmed and Kim (2003) have directed research with statistics to analyze Bangladesh’s daily pattern of summer monsoon rainfall. Another researcher, Abhishek et al. (2012), also analyzed a weather forecasting model using an Artificial Neural Network with the data series of some weather variables for ten years (1999-2009) at the station Toronto Lester B. Pearson Int’l A, Ontario, Canada. Valipour et al. (2013) have also researched to forecast the Dez dam reservoir’s monthly inflow and compared ARMA, ARIMA, and the auto-regression artificial neural network models. Navid and Niloy (2018) have also driven experiments predicting rainfall in Bangladesh using Multiple Linear Regression (MLR). Another researcher Bilgili (2010), has conducted research on soil temperature prediction by linear regression, non- linear regression, and artificial neural network (ANN) models. The outcome of this research has stated that using the ANN model is better than the other two models for soil temperature prediction. All these research papers have measured the accuracy of the predicted models on rainfall. However, they have not calculated the forecasted values, which encouraged us to work on this issue. The main objective of this paper is to apply the ANN model for rainfall prediction, forecast it, and implement them towards the removal of sufferings of people living in the South-West region of Bangladesh. LITERATURE REVIEW To yield accurate results, some statistical methods have been developed to forecast rainfall and other meteorological variables. Some researchers have developed rainfall forecasting models by autoregressive integrated moving averages (ARIMA), simple method regression analysis (SRA), exponential smoothing method (ES), etc. Several studies have reported that these methods are still inaccurate in forecasting rainfall because of the non-linear dataset (Haviluddin and Alfred, 2014; Shrivastava et al., 2012). Nevertheless, in some incidences, the statistical models also give accurate results of rainfall prediction (Farajzadeh et al., 2014). MATERIALS AND METHODS The daily average data of temperature, wind speed, humidity, mean sea-level pressure, cloud cover, and rainfall of 9 stations: Satkhira, Khulna, Jashore, Barishal, Bhola, Faridpur, Khepupara, Patuakhali, and Mongla from the period 1981-2018 (June-October) were collected from the Bangladesh Meteorological Department (BMD). In addition to the upliftment of computing technology, several authors have analyzed many models to study rainfall forecasting. Ara et al. (2005) have driven research on surface dry bulb temperature (DBT) and its trend in Bangladesh. This research has revealed that in the pre- monsoon season, the average DBT has decreased; in the Table 1: Stations and Meteorological Data Period General Location Station Name Period of Record Used Period of Missing Data Number of Years of Availability of Data SW Bangladesh Satkhira 1981-2018 1989, 1999 36 Khulna 1981-2018 1989, 1999 36 Jashore 1981-2018 1989, 1999 36 Barishal 1981-2018 1989, 1999 36 Bhola 1981-2018 1989, 1999 36 Faridpur 1981-2018 1989, 1999 36 Khepupara 1981-2018 1989, 1999 36 Potuakhali 1981-2018 1989, 1999 36 Mongla 1991-2018 - 28 Page 12 Table 1: Stations and Meteorological Data Period General Location Station Name Period of Record Used Period of Missing Data Number of Years of Availability of Data SW Bangladesh Satkhira 1981-2018 1989, 1999 36 Khulna 1981-2018 1989, 1999 36 Jashore 1981-2018 1989, 1999 36 Barishal 1981-2018 1989, 1999 36 Bhola 1981-2018 1989, 1999 36 Faridpur 1981-2018 1989, 1999 36 Khepupara 1981-2018 1989, 1999 36 Potuakhali 1981-2018 1989, 1999 36 Mongla 1991-2018 - 28 It is significant to mention that data for 1989 and 1999 were not available in Satkhira, Jashore, Barishal, Bhola and Faridpur stations. To make a comparison, we also didn’t consider the data of these two years for the other stations. The Mongla station was established in 1989, and the data availability was from 1991. So, we have used data from 1991 to 2018 for this study for Mongla station. Moreover, some steps have been taken to continue further analysis, as shown in figure 2. Some data remained missing, and we used the series method to fill up these missing values to process the data. The outliers have been detected by Mahalanobis distance Figure 2: Processes of methodology. Am. J. Environ. Clim. 2(1) 11-23, 2023 Am. J. Environ. Clim. 2(1) 11-23, 2023 and deleted from the data set. The Mahalanobis distance is applied to trace the outliers in statistical analysis. It is the distance between a distribution and a point in multivariant space. Once the outliers had been deleted from the data set, daily average data were obtained monthly. which stabilizes the boundary layer. Hence, precipitation decreases and vice-versa. The average temperature comes next to the order of the correlation coefficients in these stations. It showed a positive relationship with the monthly average temperature because the rise of the average temperature of the earth’s surface causes more evaporation and increases overall rainfall. Correlation between Rainfall and Climate Parameters Table 2. shows that the order of the spearman’s rank correlation coefficient between monthly average rainfall and meteorological variables remained the same for the stations Satkhira, Khulna, Jashore, Barishal, Bhola, and Faridpur. Spearman rank correlation is performed to measure how these data are correlated with the monthly average rainfall. Since our data are not normally distributed, we used Spearman rank correlation in this study. In Khepupara, the order of spearman’s rank correlation coefficient between monthly average rainfall and meteorological variables is Humidity>Mean Sea-level Pressur>Wind Speed>Cloud Cover>Average Temperature. Here, all the variables have a positive relationship with the monthly average precipitation except mean sea-level pressure. In Potuakhali, the order of spearman’s rank correlation coefficient between monthly average rainfall and meteorological variables is CloudCover>Mean Sea-Level pressure>Humidity>Wind Speed> Average Temperature. Here all the variables showed a positive relationship with the monthly average precipitation except mean sea-level pressure and average temperature. Since it is the entrance of the beach of the Kuakata, the rain has a cooling effect on the sea surface by decreasing the near- surface air temperature (Zuidema, 2007). In Mongla, the order of spearman’s rank correlation coefficient between monthly average rainfall and meteorological variables is Humidity>Wind Speed> Cloud Cover> Average Temperature>Mean Sea-Level Pressure. Here, all the variables have a positive relationship with the monthly average precipitation. The positive correlation with the monthly average precipitation suggests that rainfall is related to a westerly airflow on the Mongla (Rogers and Dowla, 1994). Where ρis the Spearman rank correlation coefficient,dis the difference between two ranks of the different observations, and n is the number of data. Where ρis the Spearman rank correlation coefficient,dis the difference between two ranks of the different observations, and n is the number of data. Am. J. Environ. Clim. 2(1) 11-23, 2023 The order is Cloud Cover>Humidity>Mean Sea-Level Pressure>Wind Speed>Average Temperature. The correlation between cloud cover and average rainfall is vital because when the hot air of the atmosphere evaporates the moisture (water) by the divergence and convergence of air, it slowly cools down and is condensed. As much as the clouds are thickened, the friction between them occurs, as well as the rainfall. Humidity is next to the cloud cover in this correlation part because as much as the humidity, the more water vapor, the more significant the precipitation. It showed less impact on rainfall than cloud cover in these stations. Next comes the mean sea- level pressure, which exposed a slightly strong negative correlation with the monthly average rain, i.e., lower values in the mean sea-level pressure in these stations are associated with an increase in rainfall and vice- versa. After mean sea-level pressure into the order of correlation coefficients, wind speed comes. It displayed a positive, slightly weak relation with the monthly average precipitation, i.e., higher values in the wind speed in these stations are accompanied by an increase in rainfall and vice-versa. Lower wind speeds prefer less evaporation, It is eminent that rainfall is related to several meteorological variables. The above-mentioned meteorological variables are chosen for this research analysis because of their unique relationship with monthly average monsoon rainfall. MATERIALS AND METHODS Page 12 Table 1: Stations and Meteorological Data Period General Location Station Name Period of Record Used Period of Missing Data Number of Years of Availability of Data SW Bangladesh Satkhira 1981-2018 1989, 1999 36 Khulna 1981-2018 1989, 1999 36 Jashore 1981-2018 1989, 1999 36 Barishal 1981-2018 1989, 1999 36 Bhola 1981-2018 1989, 1999 36 Faridpur 1981-2018 1989, 1999 36 Khepupara 1981-2018 1989, 1999 36 Potuakhali 1981-2018 1989, 1999 36 Mongla 1991-2018 - 28 It is significant to mention that data for 1989 and 1999 were not available in Satkhira, Jashore, Barishal, Bhola and Faridpur stations. To make a comparison, we also didn’t consider the data of these two years for the other stations. The Mongla station was established in 1989, and the data availability was from 1991. So, we have used data from 1991 to 2018 for this study for Mongla station. Moreover, some steps have been taken to continue further analysis, as shown in figure 2. Some data remained missing, and we used the series method to fill up these missing values to process the data. The outliers have been detected by Mahalanobis distance Figure 2: Processes of methodology. Table 1: Stations and Meteorological Data Period Some data remained missing, and we used the series method to fill up these missing values to process the data. The outliers have been detected by Mahalanobis distance https://journals.e-palli.com/home/index.php/ajec Artificial Neural Network (ANN) The ANN is an engineering notion of learning in the field of artificial intelligence, which is similar to the human brain, shown in figure 3 (Gogoi, 2017). Here, the input layer is compared with the dendrites of the human brain as it receives the signals. Figure 3: Neuron vs. ANN Page 13 https://journals.e-palli.com/home/index.php/ajec Figure 3: Neuron vs. ANN Figure 3: Neuron vs. ANN https://journals.e-palli.com/home/index.php/ajec Am. J. Environ. Clim. 2(1) 11-23, 2023 After that, these signals are moved through the neural network. Each neuron transmits the information to other neurons to manage the issue. It is constructed with many elements, called neurons which distribute processing information right away. Yet the ANN can be trained with many processes along with Backpropagation, Perceptron, Delta, and Self-Organizing Map (SOM) (Shrivastava et al., 2012; Farajzadeh et al., 2014). Hence, this research paper proposed a gradient descent algorithm to predict rainfall to acquire more accurate outcomes. Table 2: Spearman’s rank correlation coefficient between monthly average rainfall and meteorological variables of the study areas Meteorological Station Average Temperature Wind Speed Humidity Mean Sea-Level Pressure Cloud Cover Satkhira 0.078 0.152 0.481 -0.446 0.578 Khulna 0.096 0.211 0.533 -0.468 0.628 Jashore 0.136 0.413 0.478 -0.500 0.545 Barishal 0.068 0.434 0.562 -0.557 0.658 Bhola 0.04 0.427 0.57 -0.557 0.676 Faridpur 0.112 0.37 0.665 -0.501 0.668 Khepupara 0.045 0.373 0.587 -0.563 0.089 Potuakhali -0.023 0.283 0.533 -0.536 0.703 Mongla 0.558 0.671 0.825 0.317 0.577 nk correlation coefficient between monthly average rainfall and meteorological variables of A three-layer neural network has input, hidden, and output layers. Each neuron in every layer is associated with a neuron of the neighboring layer with several weights. Each neuron gets gestures from the neurons of the former layer without the input layer. An output signal is then generated by transiting the summed signal through an activation function (Maqsood et al., 2015). Ximin is the minimum value. By normalization, the training data has improved. The ANN was run into the MATLAB software, and the number of hidden layers with hidden neurons from 4 to 10 was taken. Testing data tested the model. The gradient descent algorithm was used to train the data set, and the logistic sigmoid function was used as an activation function for both the hidden and output layers. In the ANN, different layers may execute different works. But the main objective of the ANN is to solve a problem as the human brain does. Artificial Neural Network (ANN) The initial processing component of an ANN is a neuron. This fundamental processing element is mathematically described in the undermentioned equations (Haykin, 1994); ANN architecture of Satkhira, Khulna, Jashore, Barishal, Bhola, Faridpur, Khepupara, Patuakhali, and Mongla districts used in this research paper is shown in figure 4, where it includes an input layer with five neurons, one hidden layer with 8,7,8,7,8,8,9,8,5 neurons respectively, and an output layer with one neuron. After taking the learning rate (lr)0.01, the best fit result of the monthly average rainfall of the monsoon season was obtained after 3500, 3000, 2500, 2000, 1000, 1000, 2000, 2500, and 1500 epochs for Satkhira, Jashore, Khulna, Barishal, Bhola, Faridpur, Khepupara, Potuakhali and Mongla stations, respectively. Where n is the number of input nodes, X is the signal, w is the weights,θis the bias of the hidden node. The nodes in the hidden layers receive a signal (X) with the weights (w) and calculate a weighted sum (v). Then, this weighted sum passes through an activation function which indicates which nodes are activated. Then the activated nodes in the hidden layer pass the signal to each node of the output layer like the hidden layer, and finally, the estimated outputs (Y) are received. In this research paper, we trained the ANN model with more than one hidden layer. As a result, fewer errors in the prediction model were found, but it overfitted the model. Hence, one hidden layer was taken for the ANN model of monthly average monsoon rainfall. The R2value has been performed on the tested data to see the model’s accuracy. The formula of R2 is Page 14 Table 3: R2 value of prediction results Table 3: R2 value of prediction results shows the actual and ANN predicted rainfall, where this graph led to the highest rainfall in 2019, with4.46mm. This amount will plummet by 0.04-0.05mm for the rest of the years. After this downward trend in the predicted average monsoon rainfall, the lowest amount will be seen in 2028, around4.06mm. be in 2019, with around 6.26mm. This highest rainfall will fall slightly by 0.03-0.09mm up to 2028. Table 6 has represented the predicted rainfall of Satkhira station for each month with their average. In August, the average monsoon rainfall will be highest for all the mentioned predictable years except 2019, 2020, and 2021, whereas the lowest amount will be seen in June, except 2019. Moreover, the lowest precipitation will be around 7.07mm in 2019, which will minimally escalate by 0.55mm up to 2024. The amount of average monsoon rainfall of Khulna station has been shown in detail in table 4. The highest monsoon precipitation will be seen in September 2019, with 8.62mm, in August 2020 and 2021, with 8.80mm and 9.27mm, respectively, and in July for the rest of the years. These pieces of information are mentioned in table 4. J p In figure 5(d), the fluctuation of actual rainfall and predicted rainfall has been seen. It is manifested that the predicted monsoon average rainfall at Barishal station will have a decreasing trend. This amount will fall by 0.17mm from 2019 to 2024. Table 7 also shows that the highest monsoon average precipitation will be on July 2019, with around6.52mm. The most striking feature is that this amount will remain almost the same in August and October.l p In table 5, the data of the ANN predicted rain of Jashore station has been shown clearly. This table has told us that the highest rainfall would be in July and the lowest in October for all the predicted years in this paper. However, in some years, it can fluctuate; this amount should not be considered because of its very few changes. Among the other monsoon months, the precipitation will soar by an amount ranging from 0.22-0.88mm from 2019-2028, except September. Approximately 0.22mm of rain will be increased from 2019 to 2028. In Bhola station, The actual rainfall and predicted rainfall have been shown in figure 5(c). RESULTS AND DISCUSSIONS Between 2019-2028, the monsoon average rainfall will be seen, with an a of approximately 5.74mm in 2028, while the highe Table 4: Predicted rainfall (mm) of Khulna station 2019 2020 2021 2022 June 8.02 7.63 7.26 6.97 July 6.84 7.77 8.59 9.26 Aug 8.17 8.80 9.27 9.57 Sept 8.26 8.14 7.97 7.79 Oct 4.09 4.07 4.06 4.05 Avg 7.07 7.28 7.43 7.53 Table 5: Predicted rainfall (mm) of Jashore station 2019 2020 2021 2022 June 3.36 3.35 3.35 3.36 July 6.03 6.04 6.06 6.07 Aug 5.04 4.89 4.75 4.61 Sept 5.09 4.97 4.86 4.74 Oct 2.78 2.80 2.81 2.82 Avg 4.46 4.41 4.37 4.32 Khepupara( 8 2 % ) > Potuakhali(92%) > Bhola(93%) Barishal(89%)>Khulna(89%)>Satkhira(88%) Faridpur(92%)> Jashore(89%).i calculating the average values of the other independent weather variables from their changing pattern. A massive fluctuation in rainfall has been seen in each of mentioned stations. In Khulna station, the average highest monsoon rainfall of about 7.62mm will be seen in 2024, which will remain stable in 2024. After that, this value will slightly decrease by approximately 0.13mm in 2028. Figure 5(b) In figure 5 (a, b, c, d, e, f, g, h, i), each South-West region meteorological station’s predicted and actual monthly average monsoon rainfall has been shown. ANN model predicted the precipitation of these stations by https://journals.e-palli.com/home/index.php/ajec RESULTS AND DISCUSSIONS Where yi is the predicted rainfall, yi the actual rainfall, and y- the mean of the actual rain. The Mongla Station showed a R2value of 79%. The other eight stations have the following order of monthly average monsoon In the ANN technique, the input data were normalized in equation 6 in the range of [0,1].XN is the normalized value, Xi is the actual value, Ximax is the maximal value, and https://journals.e-palli.com/home/index.php/ajec Am. J. Environ. Clim. 2(1) 11-23, 2023 Am. J. Envi Khepupara( 8 2 % ) > Potuakhali(92%) > Bhola Barishal(89%)>Khulna(89%)>Satkhira( Faridpur(92%)> Jashore(89%). In figure 5 (a, b, c, d, e, f, g, h, i), each South-West meteorological station’s predicted and actual m average monsoon rainfall has been shown. model predicted the precipitation of these statio Table 3: R2 value of prediction results Stations Satkhira Khulna Jashore Barish ANN 0.8768 0.8916 0.8923 0.886 shows the actual and ANN predicted rainfall, whe graph led to the highest rainfall in 2019, with4.4 This amount will plummet by 0.04-0.05mm for th of the years. After this downward trend in the pre average monsoon rainfall, the lowest amount will b in 2028, around4.06mm. Moreover, the lowest precipitation will be around 7 in 2019, which will minimally escalate by 0.55mm 2024. The amount of average monsoon rainfall of K station has been shown in detail in table 4. The h monsoon precipitation will be seen in September with 8.62mm, in August 2020 and 2021, with 8.80m 9.27mm, respectively, and in July for the rest of the These pieces of information are mentioned in table In table 5, the data of the ANN predicted rain of J station has been shown clearly. This table has told the highest rainfall would be in July and the low October for all the predicted years in this paper. The actual rainfall and predicted rainfall have shown in figure 5(c). Table 3: R2 value of prediction results Between 2019-2028, the lowest monsoon average rainfall will be seen, with an amount of approximately 5.74mm in 2028, while the highest will Table 4: Predicted rainfall (mm) of Khulna station 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 June 8.02 7.63 7.26 6.97 6.77 6.65 6.60 6.60 6.64 6.70 July 6.84 7.77 8.59 9.26 9.80 10.19 10.44 10.57 10.58 10.48 Aug 8.17 8.80 9.27 9.57 9.72 9.73 9.62 9.43 9.18 8.92 Sept 8.26 8.14 7.97 7.79 7.64 7.50 7.41 7.34 7.31 7.29 Oct 4.09 4.07 4.06 4.05 4.04 4.04 4.04 4.05 4.06 4.08 Avg 7.07 7.28 7.43 7.53 7.59 7.62 7.62 7.60 7.55 7.49 Table 5: Predicted rainfall (mm) of Jashore station 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 June 3.36 3.35 3.35 3.36 3.37 3.39 3.40 3.42 3.45 3.47 July 6.03 6.04 6.06 6.07 6.07 6.07 6.07 6.05 6.04 6.02 Aug 5.04 4.89 4.75 4.61 4.47 4.34 4.22 4.10 3.98 3.87 Sept 5.09 4.97 4.86 4.74 4.62 4.50 4.38 4.26 4.13 4.01 Oct 2.78 2.80 2.81 2.82 2.84 2.85 2.86 2.88 2.89 2.89 Avg 4.46 4.41 4.37 4.32 4.28 4.23 4.19 4.15 4.10 4.06 https://journals.e-palli.com/home/index.php/ajec https://journals.e-palli.com/home/index.php/ajec Am. J. Environ. Clim. 2(1) 11-23, 2023 Am. J. Environ. Clim. 2(1) 11-23, 2023 Am. J. Environ. Clim. 2(1) 11-23, 2023 the average precipitation will decrease in all the monsoon months except October, shown in figure 5(e). Around 0.23mm of rainfall will rise, comparing the rain of 2019 and 2028. The maximum decrease of precipitation will be seen in June and July by1.8mm, among the years of the prediction. These pieces of information have been given more clearly in table 8. In this table, the forecasted rainfall has been provided with its approximate values Page 16 https://journals.e-palli.com/home/index.php/ajec Figure 4: ANN architecture of 9 stations Figure 4: ANN architecture of 9 stations Figure 4: ANN architecture of 9 stations Page 16 https://journals.e-palli.com/home/index.php/ajec https://journals.e-palli.com/home/index.php/ajec Am. J. Environ. Clim. 2(1) 11-23, 2023 Figure 5: ANN predicted rainfall of Khulna (a), Jashore (b), Satkhira (c), Barishal (d), Bhola (e), Faridpur (f), Khepupara (g), and Potuakhali (h) Figure 5: ANN predicted rainfall of Khulna (a), Jashore (b), Satkhira (c), Barishal (d), Bhola (e), Faridpur (f), Khepupara (g), and Potuakhali (h) Table 6: Predicted rainfall (mm) of Satkhira station Table 6: Predicted rainfall (mm) of Satkhira station 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 June 3.39 2.94 2.53 2.15 1.8 1.49 1.22 0.98 0.78 0.62 July 8.34 8.2 8.08 7.95 7.84 7.73 7.64 7.57 7.53 7.51 Aug 8.27 8.16 8.07 7.98 7.92 7.87 7.84 7.83 7.84 7.87 Sept 7.95 7.88 7.82 7.77 7.72 7.68 7.64 7.61 7.56 7.52 Oct 3.37 3.67 3.97 4.26 4.52 4.74 4.91 5.04 5.14 5.21 Avg 6.26 6.17 6.09 6.02 5.96 5.9 5.85 5.81 5.77 5.74 The predicted rainfall of Faridpur station has been seen in an increasing trend between the years 2019 and 2021 in June from 5.94mm to 5.98mm. These data are shown in table 9, whose graphical representation is also given in The predicted rainfall of Faridpur station has been seen in an increasing trend between the years 2019 and 2021 in June from 5.94mm to 5.98mm. These data are shown in table 9, whose graphical representation is also given in https://journals.e-palli.com/home/index.php/ajec https://journals.e-palli.com/home/index.php/ajec Am. J. Environ. Clim. Am. J. Environ. Clim. 2(1) 11-23, 2023 2(1) 11-23, 2023 Table 7: Predicted rainfall (mm) of Barishal station 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 June 5.24 5.18 5.1 5.02 4.92 4.82 4.71 4.59 4.48 4.36 July 6.52 6.49 6.46 6.44 6.41 6.39 6.36 6.34 6.32 6.3 Aug 5.88 5.88 5.87 5.87 5.87 5.87 5.87 5.87 5.87 5.87 Sept 5.37 5.39 5.42 5.44 5.47 5.49 5.52 5.54 5.57 5.59 Oct 2.73 2.73 2.74 2.74 2.74 2.74 2.75 2.75 2.75 2.75 Avg 5.15 5.13 5.12 5.1 5.08 5.06 5.04 5.02 5 4.98 data. These eight stations have the following monthly average monsoon rainfall from 1981-2018:Khepupara> Potuakhali>Bhola>Barishal>Khulna>Satkhira> Faridpur>Jashore. The average monsoon rainfall of the Satkhira station from June-October is 9.80 mm, 11.37 mm, 9.67 mm, 9.51 mm, and 4.66 mm gradually. For the Khulna station, it is 10.92 mm, 11.65 mm, 10.53 mm, 9.20 mm, and 4.37 mm; for the Jashore station, it is 10.22 mm, 11.48 mm, 8.76 mm, 8.68 mm, 4.40 mm; for the Barishal station 13.55mm, 13.56mm, 11.41mm, 9.43mm, 5.42mm; for the Bhola station 14.98mm, 14.41mm, 12.43mm, 9.64mm, 5.34mm; for Faridpur station 10.47mm, 11.15mm, 9.48mm, 7.85mm, 4.39mm; for Khepupara 17.17mm, 22.11mm, 15.48mm, 13.04mm, 8.29mm; for Potuakhali 17.02mm, 18.93mm, 14.85mm, 12.02mm, 7.21mm; for Mongla 11.59mm, 13.16mm, 10.79mm, 10.33mm, 5.39mm from June-October respectively. From this analysis, it is evident that for each of the nine stations of our research paper, the monthly average monsoon rainfall has the following order: July>June>August>September>October, except Bhola. In Bhola, the order is June>July>August>September>October because of the variability of the recorded data. rainfall. The same characteristics will be seen for the rainfall of June at this station. This amount will increase and decrease in July and October without any ups and downs, respectively. However, an exception may be seen in August and September, where the rainfall may increase and decrease with fluctuations between 2019 and 2028. The second-highest monsoon average precipitation may occur in August 2028. The most striking feature is that the lowest rainfall at Khepupara station may be seen in October, but this amount may rise over one decade from 2019 to 2028.All these information have been given into figure 5(g) and table 10.i In figure 5(h) and table 11, the average monsoon rainfall of Potuakhali station has been shown to increase by 0.72mm between 2019 and 2028. The highest rainfall may occur in 2019, with an amount of 8.02mm. Am. J. Environ. Clim. 2(1) 11-23, 2023 The maximum monsoon rainfall may occur in July in the given period. Apart from that, the second highest rainfall may happen in June in the mentioned period, except in 2028. An increasing trend may be seen in the average monsoon rainfall of the Mongla station, shown in figure 6 and table 12. The most striking feature is that all the monsoon months may show a plummeting trend in the average rainfall, except September. Despite this, the maximum rainfall may be seen in July, ranging from 5.63mm to 5.86mm. In this station, the rainfall of June, July, August, and September may not differ that much; however, an exception may be seen in the rainfall of October with a minimum amount. Figure 6: ANN predicted rainfall of Mongla (i) station https://journals.e-palli.com/home/index.php/ajec Comparison of Rainfall Pattern The average monsoon rainfall of Satkhira, Khulna, Jashore, Barishal, Bhola, Faridpur, Khepupara, Potuakhali, and Mongla stations has been calculated as 9.00 mm, 9.33 mm, 8.64 mm, 10.68mm, 11.36mm 8.67mm, 15.22mm, 14.01mm and 10.25mm respe-ctively in the period 1981-2018. The Mongla Stations will be out of comparison because of the mismatch of the recorded Figure 6: ANN predicted rainfall of Mongla (i) station Table 8: Predicted rainfall (mm) of Bhola station 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 June 7.88 7.6 7.36 7.13 6.93 6.74 6.56 6.4 6.24 6.08 July 10.89 10.65 10.43 10.21 10.01 9.81 9.62 9.44 9.26 9.09 Aug 8.38 8.29 8.2 8.13 8.05 7.98 7.91 7.84 7.77 7.7 Sept 6.76 6.67 6.59 6.5 6.42 6.33 6.24 6.15 6.05 5.96 Oct 1.91 1.93 1.95 1.97 2 2.02 2.05 2.08 2.11 2.14 Avg 7.16 7.03 6.91 6.79 6.68 6.58 6.48 6.38 6.29 6.19 Am. J. Environ. Clim. 2(1) 11-23, 2023 Table 9: Predicted rainfall (mm) of Faridpur station 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 June 5.94 5.97 5.98 5.96 5.9 5.8 5.65 5.46 5.22 4.95 July 5.17 5.3 5.44 5.6 5.78 5.96 6.14 6.32 6.48 6.62 Aug 5.98 6.09 6.2 6.3 6.38 6.45 6.48 6.48 6.44 6.34 Sept 6.05 6.11 6.16 6.2 6.22 6.23 6.2 6.16 6.08 5.97 Oct 2.85 2.83 2.81 2.8 2.78 2.76 2.74 2.73 2.71 2.69 Avg 5.32 5.39 5.43 5.44 5.44 5.41 5.37 5.32 5.26 5.2 1981-2018, the average monsoon rainfall increased by 0.0462mm/year; in 2019-2028, it escalated by 0.0462mm/ year. In figure 7(b), a comparison has been made between the actual and ANN-predicted average monsoon rainfall of Jashore station. It is evident that both periods have a decreasing trend over the rainfall of this station, with the coefficient of determination of 0.0008 and 0.9998, respectively. In 1981-2018 and 2029-2028, the average monsoon rainfall fell by 0.0055mm/year and 0.0449mm/ year, respectively The trend has been calculated for the average monsoon rainfall from 1981-2018 and the predicted average monsoon rainfall of 2019-2028 to justify the trend of the predicted values. In this section, we have drawn a linear trend line, and from where we found out the equation of this line for the rainfall of each station of the South- West region. https://journals.e-palli.com/home/index.php/ajec Comparison of Rainfall Pattern Figure 7(a) has manifested that the rainfall has an increasing trend over Khulna station in the period of 1981-2018 and 2019-2028, with the coefficient of determination of 0.0565 and 0.5017, respectively. In Table 10: Predicted rainfall (mm) of Khepupara station 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 June 8.13 7.93 7.73 7.54 7.36 7.2 7.05 6.91 6.79 6.65 July 10.69 10.66 10.64 10.62 10.59 10.57 10.55 10.53 10.51 10.49 Aug 8.34 8.41 8.49 8.57 8.65 8.73 8.81 8.89 8.97 9.05 Sept 7.7 7.7 7.71 7.73 7.75 7.78 7.82 7.88 7.94 8.01 Oct 3.9 3.94 3.99 4.04 4.09 4.15 4.21 4.27 4.33 4.39 Avg 7.72 7.71 7.7 7.69 7.69 7.69 7.7 7.71 7.73 7.75 Table 10: Predicted rainfall (mm) of Khepupara station Table 11: Predicted rainfall (mm) of Potuakhali station 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 June 9.86 9.67 9.43 9.16 8.86 8.55 8.26 7.98 7.75 7.56 July 9.99 9.88 9.76 9.62 9.47 9.32 9.16 9 8.84 8.68 Aug 8.8 8.71 8.62 8.53 8.44 8.34 8.25 8.16 8.06 7.98 Sept 6.76 6.81 6.86 6.91 6.96 7 7.06 7.11 7.16 7.21 Oct 4.68 4.72 4.77 4.81 4.84 4.88 4.92 4.96 5 5.05 Avg 8.02 7.96 7.89 7.8 7.71 7.62 7.53 7.44 7.36 7.3 Table 12: Predicted rainfall (mm) of Mongla station 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 June 5.22 5.09 4.94 4.8 4.67 4.56 4.48 4.42 4.39 4.37 July 5.86 5.85 5.84 5.81 5.78 5.75 5.71 5.68 5.66 5.63 Aug 5.77 5.74 5.71 5.68 5.66 5.63 5.61 5.6 5.58 5.56 Sept 5.24 5.25 5.27 5.3 5.33 5.36 5.39 5.41 5.42 5.41 Oct 4.72 4.48 4.25 4.04 3.85 3.69 3.56 3.46 3.38 3.32 Avg 4.86 4.88 4.91 4.95 5 5.06 5.13 5.2 5.28 5.36 A comparison between the actual and predicted rainfall of Satkhira station has been shown in figure 7(c). This graph tells about the decreasing trend of the given periods. The changing amount of rainfall at this station is -0.02mm/ A comparison between the actual and predicted rainfall of Satkhira station has been shown in figure 7(c). This graph tells about the decreasing trend of the given periods. The changing amount of rainfall at this station is -0.02mm/ https://journals.e-palli.com/home/index.php/ajec Am. J. Environ. Clim. Comparison of Rainfall Pattern 2(1) 11-23, 2023 Figure 7(a): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon Page 20 Figure 7(a): Pattern comparison between average monsoon rainfall from 1981 2018 and predicted average monsoon rainfall of Khulna station Figure 7(b): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall of Jashore station Figure 7(c): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall of Satkhira station Figure 7(d): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall of Barishal station Figure 7(b): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon Fi 7( ) P i b i f ll f 1981 2018 d di d e 20 Fig re 7(d): P tt rn mp ri n b t n r m n n r inf ll fr m 1981 2018 nd pr di t d r m n n Page Figure 7(d): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall of Barishal station https://journals.e-palli.com/home/index.php/ajec Am. J. Environ. Clim. 2(1) 11-23, 2023 Am. J. Environ. Clim. 2(1) 11-23, 2023 From figure 7(d), it is clear that the rainfall has a decreasing trend over Barishal station in the period of 1981-2018 and 2019-2028, with the coefficient of determination of 0.0035 and 0.9932, respectively. In 1981-2018, the average monsoon rainfall increased by 0.0119mm/year; in 2019- 2028, it escalated by 0.0194mm/year, with ups and downs in rainfall amount in several years. of this station, with the coefficient of determination of 0.0275 and 0.3696, respectively. In 1981-2018 and 2029- 2028, the average monsoon rainfall fell by 0.03mm/year and 0.0165mm/year, respectively. Figure 7(g) has explained that the rainfall has an increasing trend over Khepupara station in the period of 1981-2018 and 2019-2028, with the coefficient of determination of 0.066 and 0.2444, respectively. In 1981-2018, the average monsoon rainfall increased by 0.0548mm/year; in 2019- 2028, it escalated by 0.0034mm/year. A comparison between the actual and predicted rainfall of Bhola station has been shown in figure 7(e), which describes the decreasing trend of the given periods. The changing amount of rainfall of this station is -0.0393mm/ year and -0.1066mm/year between 1981-2018 and 2019- 2028, respectively. Comparison of Rainfall Pattern In 1981-2018 and 2029-2028, the average monsoon rainfall fell by 0.0615mm/year and 0.0569mm/year, respectively. Graphs 6(a, b, c, d, e, f, g, h, i) have shown that the actual and the predicted rainfall have the same changing pattern. From these graphs, it has been justified that the forecasted rainfall should be almost accurate. The predicted outcome may vary a little because it has been predicted based on the average data of the independent weather variables. Figure 7(i) has clearly explained whether there is any discrepancy between the actual and ANN-predicted average monsoon rainfall of Mongla station. It is evident that both periods have an increasing trend over the precipitation of this station, with the coefficient of determination of 0.0616 and 0.9666, respectively. In 1981-2018 and 2029-2028, the average monsoon rainfall fell by 0.0615mm/year and 0.0569mm/year, respectively. Graphs 6(a, b, c, d, e, f, g, h, i) have shown that the actual and the predicted rainfall have the same changing pattern. From these graphs, it has been justified that the forecasted rainfall should be almost accurate. The predicted outcome may vary a little because it has been predicted based on the average data of the independent weather variables. (2) The ANN model has proved to be almost accurate for predicting the monthly average monsoon rainfall of each station of the South-West region in Bangladesh by calculating the R2 value. (3) The order of the average ANN predicted monsoon rainfall of one decade from 2019 has been calculated as (7.48mm)> Bhola(6.65mm)> Satkhira(5.96mm)> Faridpur(5.36mm)> Barishal(5.07mm)> Mongla(5.06mm)> Jashore(4.26mm). These values have been calculated by taking the average value of the independent weather variables with their changing rate per year. (4) The order of the average rainfall of the monsoon season for the period 1981-2018 is found as Khepupara(15.22mm)> Potuakhali(14.01mm)> Bhola(11.36mm)> Barishal(10.68mm)> Khulna(9.33mm)> Sat-khira(9.00mm)> Faridpur(8.67mm)> Jashore(8.64mm). For Mongla, it is 10.25mm. This station is not considered in the comparison because of the non- viability of the recorded data. Comparison of Rainfall Pattern The coefficient of determination has also been seen by 0.0278 and 0.9964, respectively.i A comparison between the secondary and predicted rainfall of Potuakhali station, which has been shown in figure 7(h) asserts the decreasing trend of the given periods. This station’s changing amount of rainfall is -0.0316mm/year and -0.0838mm/year between 1981- 2018 and 2019-2028, respectively. The coefficient of determination has also been seen by 0.0184 and 0.9975, respectively. In figure 7(f), an analogy has been made between the real and ANN-predicted average monsoon rainfall of Faridpur station. An obvious result has been seen in this figure, that both periods have a decreasing trend over the rainfall Page 21 Figure 7(e): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall of Bhola station Figure 7(f): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall of Faridpur station Figure 7(g): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall of Khepupara station Figure 7(e): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall of Bhola station Figure 7(f): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon r i f ll f F ridp r t ti age 21 Figure 7(g): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon Page Figure 7(g): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall of Khepupara station https://journals.e-palli.com/home/index.php/ajec https://journals.e-palli.com/home/index.php/ajec Am. J. Environ. Clim. 2(1) 11-23, 2023 Am. J. Environ. Clim. 2(1) 11-23, 2023 Figure 7(h): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall of Potuakhali station Figure 7(i): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall of Mongla station Figure 7(h): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall of Potuakhali station Figure 7(i): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall of Mongla station Figure 7(i): Pattern comparison between average monsoon rainfall from 1981-2018 and predicted average mo rainfall of Mongla station structure. Figure 7(i) has clearly explained whether there is any discrepancy between the actual and ANN-predicted average monsoon rainfall of Mongla station. It is evident that both periods have an increasing trend over the precipitation of this station, with the coefficient of determination of 0.0616 and 0.9666, respectively. https://journals.e-palli.com/home/index.php/ajec CONCLUSIONS (8) Rainfall pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall has shown almost the same changing trend for each mentioned station. Khulna, Jashore, Barishal, Bhola, Faridpur, Khepupara, Potuakhali, and Mongla is -0.02, 0.0462, -0.0055, -0.0119, -0.0393, -0.03,0.0548, -0.0316, 0.0497mm/year, respectively. The negative sign indicates the decrease, and the positive sign indicates the increasing rainfall trend. Khulna, Jashore, Barishal, Bhola, Faridpur, Khepupara, Potuakhali, and Mongla is -0.02, 0.0462, -0.0055, -0.0119, -0.0393, -0.03,0.0548, -0.0316, 0.0497mm/year, respectively. The negative sign indicates the decrease, and the positive sign indicates the increasing rainfall trend. Haviluddin, & Alfred, R., (2014). Forecasting Network Activities Using ARIMA Method. Journal of Advances in Computer Network, 2(3), 173-179. Activities Using ARIMA Method. Journal of Advances in Computer Network, 2(3), 173-179. Haykin S., (1994) Neural Networks, A Comprehensive Foundation. Prentice-Hall, Inc., New Jersey (8) Rainfall pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall has shown almost the same changing trend for each mentioned station. Maqsood, I., Khan, M.R., Huang, G.H., Abdalla, R., (2005). Application of Soft Computing Models to Hourly Weather Analysis in Southern Saskatchewan, Canada. Engineering Application of Artificial Intelligence, 18, 115–125. Acknowledgements The authors would like to thank Bangladesh Meteorological Department (BMD), the Ministry of Defence for yielding the essential data, and Khulna University Research Cell for funding this project. Navid, M. A. I., & Niloy, N. H., (2018). Multiple Linear Regressions for Predicting Rainfall for Bangladesh. Communications, 6(1), 1-4. Rahman, M., Salehin, M., & Matsumoto, J., (1997). Trends of Monsoon Rainfall Pattern in Bangladesh. Bangladesh Journal of Water Resources, 14,121-138. CONCLUSIONS This study aims to predict the monthly average monsoon rainfall of the South-West Region of Bangladesh by using several monthly average meteorological variables. The following conclusions have been drawn based on this research. (5) The maximum average predicted rainfall has been calculated in July in all the given stations except Satkhira and Faridpur. On the other hand, in all the mentioned meteorological stations, the minimum predicted average monsoon has been calculated in October, except Satkhira. (6) The order of the monthly average monsoon rainfall for each station is found as July>June>August>September> October, except Bhola. For Bhola station, the order is June> July>August>September> October. (1) The monthly average monsoon cloud cover has solid and positive, and the mean sea-level pressure negatively correlates with the monthly average rainfall for our study areas. Most of the mentioned stations follow the exact relationship between rainfall and other variables, where the order of correlation between them has been measured as Cloud Cover>Humidity>Mean Sea-level Pressure>Wind Speed>AverageTemperature, except Jashore, Potuakhali, Khepupara, and Mongla. The other stations have not maintained the same order because of the topographical (1) The monthly average monsoon cloud cover has solid and positive, and the mean sea-level pressure negatively correlates with the monthly average rainfall for our study areas. Most of the mentioned stations follow the exact relationship between rainfall and other variables, where the order of correlation between them has been measured as Cloud Cover>Humidity>Mean Sea-level Pressure>Wind Speed>AverageTemperature, except Jashore, Potuakhali, Khepupara, and Mongla. The other stations have not maintained the same order because of the topographical (7) Average monsoon rainfall for each station of the South-West region of Bangladesh has also been analyzed. The average monsoon rainfall change for Satkhira, https://journals.e-palli.com/home/index.php/ajec Am. J. Environ. Clim. 2(1) 11-23, 2023 Gogoi, P., (2017). First Interaction Artificial Neural Network. Knoldus. Gogoi, P., (2017). First Interaction Artificial Neural Network. Knoldus. Khulna, Jashore, Barishal, Bhola, Faridpur, Khepupara, Potuakhali, and Mongla is -0.02, 0.0462, -0.0055, -0.0119, -0.0393, -0.03,0.0548, -0.0316, 0.0497mm/year, respectively. The negative sign indicates the decrease, and the positive sign indicates the increasing rainfall trend. (8) Rainfall pattern comparison between average monsoon rainfall from 1981-2018 and predicted average monsoon rainfall has shown almost the same changing trend for each mentioned station. Khulna, Jashore, Barishal, Bhola, Faridpur, Khepupara, Potuakhali, and Mongla is -0.02, 0.0462, -0.0055, -0.0119, -0.0393, -0.03,0.0548, -0.0316, 0.0497mm/year, respectively. The negative sign indicates the decrease, and the positive sign indicates the increasing rainfall trend. REFERENCES Abhishek, K., Singh, M.P., Ghosh, S., & Abhishek, A., (2012). Weather Forecasting Model Using Artificial Neural Network. Procedia Technology, 4, 311-318. Rogers, L., & Dowla, F., (1994). Optimization of Groundwater Remediation Using Artificial Neural Networks with Parallel Solute Transport Modeling. Water Resources Research, 30(2), 457- 481. Ahmed, R., & Kim, I.-K., (2003). Patterns of Daily Rainfall in Bangladesh During the Summer Monsoon Season: Case Studies at Three Stations. Physical Geography, 24(4), 295–318. Shahid, S., (2010). Recent Trends in the Climate of Bangladesh. Climate Research, 42(3), 185-193. Shaibur, M. R., Khan, M. H., & Rashid, M. S., (2017), Climate Change may Cause Natural Disasters in Shyamnagar, Satkhira: the Southwestern Parts of Bangladesh. Bangladesh. Journal of Environmental Science, 32, 101-110. Ahasan, M. N., Chowdhary, M. A., & Quadir, D. ,(2010). Variability and Trends of Summer Monsoon Rainfall over Bangladesh. Journal of Hydrology and Meteorology, 7(1), 1-17. Ara, M., Hossain, M. A., & Alam, M. M., (2005). Surface Dry Bulb Temperature and Its Trend Over Bangladesh. Journal of Bangladesh Academy of Sciences, 29(1), 29-40. Shrivastava, G., Karmakar, S., & Kowar, M. K., (2012). Application of Artificial Neural Networks in Weather Forecasting: a Comprehensive Literature Review. International Journal of Computer Applications, 51(18), 17-29. Bilgili, M., (2010). Prediction of Soil Temperature Using Regression and Artificial Neural Network Models. Meteorology and Atmospheric Physics, 110(1-2), 59–70.i Valipour, M., Banihabib, M. E., & Behbahani, S. M. R., (2013). Comparison of the ARMA, ARIMA, and the Autoregressive Artificial Neural Network Models in Forecasting the Monthly Inflow of Dez Dam Reservoir. Journal of Hydrology, 476, 433-441. Climate Change Profile: Bangladesh. (2018). Ministry of Foreign Affairs of the Netherlands. 4-6.i Farajzadeh, J., Fard, A.F., & Lotfi, S., (2014). Modeling of Monthly Rainfall and Runoff of Urmia Lake Basin Using “Feed-Forward Neural Network” and “Time Series Analysis” Model. Water Resources and Industry, 7(8), 38–48. Zuidema, P., Fairall, C., Hartten,, L., Hare., J., & Wolfe, D., (2007). On Air-Sea Interaction at the Mouth of the Gulf of California. Journal of Climate, America, 20, 1649-1661. Page 23 https://journals.e-palli.com/home/index.php/ajec
https://openalex.org/W1799018966	https://bmcbioinformatics.biomedcentral.com/counter/pdf/10.1186/1471-2105-5-206	English	null	An empirical analysis of training protocols for probabilistic gene finders.	BMC bioinformatics	2,004	cc-by	8,153	BioMed Central BioMed Central Open Research article An empirical analysis of training protocols for probabilistic gene finders William H Majoros* and Steven L Salzberg Open Access Address: The Institute for Genomic Research, 9712 Medical Center Drive, Rockville, MD 20850, USA Email: William H Majoros* - bmajoros@tigr.org; Steven L Salzberg - salzberg@tigr.org * Corresponding author Received: 05 November 2004 Accepted: 21 December 2004 Received: 05 November 2004 Accepted: 21 December 2004 Published: 21 December 2004 BMC Bioinformatics 2004, 5:206 doi:10.1186/1471-2105-5-206 Published: 21 December 2004 BMC Bioinformatics 2004, 5:206 doi:10.1186/1471-2105-5-206 This article is available from: http://www.biomedcentral.com/1471-2105/5/206 j g; This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Abstract GHMMs offer a number of clear advantages which would seem to explain this While the decoding problem for GHMM gene finders is arguably well understood, being a relatively straightfor- ward extension of the same problem for traditional Page 1 of 12 (page number not for citation purposes) Page 1 of 12 (page number not for citation purposes) http://www.biomedcentral.com/1471-2105/5/206 BMC Bioinformatics 2004, 5:206 http://www.biomedcentral.com/1471-2105/5/206 would be to maximize the average probability of the gene parses in the training set: HMMs and amenable to a Viterbi-like solution (albeit a more complex one), methods for optimally training a GHMM gene finder have received scant attention in the gene-finding literature to date. What information is avail- able (e.g., [2,4]) seems to indicate that the common prac- tice is to optimize the submodels of the GHMM independently, without regard for the optimality of the composite model. HMMs and amenable to a Viterbi-like solution (albeit a more complex one), methods for optimally training a GHMM gene finder have received scant attention in the gene-finding literature to date. What information is avail- able (e.g., [2,4]) seems to indicate that the common prac- tice is to optimize the submodels of the GHMM independently, without regard for the optimality of the composite model. ˘ arg max ( \| , ) \| \| arg max ( , \| ) ( , ) ( , θ θ φ θ θ φ θ φ =         = ∈ ∑P S T P S S T S φ θ ) ( \| ) , ( ) ∈ ∑         T P S 2 ( ) 2 where the collection of model parameters making up the GHMM is denoted θ and the elements (S, φ) of the train- ing set T comprise pairs of sequences S and their known The training of HMMs and GHMMs has traditionally been carried out using some form of maximum likelihood estima- tion (MLE). Baum-Welch training [13], which is an instance of the well-known expectation maximization (EM) procedure, is itself a form of MLE [14]. In the case of GHMM gene finders, one typically applies some form of MLE to each of the submodels (states) in the GHMM so as to render training features of each type (e.g., exon, intron, donor site) maximally likely under the induced (sub)model; i.e., maximizing: parses φ. Abstract Background: Generalized hidden Markov models (GHMMs) appear to be approaching acceptance as a de facto standard for state-of-the-art ab initio gene finding, as evidenced by the recent proliferation of GHMM implementations. While prevailing methods for modeling and parsing genes using GHMMs have been described in the literature, little attention has been paid as of yet to their proper training. The few hints available in the literature together with anecdotal observations suggest that most practitioners perform maximum likelihood parameter estimation only at the local submodel level, and then attend to the optimization of global parameter structure using some form of ad hoc manual tuning of individual parameters. Results: We decided to investigate the utility of applying a more systematic optimization approach to the tuning of global parameter structure by implementing a global discriminative training procedure for our GHMM-based gene finder. Our results show that significant improvement in prediction accuracy can be achieved by this method. Conclusions: We conclude that training of GHMM-based gene finders is best performed using some form of discriminative training rather than simple maximum likelihood estimation at the submodel level, and that generalized gradient ascent methods are suitable for this task. We also conclude that partitioning of training data for the twin purposes of maximum likelihood initialization and gradient ascent optimization appears to be unnecessary, but that strict segregation of test data must be enforced during final gene finder evaluation to avoid artificially inflated accuracy measurements. growth in popularity. Chief among these is the fact that the GHMM framework, being (in theory) purely probabi- listic, allows for principled approaches to constructing, utilizing, and extending models for accurate prediction of gene structures. g The number of generalized hidden Markov model (GHMM) gene finders reported in the literature has increased fairly dramatically of late [1-8], and the commu- nity is now contemplating various ways to extend this attractive framework in order to incorporate homology information, with a handful of such systems having already been built (e.g., [9-12]). Abstract This argmax gives us the parameterization under which the full gene parses (rather than the sequences) in the training set will be maximally likely (on average). Decomposing each parse φ into a series of (qi, di) pairs, for state qi and state duration (i.e., feature length) di, we get: θ ˆ argmax ( \| , ) ( \| ) ( \| ) ( \| ) ( , θ θ θ = − = − ∏P S q d P q q P d q P S e i i i t i i d i i i n S 1 1 1 φ) , ( ) ∈ ∑               T 3 ( ) 3 P S q d i i S T i ( \| , ), ( ) 1 ∈∑ ( ) 1 where Pe, Pt, and Pd represent the emission, transition, and duration probabilities of the GHMM, respectively. Whereas the common MLE training procedure for GHMMs as described above optimizes the individual terms in the numerator of Equation 3 independently, the argmax above calls instead for these terms to be jointly tuned so as to optimize the entire ratio in parentheses. Intuitively, one can think of this alternate formulation as attempting to account for the process in the Viterbi algo- rithm (during later decoding) whereby the individual sub- models "compete" for nucleotides (in the sense that each nucleotide can be emitted by only one submodel in any given parse, and the Viterbi algorithm chooses the final, predicted parse based on the values of the model parame- ters). Our hope is that by addressing the issue of sub- model competition explicitly during parameter estimation, we will thereby empower the gene finder to better discriminate at a global sequence level between the features modeled by individual submodels in the GHMM, thereby producing more accurate gene predictions. for state q and for Si a feature of length di from the state-q- specific training set T. The submodels are then merged into a composite model (i.e., the full GHMM) by observ- ing transition probabilities between features in the train- ing data corresponding to each of the GHMM states. for state q and for Si a feature of length di from the state-q- specific training set T. Page 2 of 12 (page number not for citation purposes) http://www.biomedcentral.com/1471-2105/5/206 BMC Bioinformatics 2004, 5:206 http://www.biomedcentral.com/1471-2105/5/206 edged by those with experience training GHMM-based gene finders (including our own systems). The practice of performing such tuning on the training set, especially when done iteratively, can be viewed as a manual form of gradient ascent optimization using the percentages of cor- rectly predicted nucleotides, exons, and whole genes as surrogates for the Σ(S,φ)∈T P(φ\|S,θ) term in Equation 2. methods for optimizing competing stochastic grammar models may improve the accuracy of systems at the level of whole-sentence parses [21]. Maximum discrimination HMMs have already been applied successfully to prob- lems in the realm of biological sequence analysis [22], though their use in gene finding has apparently not yet seen widespread adoption. To our knowledge, the only gene finder reported to use discriminative training is HMMgene [23], a gene finder based on a non-generalized HMM. We therefore decided to investigate the use of a simple form of global discriminative training for gene-finding. We did this by building a rudimentary gradient ascent optimizer and applying it to a subset of the model param- eters for our GHMM-based gene finder, TigrScan, as described in the Methods. In light of these considerations, it is worth contemplating the possible gains in gene finder accuracy that might be obtained through the use of some form of discriminative training applied to a GHMM – that is, training aimed more directly at optimizing the ability of the gene finder to discriminate between exons and non-exons, thereby improving the expected accuracy of the gene finder's pre- dictions. Anecdotal evidence already suggests that investi- gation of such methods may indeed be fruitful, as the process of manual tuning of GHMM parameters (i.e., "tweaking") after MLE training is commonly acknowl- See Table 1 for legend. Abstract The submodels are then merged into a composite model (i.e., the full GHMM) by observ- ing transition probabilities between features in the train- ing data corresponding to each of the GHMM states. For example, an exon state in a GHMM can be trained by collecting n-gram statistics (i.e., counts of n-letter sub- strings) from known exon sequences and normalizing these into transition probabilities for an (n-1)th-order Markov chain [15]. Similarly, intron, intergenic, and untranslated region (UTR) states can be modeled by col- lecting appropriate statistics from corresponding sample features and using these to train individual content-scor- ing models, such as Markov chains, neural networks, deci- sion trees, etc. Signal sensors for donor and acceptor splice sites and start and stop codons can be trained by aligning known signals of the appropriate type and counting nucleotide frequencies at each position within a fixed window around the signal; converting these counts to rel- ative frequencies produces probability estimates for use in a weight matrix or similar type of model. Transition and duration probabilities can likewise be estimated by observing appropriate frequencies in training data. All of these estimation activities can be performed independ- ently, resulting in a GHMM consisting of distinct subsets of maximum likelihood parameters. A similar optimization problem occurs in the field of speech recognition, in which systems of interacting acous- tic models and language models are employed to opti- mally parse an audio stream into a series of discrete words. Interestingly, the trend in that field, starting with Bahl et al. in 1986 [16], has increasingly been away from the sole use of MLE and toward an alternative approach very similar to that prescribed by Equation 2 known as glo- bal discriminative training [17-19] or conditional maximum likelihood [20]. The problem also appears in a slightly dif- ferent form in the related field of statistical natural lan- guage parsing, in which it has been suggested that global Such an approach does not, however, attend to the global optimality of the GHMM as a whole. Ideally, one would like to maximize the expected accuracy of the gene finder on unseen data. A reasonable approximation to this ideal Page 2 of 12 (page number not for citation purposes) Page 2 of 12 (page number not for citation purposes) http://www.biomedcentral.com/1471-2105/5/206 Page 3 of 12 (page number not for citation purposes) Maximum likelihood versus discriminative training Maximum likelihood versus discriminative training g Results for Arabidopsis thaliana are shown in Table 1 and those for Aspergillus fumigatus are shown in Table 2. The two methods being compared are maximum likelihood estimation (MLE) versus maximum likelihood followed by gradient ascent parameter estimation (GRAPE). Table 1: Results on Arabidopsis thaliana method train test nucAcc exonF geneSn GRAPE CV CV 95 ± 1% 82 ± 2% 49 ± 3% GRAPE CV H 93 ± 1% 80 ± 2% 44 ± 3% GRAPE T T 95% 86% 57% GRAPE T H 94% 81% 48% MLE CV CV 90 ± 1% 72 ± 2% 33 ± 4% MLE T T 91% 75% 36% MLE T H 90% 71% 33% GRAPE = GRadient Ascent Parameter Estimation, MLE = Maximum Likelihood Estimation only. CV=cross validation, T = training set, H = 1000- gene hold-out ("test") set. CV in the train column means training on 800 genes from T. CV in test column means testing on 200 genes from T. In rows with a CV in either column, numbers are averages from 5 runs. nucAcc = nucleotide accuracy, exonF = exon F score, geneSn = gene sensitivity. F = 2SnSp/(Sn+Sp) for Sn = sensitivity and Sp = specificity. CV averages are reported ± SD. Table 1: Results on Arabidopsis thaliana GRAPE = GRadient Ascent Parameter Estimation, MLE = Maximum Likelihood Estimation only. CV=cross validation, T = training set, H = 1000- gene hold-out ("test") set. CV in the train column means training on 800 genes from T. CV in test column means testing on 200 genes from T. In rows with a CV in either column, numbers are averages from 5 runs. nucAcc = nucleotide accuracy, exonF = exon F score, geneSn = gene sensitivity. F = 2SnSp/(Sn+Sp) for Sn = sensitivity and Sp = specificity. CV averages are reported ± SD. Table 2: Results on Aspergillus fumigatus method train test nucAcc exonF geneSn GRAPE CV CV 88 ± 1% 54 ± 4% 35 ± 4% GRAPE CV H 88 ± 1% 51 ± 2% 29 ± 1% GRAPE T T 92% 65% 48% GRAPE T H 87% 51% 31% MLE CV CV 81 ± 3% 27 ± 8% 16 ± 5% MLE T T 88% 42% 28% MLE T H 83% 30% 18% See Table 1 for legend. Maximum likelihood versus discriminative training Table 2: Results on Aspergillus fumigatus BMC Bioinformatics 2004, 5:206 http://www.biomedcentral.com/1471-2105/5/206 Maximum likelihood versus gradient ascent Figure 1 Maximum likelihood versus gradient ascent Gradient ascent parameter estimation (GRAPE) improves accuracy over MLE at the nucleotide, exon, and whole gene levels. arab = Arabidopsis thaliana, asp = Aspergillus fumigatus. ACCURACY 0% 20% 40% 60% 80% 100% NUCLEOTIDE EXON GENE 90 94 71 81 33 48 83 87 arab asp 30 51 18 31 arab asp arab asp MLE GRAPE ACCURACY 0% 20% 40% 60% 80% 100% NUCLEOTIDE EXON GENE 90 94 71 81 33 48 83 87 arab asp 30 51 18 31 arab asp arab asp MLE GRAPE MLE GRAPE Maximum likelihood versus gradient ascent Figure 1 Maximum likelihood versus gradient ascent Gradient ascent parameter estimation (GRAPE) improves accuracy over MLE at the nucleotide, exon, and whole gene levels. arab = Arabidopsis thaliana, asp = Aspergillus fumigatus. The train column indicates whether training (i.e., parame- ter estimation) was performed on the entire training set (T) or on separate 800-gene cross-validation partitions (CV). The test column indicates whether accuracy was measured on the full training set (T), on one-fifth of the training set (CV), or on the unseen data (H). We will con- sider the evaluation on H to be the most reliable measure of gene finder accuracy. For any row containing a CV, we report the average of five runs, where each run used a dif- ferent 800-gene subset of the training data for parameter estimation. over the MLE method from 71% to 81% at the level of exons and 33% to 48% at the level of whole genes. In Aspergillus the improvement was even more dramatic: 30% to 51% at the exon level and 18% to 31% for whole genes. A gain of 4% nucleotide accuracy was measured for both organisms. Page 4 of 12 (page number not for citation purposes) Data partitioning and cross validation A tangible improvement was still seen when a cross-vali- dation design was used to split the training set so as to sep- arate the data used for maximum likelihood estimation (800 genes) and subsequent gradient ascent (200 genes). However, results from both organisms suggest that this separation did not improve the accuracy of the gene finder, as shown in Figure 2. Indeed, on Arabidopsis, gradi- Both tables give compelling evidence for the value of gra- dient ascent training, as shown in Figure 1. In Arabidopsis, gradient ascent applied to the full training set improved Page 4 of 12 (page number not for citation purposes) Page 4 of 12 (page number not for citation purposes) BMC Bioinformatics 2004, 5:206 http://www.biomedcentral.com/1471-2105/5/206 Data partitioning for gradient ascent Figure 2 Data partitioning for gradient ascent Separating the training set into an 800-gene MLE set and a 200-gene gradient ascent set provides no improvement over simply performing MLE and GRAPE on the full training set. ACCURACY 0% 20% 40% 60% 80% 100% NUCLEOTIDE EXON GENE 93 94 80 81 44 48 8887 arab asp 5151 2931 arab asp arab asp 800/200 split full training set ACCURACY 0% 20% 40% 60% 80% 100% NUCLEOTIDE EXON GENE 93 94 80 81 44 48 8887 arab asp 5151 2931 arab asp arab asp 800/200 split full training set 800/200 split full training set Data partitioning for gradient ascent Figure 2 Data partitioning for gradient ascent Separating the training set into an 800-gene MLE set and a 200-gene gradient ascent set provides no improvement over simply performing MLE and GRAPE on the full training set. refinement of 29 parameters is therefore not justified (according to the above results), cross-validation does seem to have some value in terms of predicting how well the gene finder will perform on unseen data, as suggested by Figure 3. ent ascent training produced greater gains in accuracy when performed on the entire training set rather than using the cross-validation structure, while on Aspergillus the improvement due to using a cross-validation structure was either small (nucleotide level: 1%), zero (exon level), or negative (gene level: -2%). Thus, the recommended training protocol would be to apply MLE to the entire training set followed by gradient ascent on the full train- ing set as well. Data partitioning and cross validation On both genomes and at all levels (nucleotide, exon, gene), accuracy measurements obtained through cross- validation were closer to the accuracy measured on unseen data than were the measurements taken from the full training set, as we expected. This was true both with and without gradient ascent, though when gradient ascent was applied, even the cross-validation results were slightly Although use of a cross-validation structure to split the training set for the twin purposes of maximum likelihood estimation of ~90,000 parameters and gradient ascent Page 5 of 12 (page number not for citation purposes) Page 5 of 12 (page number not for citation purposes) BMC Bioinformatics 2004, 5:206 http://www.biomedcentral.com/1471-2105/5/206 Cross-validation versus testing on unseen data Figure 3 Cross-validation versus testing on unseen data Cross-validation scores provide a reasonably accurate prediction of per- formance on unseen data. Results shown for A. thaliana only; results for A. fumigatus are given in Table 2. ACCURACY 0% 20% 40% 60% 80% 100% NUCLEOTIDE EXON GENE 9090 9594 7271 82 81 3333 4948 MLE cross-val MLE tested on H GRAPE cross-val GRAPE tested on H BMC Bioinformatics 2004, 5:206 http://www.biomedcentral.com/1471-2105/5/206 ACCURACY 0% 20% 40% 60% 80% 100% NUCLEOTIDE EXON GENE 9090 9594 7271 82 81 3333 4948 MLE cross-val MLE tested on H GRAPE cross-val GRAPE tested on H MLE cross-val MLE tested on H GRAPE cross-val GRAPE tested on H Cross-validation versus testing on unseen data Figure 3 Cross-validation versus testing on unseen data Cross-validation scores provide a reasonably accurate prediction of per- formance on unseen data. Results shown for A. thaliana only; results for A. fumigatus are given in Table 2. Discussion inflated. The latter observation is presumably attributable to the "peeking" that was permitted (see Methods), whereby the gradient ascent procedure received feedback from the 200 evaluation genes held out from the training set, T. This suggests that estimating even small numbers of parameters (in this case 29) from the test set can artifi- cially inflate accuracy measurements on that set. The results presented above provide a clear demonstration that independent maximum likelihood estimation of sub- model parameters is sufficiently neglectful of global GHMM behavior as to compromise gene finder accuracy. Even such a crude method as our 29-parameter gradient ascent procedure proved to be effective at significantly improving accuracy over that achievable by simple MLE training. The potential for more sophisticated global dis- criminative training methods to produce even greater improvements is surely worthy of investigation. Figure 4 illustrates the effects of testing the gene finder on the training set. As can be seen from the figure, the accu- racy measurements taken from the training set can be sub- stantially inflated relative to the more objective measurements taken from the hold-out set, thereby pro- moting overly optimistic expectations for how the gene finder will perform on unseen data. It is interesting to observe that the natural language processing and speech recognition communities, from whom HMM-based methods were originally borrowed for Page 6 of 12 (page number not for citation purposes) Page 6 of 12 (page number not for citation purposes) BMC Bioinformatics 2004, 5:206 http://www.biomedcentral.com/1471-2105/5/206 Evaluation on the training set Figure 4 Evaluation on the training set Accuracy measurements taken from the training set were artificially inflated, as expected. Results are shown only for A. thaliana; results for A. fumigatus were even more extreme. ACCURACY 0% 20% 40% 60% 80% 100% NUCLEOTIDE EXON GENE 9190 9594 75 71 8681 36 33 57 48 MLE tested on T MLE tested on H GRAPE tested on T GRAPE tested on H ACCURACY 0% 20% 40% 60% 80% 100% NUCLEOTIDE EXON GENE 9190 9594 75 71 8681 36 33 57 48 MLE tested on T MLE tested on H GRAPE tested on T GRAPE tested on H Evaluation on the training set Figure 4 Evaluation on the training set Accuracy measurements taken from the training set were artificially inflated, as expected. Results are shown only for A. thaliana; results for A. fumigatus were even more extreme. Discussion adjusted simultaneously, so that the optimizer is less con- strained in following the direction of steepest gradient in parameter space. This may reduce the number of steps required for convergence. Indeed, more rapid conver- gence (in terms of numbers of re-evaluation steps) has been cited as a concrete advantage of these EM-style for- mulations over more generalized gradient ascent methods [23]. However, EM-style approaches to the discriminative training problem for HMMs have typically involved a number of simplifying assumptions and/or heuristics, thereby voiding formal assurances of optimality (e.g., [17,24,25,18,26]). Furthermore, as with more generalized gradient ascent procedures, EM often tends to find only a local optimum rather than a global one [13]. use in bioinformatics, have been moving toward global discriminative training methods for some time. The two most popular forms of discriminative training for speech recognition are Maximum Mutual Information (MMI) and Minimum Classification Error (MCE). Both methods can be implemented using an iterative gradient ascent/ descent algorithm. Our approach is most similar in spirit to that of MCE. In the case of "pure" (i.e., non-generalized) HMMs, expec- tation-maximization (EM) update formulas have been derived for both MMI and MCE. These formulas allow model parameters to be updated in an axis-oblique (rather than axis-parallel) manner; i.e., multiple parameters can be Page 7 of 12 (page number not for citation purposes) Page 7 of 12 (page number not for citation purposes) BMC Bioinformatics 2004, 5:206 http://www.biomedcentral.com/1471-2105/5/206 In the case of GHMM-based gene finders, the advantages of EM over a generalized gradient ascent procedure may indeed be rather slim. The very flexibility which we find attractive in GHMMs can be expected to complicate the derivation of such EM-like update formulas for arbitrary GHMM-based gene finders, likely requiring additional assumptions and approximations that would further compromise the optimality of the EM procedure. It was for this reason that we decided to employ a more general- ized gradient ascent method for the present study. A rudi- mentary gradient ascent optimizer is simple to implement, and the use of prediction accuracy as an objective function affords great convenience in approxi- mating Σ(S,φ)∈TP(φ\|S,θ). Although P(φ\|S,θ) can be more directly computed using a modified Forward algorithm [23], to do so would in theory be no more efficient than running the full gene finder, since the asymptotic run times of the Forward and Viterbi algorithms for GHMMs are equivalent. Discussion Nevertheless, inasmuch as the Forward algorithm provides a more direct approximation of P(φ\|S,θ), its use for this purpose is worthy of investigation. of the ~90,000 GHMM parameters on the 200-gene "test" set suggests that the phenomenon may conceivably occur to some degree even when an automated procedure is not employed. Finally, we would like to make note of an unfortunate consequence of discriminative training of HMMs for bio- logical sequence analysis, namely, that while the resulting models may possess improved ability for discrimination and therefore greater utility for specific tasks such as gene prediction, their suitability as representative models of biological knowledge (especially probabilistic knowl- edge) may well be reduced relative to models induced with simple MLE techniques. Indeed, some authors in the field of speech recognition (e.g., [20]) have noted that more accurate discrimination can sometimes be obtained by relaxing sum-to-one constraints for probability distri- butions, thereby permitting the gradient ascent procedure to automatically discover appropriate weightings between states or inputs. This is reminiscent of the exon "opti- mism" parameter which we employ and which seems to have no principled justification (and indeed, we might speculate that this extraneous parameter proved useful precisely because it enabled a primitive form of discrimi- native training by providing an explicit "correction factor" or weighting between submodels). Thus, despite the apparent value of discriminative training in improving gene finder accuracy, our ability to extract biological knowledge by inspecting the parameters of a gene finder trained in this way may be somewhat hindered. For the present, this does not seem to be of great practical signifi- cance, but it is a consideration worthy at least of mention. There are a number of other variations and enhancements which we are at present contemplating for our discrimina- tive trainer. One of these involves the joint training of pairs of submodels in the GHMM using a maximum dis- crimination criterion rather than the usual one based on maximum likelihood. Although such an approach would not in itself directly attend to the global optimality of the GHMM (indeed, we already apply such an approach to our signal sensors during our so-called "MLE" training regime, as remarked earlier), it would at least seem to offer a promising direction for improving our existing opti- mizer and may be feasible without increasing the compu- tational cost beyond what is practical. Conclusions We have shown that discriminative training for GHMM- based gene finders is feasible using a rudimentary gradient ascent approach, and have briefly explored the relation between this method and the EM-like techniques which have been proposed in the field of speech recognition. Our experiments show that the gradient ascent method can result in a gene finder with substantially greater pre- diction accuracy. It is our hope that even greater gains in accuracy will result from extension and refinement of dis- criminative training techniques applied to GHMM-based gene finders. For the present, we feel confident in making the recom- mendation that others tasked with the training of GHMM gene finders consider applying an automated gradient ascent procedure like that described here as a more sys- tematic alternative to manual tuning of parameters fol- lowing maximum likelihood training of individual submodels. Beyond the obvious advantage of likely improving gene finder accuracy, such an automated method may offer some degree of reproducibility (not- withstanding the typically stochastic nature of such meth- ods) and uniformity for the purposes of comparing gene finders and gene finding algorithms. In addition, we urge those practicing manual tuning on their final "test" set to consider that their reported accuracy results may well be inflated as a result of "peeking" at the test set before the final evaluation – a practice that has been criticized in the field of machine learning (eg., [27]). That significant infla- tion was seen in our studies as a result of tuning only 29 Objective function and optimization procedure As an objective function for use by the gradient ascent pro- cedure, we decided to measure the accuracy of the current parameterization by running the gene finder on a subset of the training genes. Our hope was that this accuracy measure would provide a reasonable approximation of Σ(S,φ)∈T P(φ\|S,θ) by indicating roughly how often the cur- rent model θ would cause the correct parse φ to be pre- dicted for training sequence S. We defined the nucleotide accuracy Anuc as the percentage of nucleotides correctly classified as coding vs. noncoding; Aexon was defined as an average of exon sensitivity and specificity (where a pre- dicted exon is considered correct only if both boundary coordinates were predicted correctly); and Agene was defined as the percentage of training genes which were predicted exactly correctly. These were all rounded to inte- gral percentages between 0 and 100%. The objective func- tion was then defined as: Description of the GHMM The gene finder TigrScan [8] is a GHMM-based program similar to Genie [1] and Genscan [2,28]. The forward- strand model contains six signal states (donor and acceptor sites; start and stop codons; promoter; poly-A sig- nal) and eight content states (intron; intergenic; 5' and 3' UTR; initial, internal, final, and single exons). The reverse- strand model mirrors that of the forward strand. Four Page 8 of 12 (page number not for citation purposes) Page 8 of 12 (page number not for citation purposes) BMC Bioinformatics 2004, 5:206 http://www.biomedcentral.com/1471-2105/5/206 http://www.biomedcentral.com/1471-2105/5/206 relative frequency histograms are used to estimate the duration probabilities of the four exon types; the four noncoding states are assumed to have geometric duration distributions and are therefore each parameterized by a single value representing the mean duration. Each content state is scored using a separate fifth-order Interpolated Markov Model (IMM) [29]. TigrScan offers a number of signal sensors, including WMMs, WAMs, WWAMs, and MDD trees [28] having any of the foregoing signal sensors as leaf models; for this study we used only (non-MDD) WAMs, though the order of the Markov chains within the WAMs was allowed to vary. Putative signals scoring below a given signal threshold are ignored by TigrScan. This threshold is chosen separately for each signal sensor so as to achieve a desired sensitivity Sn (Sn = TP/(TP+FN), TP = true positive count, FN = false negative count) on a train- ing set of true and "decoy" signals. "Boosting" of signal sensors was performed by iteratively retraining each signal sensor on sets of training features in which the lowest scoring features were duplicated so as to focus the training procedure on the most difficult examples. Boosting has been found to improve signal detection in other applica- tion areas [30]. Most transitions in the GHMM are oblig- atory (such as "donor site → acceptor site"); of the non- obligatory transitions, sum-to-one constraints and the for- ward/reverse strand equivalence reduce the number which can be independently varied to just four. Transi- tions into exon states are modified by an exon "optimism" multiplier (similar to that described in [6]) which has been seen anecdotally to be useful in improving predic- tion accuracy (unpublished data). Description of the GHMM • WAM order (4) • WAM order (4) relative frequency histograms are used to estimate the duration probabilities of the four exon types; the four noncoding states are assumed to have geometric duration distributions and are therefore each parameterized by a single value representing the mean duration. Each content state is scored using a separate fifth-order Interpolated Markov Model (IMM) [29]. TigrScan offers a number of signal sensors, including WMMs, WAMs, WWAMs, and MDD trees [28] having any of the foregoing signal sensors as leaf models; for this study we used only (non-MDD) WAMs, though the order of the Markov chains within the WAMs was allowed to vary. Putative signals scoring below a given signal threshold are ignored by TigrScan. This threshold is chosen separately for each signal sensor so as to achieve a desired sensitivity Sn (Sn = TP/(TP+FN), TP = true positive count, FN = false negative count) on a train- ing set of true and "decoy" signals. "Boosting" of signal sensors was performed by iteratively retraining each signal sensor on sets of training features in which the lowest scoring features were duplicated so as to focus the training procedure on the most difficult examples. Boosting has been found to improve signal detection in other applica- tion areas [30]. Most transitions in the GHMM are oblig- atory (such as "donor site → acceptor site"); of the non- obligatory transitions, sum-to-one constraints and the for- ward/reverse strand equivalence reduce the number which can be independently varied to just four. Transi- tions into exon states are modified by an exon "optimism" multiplier (similar to that described in [6]) which has been seen anecdotally to be useful in improving predic- tion accuracy (unpublished data). • signal sensitivity (1) • signal sensitivity (1) • number of signal boosting iterations (8) • skew and kurtosis of exon length distributions Modifications to skew and kurtosis of exon length distri- butions were found during early exploration to produce no improvements; these parameters were therefore left unchanged in all further experiments. All remaining parameters were estimated using standard MLE techniques. For those runs in which gradient ascent was disabled (see below), the following methods were used to estimate the above 29 parameters: mean intron and UTR lengths as well as transition probabilities were estimated using MLE from training data; mean intergenic length was set to a fixed value based on the known intergenic lengths in the test set; exon optimism was set to zero; remaining param- eters were selected so as to minimize the misclassification rate on a set of true and "decoy" signals selected from the training set. Page 9 of 12 (page number not for citation purposes) Parameters to be optimized Of 29 parameters modified by gradient ascent, some (e.g., WAM size) were used to control the MLE estimation procedure, while others (e.g., mean intron length) were used directly as parameters to the GHMM. Testing of the gradient direction was performed on the 200-gene cross-vali- dation set, which was part of the 1000-gene training set, T. Gradient ascent training Figure 5 Gradient ascent training Schematic diagram of gradient ascent training procedure. Of 29 parameters modified by gradient ascent, some (e.g., WAM size) were used to control the MLE estimation procedure, while others (e.g., mean intron length) were used directly as parameters to the GHMM. Testing of the gradient direction was performed on the 200-gene cross-vali- dation set, which was part of the 1000-gene training set, T. Parameters were optimized using an iterative gradient ascent procedure operating in the selected 29-dimen- sional parameter space, as illustrated schematically in Fig- ure 5. Steps were taken in an axis-parallel manner (one step per axis per iteration), with the step size for each axis decreasing by half whenever a local maximum was reached on that axis. sis thaliana and the pathogenic fungus Aspergillus fumiga- tus. Five-fold cross-validation was employed, so that the entire optimization procedure was carried out five times on four-fifths of the data (800 genes) and each time eval- uated on the remaining one-fifth (200 genes); accuracy results reported here were obtained by averaging the five sets of accuracy numbers obtained from the cross-valida- tion. Parameters to be optimized The total number of parameters which need to be esti- mated when training TigrScan is roughly 90,000; the large bulk of these are the n-gram statistics comprising the IMMs used for the content sensors. As an initial attempt at applying discriminative training to TigrScan, we selected 29 of these ~90,000 parameters to subject to gradient ascent optimization. Although this is a miniscule propor- tion of the available parameters, our previous experiences with hand-tuning our GHMM on other data sets suggested that these 29 parameters exert a disproportionately large influence on the accuracy of the gene predictions. By lim- iting the number of parameters to be optimized we hoped to both accelerate the training procedure and also reduce the risk of overtraining. The selected parameters were: f(θ) = 100Anuc+Aexon+Agene. (4) The Anuc and Aexon terms were included in an effort to smooth the function, which would otherwise have been insensitive to changes not reflected in the number of genes predicted exactly correctly – i.e., a step function. Though the Anuc term was given much greater weight for this study, additional work needs to be undertaken to determine the most suitable set of weights for our objec- tive function. • mean intron, intergenic, and UTR lengths (3) • mean intron, intergenic, and UTR lengths (3) • transition probabilities (4) • exon optimism (1) • WAM size and relative positioning (8) • transition probabilities (4) Page 9 of 12 (page number not for citation purposes) Page 9 of 12 (page number not for citation purposes) http://www.biomedcentral.com/1471-2105/5/206 BMC Bioinformatics 2004, 5:206 Gradient ascent training Figure 5 Gradient ascent training Schematic diagram of gradient ascent training procedure. Of 29 parameters modified by gradient ascent, some (e.g., WAM size) were used to control the MLE estimation procedure, while others (e.g., mean intron length) were used directly as parameters to the GHMM. Testing of the gradient direction was performed on the 200-gene cross-vali- dation set, which was part of the 1000-gene training set, T. train (800) test (200) T (1000 genes) final evaluation reported accuracy MLE model files control parms gradient ascent evaluation accuracy final model files x5 “peeking” T (1000 genes) final model files Gradient ascent training Figure 5 Gradient ascent training Schematic diagram of gradient ascent training procedure. Data and experimental design The held-out one-fifth was also used by the gradient ascent procedure to tune the selected 29 parameters. The practice of using a held-out set for smoothing or to esti- mate a small number of additional parameters is common in the natural language processing field [31], where it is The quality of a given parameterization θ was measured by evaluating the objective function f(θ) on a held-out subset of the training set. The training set was limited to 1000 genes, and all experiments were repeated separately on two highly divergent species, the model plant Arabidop- Page 10 of 12 (page number not for citation purposes) Page 10 of 12 (page number not for citation purposes) BMC Bioinformatics 2004, 5:206 http://www.biomedcentral.com/1471-2105/5/206 Cross-validation experiments Figure 6 Cross-validation experiments Five-fold cross-validation was used both in the gradient ascent and in the MLE-only experi- ments. For gradient ascent training, MLE was performed on four-fifths of the training set (T) and then gradient ascent was per- formed on the other one-fifth. A separate hold-out set (H) of 1000 genes was used to obtain an unbiased evaluation of all final models. set T (training) set H (holdout set) 5-fold cross validation evaluation on holdout set mean & variance accuracy on holdout set average 5 accuracy scores 1000 genes 1000 genes 200 genes 200 genes 200 genes 200 genes 200 genes 1000 genes set H (holdout set) 1000 genes set T (training) 1000 genes accuracy on holdout set Page 11 of 12 (page number not for citation purposes) Authors' contributions Software implementation and computational experi- ments were performed by WHM. The manuscript was written by WHM with assistance from SLS. 21. Toutanova K, Mitchell M, Manning CD: Optimizing local probabil- ity models for statistical parsing. In In Proceedings of the Four- teenth European Conference on Machine Learning (ECML 2003) New York: Springer Verlag; 2003:409-420. References 1. Kulp D, Haussler D, Reese MG, Eeckman FH: A generalized hid- den Markov model for the recognition of human genes in DNA. In In Proceedings of the Fourth International Conference on Intelli- gent Systems for Molecular Biology: 12–15 June 1996 St. Louis Edited by: States DJ, Agarwal P, Gaasterland T, Hunter L, Smith RF. Menlo Park: American Association for Artificial Intelligence; 1996:134-142. 24. g 24. Reichl W, Ruske G: Discriminative training for continuous speech recognition. In In Proceedings of the Fourth European Confer- ence on Speech Communication and Technology (EUROSPEECH-95): 18– 21 September 1995 Madrid Amsterdam: Institute of Phonetic Sciences; 1995:537-540. 25. Normandin Y: Maximum mutual information estimation of hidden Markov models. In In Automatic Speech and Speaker Recog- nition Edited by: Lee C-H, Soong FK, Paliwal KK. Norwell: Klewer Academic Publishers; 1996:58-81. g 2. Burge C: Identification of genes in human genomic DNA. In PhD thesis Stanford University, Mathematics Department; 1997. y p 3. Salamov A, Salovyev V: Ab initio gene finding in Drosophila genome DNA. Genome Res 2000, 10:516-522. 26. 26. Markov K, Nakagawa S, Nakamura S: Discriminative training of HMM using maximum normalized likelihood algorithm. In In Proceedings of the International Conference on Acoustics, Speech and Sig- nal Processing: 7–11 May 2001 Salt Lake City IEEE Computer Society Press; 2001:497-500. genome DNA. Genome Res 2000, 10:516 522. 4. Cawley SE, Wirth AI, Speed TP: Phat – a gene finding program for Plasmodium falciparum. Mol Biochem Parasitol 2001, 118:167-174. 5. Majoros WM, Pertea M, Antonescu C, Salzberg SL: GlimmerM, Exonomy and Unveil: three ab initio eukaryotic genefinders. Nucleic Acids Res 2003, 31:3601-3604. 27. Salzberg SL: On comparing classifiers: a critique of current research and methods. Data Mining and Knowledge Discovery 1999, 1:1-12. 6. Stanke M, Waack S: Gene prediction with a hidden Markov model and a new intron submodel. Bioinformatics 2003, 19:II215-II225. 28. Burge C, Karlin S: Prediction of complete gene structures in human genomic DNA. J Mol Biol 1997, 268:78-94. 7. Korf I: Gene finding in novel genomes. BMC Bioinformeltics 2004, 5:59. g J 29. Salzberg SL, Pertea M, Delcher AL, Gardner MJ, Tettelin H: Interpo- lated Markov models for eukaryotic gene finding. Genomics 1999, 59:24-31. 8. Majoros WM, Pertea M, Salzberg SL: TigrScan and Glimmer- HMM: two open-source ab initio eukaryotic gene finders. Bio- informatics 2004, 20:2878-2879. 30. Friedman J, Hastie T, Tibshirani R: Additive logistic regression: a statistical view of boosting. http://www.biomedcentral.com/1471-2105/5/206 BMC Bioinformatics 2004, 5:206 the above experimental design as a convenient opportu- nity to verify our expectation that it would also prove use- ful for objective analysis of gene finder accuracy. 18. Jelinek F: Statistical Methods for Speech Recognition Cambridge: Brad- ford Books; 1997. 18. Jelinek F: Statistical Methods for Speech Recognition Cambridge: Brad- ford Books; 1997. 19. Schlüter R, Macherey W, Müller B, Ney H: Comparison of dis- criminative training criteria and optimization methods for speech recognition. Speech Communication 2001, 34:287-310. 20. p g p 20. Johansen FT: A comparison of hybrid HMM architectures using global discriminative training. In In Proceedings of the Fourth International Conference on Spoken Language Processing: 3–4 October 1996 Philadelphia Piscataway IEEE Computer Society Press; 1996:498-501. Cross-vali Figure 6 Cross validation experiments Figure 6 Cross-validation experiments Five-fold cross-validation was used both in the gradient ascent and in the MLE-only experi- ments. For gradient ascent training, MLE was performed on four-fifths of the training set (T) and then gradient ascent was per- formed on the other one-fifth. A separate hold-out set (H) of 1000 genes was used to obtain an unbiased evaluation of all final models. the nucleotide level. This training protocol is illustrated in Figure 6. recognized that such "peeking" at the test set (by which we mean iterative re-estimation of model parameters from the training set after receiving accuracy feedback on the test set) by the training procedure can (unfortunately) artificially inflate reported accuracy numbers. For this rea- son, an additional 1000 genes were used for testing the gene finder after each cross-validation run. The results of this final testing were not made available to the optimizer, but are instead reported here as a more objective assess- ment of final model accuracy. We will refer to the training set as T and the additional 1000 genes for testing as H. BLAST [32] was used to ensure that no two genes in T∪H were more than 80% similar over 80% of their lengths at Several variations of this experiment were also performed. To evaluate the utility of splitting the training set and per- forming MLE and gradient ascent parameter estimation on separate subsets (as described above), we also per- formed MLE followed by gradient ascent training on the full training set T and again evaluated the induced models on H. To assess whether gradient ascent provided any improvement in accuracy we also trained a model on T using only MLE and evaluated that model on H. Although the virtues of cross-validation have been well explored in the context of many other applications, we decided to use Page 11 of 12 (page number not for citation purposes) http://www.biomedcentral.com/1471-2105/5/206 Acknowledgements This work was supported in part by NIH grants R01-LM06845 and R01- LM007938. Thanks to I. Korf, D. Kulp, M. Brent, J. Allen, M. Pertea, M. Pop, This work was supported in part by NIH grants R01-LM06845 and R01- LM007938. Thanks to I. Korf, D. Kulp, M. Brent, J. Allen, M. Pertea, M. Pop, p g g 22. Eddy S, Mitchison G, Durbin R: Maximum discrimination hidden Markov models of sequence consensus. J Comput Biol 1995, 2:9-23. A. Delcher, and L. Pachter for useful discussions. I. Korf provided valuable comments on a previous version of the manuscript and suggested the use of boosting in the training of signal sensors. A. Delcher, and L. Pachter for useful discussions. I. Korf provided valuable comments on a previous version of the manuscript and suggested the use of boosting in the training of signal sensors. 23. Krogh A: Two methods for improving performance of an HMM and their application for gene finding. In In Proceedings of the Fifth International Conference on Intelligent Systems for Molecular Biol- ogy: 21–25 June 1997 Halkidiki, Greece Edited by: Gaasterland T, Karp P, Karplus K, Ouzounis C, Sander C, Valencia A. Menlo Park: Amer- ican Association for Artificial Intelligence; 1997:179-186. References Annals of Statistics 2000, 38:337-374. f , 9. Korf I, Flicek P, Duan D, Brent MR: Integrating genomic homol- ogy into gene structure prediction. Bioinformatics 2001:140-148. g f 31. Manning CD, Schütze H: Foundations of statistical natural language processing Cambridge: MIT Press; 1999. gy g p f 10. Yeh R-F, Lim LP, Burge CB: Computational inference of homol- ogous gene structures in the human genome. Genome Res 2001, 11:803-816. p g g 32. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ: Basic local alignment search tool. J Mol Biol 1990, 215:403-410. 11. Alexandersson M, Cawley S, Pachter L: SLAM: Cross-species gene finding and alignment with a generalized pair hidden Markov model. Genome Res 2003, 13:496-502. 12. Zhang L, Pavlovic V, Cantor CR, Kasif S: Human-mouse gene identification by comparative evidence integration and evo- lutionary analysis. Genome Res 2003, 13:1190-1202. y y 13. Baum LE: An inequality and associated maximization tech- nique in statistical estimation for probabilistic functions of Markov processes. Inequalities 1972, 3:1-8. Publish with BioMed Central and every scientist can read your work free of charge "BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime." Sir Paul Nurse, Cancer Research UK Your research papers will be: available free of charge to the entire biomedical community peer reviewed and published immediately upon acceptance cited in PubMed and archived on PubMed Central yours — you keep the copyright Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp BioMedcentral Publish with BioMed Central and every scientist can read your work free of charge "BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime." Sir Paul Nurse, Cancer Research UK Your research papers will be: available free of charge to the entire biomedical community peer reviewed and published immediately upon acceptance cited in PubMed and archived on PubMed Central yours — you keep the copyright Submit your manuscript here: http://www.biomedcentral.com/info/publishing_adv.asp BioMedcentral Publish with BioMed Central and every scientist can read your work free of charge Publish with BioMed Central and every scientist can read your work free of charge p q 14. Rabiner LR: A tutorial on hidden Markov models and selected applications in speech recognition. Proc of the IEEE 1989, 77:257-285. 15. Krogh A: An introduction to hidden Markov models for bio- logical sequences. In In Computational Methods in Molecular Biology Edited by: Salzberg SL, Searls DB, Kasif S. Amsterdam: Elsevier Science BV; 1998:45-62. 16. Bahl LR, Brown PF, de Souza PV, Mercer RL: Maximum mutual information estimation of hidden Markov model parameters for speech recognition. In In Proceedings of the International Confer- ence on Acoustics, Speech and Signal Processing IEEE Computer Society Press; 1986:49-52. 17. Krogh A: Hidden Markov models for labeled sequences. In Pro- ceedings of the Twelfth IAPR International Conference on Pattern Recogni- tion: 9–12 October 1994 Jerusalem. Piscataway: IEEE Computer Society Press; 1994:140-144.
https://openalex.org/W4226417256	https://library.oapen.org/bitstream/20.500.12657/54045/1/978-981-16-9797-5.pdf	English	null	R&D Management Practices and Innovation: Evidence from a Firm Survey	SpringerBriefs in economics	2,022	cc-by	42,450	SpringerBriefs in Economics SpringerBriefs present concise summaries of cutting-edge research and practical applications across a wide spectrum of ﬁelds. Featuring compact volumes of 50 to 125 pages, the series covers a range of content from professional to academic. Typical topics might include: • A timely report of state-of-the art analytical techniques • A bridge between new research results, as published in journal articles, and a contextual literature review • A snapshot of a hot or emerging topic • An in-depth case study or clinical example • A presentation of core concepts that students must understand in order to make independent contributions • A presentation of core concepts that students must understand in order to make independent contributions SpringerBriefs in Economics showcase emerging theory, empirical research, and practical application in microeconomics, macroeconomics, economic policy, public ﬁnance, econometrics, regional science, and related ﬁelds, from a global author community. SpringerBriefs in Economics showcase emerging theory, empirical research, and practical application in microeconomics, macroeconomics, economic policy, public ﬁnance, econometrics, regional science, and related ﬁelds, from a global author community. Briefs are characterized by fast, global electronic dissemination, standard publishing contracts, standardized manuscript preparation and formatting guidelines, and expedited production schedules. More information about this series at https://link.springer.com/bookseries/8876 Shoko Haneda · Arito Ono Shoko Haneda Faculty of Commerce Chuo University Tokyo, Japan Arito Ono Faculty of Commerce Chuo University Tokyo, Japan This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore Shoko Haneda Faculty of Commerce Chuo University Tokyo, Japan ISSN 2191-5504 ISSN 2191-5512 (electronic) SpringerBriefs in Economics ISBN 978-981-16-9796-8 ISBN 978-981-16-9797-5 (eBook) https://doi.org/10.1007/978-981-16-9797-5 ISSN 2191-5504 ISSN 2191-5512 (electronic) SpringerBriefs in Economics ISBN 978-981-16-9796-8 ISBN 978-981-16-9797-5 (eBook) https://doi.org/10.1007/978-981-16-9797-5 © The Author(s) 2022. This book is an open access publication. Open Access This book is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribu- tion and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made. The images or other third party material in this book are included in the book’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the book’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a speciﬁc statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore Contents 1 R&D Management Practices and Innovation: Evidence from a Firm Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Key Questions and Related Literature . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2.1 Organizational Structure of R&D Activities . . . . . . . . . . . . . . . 3 1.2.2 Staged Project Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.2.3 Compensation and Incentive Schemes for R&D Personnel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.2.4 Risk Preferences and Corporate Culture . . . . . . . . . . . . . . . . . . 9 1.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.3.1 Survey Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.3.2 Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Contents . . . . . . . 13 1.4 R&D Outcomes and Inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 1.4.1 R&D Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 1.4.2 R&D Inputs: R&D Expenditures . . . . . . . . . . . . . . . . . . . . . . . . 19 1.4.2.1 Amount of R&D Expenditure . . . . . . . . . . . . . . . . . . . 19 1.4.2.2 Funding Sources for R&D Expenditure . . . . . . . . . . 21 1.4.2.3 Determinants of R&D Expenditure . . . . . . . . . . . . . . 23 1.4.3 R&D Inputs: R&D Personnel . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 1.5 Organizational Structure of R&D Activities . . . . . . . . . . . . . . . . . . . . . 31 1.5.1 Centralized/Decentralized R&D Structure . . . . . . . . . . . . . . . . 31 1.5.2 Allocation of R&D Expenditure and R&D Personnel in Firms with a Hybrid R&D Structure . . . . . . . . . . . . . . . . . . . 35 1.5.3 Initiative in Hiring R&D Personnel . . . . . . . . . . . . . . . . . . . . . . 38 1.6 R&D Project Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 1.6.1 Overview on R&D Projects . . . . . . . . . . . . . . . . . . . . . . . . Preface This monograph provides a detailed account of what ﬁrms do in their R&D activities. In particular, using a unique survey of ﬁrms in Japan, we focus on the following four aspects of R&D management: the organizational structure of R&D, staged project management for R&D projects, compensation and incentive schemes for R&D personnel, and a ﬁrm’s risk preferences and corporate culture. We also examine whether and how R&D management practices are linked to the likelihood of ﬁrms’ success in making product innovations and the choice between explorative (radical) and exploitive (incremental) innovation. While previous studies recognize that R&D management practices are important drivers of innovation, most take the form of case studies that focus on a partic- ular aspect of R&D management, and there are few studies that systematically and quantitatively examine the link between various R&D management practices and innovation. To ﬁll the gap in the literature, we designed and conducted the orig- inal ﬁrm survey, the “Survey of R&D Management Practices,” in January–February 2020. This monograph presents our ﬁrst step in examining how R&D management is associated with corporate innovation using the survey. We hope that this monograph is useful for readers interested in a detailed analysis of the relationship between R&D management and innovation using quantitative data. ThismonographistheproductofaresearchprojectfundedbyaJSPSGrant-in-Aid for Scientiﬁc Research (B) No. 19H01488). We would like to thank an anonymous referee, Christian Rammer, Ralph Paprzycki, and participants of the 2021 Annual Meeting of the Japan Society for Research Policy and Innovation Management for their useful comments, Yuya Ikeda and Tomohiko Inui for conducting the survey with us, and Koki Kurihara for superb research assistance. We gratefully acknowledge the cooperation of the National Institute of Science and Technology Policy (NISTEP) in conducting the “Survey of R&D Management Practices.” Ono gratefully acknowl- edges that this monograph was prepared while he was a visiting researcher at NISTEP, v v Preface Preface vi while Haneda acknowledges the hospitality she received while she was a visiting researcher at the ZEW (Leibniz Centre for European Economic Research). while Haneda acknowledges the hospitality she received while she was a visiting researcher at the ZEW (Leibniz Centre for European Economic Research). Shoko Haneda Arito Ono Tokyo, Japan Tokyo, Japan Contents . . . . . 40 1.6.1.1 Number of R&D Projects . . . . . . . . . . . . . . . . . . . . . . 40 1.6.1.2 Duration of R&D Projects . . . . . . . . . . . . . . . . . . . . . . 42 1.6.1.3 Termination or Suspension of R&D Projects . . . . . . 42 1.6.2 Staged Project Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 1.6.2.1 Milestones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 vii viii Contents 1.6.2.2 Feedback . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 1.7 Evaluation of R&D Personnel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 1.7.1 Salary Schemes for R&D Personnel . . . . . . . . . . . . . . . . . . . . . 57 1.7.2 Performance- and Ability-Based Evaluation . . . . . . . . . . . . . . . 61 1.7.2.1 Weights on Performance and Ability in Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 1.7.2.2 Criteria for the Evaluation of R&D Personnel . . . . . 62 1.7.3 Incentive Schemes for R&D Personnel . . . . . . . . . . . . . . . . . . . 65 1.7.4 Incentives for Long-Term Success . . . . . . . . . . . . . . . . . . . . . . . 69 1.7.4.1 Rewards for Long-Term Success . . . . . . . Contents . . . . . . . . . 69 1.7.4.2 Possibility of Promotion . . . . . . . . . . . . . . . . . . . . . . . 71 1.8 Risk Preferences and Corporate Culture . . . . . . . . . . . . . . . . . . . . . . . . 72 1.8.1 Risk Preferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 1.8.2 Corporate Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 1.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 Survey of R&D Management Practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 Glossary of Terms in the “Survey of R&D Management Practices” . . . . . . 94 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 About the Authors Shoko Haneda is Professor at the Faculty of Commerce, Chuo University, Japan. She also served as visiting researcher at the National Institute of Science and Technology Policy (2011–2020). Her main ﬁelds of research are innovation, business economics, and management. She has published papers in Research Policy, Economics of Inno- vation and New Technology, and other scholarly journals. She received a B.A. in Mathematics from Tsuda University and a Ph.D. in Economics from Tsukuba University. Arito Ono is Professor at the Faculty of Commerce, Chuo University, Japan. Prior to joining Chuo University in 2015, he was a senior economist at the Mizuho Research Institute, and a senior economist at the Institute for Monetary and Economic Studies, Bank of Japan (2009–2011). He also served as a member of several working groups at the Financial System Council, Financial Services Agency (2011–2015) and as an advisor at the Research and Statistics Department, Bank of Japan (2015). His main ﬁelds of research are banking and corporate ﬁnance. He is a coauthor and coeditor of the book titled The Economics of Interﬁrm Networks and has published academic articles in the International Economic Review, the Journal of Banking & Finance, the Journal of Financial Stability, the Journal of Money, Credit, and Banking, and Real Estate Economics, among others. He received a B.A. in economics from the University of Tokyo in 1991 and a Ph.D. in economics from Brown University in 2001. ix ix © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5_1 1.1 Introduction Innovation plays an important role in increasing productivity and economic growth. Muchofthisinnovationisdrivenbytheresearchanddevelopment(R&D)activitiesof business ﬁrms as part of their efforts to develop new products and processes. Against this background, how ﬁrms manage their R&D activities has become an increasingly important issue (see, e.g., Teece 1996; Azoulay and Lerner 2012). Meanwhile, there is a growing literature which argues that management practices are important factors in explaining differences in productivity across ﬁrms (e.g., Bloom and Van Reenen 2007, 2010; Bloom et al. 2019). Because innovation is a key determinant of a ﬁrm’s productivity, this literature suggests that there is a link between R&D management practices, innovation, and productivity. Yet, while there are many case studies and small-sample studies describing how well-articulated R&D management practices create innovation (e.g., Hartmann and Hassan 2006; Hullova et al. 2019; Smolnik and Bergmann 2020), there are relatively few studies that empirically examine the link between R&D management practices and innovation using large-scale data (notable exceptions are Laursen and Foss 2003; Haneda and Ito 2018). Moreover, there are even fewer studies that systematically investigate from a variety of angles how ﬁrms manage R&D activities in practice and examine which R&D management practices are beneﬁcial for, or detrimental to, innovation. To ﬁll the gap in the literature, this monograph seeks to better understand what ﬁrms do in their R&D activities using data from a unique survey of ﬁrms in Japan, the “Survey of R&D Management Practices,” which was implemented by a research team including the authors in January–February 2020. This survey focuses on the following four aspects of R&D management: the organizational structure of R&D, staged project management for R&D projects, compensation and incentive schemes for R&D personnel, and a ﬁrm’s risk preferences and corporate culture. Using an original data set that matches the survey data with ﬁrm-level micro data, we provide detailedinformationonﬁrms’R&Dmanagementpractices.Wealsoexaminewhether 1 © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5_1 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 2 and how various R&D management practices are linked to the likelihood of ﬁrms’ success in making product innovations. 1 Note that our analyses are descriptive in nature and do not provide evidence of any causal rela- tionships. While we discuss possible mechanisms that may explain our results, we do not formally test them, so that these discussions should be regarded as conjectures. 1.1 Introduction Further, reﬂecting the fact that much of the literature on innovation focuses on the tension between explorative (radical) innova- tions and exploitative (incremental) innovations (e.g., March 1991; Manso 2011), we also investigate whether and how R&D management practices are associated with the choice between explorative and exploitative product innovation. The remainder of this monograph is organized as follows. Section 1.2 explains the key question about R&D management practices asked in the survey and provides a review of the related literature motivating the questions. Section 1.3 outlines the survey design and presents basic summary statistics. Section 1.4 explains the vari- ables we use for R&D outcomes and R&D inputs. Next, Sects. 1.5 to 1.8 are the main parts of this monograph. They provide detailed information on ﬁrms’ R&D management practices, namely, on the organizational structure of R&D activities (Sect. 1.5), staged project management (Sect. 1.6), compensation and incentive schemes for R&D personnel (Sect. 1.7), and ﬁrms’ risk preferences and corporate culture (Sect. 1.8). We also conduct a simple statistical analysis (two-sample equal variance t-tests) in these sections to examine the relationship between R&D manage- ment practices and R&D outcomes and discuss whether the results are consistent with the literature discussed in Sect. 1.2.1 While we need to control for a range of factors such as ﬁrm size and industry to examine the link between R&D management prac- tices and R&D outcomes more rigorously, we believe that our simple analysis serves as a useful ﬁrst step for future studies. Section 1.9 concludes. 1.2.1 Organizational Structure of R&D Activities Organizational economics theory suggests that a ﬁrm’s internal organizational struc- ture affects its performance (e.g., Gibbons and Roberts 2012). We examine how a ﬁrm’s internal organizational structure of R&D activities affects its innovation from two separate but intertwined aspects. Speciﬁcally, we focus on how the delegation of authority to R&D organizations and the centralization or decentralization of R&D organization structures are linked with innovation outcomes. First, to investigate the effect of delegation on innovation, we measure to what extent the authority to hire employees (R&D personnel) and to terminate/suspend or continue ongoing R&D projects is allocated between R&D organizations and corporate headquarters. Aghion and Tirole (1997) argue that the basic trade-off in delegating authority is between initiative and control. That is, the transfer of authority to an agent (an R&D organization in our case) increases the ability of the agent to take the initiative to acquire relevant knowledge for the project, but this comes at the expense of the principal’s (the headquarters’) control over the choice and manage- ment of projects. Aghion and Tirole’s (1997) argument suggests that to what extent ﬁrms allocate authority to R&D organizations depends on whether their corporate headquarters have sufﬁcient knowledge about choosing R&D personnel and running R&D projects. Consistent with Aghion and Tirole’s (1997) prediction, Acemoglu et al. (2007) ﬁnd that ﬁrms closer to the technological frontier are more likely to delegate authority to the manager (agent) of the ﬁrm’s “proﬁt center” business unit because there is less public information about the new technology from which corpo- rate headquarters (the principal) can learn. Meanwhile, Kastl et al. (2013) empirically examine the link between delegation and R&D expenditure. Speciﬁcally, using a ﬁrm survey of Italian manufacturing ﬁrms, which asks a respondent ﬁrm whether R&D- related decisions, as well as administrative, ﬁnancial, and business decisions, are autonomously made by separate divisions, they construct several measures of dele- gation. They ﬁnd a positive correlation between the delegation measures and R&D expenditures, which suggests that ﬁrms in which more authority is delegated to the R&D division tend to spend more on R&D. In this study we measure the delegation of authority to hire researchers and manage on-going R&D projects to R&D orga- nizations and examine the statistical association of such delegation with innovation outcomes. Second, we measure whether a ﬁrm’s R&D activities are organized in a centralized or decentralized manner. 1.2 Key Questions and Related Literature This section outlines the key questions about R&D management practices that we examine in this monograph. We formulate our key questions focusing on the following aspects: the organizational structure of R&D activities, staged projects management, compensation and incentive schemes for R&D personnel, and ﬁrms’ risk preferences and corporate culture. We also review the related literature to which we refer in constructing these questions. In his seminal study, March (1991) argues that for many businesses innovation is difﬁcult because there is a trade-off between allocating resources to the exploration of new possibilities and the exploitation of well-known approaches. The tension between exploration and exploitation is analyzed more formally by Manso (2011), who constructs a principal-agent model in which he embeds a Bayesian decision model known as the bandit problem. Manso (2011) shows that the optimal scheme for promoting innovation (exploration) is one that exhibits substantial tolerance for early failure, reward for long-term success, and timely feedback on performance. In formulating our key questions, we rely not only 2 Key Questions and Related Literature 3 1.2 on the empirical implications of March’s (1991) and Manso’s (2011) analyses but also those of other studies that we discuss in detail below. on the empirical implications of March’s (1991) and Manso’s (2011) analyses but also those of other studies that we discuss in detail below. 1.2.1 Organizational Structure of R&D Activities We deﬁne a ﬁrm’s R&D organization structure as central- ized if R&D activities are highly independent of business units and as decentralized if R&D activities are directly controlled by separate business units. The literature indi- cates that there is a trade-off between centralized and decentralized R&D structures (Azoulay and Lerner 2012). The advantage of adopting a decentralized R&D struc- ture is that managers of R&D organizations (such as a pharmaceuticals development 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 4 division) are likely to have superior knowledge about the local market and the need of customers and be better placed to prevent R&D employees from losing sight of market imperatives. On the other hand, decentralized R&D may prevent the pooling of knowledge and spillovers from R&D activities within other units of the ﬁrm, and managers of speciﬁc business units may lack the knowledge and skills for R&D activ- ities that are non-local and/or explorative in nature. Centralized R&D (such as in a central research laboratory) can potentially overcome these drawbacks by providing a place for nonlocal research activities and long-term and explorative projects but runs the risk of losing information about customer needs and choosing R&D projects for their scientiﬁc interest. This trade-off suggests that decentralized R&D structures are more suitable for incremental innovation, while centralized R&D structures are more conducive to radical innovation. A number of studies report empirical ﬁndings providing evidence for this trade-off. For instance, Argyres and Silverman (2004) ﬁnd that ﬁrms with centralized R&D structures generate innovations that have a higher level of impact than ﬁrms with decentralized R&D structures.2 Using a sample of 71 large U.S. corporations taken from a survey of R&D executives, they construct measures of R&D organization structures and ﬁnd a positive association between the degree of centralization and the number of patent citations received, which is a conventional proxy for the impact of innovations. Their ﬁnding suggests that central- ized R&D structures are more likely to generate radical innovations than decentral- ized structures. In this study, we construct a centralization measure of R&D structures that is similar to Argyres and Silverman’s (2004) and examine its association with the degree to which innovations are radical or incremental in nature.3,4 3 Argyres and Silverman (2004) and Argyres et al. (2020) use the share of the R&D budget that is allocated by corporate headquarters as another measure for the degree of centralization of R&D structures. Meanwhile, Arora et al. (2011, 2014) develop yet another empirical measure of the decentralization of R&D, namely, the share of patents assigned to afﬁliates (as opposed to the parent). 2 Also see Argyres et al. (2020), who ﬁnd that the positive link between centralized R&D and the impact and depth of innovation works through the increase in the connectedness of internal inventor networks: researchers in centralized R&D structures are likely to undertake technological search of greater breath so as to produce more radical innovations that beneﬁt multiple divisions within the ﬁrm. 4 Although we deﬁne the degree of decentralization/centralization of R&D structures in terms of the degree of independence from a ﬁrm’s business units, it should be noted that R&D decentralization and the delegation of authority over R&D activities may be closely intertwined. In fact, some studies (e.g., Argyres and Silverman 2004; Acemoglu et al. 2007) deﬁne decentralization in terms of the degree of delegation. Using our survey, we examine the correlation between delegation and decentralization in footnote 19 in Sect. 1.5.3. 2 Also see Argyres et al. (2020), who ﬁnd that the positive link between centralized R&D and the impact and depth of innovation works through the increase in the connectedness of internal inventor networks: researchers in centralized R&D structures are likely to undertake technological search of greater breath so as to produce more radical innovations that beneﬁt multiple divisions within the ﬁrm. 3 Argyres and Silverman (2004) and Argyres et al. (2020) use the share of the R&D budget that is allocated by corporate headquarters as another measure for the degree of centralization of R&D structures. Meanwhile, Arora et al. (2011, 2014) develop yet another empirical measure of the decentralization of R&D, namely, the share of patents assigned to afﬁliates (as opposed to the parent). 4 Although we deﬁne the degree of decentralization/centralization of R&D structures in terms of the degree of independence from a ﬁrm’s business units, it should be noted that R&D decentralization and the delegation of authority over R&D activities may be closely intertwined. In fact, some studies (e.g., Argyres and Silverman 2004; Acemoglu et al. 2007) deﬁne decentralization in terms of the degree of delegation. Using our survey, we examine the correlation between delegation and decentralization in footnote 19 in Sect. 1.5.3. 1.2.2 Staged Project Management The management and funding of R&D projects often proceeds in stages. For example, the “Stage-Gate” method proposed by Cooper (1988, 2017) sets concrete interim 2 Also see Argyres et al. (2020), who ﬁnd that the positive link between centralized R&D and the impact and depth of innovation works through the increase in the connectedness of internal inventor networks: researchers in centralized R&D structures are likely to undertake technological search of greater breath so as to produce more radical innovations that beneﬁt multiple divisions within the ﬁrm. 4 Although we deﬁne the degree of decentralization/centralization of R&D structures in terms of the degree of independence from a ﬁrm’s business units, it should be noted that R&D decentralization and the delegation of authority over R&D activities may be closely intertwined. In fact, some studies (e.g., Argyres and Silverman 2004; Acemoglu et al. 2007) deﬁne decentralization in terms of the degree of delegation. Using our survey, we examine the correlation between delegation and decentralization in footnote 19 in Sect. 1.5.3. 2 Key Questions and Related Literature 5 1.2 goals, referred to as “gates” or “milestones,” in each stage of an R&D project, and the project is continued only if the milestones are met. It is also well known that venture capital (VC) investors typically make staged investments in venture ﬁrms, holding open the option of abandoning a venture ﬁrm if it fails to meet milestones (Sahlman 1990). The literature on VC ﬁnds that staging is a way for VC investors (principals) to monitor ﬁrms (agents) and mitigate agency problems (Gompers 1995; Kaplan and Strömberg 2003; Tian 2011) and that staging allows VC investors to learn about the agent over time and sort good projects from bad ones (Dahiya and Ray 2012). On the other hand, staging may lead to underinvestment by VC investors at the early stage (Wang and Zhou 2004) and exacerbate venture ﬁrms’ focus on short- term success to continually look attractive to VC investors (Cornelli and Yosha 2003; Yung 2019). In our view, the two-period model of the innovation process presented by Manso (2011) captures the advantages and disadvantages of staging well. In the model, the agent chooses between two actions in each stage: exploration or exploitation. Exploitation consists of well know actions or work methods to achieve incremental innovations with a known probability of success, while exploration consists of new untested actions or work methods to achieve radical innovations. 1.2.2 Staged Project Management The probability of success of radical innovations is unknown and the agent updates his/her beliefs about the probability of success once he/she has attempted radical innovation in the ﬁrst stage. Because both actions entail private costs to the agent, the agent has an incentive to shirk. Manso’s (2011) model makes two predictions with respect to staging. First, the effect that the threat of termination has on exploration is ambiguous because it prevents the agent from shirking but encourages the agent to choose a project with a higher probability of success, i.e., exploitation. Depending on which of these two effects is more important, staging of innovation projects may either encourage or discourage an agent from choosing exploration. Second, feedback on interim outcomes of the project provides incentives for exploration because it allows interim adjustments by the agent and increases the probability of success of the project. Several empirical studies examine Manso’s (2011) predictions. Ederer and Manso (2013) provide experimental evidence on the effects of termination. Speciﬁcally, they conducted a laboratory experiment in which participants operate a hypothetical computerized lemonade stand and choose between exploitation, i.e., making minor adjustments to the business strategy (e.g., ﬁne-tuning the product mix of lemonade), or exploration, i.e., making major adjustments to the business strategy (e.g., changing the location of the lemonade stand). To study the effect of termination, they divide participants into two groups: one whose lemonade stands were eliminated if they underperformed in the ﬁrst half of the experiment and another whose lemonade stands continued regardless of the performance in the ﬁrst half. Ederer and Manso (2013) ﬁnd that participants in the latter group were more likely to choose an explo- rative strategy, suggesting that the threat of termination undermines the incentives for explorative innovation. Meanwhile, using a sample of VC-backed initial public offering (IPO) ﬁrms, Mao et al. (2014) ﬁnd that IPO ﬁrms were less innovative, as measuredbythenumberofpatentsgrantedandthenumberoffuturecitationsreceived 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 6 by each patent, when VC investors held a larger number of VC ﬁnancing rounds (stages). Further, in the realm of scientiﬁc research, Azoulay et al. (2011) examine whether the funding program of the Howard Hughes Medical Institute (HHMI), which tolerates early failure and provides detailed and high-quality feedback to the researcher, encourages exploration more than the funding program of the National Institutes of Health (NIH), which imposes more stringent interim reviews. 1.2.2 Staged Project Management They ﬁnd that researchers who used HHMI grants produced higher-impact articles than NIH- funded researchers. The empirical results obtained by Azoulay et al. (2011) suggest that more “forgiving” scientiﬁc research grants with extensive feedback lead to more explorative innovations than grants with stricter interim reviews. As explained above, staging is prevalent in the management and ﬁnancing of R&D projects and VC investment in start-up ﬁrms, and there are several theoretical and empirical studies that examine the determinants and effects of staging in the context of such ﬁrms. However, as far as we are aware, there are few empirical studies that investigate staging in the context of R&D projects. An exception is the study by Andries and Hünermund (2020), which examines the impact of staging on the initiation/abandonment of innovation projects. Our survey contributes to the literature by providing a systematic description of the staging in R&D projects. Concretely, we ﬁrst ask respondent ﬁrms whether they implement staged project management of their R&D projects and examine the statistical association between staging and the likelihood of making product innovations as well as the association between staging and the choice between explorative and exploitative innovation. Since the VC literature surveyed above suggests that there are both advantages and disadvantages to staging, it is an empirical matter whether the correlation between staging and making product innovations is positive or negative and whether staging is correlated with exploration or exploitation. In addition, we ask the following questions to examine whether our data are consistent with Manso’s (2011) predictions with respect to staging. First, to examine the effect of the threat of termination on exploration, we ask whether a ﬁrm sets intermediate goals (“milestones”) for the interim evaluation of a project. If the answer to this question is positive, we then ask to what extent the ﬁrm considers whether the milestones were achieved when assessing whether to terminate/suspend or continue the R&D project. Second, to examine the effect of feedback on exploration, we ask whether a ﬁrm provides feedback on the interim evaluation results to the R&D personnel in charge of an R&D project. If the answer to this question is positive, we then ask who provides feedback: other research teams in the R&D organizations within the ﬁrm, non-R&D business units within the ﬁrm and the head ofﬁce, or external experts outside the ﬁrm. 1.2.3 Compensation and Incentive Schemes for R&D Personnel Incentives play an in important role in the organization of R&D activities. A key issue therefore is how ﬁrms assess and compensate/reward their R&D personnel. Since the interests of employees and their employers are not always aligned, many studies have examined how ﬁrms design compensation contracts and provide incen- tives to induce employees to work in the ﬁrm’s interest (see Prendergast (1999) for a survey). However, these studies also highlight that providing incentives for innova- tion is especially difﬁcult. Because innovative projects are risky and their outcomes are unpredictable, standard pay-for-performance compensation is less effective in inducing effort in the case of R&D than other employees. Worse still, pay-for- performance compensation may be detrimental in getting R&D personnel and/or managers to choose explorative R&D projects because they are, by deﬁnition, more likely to fail (Holmström 1989; Manso 2011).5 To deal with such problems, Manso (2011) argues, tolerance for early failure and reward for long-term success is essen- tial for motivating radical innovation. However, while Manso lists several long-term compensations plans for executives and managers (e.g., stock options with long vesting period), he does not discuss long-term incentive schemes for employees. Although providing incentive schemes for R&D employees is fraught with difﬁ- culties, previous studies—which we outline below—as well as pre-interviews we conducted with managers of R&D organizations and human resources departments in Japanese ﬁrms suggest that many ﬁrms try to devise compensation and incentive schemes for R&D personnel to motivate innovation. In our survey, we focus on the following aspects of human resource management practices for R&D personnel to understand what Japanese ﬁrms do. First, we ask about the relative weights given to ability and performance in R&D employee evaluations, where ability refers to the ability demonstrated in performing a job and performance refers to the level of achievement in performing the job. Stan- dard agency theory predicts that a ﬁrm will not adopt pay-for-performance when performance measures are noisy in the sense that they do not adequately reﬂect employees’ input of effort (Holmström and Milgrom 1991). Because more ambi- tious projects entail larger risk and uncertainty, theory suggests that ﬁrms are less likely to employ performance evaluation in the case of explorative R&D projects. However, there is little empirical support for a negative relationship between the risk (uncertainty) of projects and the use of pay-for-performance (Prendergast 2011). 5 Because incentives, including pay-for-performance, may not be effective in inducing innovation, some studies postulate contractual incompleteness, based on the recognition that writing incentive contracts is too complex and costly. See, for example, Aghion and Tirole (1994) and Hellman and Thiele (2011). 1.2.2 Staged Project Management Using these two questions in our survey, we examine how the threat of termination and feedback is associated with the success of R&D projects and the choice between radical/incremental innovation. 7 1.2 Key Questions and Related Literature 7 1.2.3 Compensation and Incentive Schemes for R&D Personnel To explain why this is the case, Prendergast (2002) constructs a theoretical model that predicts that uncertainty may be positively related to pay-for-performance through a different mechanism. Speciﬁcally, he argues that uncertain environments lead to the delegation of responsibility to employees because in very uncertain settings, it is hard 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 8 to tell employees what to do. Because the principal delegates control to employees, performance-based pay becomes the only way to compensate for employees’ unob- servable effort. Consistent with Prendergast’s (2002) prediction, Foss and Laursen (2005) ﬁnd that the extent to which ﬁrms innovate is positively correlated with the use of pay-for-performance. Using our survey, we examine whether there is a link between the use of performance evaluation on the one hand and the likelihood of product innovations and of explorative/exploitative innovations on the other. p p p Second, to further hone in on human resource management practices in more detail, we ask ﬁrms whether they employ various practices for the evaluation of R&D personnel from a list we provide in our questionnaire. Some of the items on the list refer to R&D employees’ performance (e.g., patent applications/registrations), while others refer to their ability (e.g., the acquisition of qualiﬁcations/degrees). Using the responses, we examine the link between performance-based and ability- based evaluation on the one hand and innovation outcomes on the other from a slightly different perspective than the previous question. Third, we ask ﬁrms about pecuniary and non-pecuniary incentive schemes for R&D personnel. Pecuniary incentives are monetary rewards based on the outcomes of innovative activities by R&D personnel (e.g., rewards based on the number of patent applications), while non-pecuniary incentives are non-monetary rewards/subsidies (e.g., dispatch to university and/or support for studying abroad), which may increase R&D employees’ intrinsic motivation for innovation. Several studies suggest that pecuniary incentives that provide extrinsic motivation for workers may adversely affect their intrinsic motivation, such as pride in their work and enjoyment in carrying out tasks (Bénabou and Tirole 2003; Kreps 1997). In the context of R&D manage- ment, such intrinsic motivation includes, for instance, the intellectual challenge of contributing to scientiﬁc and technological progress. Previous empirical studies that examine the link between pecuniary and non-pecuniary incentives and innovation outputs report mixed results. 1.2.3 Compensation and Incentive Schemes for R&D Personnel Using ﬁrm-level panel data for listed ﬁrms in Japan, Onishi (2013) reports a positive association between monetary compensation plans for employee inventions and the number of patent citations but no association with the number of patent applications. In contrast, using a 2001 court decision that effectively forced Japanese ﬁrms to strengthen pecuniary incentives based on the commercial success of an invention as an exogenous instrument for pecuniary incen- tives, Onishi et al. (2021) ﬁnd that pecuniary incentives decreased the number of science-based patents. Using individual-level data for R&D employees, Sauermann and Cohen (2010) examine the relationship between extrinsic and intrinsic motives of R&D personnel on the one hand and their innovative performance on the other hand. They ﬁnd that R&D employees with stronger motives such as pay (extrinsic motive) and intellectual challenge and independence (intrinsic motives) produce a larger number of patent applications (innovative outputs). Sauermann and Cohen’s ﬁndings (2010) suggest that both extrinsic and intrinsic motives are important for 1.2 Key Questions and Related Literature 9 1.2 innovation.6 Our study differs from Sauermann and Cohen’s (2010) in that we ask ﬁrms about the pecuniary and non-pecuniary incentive schemes they employ, while Sauermann and Cohen (2010) ask employees about their subjective motives. Finally, we use some of the questions outlined above to examine whether reward for long-term success (Manso 2011) is correlated with innovation. Speciﬁcally, the question about the evaluation of R&D personnel contains as one of the possible criteria the “amount of sales generated by new products to which the R&D employee contributed,” while the question about incentive schemes for R&D personnel contains as one of the possible incentives “rewards based on the amount of proﬁts from inven- tions and patents (invention reward schemes).” In addition, we regard promotion as another potential reward for long-term success. The questionnaire contains a ques- tion asking whether any of the directors on the board of the ﬁrm belonged to an R&D organization within the ﬁrm in the past, and we regard the answer to this question as an indicator of whether promotion to the board is possible and hence a poten- tial incentive for R&D personnel. Promotion is one of the most common means of rewarding white-collar workers for effort and long-term outcomes, and a substantial share of directors on the boards of Japanese ﬁrms are promoted internally. 6 Dewett (2007) reports that the intrinsic motivation of R&D employees is positively associated with their willingness to take risk, but that the statistical association between intrinsic motivation and employee creativity depends on the proxy used for creativity. 1.2.3 Compensation and Incentive Schemes for R&D Personnel However, to what level in the hierarchy R&D personnel can be promoted may vary across ﬁrms. We use this question to examine whether the possibility of promotion to top-level management works as an effective incentive scheme for innovation. 1.2.4 Risk Preferences and Corporate Culture Some studies suggest that corporate culture, which is deﬁned as “a set of norms and values that are widely shared and strongly held throughout the organization” (O’Reilly and Chatman 1996), may be an important driver of ﬁrm performance. The empirical study by Guiso et al. (2015) shows that corporate culture impacts a ﬁrm’s performance more than corporate governance. Manso (2011) argues that nurturing a corporate culture that encourages experimentation and tolerates early failure is important for innovation because it is difﬁcult, if not impossible, to devise compen- sation and incentive schemes that credibly motivate innovation among employees, as discussed in Sect. 1.2.2. Some studies ﬁnd empirical evidence that is consistent with Manso’s (2011) argument. For instance, Ederer and Manso (2013) show that risk aversion plays an important role in explaining differences in participants’ behavior in their hypothetical lemonade stand experiment (see Sect. 1.2.2). Meanwhile, using data on large pharmaceutical ﬁrms’ drug development decisions, Krieger et al. (2022) show that risk aversion leads ﬁrms to underinvest in radical innovation. Further, in experiments with master’s degree students, Carson et al. (2020) ﬁnd that partici- pants that are more tolerant of risk are more likely to choose higher-risk projects. In 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 10 contrast, ﬁnancial rewards that encourage participants to disproportionately under- take higher-risk projects do not induce most participants to invest in such projects. In the realm of the venture capital (VC) industry, Tian and Wang (2014) report that IPO ﬁrms backed by more failure-tolerant VC investors are more innovative. Finally, based on a survey of large North American ﬁrms, Graham et al. (2021) report that 57% of senior executives surveyed think that corporate culture has a “big effect” on their ﬁrms’ creativity while 41% think that corporate culture has a “big effect” on their ﬁrms’ willingness to take on risky projects. Constructing measures of cultural values from their survey, Graham et al. (2021) ﬁnd that “adaptability” is positively correlated with creativity while “results-orientation” is negatively correlated with creativity. In our survey, we ask several questions to measure a ﬁrm’s risk preferences and corporate culture and examine their impact on innovation. To measure a ﬁrm’s risk preferences, we included the following three questions in our survey. First, we asked ﬁrms whether they were taking appropriate risks in their R&D projects. Asking a similar question in their survey, Graham et al. 1.2 Key Questions and Related Literature 1.2 Key Questions and Related Literature 1.2 Key Questions and Related Literature 11 Long-term change Individuality and ﬂexibility TransformaƟonal change Culture Type: Clan Culture Type: Adhocracy OrientaƟon: Collaborate OrientaƟon: Create Leader type: Facilitator Leader type: Innovator Mentor Entrepreneur Team builder Visionary Value drivers: Commitment Value drivers: InnovaƟve outputs CommunicaƟon TransformaƟon Development Agility Theory of eﬀecƟveness Human development and high commitment produce eﬀecƟveness Theory of eﬀecƟveness InnovaƟveness, vision, and constant change produce eﬀecƟveness Culture Type: Hierarchy Culture Type: Market OrientaƟon: Control OrientaƟon: Compete Leader type: Coordinator Leader type: Hard-driver Monitor CompeƟtor Organizer Producer Value drivers: Eﬃciency Value drivers: Market share Timeliness Goal achievement Consistency and uniformity Proﬁtability Theory of eﬀecƟveness Control and eﬃciency with capable processes produce eﬀecƟveness Theory of eﬀecƟveness Aggressively compeƟng and customer focus produce eﬀecƟveness Incremental change Stability and control Fast change Internal maintenance External posiƟoning Fig. 1.1 The competing values framework (Source Cameron et al. 2014, Fig. 3.1) Long-term change Individuality and ﬂexibility TransformaƟonal change Culture Type: Clan Culture Type: Adhocracy OrientaƟon: Collaborate OrientaƟon: Create Leader type: Facilitator Leader type: Innovator Mentor Entrepreneur Team builder Visionary Value drivers: Commitment Value drivers: InnovaƟve outputs CommunicaƟon TransformaƟon Development Agility Theory of eﬀecƟveness Human development and high commitment produce eﬀecƟveness Theory of eﬀecƟveness InnovaƟveness, vision, and constant change produce eﬀecƟveness Culture Type: Hierarchy Culture Type: Market OrientaƟon: Control OrientaƟon: Compete Leader type: Coordinator Leader type: Hard-driver Monitor CompeƟtor Organizer Producer Value drivers: Eﬃciency Value drivers: Market share Timeliness Goal achievement Consistency and uniformity Proﬁtability Theory of eﬀecƟveness Control and eﬃciency with capable processes produce eﬀecƟveness Theory of eﬀecƟveness Aggressively compeƟng and customer focus produce eﬀecƟveness Incremental change Stability and control Fast change Internal maintenance External posiƟoning Fig. 1.1 The competing values framework (Source Cameron et al. 2014, Fig. 3.1) Fig. 1.1 The competing values framework (Source Cameron et al. 2014, Fig. 3.1) dimension focuses on the orientation toward individuality and ﬂexibility on the one hand versus stability and control on the other. Turning to the quadrants, ﬁrms with a collaborate-oriented culture attempt to develop human competencies and coop- erative processes by building consensus. In contrast, ﬁrms with a control-oriented culture attempt to improve internal organizational efﬁciency through control mecha- nisms. These are two internally oriented culture types that are differentiated in terms of the second dimension based on their individuality and ﬂexibility (collaborate- orientation) or stability and control (control-orientation). 1.2 Key Questions and Related Literature Next, ﬁrms with a compete- oriented culture seek to enhance their competitiveness and prioritize customers and shareholders. As a result, ﬁrms with a compete-oriented culture tend to judge success in terms of their market share, revenue, and proﬁtability. In contrast, ﬁrms with a create-oriented culture seek to create future opportunities in the marketplace through innovation and encourage entrepreneurship and changes. As a result, ﬁrms with a create-oriented culture tend to judge success in terms of the development of new products, services and technologies. These are two externally oriented culture types, differentiated again in terms of the second dimension based on their individuality and ﬂexibility (create-orientation) or stability and control (compete-orientation). In our survey, we asked respondent ﬁrms to choose up to three words to describe their corporate culture from a total of eight words corresponding to the four quadrants. Fiordelisi and Ricci (2014) use the CVF to examine the effect of corporate culture on the relationship between ﬁrm performance and CEO turnover. However, as far as we are aware, there are no empirical studies that examine the effect of corporate culture on innovation using the CVF. dimension focuses on the orientation toward individuality and ﬂexibility on the one hand versus stability and control on the other. Turning to the quadrants, ﬁrms with a collaborate-oriented culture attempt to develop human competencies and coop- erative processes by building consensus. In contrast, ﬁrms with a control-oriented culture attempt to improve internal organizational efﬁciency through control mecha- nisms. These are two internally oriented culture types that are differentiated in terms of the second dimension based on their individuality and ﬂexibility (collaborate- orientation) or stability and control (control-orientation). Next, ﬁrms with a compete- oriented culture seek to enhance their competitiveness and prioritize customers and shareholders. As a result, ﬁrms with a compete-oriented culture tend to judge success in terms of their market share, revenue, and proﬁtability. In contrast, ﬁrms with a create-oriented culture seek to create future opportunities in the marketplace through innovation and encourage entrepreneurship and changes. As a result, ﬁrms with a create-oriented culture tend to judge success in terms of the development of new products, services and technologies. These are two externally oriented culture types, differentiated again in terms of the second dimension based on their individuality and ﬂexibility (create-orientation) or stability and control (compete-orientation). 1.2.4 Risk Preferences and Corporate Culture (2021) report that 60% of respondent ﬁrmsfelttheytookthe“rightamountofrisk,”while29%saidtheytook“toolittlerisk” and 11% said they took “too much risk.” Second, we set a hypothetical question about anR&Dprojectwhichwasexpectedtogenerategrosssalesof100millionyenifitwas successful but gross sales of 0 yen if it failed and asked about the maximum amount that respondent ﬁrms would be willing to invest in this project (i.e., their reservation price). It was assumed that the probability of success was 10% and the expected payoff of the project accordingly was 10 million yen. In our analysis below, we classify respondent ﬁrms that were willing to invest 10 million yen as “risk-neutral,” those willing to invest less than 10 million yen as “risk-averse,” and those willing to invest more than 10 million yen as “risk-tolerant.” Using a similar survey question about a hypothetical lottery ticket, Cramer et al. (2002) construct a measure of risk aversion to examine whether risk aversion affects individuals’ choice of becoming an entrepreneur. Third, we measure ﬁrms’ risk preferences by asking them to choose between two otherwise identical projects. Project 1 has a greater net present value (NPV) but negative cash ﬂow for the ﬁrst few years, whereas Project 2 has positive cash ﬂow throughout its duration but has a smaller NPV. Graham et al. (2021) ask a similar question in their survey and report that, somewhat surprisingly, 41% of ﬁrms chose the NPV-inferior project (which would be Project 2 in our case). They also show that about 80% of ﬁrms that chose the NPV-superior project (corresponding to Project 1 in our case) say that corporate culture plays a role in their preference for the NPV-superior project. Tomeasureﬁrms’corporateculture,weemploytheCompetingValuesFramework (CVF) proposed by Cameron et al. (2014), who argue that the CVF provides a way to characterize organizational culture in simple terms. Speciﬁcally, the framework consists of two dimensions that express the tensions (“competing values”) in orga- nizations, which result in four categories (quadrants): Collaborate (Clan), Control (Hierarchy), Compete (Market), and Create (Adhocracy). Figure 1.1 provides a schematic representation. One dimension focuses on the orientation toward internal maintenance versus external positioning, while the other 1.3.1 Survey Design Before designing the “Survey of R&D Management Practices,” we visited ﬁve R&D organizations on-site to meet with the managers and directors of R&D organizations and/or human resources departments and asked about the ﬁrms’ R&D management practices. We then incorporated their insights, as well as those from the related literature described in Sect. 1.2, in the ﬁrst draft of the survey questionnaire. Next, we circulated the initial draft among them and asked for comments. In addition, we asked three other R&D employees at different ﬁrms to answer the draft survey questionnaire and provide feedback. Based on the comments and feedback, we then made numerous changes to the questionnaire. WefocusonR&Dmanagementpracticesamongbusinessenterpriseswithsystem- atic R&D operations. Speciﬁcally, we target business enterprises with paid-in capital of 100 million yen or more that undertake R&D activities. Because many ﬁrms in service industries do not conduct R&D, we target ﬁrms in manufacturing (Japan Standard Industrial Classiﬁcation [JSIC]: 09–32), information and communications (JSIC: 37–41), and wholesale and retail trade (JSIC: 50–55). We construct our sample byidentifyingﬁrmsinthe2017and2018roundsoftheSurveyofResearchandDevel- opment conducted by the Statistics Bureau of Japan, Ministry of Internal Affairs and Communications meeting these criteria.7 There were 3,456 such ﬁrms. It should be noted that since we do not include ﬁrms with paid-in capital of less than 100 million yen and ﬁrms in industries other than manufacturing, information and communica- tions, and wholesale and retail trade, our results may not reﬂect R&D management practices and innovation among Japanese ﬁrms overall. We then conducted our “Survey of R&D Management Practices” in January– February 2020.” Survey questionnaires were sent out to the 3,456 ﬁrms and we asked for questions to be answered by the person(s) most qualiﬁed to respond with regard to the following: (1) R&D expenditure, R&D personnel, and R&D organizational structure, (2) R&D project management, (3) personnel evaluation of researchers and engineers, and (4) R&D outputs. In addition to these questions, we asked ﬁrms to provide some details on the person who responded to our questionnaire (such as the job title and number of years with the company). Many respondents were in managerial positions of R&D organizations or general affairs divisions and had worked for their company for “more than 20 years.” The survey consists of up to 32 questions (depending on the survey path taken), and we checked that it took about 25–35 min to complete all questions. 7 For details of the Survey of Research and Development, see the following website: https://www. stat.go.jp/english/data/kagaku/index.html (accessed 16 November 2021). 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 12 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 1.2 Key Questions and Related Literature In our survey, we asked respondent ﬁrms to choose up to three words to describe their corporate culture from a total of eight words corresponding to the four quadrants. Fiordelisi and Ricci (2014) use the CVF to examine the effect of corporate culture on the relationship between ﬁrm performance and CEO turnover. However, as far as we are aware, there are no empirical studies that examine the effect of corporate culture on innovation using the CVF. 8 There are several ﬁrms for which data from the 2019 Survey of Research and Development were unavailable. We use data from either the 2018 or 2017 Survey of Research and Development for these ﬁrms. 1.3 Methodology 13 The ﬁlled-in surveys were returned by mail or via a website by the end of March 2020. To increase the survey response rate, we sent reminder emails and/or made reminder telephone calls to ﬁrms that had not responded to the questionnaire. We also checked the accuracy of the answers to the questions by examining whether they were mutually and/or logically consistent and asked respondents for conﬁrmation when necessary. Following this procedure, we were left with 611 valid responses for a response rate of 17.7%. Of these 611 responses, 150 (24.5%) were received by mail and 461 (75.5%) online. We summarize the results of the survey in Ono et al. (2020) and in September 2020 sent this summary to ﬁrms that had responded to the questionnaire. 1.3.1 Survey Design We asked respondents to provide answers as of ﬁscal year (FY) 2018 and on a non-consolidated basis, unless stated otherwise. The survey questionnaire that we sent to respondent ﬁrms and a glossary of terms attached to the survey for respondents’ reference are provided in the Appendix. 1.3 Methodology 1.3.2 Summary Statistics This subsection provides summary statistics of the sample that we use in the subse- quent analyses. In constructing the sample, we match our survey data with data taken from the Survey of Research and Development. Speciﬁcally, we use the 2019 Survey of Research and Development, which reports the basic characteristics of the ﬁrms as of FY 2018.8 We use the Survey of Research and Development for several important variables that characterize respondent ﬁrms, such as their sales turnover and R&D expenditure as well as the total number of employees, number of R&D employees, and employees with a doctorate degree—information that is not included in the survey. Table 1.1 presents the number of ﬁrms in the sample overall by industry and ﬁrm size. The latter is measured in terms of the number of employees. We classify the sample into small ﬁrms (with 300 or fewer employees), medium-sized ﬁrms (with 301 to 1,000 employees), and large ﬁrms (with more than 1,000 employees). Small ﬁrms make up the largest share of the sample (51.4%), followed by medium-sized ﬁrms (31.9%), and large ﬁrms (16.7%). In terms of industry, the sample consists of 558 manufacturing ﬁrms (91.3%) and 53 non-manufacturing ﬁrms (8.7%). Table 1.1 breaks down manufacturing into ﬁve subcategories and non-manufacturing into two subcategories. Within manufacturing, the top two industries in terms of the number of observations are the machinery and equipment industry with 226 ﬁrms (37.0%) and the chemical, petroleum, coal, and plastic products industry with 161 ﬁrms (26.4%). The former includes manufacturers of motor vehicles, parts and accessories, while the latter includes manufacturers of pharmaceuticals and medicinal chemicals, two industries that spend a large amount on R&D. The ﬁrm size distribution varies across industries. The share of small ﬁrms is larger in non-manufacturing (73.6%) than that in manufacturing (49.3%). 9 We deﬁne “process innovation” as new or signiﬁcantly improved production processes and methods of providing services and/or delivering products or support activities that include signif- icant improvements in techniques, equipment, and/or software. We deﬁne “product innovation” as new or signiﬁcantly improved goods or services with respect to their technical speciﬁcations, components and materials, software in the product, user friendliness, or other functional character- istics that include new combinations of existing technologies or technology upgrades of existing goods or services. For details, see the glossary of terms in the Appendix. The deﬁnitions of process 1.3.2 Summary Statistics Within manufacturing, small ﬁrms account for the largest share in the chemical, petroleum, 14 1 R&D Management Practices and Innovation: Evidence from a Firm Survey Table 1.1 Number of ﬁrms in the sample Entire sample By ﬁrm size: Small Medium Large N Share (%) N Share (%) N Share (%) N Share (%) Entire sample 611 (100.0) 314 [51.4] 195 [31.9] 102 [16.7] By industry Manufacturing industries 558 (91.3) 275 [49.3] 186 [33.3] 97 [17.4] Food, beverages, and tobacco 60 (9.8) 28 [46.7] 22 [36.7] 10 [16.7] Chemical, petroleum, coal, and plastic products 161 (26.4) 94 [58.4] 47 [29.2] 20 [12.4] Iron, steel, and non-ferrous metals products 56 (9.2) 23 [41.1] 23 [41.1] 10 [17.9] Machinery and equipment 226 (37.0) 103 [45.6] 76 [33.6] 47 [20.8] Miscellaneous manufacturing 55 (9.0) 27 [49.1] 18 [32.7] 10 [18.2] Non-manufacturing industries 53 (8.7) 39 [73.6] 9 [17.0] 5 [9.4] Information and communications 31 (5.1) 21 [67.7] 5 [16.1] 5 [16.1] Wholesale and retail trade 22 (3.6) 18 [81.8] 4 [18.2] 0 [0.0] Note Figures in parentheses () represent the percentage share of ﬁrms in each industry in the total number of ﬁrms, whereas ﬁgures in square brackets [] represent the percentage share of ﬁrms of a particular size in the total number of ﬁrms in that industry 15 1.3 Methodology Table 1.2 Number of employees Table 1.2 Number of employees N Mean Median S.D. Entire sample 611 764.6 287.0 1,658.0 By industry Manufacturing industries 558 792.4 310.5 1,704.5 Food, beverages, and tobacco 60 673.7 334.5 929.9 Chemical, petroleum, coal, and plastic products 161 514.1 238.0 735.2 Iron, steel, and non-ferrous metals products 56 1,013.9 401.5 2,309.8 Machinery and equipment 226 1,012.4 335.0 2,236.5 Miscellaneous manufacturing 55 607.1 308.0 782.6 Non-manufacturing industries 53 471.4 122.0 1,014.5 Information and communications 31 678.0 103.0 1,289.5 Wholesale and retail trade 22 180.3 167.0 146.6 coal, and plastic products industries (58.4%). In contrast, large ﬁrms account for the largest share in the machinery and equipment industry (20.8%). coal, and plastic products industries (58.4%). In contrast, large ﬁrms account for the largest share in the machinery and equipment industry (20.8%). Table 1.2 provides descriptive statistics on the number of employees by industry. The overall sample mean is 765 employees, while the median is 287 for the whole sample, suggesting that the distribution of the number of employees is highly skewed. 1.3.2 Summary Statistics This is because the sample includes nine ﬁrms that have more than 5,000 employees. Firms in manufacturing industries have a larger number of employees than those in non-manufacturing industries. In particular, ﬁrms in the iron, steel, and non-ferrous metal products industry (mean: 1,014) and the machinery and equipment industry (mean: 1,012) have a larger number of employees than their counterparts in other industries. 1.4.1 R&D Outcomes In the survey, we deﬁne the success of R&D outcomes in terms of whether a ﬁrm developed process innovations or product innovations during the past three years, from FY2016 to FY2018.9 Table 1.3 provides an overview of the share of ﬁrms that 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 16 Table 1.3 Firms that introduced process and product innovations N Process innovation Product innovation Share (%) S.D. Share (%) S.D. Entire sample 609 44.5 49.7 54.4 49.9 By ﬁrm size (a) Small 313 36.4 48.2 47.0 50.0 (b) Medium 194 44.9 49.9 56.7 49.7 (c) Large 102 68.6 46.6 72.5 44.8 By industry Manufacturing industries 558 46.8 49.9 55.0 49.8 Food, beverages, and tobacco 60 51.7 50.4 71.7 45.4 Chemical, petroleum, and plastic products 161 44.1 49.8 49.7 50.2 Iron, steel, and non-ferrous metals products 56 42.9 49.9 55.4 50.2 Machinery and equipment 226 46.0 50.0 52.7 50.0 Miscellaneous manufacturing 55 56.4 50.1 61.8 49.0 Non-manufacturing industries 51 19.6 40.1 47.1 50.4 Information and communications 30 20.0 40.7 46.7 50.7 Wholesale and retail trade 21 19.0 40.2 47.6 51.2 Difference N Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 507 −8.4* 4.5 −9.7 4.6 (b)–(c), Medium vs. Large 296 −23.8* 6.0 −15.8* 5.9 (a)–(c), Small vs. Large 415 −32.2* 5.5 −25.6* 5.6 Note , , and indicate signiﬁcance at the 1, 5, and 10% levels respectively Note *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively introduced process and product innovations by ﬁrm size and industry. This and all following tables report the mean and the standard deviation of the variables and test differences between subsamples using a two-sample equal variance t-test.10 Among ﬁrm size categories, we choose two out of three categories and test the difference between the two subsample means. , , and ** in tables denote statistical signiﬁ- cance at the 10%, 5%, and 1% levels, respectively. For some highly skewed variables, we also report the median. introduced process and product innovations by ﬁrm size and industry. This and all following tables report the mean and the standard deviation of the variables and test differences between subsamples using a two-sample equal variance t-test.10 Among ﬁrm size categories, we choose two out of three categories and test the difference between the two subsample means. and product innovation and the novelty of product innovation (Table 1.4) are based on the Oslo Manual 2018 by the Organisation for Economic Co-operation and Development, which provides international guidelines on innovation statistics. 10 Weuset-testsforallvariablesanddonotusebinomialtestsforcategoricalvariables.Asrobustness checks, we conducted binomial tests for several categorical variables and conﬁrmed that there is no substantial difference between the t-tests and the binomial tests for our sample sizes. and product innovation and the novelty of product innovation (Table 1.4) are based on the Oslo Manual 2018 by the Organisation for Economic Co-operation and Development, which provides international guidelines on innovation statistics. 10 Weuset-testsforallvariablesanddonotusebinomialtestsforcategoricalvariables.Asrobustness 1.4.1 R&D Outcomes , , and in tables denote statistical signiﬁ- cance at the 10%, 5%, and 1% levels, respectively. For some highly skewed variables, we also report the median. 17 1.4 R&D Outcomes and Inputs 17 Table 1.4 Product innovation by novelty of products N New-to-market products only New-to-ﬁrm products only Both types of products Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 329 17.3 37.9 41.3 49.3 41.3 49.3 By ﬁrm size (a) Small 147 25.2 43.5 36.1 48.2 38.8 48.9 (b) Medium 110 10.9 31.3 48.2 50.2 40.9 49.4 (c) Large 72 11.1 31.6 41.7 49.6 47.2 50.3 By industry Manufacturing industries 305 17.4 38.0 41.0 49.3 41.6 49.4 Food, beverages, and tobacco 43 23.3 42.7 44.2 50.2 32.6 47.4 Chemical, petroleum, and plastic products 80 13.8 34.7 38.8 49.0 47.5 50.3 Iron, steel, and non-ferrous metals products 31 12.9 34.1 48.4 50.8 38.7 49.5 Machinery and equipment 118 18.6 39.1 39.0 49.0 42.4 49.6 Miscellaneous manufacturing 33 18.2 39.2 42.4 50.2 39.4 49.6 Non-manufacturing industries 24 16.7 38.1 45.8 50.9 37.5 49.5 Information and communications 14 21.4 42.6 35.7 49.7 42.9 51.4 Wholesale and retail trade 10 10.0 31.6 60.0 51.6 30.0 48.3 Difference N Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 257 14.26* 4.89 −12.13* 6.18 −2.13 6.19 (b)–(c), Medium vs. Large 182 −0.20 4.77 6.52 7.58 −6.31 7.54 (a)–(c), Small vs. Large 219 14.06 5.76 −5.61 7.00 −8.45 7.10 Note , , and indicate signiﬁcance at the 1, 5, and 10% levels respectively 18 1 R&D Management Practices and Innovation: Evidence from a Firm Survey Table 1.3 shows that the percentage shares of ﬁrms that developed process inno- vations and product innovations are 44.5% and 54.4%, respectively. Larger ﬁrms are more likely to develop both process and product innovations. By industry, ﬁrms in the information and communications industry show the lowest propensity to produce process innovations (20.0%) and product innovations (46.7%), while those in the food, beverage, and tobacco industry show the highest propensity to produce product innovations (71.7%). For ﬁrms that introduced product innovations in the market, the survey asked follow-up questions on the novelty of the products, and we deﬁne two types of product novelty. 1.4.1 R&D Outcomes The ﬁrst refers to new or signiﬁcantly improved goods or services that no competitorswereoffering(referredtoas“new-to-market”products),whilethesecond refers to new or improved goods or services that were the same as or very similar to ones already offered by competitors (referred to as “new-to-ﬁrm” products). Table 1.4 reports the percentage shares of ﬁrms that introduced new-to-market and/or new-to-ﬁrm products during FY 2016–2018. Of the 329 product innovators, 17.3% (57 ﬁrms) developed new-to-market products but did not develop new-to-ﬁrm products (“new-to-market products only”). Meanwhile, 41.3% of product innovators (136 ﬁrms) developed new-to-ﬁrm products but did not develop new-to-market prod- ucts (“new-to-ﬁrm products only”), and 41.3% (136 ﬁrms) developed both types of products (“both types of products”). The percentage shares of “new-to-market products only,” “new-to-ﬁrm products only,” and “both types of products” ﬁrms are similar for medium-sized and large ﬁrms. By contrast, a considerably larger share of small ﬁrms (25.2%) than large and medium-sized ﬁrms (ca. 11%) generate new-to-market product innovations. On the other hand, the share of small ﬁrms generating new-to-ﬁrm product innovations (36.1%) is lower than that of medium-sized ﬁrms (48.2%) and large ﬁrms (41.7%). These results suggest that small ﬁrms are more oriented toward developing new-to- market products. This may be because small ﬁrms have a comparative disadvantage in accumulating intangible assets such as reputation and brand recognition and therefore seek to build an advantage through the introduction of novel products. In contrast, large ﬁrms tend to pursue both types of product innovation: the percentage share of “both types of products” for large ﬁrms is 47.2%, which is higher than that for small ﬁrm (38.8%) and medium-sized ﬁrms (40.9%). It should be noted, however, that the differences among the three types of ﬁrms are not statistically signiﬁcant. The distribution of ﬁrms in terms of the type of product innovation also varies across industries. The food, beverage, and tobacco industry has the highest share of ﬁrms generating new-to-market products only (23.3%), while the wholesale and retail trade industry has the highest share of ﬁrms generating new-to-ﬁrm products only (60.0%), and the chemical, petroleum, coal, and plastic products industry has the highest share of ﬁrms generating both types of product innovation (47.5%). 1.4.2 R&D Inputs: R&D Expenditures This subsection reviews respondent ﬁrms’ total R&D expenditure in FY 2018, their funding sources for R&D expenditure, and the importance of various factors taken into account when determining the level of R&D expenditure.13 1.4.1 R&D Outcomes In the tables below we report summary statics for the responses to our survey questions by ﬁrm size, by whether a ﬁrm has made product innovations (“innovator” or “non-innovator”), and in terms of the novelty of product innovations that a ﬁrm has made (“new-to-market” or “new-to-ﬁrm”). Because we focus on the interac- tion between R&D management practices and innovation, we examine whether the 1.4 R&D Outcomes and Inputs 19 1.4 summary statistics for various R&D management practices differ between product innovators and non-innovators by conducting two-sample equal variance t-tests.11 We recognize that differences between product innovators and non-innovators may reﬂect differences in ﬁrm characteristics such as ﬁrm size. While it is beyond the scope of this study to examine this possibility in detail, we add footnotes where we suspect that correlations may be spurious. Because our sample consists of ﬁrms with systematic R&D operations, it is likely that non-innovators, at least some of them, have tried to create product innovations but failed in the past three years. In the analysis below, we therefore assume that non- innovatorshaveattemptedtomakeproductinnovationsbutfailedtodoso.Inaddition, we use the novelty of product innovations among product innovators to examine the tension between explorative and exploitative innovation. For this purpose, we assume that ﬁrms that introduced new-to-market products in the preceding three years pursued explorative R&D, whereas those that introduced new-to-ﬁrm products in the preceding three years pursued exploitative R&D. More speciﬁcally, in the analysis below, we regard ﬁrms that introduced “new-to-market products only” as ﬁrms creating explorative innovations and ﬁrms that introduced “new-to-ﬁrm prod- ucts only” as ﬁrms creating exploitative innovations.12 Based on these assumptions with regard to product non-innovators and product novelty, we examine whether the results of the survey are consistent with the predictions of the related literature we outlined in Sect. 1.2. 11 We do not report summary statistics for process innovations in the tables below because in many cases the results are qualitatively similar to those for product innovations. Because we also focu on product novelty, we chose to report the results for product innovation. 13 “R&D expenditure” in the survey refers to the total amount of expenses on R&D, irrespective of whether such expenses are funded internally (e.g., through retained earnings) or externally (e.g., through grants from the government). R&D expenditure includes expenditures spent both within and outside respondent ﬁrms. 12 Firms that developed “both types of products” are likely to pursue both explorative and exploita- tive R&D. We do not report summary statistics for these ﬁrms because we cannot examine whether R&D management practices differ between ﬁrms that pursue explorative R&D and ﬁrms that pursue exploitative R&D when we use this subsample. 11 We do not report summary statistics for process innovations in the tables below because in many cases the results are qualitatively similar to those for product innovations. Because we also focus on product novelty, we chose to report the results for product innovation. 12 Firms that developed “both types of products” are likely to pursue both explorative and exploita- tive R&D. We do not report summary statistics for these ﬁrms because we cannot examine whether R&D management practices differ between ﬁrms that pursue explorative R&D and ﬁrms that pursue exploitative R&D when we use this subsample. 13 “R&D expenditure” in the survey refers to the total amount of expenses on R&D, irrespective of whether such expenses are funded internally (e.g., through retained earnings) or externally (e.g., through grants from the government). R&D expenditure includes expenditures spent both within and outside respondent ﬁrms. q y p on product novelty, we chose to report the results for product innovation. 12 Firms that developed “both types of products” are likely to pursue both explorative and exploita- tive R&D. We do not report summary statistics for these ﬁrms because we cannot examine whether R&D management practices differ between ﬁrms that pursue explorative R&D and ﬁrms that pursue exploitative R&D when we use this subsample. 13 “R&D expenditure” in the survey refers to the total amount of expenses on R&D irrespective 1.4.2.1 Amount of R&D Expenditure Table 1.5 reports summary statistics on ﬁrms’ total amount of R&D expenditure and its ratio to total sales (referred to as the R&D-to-sales ratio). The mean of R&D 11 We do not report summary statistics for process innovations in the tables below because in many cases the results are qualitatively similar to those for product innovations. Because we also focus on product novelty, we chose to report the results for product innovation. 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 20 20 Table 1.5 R&D expenditure R&D expenditure (million yen) R&D-to-sales ratio (%) N Mean S.D. Median N Mean S.D. Median Entire sample 611 1,669.8 7,446.8 186.1 610 4.3 13.2 1.8 By ﬁrm size (a) Small 314 159.7 280.6 74.6 313 5.2 16.9 1.8 (b) Medium 195 640.5 845.4 335.1 195 2.9 6.3 1.6 (c) Large 102 8,285.9 16,732.2 1,930.1 102 4.3 9.6 1.7 Non-innovators vs. Innovators (d) Non-innovators 278 687.0 2,319.6 129.8 277 3.9 9.1 1.5 (e) Innovators 331 2,501.3 9,822.1 218.9 331 4.3 14.7 1.9 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 661.3 2,255.3 156.4 57 9.4 31.8 1.3 (e2) New-to-ﬁrm (NTF) innovators 136 1,976.0 6,072.7 203.5 136 3.0 5.6 1.8 Difference N Mean S.E. N Mean S.E. (a)–(b), Small vs. Medium 509 −480.8*** 51.7 508 2.3* 1.3 (b)–(c), Medium vs. Large 297 −7,645.3* 1,199.3 297 −1.4 0.9 (a)–(c), Small vs. Large 416 −8,126.1* 942.3 415 0.9 1.8 (d)–(e), Non-innovators vs. Innovators 609 −1,814.3* 602.8 608 −0.5 1.0 (e1)–(e2), NTM vs. NTF innovators 193 −1,314.7 828.3 193 6.4 2.8 Notes The R&D-to-sales ratio is deﬁned as the ratio of ﬁrms’ total R&D expenditure to total sales. *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 1.4 R&D Outcomes and Inputs 21 expenditure for the entire sample is 1.67 billion yen, while the median is 186 million yen, suggesting that the distribution of R&D expenditure is highly skewed. Looking at the means for small, medium-sized, and large ﬁrms, we ﬁnd, unsurprisingly, that larger ﬁrms spend a larger amount on R&D than small and medium-sized ﬁrm. We therefore also calculate the R&D-to-sales ratio to adjust for ﬁrm size and ﬁnd that the mean of the R&D-to-sales ratio is 4.3%, while the median for the entire sample is 1.8%. Interestingly, we ﬁnd that the mean of the R&D-to-sales ratio is highest for small ﬁrms (5.2%) and lowest for medium-sized ﬁrms (2.9%). 1.4.2.1 Amount of R&D Expenditure Innovating ﬁrms invest more in R&D than non-innovating ﬁrms. The mean of the R&D expenditure of innovating ﬁrms (2.50 billion yen) is 3.6 times larger than that of non-innovating ﬁrms (0.69 billion yen). However, the difference in the R&D-to- sales ratio between innovating ﬁrms (4.3%) and non-innovating ﬁrms (3.9%) turns out to be much smaller (1.1 times) and is statistically insigniﬁcant. This suggests that the difference in R&D expenditure may simply reﬂect the fact that large ﬁrms are more likely to succeed in making product innovations, as reported in Table 1.3. More importantly, the insigniﬁcant difference in the R&D-to-sales ratio between innovating ﬁrms and non-innovating ﬁrms suggests that factors other than R&D inputs, such as R&D management practices, which we will examine later, may be an important determinant of success in product innovation. The total amount of R&D expenditure of new-to-market innovators (0.66 billion yen) is two-ﬁfth as large as that of new-to-ﬁrm innovators (1.98 billion yen), but the difference between these subsamples is insigniﬁcant. By contrast, the R&D-to-sales ratiofornew-to-marketinnovators(9.4%)is3.1timeshigherthanthatfornew-to-ﬁrm innovators (3.0%) and the difference is statistically signiﬁcant. This result indicates that once we control for the effect of ﬁrm size on R&D expenditure, new-to-market innovators invest more in their R&D activities than new-to-ﬁrm innovators. 1.4.2.2 Funding Sources for R&D Expenditure The survey asked about the shares of different funding sources for R&D expenditure. Funding sources are classiﬁed into four categories. The ﬁrst is funding from head- quarters or the business unit to which the R&D organization belongs. The second is commissions received from other business units within the ﬁrm. The third is funding from outside the ﬁrm including commissions, subsidies, grants, etc. The fourth is sources other than these three categories. Table 1.6 shows that 89.2% of R&D expenditure is provided by headquarters or the business unit to which an R&D organization belongs. The tendency for funding to comeprimarilyfromheadquartersorthebusinessunittowhichtheR&Dorganization belongs is particularly pronounced among small and medium-sized ﬁrms. On the other hand, among large ﬁrms, the sources of funding are more dispersed, with headquarters accounting for 84.6%, other business units for 8.0%, sources outside the ﬁrm for 5.2%, and other sources for 2.2%. The differences in funding sources between non-innovators and innovators and betweennew-to-marketinnovatorsandnew-to-ﬁrminnovatorsaresmall.Onenotable 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 22 22 1 R&D Management Practices and Innovation: Evidence from a Firm Survey Table 1.6 Funding sources for R&D expenditure N Headquarters or the business unit to which the R&D organization belongs Other business units within the ﬁrm Outside the ﬁrm Other Mean (Share, %) S.D. Mean (Share, %) S.D. Mean (Share, %) S.D. Mean (Share, %) S.D. Entire sample 608 89.2 25.3 5.1 18.6 4.3 14.3 1.4 11.3 By ﬁrm size (a) Small 313 89.6 26.2 4.0 17.9 4.8 16.4 1.6 12.3 (b) Medium 195 90.8 22.9 5.2 18.9 3.1 10.4 0.8 7.1 (c) Large 100 84.6 26.5 8.0 19.9 5.2 13.8 2.2 14.1 Non-innovators vs. Innovators (d) Non-innovators 277 88.8 26.0 6.0 20.6 3.9 13.3 1.3 11.2 (e) Innovators 329 90.1 23.7 4.0 15.9 4.4 14.2 1.5 11.4 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 87.3 28.6 1.8 13.3 5.6 15.3 5.3 21.8 (e2) New-to-ﬁrm (NTF) innovators 135 90.1 23.4 5.8 19.8 4.2 13.3 0.0 0.0 Difference N Mean (Share, %) S.E. Mean (Share, %) S.E. Mean (Share, %) S.E. Mean (Share, %) S.E. (a)–(b), Small vs. Medium 508 −1.22 2.28 −1.22 1.67 1.64 1.31 0.80 0.97 (b)–(c), Medium vs. Large 295 6.22** 2.98 −2.80 2.37 −2.01 1.43 −1.40 1.24 (a)–(c), Small vs. Large 413 4.99* 3.02 −4.02* 2.11 −0.37 1.82 −0.60 1.47 (d)–(e), Non-innovators vs. Innovators 606 −1.32 2.02 2.03 1.48 −0.55 1.12 −0.17 0.92 (e1)–(e2), NTM vs. 1.4.2.3 Determinants of R&D Expenditure Table 1.7 reports the importance of different factors that ﬁrms take into account when determining their total R&D expenditure. The survey provides six factors as possible determinants: gross sales in the previous year, proﬁts in the previous year, R&D expenditure in the previous year, labor costs of the ﬁrm’s R&D organization(s), cumulative costs of individual R&D projects, and annual sales goals for new products as a share of total sales. Firms were asked to answer whether each of these factors was “fully taken into account,” “to some extent taken into account,” “not very much taken into account,” or “not taken into account at all.” Table 1.7 reports the share of ﬁrms thattookeachofthefactorseither“fully”or“tosomeextent”intoaccount.Weassume that these factors are linked with the ﬂexibility of the R&D budget. For example, if a ﬁrm puts more weight on gross sales or proﬁts, the ﬁrm can ﬂexibly adjust its R&D budget to its ﬁnancial situation. By contrast, if a ﬁrm puts more weight on the R&D expenditure in the previous year and the cost of labor and research projects, the ﬁrm is likely to be bound by cost-related factors, making the research budget less ﬂexible. In addition, based on anecdotal evidence that innovation-oriented ﬁrms set annual sales targets for new products as a share of total sales, the survey asked about the importance of this factor in determining the R&D budget. The top three factors that the majority of ﬁrms in the sample said they took “fully” or “to some extent” into account are the R&D expenditure in the previous year (83.1%), the labor costs of their R&D organization(s) (67.8%), and proﬁts in the previous year (67.4%). As for R&D expenditure in the previous year, 30.9% of ﬁrms “fully” take this into account as a determinant, and this share is much higher than those for the other factors. In contrast, the share of ﬁrms that “fully” or “to some extent” take into account annual sales goals for new products is only about 50%, making this the least important factor among the different options provided. To sum up, these results indicate that ﬁrms take several factors into account when determining their R&D budget, and cost-related factors are more important than performance-related factors. This tendency is more prominent among large ﬁrms. 1.4.2.2 Funding Sources for R&D Expenditure NTF innovators 192 −2.81 3.96 −3.92 2.87 1.47 2.20 5.26* 1.87 Note , , and indicate signiﬁcance at the 1, 5, and 10% levels respectively 1.4 R&D Outcomes and Inputs 23 feature is that new-to-market innovators obtain a higher share of funding from “other” sources (5.3%) than new-to-ﬁrm innovators (0.0%). However, the primary funding source for new-to-market innovators nevertheless also is the headquarters or the business unit to which the R&D organization belongs (87.3%). 1.4.2.3 Determinants of R&D Expenditure In fact, although the share of ﬁrms taking a particular factor into account is highest among large ﬁrms for all factors, the difference between large ﬁrms on the one hand and small and medium-sized ones on the other is signiﬁcant only for cost-related factors, i.e., R&D expenditure, labor costs, and the cumulative costs of individual R&D projects. Next, looking at differences between innovating and non-innovating ﬁrms shows that the former are more likely to take cost-related factors into account when deciding R&D expenditure. In addition, they are also more likely to consider performance- related factors in determining R&D expenditure. Speciﬁcally, 64.0% of innovating 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 24 24 1 R&D Management Practices and Innovation: Evidence Table 1.7 Determinants of total R&D expenditure N Gross sales in the previous year Proﬁts in the previous year R&D expenditure in the previous year Labor costs of R&D organization(s) Cumulative costs of individual R&D projects Annual sales goals for new products as a share of total sales Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 611 58.8 49.3 67.4 46.9 83.1 37.5 67.8 46.8 64.3 47.9 50.7 50.0 By ﬁrm size (a) Small 314 56.4 49.7 65.9 47.5 76.4 42.5 62.7 48.4 62.7 48.4 51.0 50.1 (b) Medium 195 59.0 49.3 67.2 47.1 89.2 31.1 68.7 46.5 61.5 48.8 49.7 50.1 (c) Large 102 65.7 47.7 72.6 44.8 92.2 27.0 81.4 39.1 74.5 43.8 52.0 50.2 Non-innovators vs. Innovators (d) Non-innovators 278 52.2 50.0 61.2 48.8 78.4 41.2 62.9 48.4 61.2 48.8 47.5 50.0 (e) Innovators 331 64.0 48.1 72.5 44.7 87.0 33.7 71.6 45.2 66.8 47.2 53.2 50.0 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 54.4 50.3 64.9 48.1 82.5 38.4 75.4 43.4 64.9 48.1 54.4 50.3 (e2) New-to-ﬁrm (NTF) innovators 136 64.7 48.0 72.8 44.7 89.7 30.5 70.6 45.7 69.1 46.4 47.8 50.1 ( i d) 25 1.4 R&D Outcomes and Inputs Table 1.7 Determinants of total R&D expenditure Difference N Gross sales in the previous year Proﬁts in the previous year R&D expenditure in the previous year Labor costs of R&D organization(s) Cumulative costs of individual R&D projects Annual sales goals for new products as a share of total sales Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. 14 “R&D personnel” in the survey refers to individuals holding at least a bachelor’s degree (or having equivalent or greater expertise) and engaged in R&D activities in their area of expertise for more than half of their working hours. 1.4.2.3 Determinants of R&D Expenditure (a)–(b), Small vs. Medium 509 −2.6 4.5 −1.3 4.3 −12.8* 3.5 −6.0 4.3 1.2 4.4 1.2 4.6 (b)–(c), Medium vs. Large 297 −6.7 6.0 −5.4 5.7 −2.9 3.6 −12.7 5.4 −13.0** 5.8 −2.2 6.1 (a)–(c), Small vs. Large 416 −9.3* 5.6 −6.6 5.3 −15.7* 4.5 −18.6* 5.3 −11.8 5.4 −1.0 5.7 (d)–(e), Non-innovators vs. Innovators 609 −11.9* 4.0 −11.4* 3.8 −8.6* 3.0 −8.7 3.8 −5.6 3.9 −5.7 4.1 (e1)–(e2), NTM vs. NTF innovators 193 −10.3 7.7 −7.9 7.2 −7.2 5.2 4.9 7.1 −4.2 7.4 6.6 7.9 Note Figures represent the percentage share of ﬁrms that responded with either “fully taken into account” or “to some extent taken into account” when determining total R&D expenditure. , , and indicate signiﬁcance at the 1, 5, and 10% levels respectively 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 26 ﬁrms fully or to some extent take gross sales in the previous year into account, compared to 52.2% of non-innovating ﬁrms—a difference of 11.9 percentage points. Similarly, the share of ﬁrms that take proﬁts in the previous year into account is 11.4 percentage points higher among innovating ﬁrms (72.5%) than non-innovating ﬁrms (61.2%). Given that innovating ﬁrms are concerned not only about the costs of R&D projects but also their gross sales and proﬁts when deciding their R&D budget, the results suggest that innovating ﬁrms are more ﬂexible in adjusting their R&D budget to their performance. The shares of new-to-market innovators that take either performance-related factors or cost-related factors into account are smaller than those of new-to-ﬁrm innovators, except in the case of labor costs and annual sales goals for new products as a share of total sales. However, for all factors, the differences between these two subsamples are not statistically signiﬁcant. 1.4.3 R&D Inputs: R&D Personnel Next, we provide an overview of another R&D input: the total number of R&D personnel and their age composition.14 Table 1.8(a) reports the number of R&D personnel and the ratio to the total number of employees and Table 1.8(b) reports the number of R&D personnel with a doctorate degree and the ratio to the total number of R&D personnel. As in the case when we used the R&D-to-sales ratio in Table 1.5, we use these ratios to adjust for ﬁrm size. Table 1.8(a) shows that the mean number of R&D personnel is 74, while the median is 16. The mean of the R&D personnel-to- total-employees ratio is 9.2%, and the median is 5.5%. While we ﬁnd that the number of R&D employees increases with ﬁrm size, no clear pattern in terms of the R&D personnel-to-total-employees ratio can be observed: it is lowest for medium-sized ﬁrms (7.7%) and highest for small ﬁrms (10.1%). The mean of the number of R&D personnel is 107 for innovating ﬁrms and 35 for non-innovating ﬁrms, which means that the total number of R&D employees of innovating ﬁrms is 3.1 times larger than that of non-innovating ﬁrms. However, the means of the R&D personnel-to-total-employees ratio are approximately nine persons for both subsamples, and the difference between the two is not signiﬁcant. These results are similar to the patterns observed for R&D expenditure and the R&D expenditure-to-sales ratio in Table 1.5. The mean of the number of R&D personnel of new-to-market innovators is 3.5 times smaller than that of new-to-ﬁrm innovators, while the R&D-to-total-employees ratios are approximately 9% for both subsamples. Interestingly, the latter result is different from that for the R&D expenditure-to-sales ratio in Table 1.5, where we observed that the R&D-to-sales ratio of new-to-market innovators was signiﬁcantly larger thanthat of new-to-ﬁrminnovators. Takentogether, theresults inTables1.5and 27 1.4 R&D Outcomes and Inputs 27 Table 1.8 R&D personnel (a) R&D personnel Number of R&D personnel Ratio to total employees (%) N Mean S.D. Median N Mean S.D. Median Entire sample 611 73.6 262.6 16.0 611 9.2 12.1 5.5 By ﬁrm size (a) Small 314 11.5 14.4 7.0 314 10.1 13.8 6.0 (b) Medium 195 43.2 66.2 27.0 195 7.7 9.8 4.9 (c) Large 102 322.6 575.3 113.5 102 9.1 9.7 5.4 Non-innovators vs. Innovators (d) Non-innovators 278 34.7 81.2 12.0 278 8.7 11.8 5.4 (e) Innovators 331 106.5 345.7 19.0 331 9.3 11.4 5.5 New-to-market vs. 1.4.3 R&D Inputs: R&D Personnel New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 26.9 45.4 11.0 57 8.9 13.8 5.1 (e2) New-to-ﬁrm (NTF) innovators 136 94.8 286.8 18.0 136 8.6 10.5 4.7 Difference N Mean S.E. N Mean S.E. (a)–(b), Small vs. Medium 509 −31.7* 3.9 509 2.4 1.1 (b)–(c), Medium vs. Large 297 −279.4* 41.7 297 −1.4 1.2 (a)–(c), Small vs. Large 416 −311.1* 32.4 416 0.9 1.5 (d)–(e), Non-innovators vs. Innovators 609 −71.8*** 21.2 609 −0.6 0.9 (e1)–(e2), NTM vs. NTF innovators 193 −67.9* 38.2 193 0.3 1.8 (continued) 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 28 Table 1.8 R&D personnel (b) R&D personnel with doctorate degree (Ph.D.) N Number of R&D personnel with Ph.D. Ratio to total R&D employees (%) Mean S.D. Median N Mean S.D. Median Entire sample 611 3.9 17.0 0.0 587 5.4 12.5 0.0 By ﬁrm size (a) Small 314 0.5 1.1 0.0 292 5.8 15.7 0.0 (b) Medium 195 2.7 9.9 0.0 194 4.8 8.6 0.0 (c) Large 102 16.6 36.8 3.0 101 5.5 7.6 2.1 Non-innovators vs. Innovators (d) Non-innovators 278 2.9 17.6 0.0 266 5.3 13.5 0.0 (e) Innovators 331 4.7 16.6 0.0 319 5.5 11.7 0.0 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 1.8 7.2 0.0 52 6.7 18.1 0.0 (e2) New-to-ﬁrm (NTF) innovators 136 4.2 13.7 0.0 132 5.5 9.8 0.0 Difference N Mean S.E. N Mean S.E. (a)–(b), Small vs. Medium 509 −2.2* 0.6 486 1.0 1.2 (b)–(c), Medium vs. Large 297 −14.0* 2.8 295 −0.8 1.0 (a)–(c), Small vs. Large 416 −16.2* 2.1 393 0.2 1.6 (d)–(e), Non-innovators vs. Innovators 609 −1.8 1.4 585 −0.2 1.0 (e1)–(e2), NTM vs. NTF innovators 193 −2.4 1.9 184 1.2 2.1 Note , , and indicate signiﬁcance at the 1, 5, and 10% levels respectively 1.4 R&D Outcomes and Inputs 29 1.8(a) suggest that new-to-market innovators spend more on R&D than new-to-ﬁrm innovators, but less of that spending is on R&D personnel. Table 1.8(b) indicates that the majority of ﬁrms in the survey do not employ R&D personnel with doctorate degrees.15 The mean of the number of Ph.D. researchers is 3.9, while the median is 0, and the mean of the Ph.D. researchers-to-total R&D personnel ratio is 5.4%. Looking at the subsample results, while the average number of Ph.D. 15 The White Paper on Science and Technology 2017 (Ministry of Education, Culture, Sports, Science and Technology) shows that the percentage share of researchers with a Ph.D. in the total R&D personnel in Japanese ﬁrms is 4.4% on average, which is lower than the corresponding shares in other OECD countries. For example, the share is 10.1% in U.S ﬁrms and 6.7% in Korean ﬁrms. 1.4.3 R&D Inputs: R&D Personnel researchers is larger for larger ﬁrms, the ratio to R&D employees is approx- imately the same across small, medium, and large ﬁrms (5–6%). Similarly, the share of Ph.D. researchers is 5–6% regardless of whether ﬁrms are product innovators or not, and whether ﬁrms are new-to-market innovators or new-to-ﬁrm innovators. Table 1.9 reports the age composition of R&D personnel. In the sample overall, researchers aged 24–34 years old accounted for 34.9%, those aged 35–44 for 27.2%, and those aged 45 or older for 37.9%. We assume that job roles and titles correspond to researchers’ age and refer to R&D personnel aged 24–34 as young researchers, those aged 35–44 as middle or project managers, and those aged 45 or over as chief or division managers. The share of chief or division managers in R&D organizations is highest in small ﬁrms (41.4%), while the share of young researchers is lowest in small ﬁrms (31.0%). There is little difference in the age composition of researchers between product innovators or non-innovators: the share of young researchers is around 35% at both product innovators and non-innovators. However, when we compare the share of young researchers between new-to-market innovators and new-to-ﬁrm innovators, the share is smaller at the former (29.6%) than the latter (36.9%). 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 30 Table 1.9 Age composition of R&D personnel N Young researchers (age: 24–34) Middle managers (age: 35–44) Chief managers (age: 45 or over) Mean (Share, %) S.D. Mean (Share, %) S.D. Mean (Share, %) S.D. Entire sample 602 34.9 21.7 27.2 17.6 37.9 23.2 By ﬁrm size (a) Small 311 31.0 23.4 27.6 21.5 41.4 26.4 (b) Medium 194 40.5 20.3 26.7 13.0 32.8 18.3 (c) Large 97 36.4 15.5 26.8 10.2 36.8 18.3 Non-innovators vs. Innovators (d) Non-innovators 275 33.6 22.9 27.4 18.1 39.0 25.0 (e) Innovators 325 36.0 20.7 27.0 17.2 37.0 21.5 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 56 29.6 23.2 30.9 26.1 39.5 28.7 (e2) New-to-ﬁrm (NTF) innovators 134 36.9 21.5 26.9 16.6 36.1 21.0 Difference N Mean (Share, %) S.E. Mean (Share, %) S.E. Mean (Share, %) S.E. (a)–(b), Small vs. Medium 505 −9.5* 2.0 0.9 1.7 8.6* 2.2 (b)–(c), Medium vs. Large 291 4.1* 2.3 −0.1 1.5 −4.0* 2.3 (a)–(c), Small vs. Large 408 −5.4 2.5 0.8 2.3 4.6 2.9 (d)–(e), Non-innovators vs. Innovators 600 −2.4 1.8 0.4 1.4 2.0 1.9 (e1)–(e2), NTM vs. 16 Using the terminology in Argyres and Silverman (2004; Fig. 2), “R&D organizations” in the survey refers to organizations (such as departments, divisions, or units) in which R&D personnel mainly conduct R&D. For the purpose of the survey, organizations that perform R&D activities are regarded as “R&D organizations” even if their name does not include the words “Research” or “Development.” 1.4.3 R&D Inputs: R&D Personnel NTF innovators 190 −7.4 3.5 4.0 3.2 3.4 3.7 Note *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 1.5 Organizational Structure of R&D Activities 31 1.5 Organizational Structure of R&D Activities In this section, we discuss the relationship between the organizational design of R&D activities and innovation from the following two perspectives based on our discussion in Sect. 1.2.1. First, we examine whether there is a link between the orga- nizational structure of R&D activities, i.e., whether the R&D structure is centralized or decentralized, and innovation outcomes. In particular, we examine whether there is a link between a centralized R&D structure and explorative (new-to-market) innova- tion. Second, we examine whether the delegation of authority to R&D organizations promotes innovation. In our survey, we ﬁrst asked several questions about the organizational struc- ture of R&D activities, such as the number of R&D organizations within a ﬁrm.16 We then asked whether ﬁrms operated a centralized or decentralized R&D structure (Sect. 1.5.1). In cases where ﬁrms had a hybrid R&D structure consisting of both centralized and decentralized R&D activities, we also asked about the percentage shares of R&D expenditure and R&D personnel devoted to centralized and decen- tralized R&D activities (Sect. 1.5.2). Finally, to investigate whether ﬁrms delegate authority to R&D organizations, we asked whether it is the R&D organization or the human resources department that takes the initiative in hiring R&D personnel (Sect. 1.5.3). In our survey, we ﬁrst asked several questions about the organizational struc- ture of R&D activities, such as the number of R&D organizations within a ﬁrm.16 1.5.1 Centralized/Decentralized R&D Structure Table 1.10 reports the number of R&D organizations a ﬁrm has and whether they are directly controlled by business units or one or more of them are highly independent of business units. The ﬁrst column indicates that the mean of the number of R&D organizations in the sample overall is 3.3, while the median is 1.0. We ﬁnd that the number of R&D organizations increases with ﬁrm size, and the differences between the different size categories are statistically signiﬁcant. g y g The mean of the number of R&D organization is 4.3 for innovating ﬁrms and 2.1 for non-innovating ﬁrms, which means that the total number of R&D organizations of the former is approximately twice as large as that of the latter. The mean of the number of R&D organizations of new-to-market innovators (2.1) is less than half of that of new-to-ﬁrm innovators (5.5), although the difference between the two subsamples is insigniﬁcant. The second to fourth columns of Table 1.10 classify ﬁrms in terms of the orga- nizational structure of their R&D activities. Speciﬁcally, following Argyres and 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 32 32 1 R&D Management Practices and Innovation: Evidenc Table 1.10 R&D organizational structure Number of R&D organizations Organizational structure of R&D activities N Highly independent of business units (centralized) Directly controlled by business units (decentralized) Both centralized and decentralized R&D entities (hybrid) N Mean S.D. Median Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 609 3.3 11.7 1.0 585 45.3 49.8 41.4 49.3 13.3 34.0 By ﬁrm size (a) Small 313 1.4 1.1 1.0 298 46.3 49.9 49.0 50.1 4.7 21.2 (b) Medium 195 2.8 2.7 2.0 190 45.3 49.9 41.1 49.3 13.7 34.5 (c) Large 101 9.9 27.5 4.0 97 42.3 49.7 18.6 39.1 39.2 49.1 Non-innovators vs. Innovators (d) Non-innovators 278 2.1 2.7 1.0 264 45.8 49.9 45.5 49.9 8.7 28.3 (e) Innovators 329 4.3 15.6 2.0 320 44.7 49.8 38.1 48.6 17.2 37.8 New-to-market vs. 1.5.1 Centralized/Decentralized R&D Structure New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 2.1 1.6 1.0 55 49.1 50.5 41.8 49.8 9.1 29.0 (e2) New-to-ﬁrm (NTF) innovators 135 5.5 23.5 1.0 131 47.3 50.1 32.1 46.9 20.6 40.6 (continued) 33 1.5 Organizational Structure of R&D Activities 33 Table 1.10 R&D organizational structure Difference Number of R&D organizations Organizational structure of R&D activities N Highly independent of business units (centralized) Directly controlled by business units (decentralized) Both centralized and decentralized R&D entities (hybrid) N Mean S.E. Mean S.E. Mean S.E. Mean S.E. (a)–(b), Small vs. Medium 508 −1.4* 0.2 488 1.0 4.6 7.9 4.6 −9.0* 2.5 (b)–(c), Medium vs. Large 296 −7.1* 2.0 287 3.0 6.2 22.5* 5.8 −25.5* 5.0 (a)–(c), Small vs. Large 414 −8.5* 1.6 395 4.0 5.8 30.4* 5.6 −34.5* 3.6 (d)–(e), Non-innovators vs. Innovators 607 −2.3 0.9 584 1.1 4.1 7.3* 4.1 −8.5*** 2.8 (e1)–(e2), NTM vs. NTF innovators 192 −3.4 3.1 186 1.8 8.1 9.8 7.7 −11.5* 6.0 Note *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 34 1 R&D Management Practices and Innovation: Evidence from a Firm Survey Silverman (2004), we classify ﬁrms into the following three categories: ﬁrms with a centralized R&D structure, where R&D organizations are highly independent of business units, ﬁrms with a decentralized structure, where R&D organizations are directly controlled by business units, and ﬁrms with a hybrid structure, i.e., ﬁrms that have both a centralized R&D entity (such as a dedicated central R&D facility) and decentralized R&D entities that are embedded in other business units. The shares of ﬁrms in the sample overall that have a centralized, decentralized, or hybrid structure are 45%, 41%, and 13%, respectively. Since the median of the number of R&D organizations in the sample overall is 1.0, this result indicates that the majority of ﬁrms in the sample have one R&D organization that is either highly independent of business units or that is directly controlled by a business unit. The percentage share of ﬁrms that have a centralized R&D structure is very similar across small, medium, and large ﬁrms (42–46%). The share of decentralized structures is highest among small ﬁrms (49%), while the share of hybrid structures is highest among large ﬁrms (39%). Innovating ﬁrms are more likely to choose a hybrid R&D structure and less likely to choose a decentralized structure. 17 Because Argyres and Silverman’s (2004) sample consists mostly of large conglomerate corpo- rations, 62.5% of their sample ﬁrms have a hybrid R&D structure. Similarly, Argyres et al. (2020) focus on 130 large corporations. 1.5.1 Centralized/Decentralized R&D Structure The share of ﬁrms that have a decentralized structure is 7 percentage points lower among product innovators (38%) than non- innovators (45%), while the share of ﬁrms that have a hybrid structure is 8 percentage points higher among product innovators (17%) than non-innovators (9%). The share of ﬁrms that have a centralized R&D structure is about 45% for both subsamples and there is little difference between the two. We ﬁnd that most new-to-market innovators do not have a hybrid structure but have either a centralized or decentralized R&D structure. The shares of new-to-market innovators that have a centralized or decentralized R&D structure are 49% and 42%, respectively, while the share of ﬁrms that have a hybrid structure is only 9%. This result may reﬂect the fact that many new-to-market innovators are small ﬁrms (Table 1.4), among which the share of ﬁrms with a hybrid structure is very small (5%). There is little difference in the share of ﬁrms with a centralized R&D structure between new-to-market innovators and new-to-ﬁrm innovators. This result is inconsistent with Argyres and Silverman’s (2004) ﬁnding that ﬁrms with a centralized R&D structure were more likely to develop explorative innovations (new-to-market innovations in our case). The difference between our and Argyres and Silverman’s (2004) results may be due to differences in the samples used: while 80% of the ﬁrms in our sample are small and medium-sized ﬁrms, the ﬁrms in Argyres and Silverman’s (2004) sample were mostly large conglomerate corporations.17 Overall, we ﬁnd that ﬁrms that developed product innovations, especially new-to- ﬁrm innovations, are more likely to have a hybrid R&D structure. Among ﬁrms that made product innovations, we ﬁnd that most new-to-market innovators have either a Organizational Structure of R&D Activities 35 1.5 centralized or a decentralized R&D structure. This result is inconsistent with studies (such as Argyres and Silverman 2004) that ﬁnd that it is ﬁrms with a centralized R&D structure that tend to generate radical innovations. 1.5.2 Allocation of R&D Expenditure and R&D Personnel in Firms with a Hybrid R&D Structure In the case of ﬁrms that had a hybrid R&D structure with both centralized and decentralized R&D activities, we additionally asked about the allocation of R&D expenditure and R&D personnel among the two. There are 77 ﬁrms that have a hybrid structure in our sample. Because the number of observations is very small, in this subsection we mainly report the summary statistics and do not discuss the statistical signiﬁcance of differences across different groups. Table 1.11(a) shows the average number of independent R&D organizations and of dependent R&D organizations that form part of another business unit for the 77 ﬁrms that have a hybrid R&D structure. The average number of independent R&D orga- nizations is 2.9, while the median is 1.0. Further, the average number of dependent R&D organizations is 8.7, while the median is 3.0. Thus, the number of dependent R&D organizations is larger than that of independent R&D organizations, presum- ably because some large ﬁrms have many business units. Looking at the median of the sample overall, the typical ﬁrm in our sample has four R&D organizations: one independent and three dependent ones. Large ﬁrms have more R&D organizations, and this is reﬂected primarily in the number of R&D organizations that are part of another business unit. The average number of dependent R&D organizations of large ﬁrms (15.3) is 8.5 times larger than thatofsmallﬁrms(1.8),whiletheaveragenumberofindependentR&Dorganizations of large ﬁrms (4.4) is 3.6 times larger than that of small ﬁrms (1.2). Turning to innovation, innovating ﬁrms have a larger average number of both independent and dependent R&D organizations than non-innovating ﬁrms. Looking at the median for the subsample of product innovators, we ﬁnd that the typical innovating ﬁrm has one independent and three or four dependent R&D organizations. Among product innovators, new-to-market innovators have a smaller number of each type of R&D organization than new-to-ﬁrm innovators. Looking at the medians, the typical new-to-market innovator has one independent and one dependent R&D organization, while the typical new-to-ﬁrm innovator has two independent and four dependent R&D organizations. Table 1.11(b) reports the percentage shares of R&D expenditure and R&D personnel accounted for by independent R&D organizations at ﬁrms with hybrid R&D structures. The mean (median) of the R&D expenditures share of indepen- dent R&D organizations is 44.5% (40.0%), while the mean (median) of the R&D personnel share of independent R&D organizations is 40.3% (30.0%). 18 In Sect. 1.6.1.3 (Table 1.15), we will examine to what extent the authority to manage R&D projects is delegated to R&D organizations. 19 Although not directly related to the key concern of this monograph—the link between R&D management practices and innovation outcomes—it should be noted that R&D decentralization and the delegation of authority over R&D activities may be closely intertwined, as pointed out in footnote 4 in Sect. 1.2.1. As an aside, we therefore examine whether there is a link between the delegation of authority in hiring R&D personnel to R&D organizations (Table 1.12) and whether ﬁrms have a centralized or decentralized R&D structure (Sect. 1.5.1, Table 1.10). We ﬁnd that the percentage share of ﬁrms where the R&D organization takes the initiative in hiring R&D personnel 1.5.2 Allocation of R&D Expenditure and R&D Personnel in Firms with a Hybrid R&D Structure NTF innovators 30 18.1 12.4 32 8.8 12.2 Note *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 38 showed that the medians of the number of independent and dependent R&D organi- zations are one and three, respectively so that the percentage share of the number of independent R&D organizations is 25%, which is smaller than the percentage shares of R&D expenditure and R&D personnel. This suggests that, among ﬁrms that have a hybrid R&D structure, independent R&D organizations account for more R&D inputs (in terms of expenditure and employees) than dependent R&D organizations. 1.5.2 Allocation of R&D Expenditure and R&D Personnel in Firms with a Hybrid R&D Structure Table 1.11(a) 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 36 Table 1.11 R&D activities in ﬁrms with a hybrid R&D structure (a) Number of R&D organizations N Independent organizations Dependent organizations Mean S.D. Median Mean S.D. Median Entire sample 77 2.9 4.6 1.0 8.7 28.7 3.0 By ﬁrm size (a) Small 14 1.2 0.8 1.0 1.8 1.2 1.0 (b) Medium 26 1.7 1.3 1.0 3.1 2.8 2.0 (c) Large 37 4.4 6.3 2.0 15.3 40.5 5.0 Non-innovators vs. Innovators (d) Non-innovators 23 2.0 2.7 1.0 4.1 5.3 2.0 (e) Innovators 54 3.3 5.3 1.0 10.7 33.9 3.5 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 5 1.4 0.5 1.0 1.6 0.9 1.0 (e2) New-to-ﬁrm (NTF) innovators 27 3.4 4.0 2.0 16.1 47.4 4.0 Difference N Mean S.E. Mean S.E. (a)–(b), Small vs. Medium 40 −0.5 0.4 −1.3* 0.8 (b)–(c), Medium vs. Large 63 −2.7** 1.3 −12.2 8.0 (a)–(c), Small vs. Large 51 −3.2* 1.7 −13.5 10.9 (d)–(e), Non-innovators vs. Innovators 77 −1.3 1.2 −6.6 7.1 (e1)–(e2), NTM vs. NTF innovators 32 −2.0 1.8 −14.5 21.5 ( i d) 37 1.5 Organizational Structure of R&D Activities 37 Table 1.11 R&D activities in ﬁrms with a hybrid R&D structure (b) Share of R&D expenditure and R&D personnel of independent R&D organizations R&D expenditure R&D employees N Mean (Share, %) S.D. Median N Mean (Share, %) S.D. Median Entire sample 73 44.5 27.6 40.0 76 40.3 26.9 30.0 By ﬁrm size (a) Small 14 45.8 26.6 45.0 14 41.8 23.1 43.0 (b) Medium 26 48.2 28.5 48.0 26 43.2 26.5 32.0 (c) Large 33 41.0 27.8 33.0 36 37.6 28.9 29.0 Non-innovators vs. Innovators (d) Non-innovators 22 38.0 28.3 31.5 23 40.2 28.6 30.0 (e) Innovators 51 47.3 27.1 49.0 53 40.3 26.4 34.0 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 5 64.0 18.2 70.0 5 46.2 25.4 50.0 (e2) New-to-ﬁrm (NTF) innovators 25 45.9 26.4 49.0 27 37.4 25.1 30.0 Difference N Mean (Share, %) S.E. N Mean (Share, %) S.E. (a)–(b), Small vs. Medium 40 −2.4 9.2 40 −1.4 8.4 (b)–(c), Medium vs. Large 59 7.3 7.4 62 5.6 7.2 (a)–(c), Small vs. Large 47 4.8 8.8 50 4.2 8.6 (d)–(e), Non-innovators vs. Innovators 73 −9.2 7.0 76 −0.2 6.8 (e1)–(e2), NTM vs. 1.5.3 Initiative in Hiring R&D Personnel Table 1.12 shows the responses to the question about who takes the initiative in hiring R&D personnel, the R&D organization or the human resources (HR) depart- ment.18 The most frequent answer is both the R&D organization and the HR depart- ment (58.1%), followed by the HR department (21.1%), and the R&D organization (15.9%). Thus, in most ﬁrms, the R&D organization and the HR department jointly take the initiative in hiring R&D personnel. The pattern is quite similar for small, medium, and large ﬁrms. However, one notable ﬁnding is that the percentage share of small ﬁrms that responded “other” (7.3%) is signiﬁcantly higher than that of medium-sized and large ﬁrms. The percentage share of ﬁrms where both the R&D organization and the HR department take the initiative is signiﬁcantly higher for innovating ﬁrms (62.5%) than for non-innovating ﬁrms (52.9%). In contrast, the share of ﬁrms where the R&D organization takes the initiative is signiﬁcantly lower for innovating ﬁrms (13.3%) than for non-innovating ﬁrms (19.1%). The latter result is inconsistent with studies that found a positive link between the delegation of authority and innovation such as Acemoglu et al. (2007) and Kastl et al. (2013). Among innovating ﬁrms, the percentage share of ﬁrms where the R&D organization takes the initiative in hiring R&D employees is signiﬁcantly higher for new-to-market innovators (19.3%) than for new-to-ﬁrm innovators (8.8%). This result suggests that there is a positive link between the delegation of authority and explorative innovation and appears consistent withtheﬁndingsbyAcemogluetal.(2007)andKastletal.(2013).Overall,Table1.12 shows that whether the delegation of authority to R&D organizations is positively associated with innovation depends on the proxy used for innovation, i.e., whether a ﬁrm introduced new products in the market, or the type of product innovation (new-to-ﬁrm or new-to-market).19 1.5 Organizational Structure of R&D Activities 39 39 Table 1.12 Department that takes the initiative in hiring R&D personnel N R&D organization HR department Both R&D and HR Other Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 611 15.9 36.6 21.1 40.8 58.1 49.4 4.9 21.6 By ﬁrm size (a) Small 314 17.5 38.1 19.8 39.9 55.4 49.8 7.3 26.1 (b) Medium 195 12.3 32.9 25.6 43.8 59.5 49.2 2.6 15.8 (c) Large 102 17.7 38.3 16.7 37.5 63.7 48.3 2.0 13.9 Non-innovators vs. is 20.0% for ﬁrms with a centralized R&D structure while it is 14.1% for ﬁrms with a decentralized R&D structure. Further, the difference between the subsamples is weakly signiﬁcant, indicating that in ﬁrms with a centralized R&D structure R&D organizations have greater authority in hiring employees. While this result suggests that R&D decentralization and delegation of authority to R&D organizations capture the same aspect of the organization of R&D activities, our analyses also showed that the empirical link between the organizational structure of R&D activities and innovation outcomes presented in Sect. 1.5.1 is different from the empirical link between the delegation of authority over R&D activities and innovation outcomes presented in this subsection (Sect. 1.5.3). We therefore think that it is safe to assume that the delegation of authority in hiring R&D personnel to R&D organizations and whether R&D activities are centralized or decentralized captures different aspects of the organization of R&D activities. 1.5.3 Initiative in Hiring R&D Personnel Innovators (d) Non-innovators 278 19.1 39.4 21.6 41.2 52.9 50.0 6.5 24.7 (e) Innovators 331 13.3 34.0 20.9 40.7 62.5 48.5 3.3 18.0 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 19.3 39.8 17.5 38.4 57.9 49.8 5.3 22.5 (e2) New-to-ﬁrm (NTF) innovators 136 8.8 28.5 22.1 41.6 64.7 48.0 4.4 20.6 Difference N Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 509 5.2 3.3 −5.9 3.8 −4.1 4.5 4.8** 2.1 (b)–(c), Medium vs. Large 297 −5.3 4.3 9.0* 5.1 −4.2 6.0 0.6 1.9 (a)–(c), Small vs. Large 416 −0.1 4.3 3.1 4.5 −8.3 5.6 5.4** 2.7 (d)–(e), Non-innovators vs. Innovators 609 5.8* 3.0 0.7 3.3 −9.7** 4.0 3.2* 1.7 (e1)–(e2), NTM vs. NTF innovators 193 10.5 5.1 −4.5 6.4 −6.8 7.7 0.9 3.3 Note , , and indicate signiﬁcance at the 1, 5, and 10% levels respectively 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 40 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 1.6 R&D Project Management This section investigates how R&D projects are managed. Speciﬁcally, we examine whether and how staged project management, in which the project proceeds in consecutive multiple stages, is implemented. Based on the discussion and litera- ture review (e.g., Manso 2011) in Sect. 1.2.2, we focus on the threat of termination and feedback in staged project management. We ﬁrst provide an overview of the ongoing R&D projects of ﬁrms in the sample, including the number of projects and their track record, i.e., whether they were suspended, terminated, or continued (Sect. 1.6.1). We then examine whether and how ﬁrms implement staged project management (Sect. 1.6.2). In particular, for ﬁrms that implement staged project management, we examine whether ﬁrms set milestones in assessing whether an R&D project should be continued and, if they do, how important these milestones are in assessing whether the R&D project is continued (Sect. 1.6.2.1). We also examine whether ﬁrms provide feedback to R&D personnel in charge of the project, and if so, whose opinions are incorporated in the feedback (Sect. 1.6.2.2). 1.6.1.1 Number of R&D Projects Table 1.13 reports the approximate number of R&D projects in progress. The mean of the number of ongoing projects for the whole sample is 23.9, while the median is 6.0. Because it is likely that the number of R&D projects is positively correlated with ﬁrm size, we also calculate the ratio of the number of ongoing projects to 100 employees (which we refer to as the project-to-employee ratio). The mean of the project-to-employee ratio is 6.9 and the median is 2.3. Table 1.13 shows that while the number of ongoing projects increases with ﬁrm size as expected, the project-to-employee ratio does not and is lowest for medium- sized ﬁrms (2.9) and highest for small ﬁrms (10.5). These patterns are similar to 41 1.6 R&D Project Management 41 Table 1.13 Number of R&D projects in progress N Number of ongoing R&D projects Project-to-employee ratio Mean S.D. Median Mean S.D. Median Entire sample 600 23.9 66.8 6.0 6.9 26.8 2.3 By ﬁrm size (a) Small 311 9.6 20.2 4.0 10.5 36.6 3.9 (b) Medium 193 15.2 19.8 10.0 2.9 4.0 1.6 (c) Large 96 87.6 145.3 32.5 3.4 5.7 1.4 Non-innovators vs. Innovators (d) Non-innovators 275 15.8 57.4 4.0 6.5 27.3 1.9 (e) Innovators 324 30.8 73.3 10.0 7.3 26.4 2.7 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 11.6 15.5 6.0 7.1 11.4 3.1 (e2) New-to-ﬁrm (NTF) innovators 133 26.7 73.0 8.0 4.4 5.5 2.3 Difference N Mean S.E. Mean S.E. (a)–(b), Small vs. Medium 504 −5.7* 1.8 7.6* 2.6 (b)–(c), Medium vs. Large 289 −72.4* 10.6 −0.5 0.6 (a)–(c), Small vs. Large 407 −78.0* 8.5 7.1* 3.8 (d)–(e), Non-innovators vs. Innovators 599 −14.9* 5.5 −0.9 2.2 (e1)–(e2), NTM vs. NTF innovators 190 −15.1 9.8 2.8 1.2 Notes The project-to-employee ratio represents the number of ongoing projects per 100 employees. *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 42 those in Table 1.5 for the R&D expenditure-to-sales ratio and Table 1.8 for the R&D personnel-to-total-employees ratio. The mean of the number of projects in progress is 30.8 for product innovators and 15.8 for non-innovators. However, the mean of the project-to-employee ratio is 7.3 for the former and 6.5 for the latter, and there is no signiﬁcant difference between the two subsamples. 1.6.1.1 Number of R&D Projects Among product innovators, the mean of the number of ongoing R&D projects for new-to-market innovators (11.6) is less than half of that for new-to-ﬁrm innovators (26.7), but the difference is statistically insigniﬁcant. By contrast, the mean of the project-to-employee ratio for new-to-market innovators (7.1) is 1.6 times higher for new-to-ﬁrm innovators (4.4), and the difference between the subsamples is statistically signiﬁcant. This result indicates that, once we adjust for ﬁrm size, new-to-market innovators undertake a larger number of R&D projects than new-to-ﬁrm innovators, presumably because new-to-market innovations require more experimentation. 1.6.1.2 Duration of R&D Projects Next, we turn to the duration of R&D projects, about which the survey asked two questions. First, it asked about the average number of years from the commencement of a project to the achievement of ﬁnal results. And second, it asked about the approximate share of current R&D projects that have continuously been ongoing for more than three years. Table 1.14 presents the results. The mean of the average duration of R&D projects of ﬁrms in the entire sample is 3.5 years. Meanwhile, the mean of the percentage share of current R&D projects that had been in progress for more than three years is 38.7%, which is in line with the ﬁnding that the average duration is 3.5 years. Looking at various subsamples, the means of the average duration of R&D projects are quite similar for small and medium ﬁrms, product innovators and non-innovators, and new-to-market innovators and new-to-ﬁrm innovators. The one subsample whose mean, at 3.9 years, is slightly larger than that of other ﬁrms is large ﬁrms, which compares to 3.4 years for small and medium-sized ﬁrms. Similar patterns are observed when looking at the percentage share of R&D projects that had been ongoing for more than three years: we ﬁnd no signiﬁcant differences among the various subsamples. 1.6.1.3 Termination or Suspension of R&D Projects Table 1.15 reports whether a ﬁrm has any R&D projects that had been terminated or suspended within the past three years (ﬁrst column) and the approximate share of projects where R&D organizations have the authority to decide whether to terminate, suspend, or continue the project (right column). The table shows that 59.5% of ﬁrms 1.6 R&D Project Management 43 Table 1.14 Duration of R&D projects Average number of year from the commencement of an R&D project to the achievement of ﬁnal results Share of R&D projects that have continuously been on going for more than 3 years N Mean S.D. Median N Mean (Share, %) S.D. Median Entire sample 597 3.5 2.5 3.0 565 38.7 31.9 33.0 By ﬁrm size (a) Small 309 3.4 2.5 3.0 290 36.6 32.5 30.0 (b) Medium 191 3.4 2.1 3.0 182 40.3 31.6 33.0 (c) Large 97 3.9 2.8 3.0 93 42.2 30.4 40.0 Non-innovators vs. Innovators (d) Non-innovators 271 3.5 2.6 3.0 249 39.1 32.9 33.0 (e) Innovators 326 3.4 2.3 3.0 315 38.6 31.1 33.0 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 3.6 2.3 3.0 56 36.1 30.3 33.0 (e2) New-to-ﬁrm (NTF) innovators 133 3.5 2.0 3.0 127 41.1 31.5 37.0 Difference N Mean S.E. N Mean (Share, %) S.E. (a)–(b), Small vs. Medium 500 0.0 0.2 472 −3.7 3.0 (b)–(c), Medium vs. Large 288 −0.5* 0.3 275 −1.9 4.0 (a)–(c), Small vs. Large 406 −0.6* 0.3 383 −5.6 3.8 (d)–(e), Non-innovators vs. Innovators 597 0.1 0.2 564 0.5 2.7 (e1)–(e2), NTM vs. NTF innovators 190 0.1 0.3 183 −5.0 5.0 Note *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively Table 1.14 Duration of R&D projects 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 44 Table 1.15 Termination or suspension of R&D projects Firms with projects that were terminated/suspended within the past 3 years Share of projects where R&D organizations have the authority to decide whether to terminate, suspend, or continue the project N Share (%) S.D. N Mean (Share, %) S.D. Median Entire sample 603 59.5 49.1 564 40.8 40.9 30.0 By ﬁrm size (a) Small 312 51.9 50.0 288 39.3 41.5 25.0 (b) Medium 193 61.7 48.7 182 43.1 41.3 30.0 (c) Large 98 79.6 40.5 94 40.8 38.4 30.0 Non-innovators vs. 1.6.1.3 Termination or Suspension of R&D Projects Innovators (d) Non-innovators 277 50.9 50.1 248 39.6 41.2 25.0 (e) Innovators 325 67.1 47.1 316 41.7 40.8 30.0 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 59.7 49.5 56 45.2 43.8 50.0 (e2) New-to-ﬁrm (NTF) innovators 134 64.2 48.1 128 40.6 41.1 29.0 Difference N Mean S.E. N Mean (Share, %) S.E. (a)–(b), Small vs. Medium 410 −9.7 4.5 470 −3.8 3.9 (b)–(c), Medium vs. Large 291 −17.9* 5.7 276 2.3 5.1 (a)–(c), Small vs. Large 410 −27.7* 5.6 382 −1.5 4.8 (d)–(e), Non-innovators vs. Innovators 602 −16.2* 4.0 564 −2.1 3.5 (e1)–(e2), NTM vs. NTF innovators 191 −4.5 7.7 184 4.6 6.7 Note *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 1.6 R&D Project Management 45 1.6 in the whole sample had terminated and/or suspended at least one of their R&D projects within past three years. We ﬁnd that the share of ﬁrms that had terminated or suspended an R&D project clearly increases with ﬁrm size, presumably because larger ﬁrms have a larger number of ongoing R&D projects (Table 1.13). The share of ﬁrms that had terminated and/or suspended projects is signiﬁcant higher for innovating ﬁrms (67.1%) than non-innovating ﬁrms (50.9%). This suggests that innovating ﬁrms face challenges during the product development process more frequently than non-innovating ﬁrms, resulting in a higher likelihood of termina- tion/suspension. However, when we look at the shares for new-to-market innova- tors (59.7%) and new-to-ﬁrm innovators (64.2%), we ﬁnd that it is slightly lower for new-to-market innovators. Because new-to-market innovators are more likely to face challenges than new-to-ﬁrm innovators, this ﬁnding is inconsistent with our interpretation above that innovating ﬁrms are more likely to terminate or suspend projects because they face challenges more frequently. We infer that innovating ﬁrms are more likely to terminate/suspend projects because both the share of innovating ﬁrms (Table 1.3) and the share of ﬁrms that had terminated or suspended an R&D project (Table 1.15) increases with ﬁrm size. This inference is consistent with the observation that the share of new-to-market innovators is highest among small ﬁrms, whereas the share of new-to-ﬁrm innovators is smallest among small ﬁrms (Table 1.4). In Sect. 1.5.3, we examined to what extent R&D organizations have the authority to hire R&D employees (Table 1.12). In Table 1.15, we examine the delegation of authority to R&D organizations from a different angle. 1.6.1.3 Termination or Suspension of R&D Projects Speciﬁcally, we examine the share of projects in a ﬁrm where the R&D organization has the authority to decide whether to terminate/suspend or continue the project. We ﬁnd that the mean of this share is 40.8% (median: 30.0%). Looking at the different subsamples, the shares are approximately 40% for small, medium, and large ﬁrms, and for product innova- tors and non-innovators. Among innovators, the share is slightly higher for new-to- market innovators (45.2%) than for new-to-ﬁrm innovators (40.6%), but the differ- ence between the two is insigniﬁcant. In contrast to Table 1.12, where we observed that the percentage share of ﬁrms whose R&D organizations had the initiative to hire R&D personnel is higher for new-to-market innovators than for new-to-ﬁrm innovators, there is no difference between the two in terms of R&D organizations’ authority to decide on the continuation of projects. 1.6.2 Staged Project Management In the survey, we deﬁne staged project management as “a method of R&D project management in which the project proceeds in consecutive multiple stages 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 46 (phases).”20 Table 1.16(a) shows the percentage shares of ﬁrms that implement staged project management for R&D projects. Moreover, ﬁrms that employed staged project management were asked to provide more details, including the average number of stages, whether they set intermediate goals (milestones) for the interim evaluation of projects, and whether they provide feedback on the interim evaluation results to R&D personnel in charge of the project. The results for these additional questions are shown in Table 1.16(b). The percentage share of ﬁrms that employ staged project management is 51.3% (313 ﬁrms). We ﬁnd that the likelihood of employing staged project management is higher for larger ﬁrms. The share of innovating ﬁrms that employed staged project management (64.7%) is higher than that of non-innovating ﬁrms (35.6%), suggesting that there is a positive link between staging and making product innovation. Among innovating ﬁrms, the share is 59.6% for new-to-market innovators and 57.4% for new-to-ﬁrm innovators. The difference between the two subsamples is small and insigniﬁcant. Turning to the average number of stages in Table 1.16(b), we ﬁnd that the mean of the average number of stages for the whole sample is 4.6. It is slightly larger for large ﬁrms (5.2) than for small ﬁrms (4.1). The differences between product innovators and non-innovators and between new-to-market product innovators and new-to-ﬁrm product innovators are small and insigniﬁcant. Next, we look at the shares of ﬁrms that set milestones for interim evaluation and that provide feedback on the evaluation results to R&D personnel. Among ﬁrms that employ staged project management, 78.6% set milestones for interim evaluation, and 85.3% provide feedback on the interim evaluation results to R&D personnel. Large ﬁrms are more likely to set milestones: the share of large ﬁrms that set mile- stones (93.3%) is signiﬁcantly higher than those of medium (79.3%) and small ﬁrms (69.7%). However, we ﬁnd that the provision of feedback does not increase with ﬁrm size. The share of ﬁrms that set milestones is signiﬁcantly higher among product inno- vators (81.3%) than non-innovators (72.7%). In addition, the share of ﬁrms that provide feedback is also signiﬁcantly higher among product innovators (88.8%) than non-innovators (77.6%). 20 Speciﬁcally, in the glossary of terms sent to respondents (see the Appendix), we deﬁned staged project management as follows: “Staged project management” refers to the management of R&D projects in consecutive stages, such as “ideation and concept development,” “preliminary assessment of commercial- ization,” “development,” “testing and validation,” and “production and marketing.” Staged project management also entails a phase-based interim evaluation that affects the decision about whether the project is continued, suspended, or abandoned, as well as a revision of the schedule. 1.6.2 Staged Project Management Among product innovators, the share that set milestones is similar for new-to-market (79.4%) and new-to-ﬁrm innovators (78.2%). The share of ﬁrms that provide feedback is slightly higher for new-to-market innovators (88.2%) than new-to-ﬁrm innovators (82.1%), but the difference between the subsamples is insigniﬁcant. 47 1.6 R&D Project Management Table 1.16 Staged project management (a) Share of ﬁrms that implement staged project management Staged project management N Share (%) S.D. Entire sample 610 51.3 50.0 By ﬁrm size (a) Small 314 42.0 49.4 (b) Medium 194 54.6 49.9 (c) Large 102 73.5 44.3 Non-innovators vs. Innovators (d) Non-innovators 278 35.6 48.0 (e) Innovators 331 64.7 47.9 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 59.6 49.5 (e2) New-to-ﬁrm (NTF) innovators 136 57.4 49.6 Difference N Mean S.E. (a)–(b), Small vs. Medium 508 −12.6* 4.5 (b)–(c), Medium vs. Large 296 −18.9* 5.9 (a)–(c), Small vs. Large 416 −31.5* 5.5 (d)–(e), Non-innovators vs. Innovators 609 −29.0* 3.9 (e1)–(e2), NTM vs. NTF innovators 193 2.3 7.8 (continued) 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 48 Table 1.16 Staged project management (b) Number of stages, milestones, and feedback Number of stages Intermediate goals for the interim evaluation of projects (milestones) Feedback on the interim evaluation results N Mean S.D. Median N Share (%) S.D. N Share (%) S.D. Entire sample 305 4.6 4.2 4 313 78.6 41.1 312 85.3 35.5 By ﬁrm size (a) Small 130 4.1 2.3 4 132 69.7 46.1 131 86.3 34.6 (b) Medium 105 4.7 4.2 4 106 79.3 40.7 106 82.1 38.5 (c) Large 70 5.2 6.4 4 75 93.3 25.1 75 88.0 32.7 Non-innovators vs. Innovators (d) Non-innovators 97 4.3 4.4 3 99 72.7 44.8 98 77.6 41.9 (e) Innovators 208 4.7 4.1 4 214 81.3 39.1 214 88.8 31.6 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 34 5.1 4.5 5 34 79.4 41.0 34 88.2 32.7 (e2) New-to-ﬁrm (NTF) innovators 76 4.6 4.5 4 78 78.2 41.6 78 82.1 38.6 Difference N Mean S.E. N Mean S.E. N Mean S.E. (a)–(b), Small vs. Medium 235 −0.5 0.4 238 −9.5* 5.7 237 4.2 4.8 (b)–(c), Medium vs. Large 175 −0.5 0.8 181 −14.1*** 5.3 181 −5.9 5.5 (a)–(c), Small vs. Large 200 −1.1* 0.6 207 −23.6*** 5.8 206 −1.7 4.9 (d)–(e), Non-innovators vs. Innovators 305 −0.4 0.5 313 −8.6* 5.0 312 −11.2* 4.3 (e1)–(e2), NTM vs. 1.6.2 Staged Project Management NTF innovators 110 0.5 0.9 112 1.2 8.5 112 6.2 7.6 Note The sample consists of ﬁrms that implement staged project management. , , and indicate signiﬁcance at the 1, 5, and 10% levels respectively 1.6 R&D Project Management 49 To sum up, we ﬁnd that ﬁrms that succeed in making product innovations are more likely to manage R&D projects in stages, set milestone for interim evaluation, and provide feedback on a project’s interim evaluation results to researchers. These results are consistent with the literature highlighting the positive effects of staging in R&D projects (Cooper 1988, 2017) and VC investments (Sahlman 1990; Gompers 1995; Kaplan and Strömberg 2003). We do not ﬁnd any differences between new-to- market innovators and new-to-ﬁrm innovators. In particular, we do not ﬁnd evidence suggesting that staging deters ﬁrms from engaging in explorative R&D projects. In the following subsections, we examine the role of milestones and feedback in more detail. 1.6.2.1 Milestones As noted, ﬁrms that employ staged project management were asked a number of further questions. In addition to whether they set milestones (Table 1.16), we also asked how important these milestones were for successfully completing a project. Speciﬁcally, we divided project management stages into “initial stages” (e.g., idea/basic research) and “late stages” (e.g., preparation for launching new goods/services) and asked to what extent ﬁrms took into account whether milestones were achieved when assessing whether to terminate/suspend or continue the R&D project. Table 1.17 reports the results for the importance of milestones, with panel (a) showing those for initial stages and panel (b) those for late stages. In the initial stages, 28.0% of ﬁrms “fully” take into account whether milestones were achieved in decidingwhethertoterminate/suspendorcontinueanR&Dproject.Incontrast,inlate stages, 63.0% of ﬁrms fully take milestones into account. These results indicate that the achievement of milestones is more important in late stages than in initial stages. In addition, if we assume that ﬁrms that “fully” take the achievement of milestones into account for initial stages are ﬁrms that employ “the threat of termination” (Manso 2011) as a way to manage their R&D projects, our result suggests that 28.0% of Japanese ﬁrms use such threat of termination. Another notable feature in Table 1.17 is that the share of ﬁrms that answered “not very much” or “not at all” to the question of whether they take milestones into account is 10.9% (8.9 + 2.0) for the initial stages. This suggests that about 10% of ﬁrms in our sample have “tolerance for early failure” (Manso 2011). For comparison, the corresponding share for later stages 2.4% (2.0 + 0.4). Next, using ﬁrms that “fully” take milestones into account and ﬁrms that take mile- stones “not very much” or “not at all” into account, we examine whether and how “the threat of termination” and “tolerance for early failure” is linked to innovation outcomes. 1.6.2.1 Milestones The shares of ﬁrms in which R&D projects face the threat of termination (i.e., milestones are “fully taken into account” in the initial stages) and the shares of ﬁrms that have a tolerance for early failure (the sum of ﬁrms saying milestones are “not very much taken into account” or “not taken into account at all” in the initial 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 50 Table 1.17 Importance of milestones (a) Initial stages N Fully taken into account To some extent taken into account Not very much taken into account Not taken into account at all Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 246 28.0 45.0 61.0 48.9 8.9 28.6 2.0 14.1 By ﬁrm size (a) Small 92 26.1 44.2 65.2 47.9 7.6 26.7 1.1 10.4 (b) Medium 84 25.0 43.6 64.3 48.2 8.3 27.8 2.4 15.3 (c) Large 70 34.3 47.8 51.4 50.3 11.4 32.0 2.9 16.8 Non-innovators vs. Innovators (d) Non-innovators 72 22.2 41.9 68.1 47.0 8.3 27.8 1.4 11.8 (e) Innovators 174 30.5 46.2 58.1 49.5 9.2 29.0 2.3 15.0 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 27 22.2 42.4 59.3 50.1 14.8 36.2 3.7 19.2 (e2) New-to-ﬁrm (NTF) innovators 61 26.2 44.4 62.3 48.9 9.8 30.0 1.6 12.8 Difference N Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 176 1.1 6.6 0.9 7.2 −0.7 4.1 −1.3 2.0 (b)–(c), Medium vs. Large 154 −9.3 7.4 12.9 8.0 −3.1 4.8 −0.5 2.6 (a)–(c), Small vs. Large 162 −8.2 7.3 13.8* 7.8 −3.8 4.6 −1.8 2.1 (d)–(e), Non-innovators vs. Innovators 246 −8.2 6.3 10.0 6.8 −0.9 4.0 −0.9 2.0 (e1)–(e2), NTM vs. NTF innovators 88 −4.0 10.1 −3.0 11.4 5.0 7.4 2.1 3.5 (continued) 1.6 R&D Project Management 51 Table 1.17 Importance of milestones (b) Late stages N Fully taken into account To some extent taken into account Not very much taken into account Not taken into account at all Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 246 63.0 48.4 34.6 47.7 2.0 14.1 0.4 6.4 By ﬁrm size (a) Small 92 59.8 49.3 35.9 48.2 4.3 20.5 0.0 0.0 (b) Medium 84 60.7 49.1 39.3 49.1 0.0 0.0 0.0 0.0 (c) Large 70 70.0 46.2 27.1 44.8 1.4 12.0 1.4 12.0 Non-innovators vs. 1.6.2.1 Milestones Innovators (d) Non-innovators 72 56.9 49.9 38.9 49.1 4.2 20.1 0.0 0 (e) Innovators 174 65.5 47.7 32.8 47.1 1.1 10.7 0.6 7.6 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 27 63.0 49.2 33.3 48.0 3.7 19.2 0.0 0.0 (e2) New-to-ﬁrm (NTF) innovators 61 60.7 49.3 39.3 49.3 0.0 0.0 0.0 0.0 Difference N Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 176 −0.9 7.4 −3.4 7.3 4.3* 2.2 0.0 0.0 (b)–(c), Medium vs. Large 154 −9.3 7.7 12.1 7.6 −1.4 1.3 −1.4 1.3 (a)–(c), Small vs. Large 162 −10.2 7.6 8.7 7.4 2.9 2.8 −1.4 1.2 (d)–(e), Non-innovators vs. Innovators 246 −8.6 6.8 6.1 6.7 3.0 2.0 −0.6 0.9 (e1)–(e2), NTM vs. NTF innovators 88 2.3 11.4 −6.0 11.3 3.7 2.4 0.0 0.0 Note Figures represent the share of ﬁrms that ticked a particular answer in response to the following question: “To what extent do you take into account whether intermediate goals (milestones) were achieved when assessing whether to terminate/suspend or continue the R&D project?” *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 52 stages) are similar among small, medium, and large ﬁrms; between product inno- vators and non-innovators; and between new-to-market and new-to-ﬁrm innovators. For example, the share of ﬁrms that fully take milestones into account in the initial stages is 22.2% for new-to-market innovators, while it is 26.2% for new-to-ﬁrm innovators. This seems consistent with Manso’s (2011) argument that the threat of termination promotes exploitation, but the difference between the two subsamples is insigniﬁcant. Similarly, the share of ﬁrms that take milestone not very much or not at all into account in the initial stages is 18.5% (14.8 + 3.7) for new-to-market inno- vators, while it is 11.4% (9.8 + 1.6) for new-to-ﬁrm innovators. Again, this seems consistent with Manso’s (2011) argument that tolerance for early failure promotes exploration, but the difference between the two is insigniﬁcant. Overall, we do not ﬁnd evidence that the threat of termination and/or tolerance for early failure affects the likelihood of engaging in explorative (new-to-market) innovation. Our results are consistent with Manso’s (2011) theoretical prediction that whether the threat of termination is beneﬁcial or detrimental to exploration is ambiguous but inconsistent with empirical ﬁndings by Ederer and Manso (2013) and Mao et al. 1.6.2.1 Milestones (2014), who ﬁnd that the threat of termination is detrimental to exploration. Finally, it should be noted that to what extent a ﬁrm considers the achievement of milestones in deciding whether to continue an R&D project may be associated not only with the extent to which ﬁrms employ the threat of termination in their project management and with their tendency to tolerate early failure, but may also be linked to other factors. For example, a ﬁrm that engages in a joint research project with other ﬁrms (external partners) may put greater emphasis on the achievement of milestones to comply with contracts governing such joint research. Other factors such as receiving funds from VC investors, who usually invest in stages and set milestones for their investment, may also be positively associated with the extent to which the achievement of milestones is taken into account. Therefore, to examine whether and how the threat of termination and tolerance for early failure are associated with innovation outcomes requires proper statistical analysis, which we leave for future studies. 1.6.2.2 Feedback Further,asmentioned,inthecaseofﬁrmsthatemploystagedprojectmanagement,we also asked about feedback to researchers. Table 1.18 reports the results for the ques- tion about whose opinions are incorporated when providing feedback on the interim evaluation results to R&D employees. In the survey questionnaire, we provided three possible, not mutually exclusive options: “Opinions from other research teams within R&D organizations,” “opinions from non-R&D organizations (business units and head ofﬁce) within the company,” and “opinions (including informal ones) from external experts outside the company.” We again divided project management stages into initial and late stages and asked respondents in which stages these opinions wereincorporated.Aﬁrmmay,forexample,incorporateopinionsfromotherresearch teams within R&D organizations for both the initial and late stages (multiple answers 53 1.6 R&D Project Management Table 1.18 Feedback on interim evaluation results: Opinions that are incorporated (a) Opinions from other research teams in the same or other R&D organizations N Incorporated in initial stages Incorporated in late stages Not incorporated Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 266 70.7 45.6 55.3 49.8 24.4 43.1 By ﬁrm size (a) Small 113 70.8 45.7 55.8 49.9 25.7 43.9 (b) Medium 87 73.6 44.4 59.8 49.3 20.7 40.7 (c) Large 66 66.7 47.5 48.5 50.4 27.3 44.9 Non-innovators vs. Innovators (d) Non-innovators 76 72.4 45.0 57.9 49.7 22.4 41.9 (e) Innovators 190 70.0 45.9 54.2 50.0 25.3 43.6 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 30 66.7 47.9 46.7 50.7 30.0 46.6 (e2) New-to-ﬁrm (NTF) innovators 64 67.2 47.3 50.0 50.4 29.7 46.0 Difference N Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 200 −2.8 6.4 −4.0 7.1 5.0 6.1 (b)–(c), Medium vs. Large 153 6.9 7.5 11.3 8.1 −6.6 6.9 (a)–(c), Small vs. Large 179 4.1 7.2 7.3 7.8 −1.6 6.9 (d)–(e), Non-innovators vs. Innovators 266 2.4 6.2 3.7 6.8 −2.9 5.9 (e1)–(e2), NTM vs. NTF innovators 94 −0.5 10.5 −3.3 11.2 0.3 10.2 (continued) 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 54 Table 1.18 Feedback on interim evaluation results: Opinions that are incorporated (b) Opinions from non-R&D organizations (business units and head ofﬁce) within the company N Incorporated in initial stages Incorporated in late stages Not incorporated Share (%) S.D. Share (%) S.D. Share (%) S.D. 1.6.2.2 Feedback If a ﬁrm did not incorporate speciﬁc opinions in any of the stages, we asked ﬁrms to answer with “not incorporated.” Table 1.18 reports the results. The most frequent answer regarding whose opin- ions are incorporated is non-R&D organizations within the company (Table 1.18(b), initial stage: 70.7%, late stage: 83.8%), followed by other research teams within R&D organizations (Table 1.18(a), initial: 70.7%, late: 55.3%), and external experts outside the company (Table 1.18(c), initial: 30.0%, late: 24.7%). We ﬁnd that the share of ﬁrms that incorporate opinions from external experts is the smallest among the three options, and 62.4% of ﬁrms in the sample do not incorporate opinions from external experts outside the company at all. Comparing the stage when opin- ions are incorporated into feedback, we ﬁnd that opinions from other teams within R&D organizations and those from non-R&D organizations are equally incorpo- rated in the initial stages, whereas opinions from the latter are more likely to be incorporated in the late stages. These results suggest that ﬁrms’ main concern is the technological feasibility of product ideas when a project is launched, and as the project progresses, concern gradually shifts to commercialization of the invention and product marketing. Looking at the subsample results in Table 1.18(b), the share of ﬁrms that incor- porate opinions from non-R&D organizations in the initial stages is smallest among large ﬁrms (60.6%). In addition, the share of ﬁrms that do not incorporate opinions from non-R&D organizations is highest among large ﬁrms (9.1%). These results suggest that large ﬁrms tend to not incorporate opinions from non-R&D organiza- tions, presumably because they have a vertically segmented organizational structure. Next, comparing product innovators and non-innovators, the share of ﬁrms that incor- porate opinions from non-R&D organizations in the initial stages is signiﬁcantly higher among innovating ﬁrms (74.2%) than among non-innovating ﬁrms (61.8%). This suggests that opinions from non-R&D organizations are important for making product innovations. Looking at the subsample results in Table 1.18(c), we ﬁnd that the share of ﬁrms that incorporate opinions from external experts is higher among new-to-market inno- vators (initial: 36.7%, late: 46.7%) than new-to-ﬁrm innovators (initial: 26.6%, late: 23.4%). In addition, the share of ﬁrms that do not incorporate opinions from external experts is signiﬁcantly higher among new-to-ﬁrm innovators (67.2%) than new-to- market innovators (46.7%). 1.6.2.2 Feedback Entire sample 266 70.7 45.6 83.8 36.9 5.3 22.4 By ﬁrm size (a) Small 113 73.5 44.4 84.1 36.8 2.7 16.1 (b) Medium 87 74.7 43.7 81.6 39.0 5.8 23.4 (c) Large 66 60.6 49.2 86.4 34.6 9.1 29.0 Non-innovators vs. Innovators (d) Non-innovators 76 61.8 48.9 88.2 32.5 2.6 16.1 (e) Innovators 190 74.2 43.9 82.1 38.4 6.3 24.4 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 30 83.3 37.9 83.3 37.9 6.7 25.4 (e2) New-to-ﬁrm (NTF) innovators 64 71.9 45.3 84.4 36.6 4.7 21.3 Difference N Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 200 −1.3 6.3 2.5 5.4 −3.1 2.8 (b)–(c), Medium vs. Large 153 14.1* 7.5 −4.8 6.1 −3.3 4.2 (a)–(c), Small vs. Large 179 12.8* 7.2 −2.3 5.6 −6.4* 3.4 (d)–(e), Non-innovators vs. Innovators 266 −12.4 6.2 6.1 5.0 −3.7 3.0 (e1)–(e2), NTM vs. NTF innovators 94 11.5 9.5 −1.0 8.2 2.0 5.0 (continued) 55 1.6 R&D Project Management Table 1.18 Feedback on interim evaluation results: Opinions that are incorporated (c) Opinions (including informal ones) from external experts outside the company N Incorporated in initial stages Incorporated in late stages Not incorporated Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 266 30.0 45.9 24.7 43.2 62.4 48.5 By ﬁrm size (a) Small 113 31.8 46.8 26.4 44.3 60.0 49.2 (b) Medium 87 26.4 44.4 24.1 43.0 65.5 47.8 (c) Large 66 31.8 46.9 22.7 42.2 62.1 48.9 Non-innovators vs. Innovators (d) Non-innovators 76 29.3 45.8 25.3 43.8 62.7 48.7 (e) Innovators 190 30.3 46.1 24.5 43.1 62.2 48.6 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 30 36.7 49.0 46.7 50.7 46.7 50.7 (e2) New-to-ﬁrm (NTF) innovators 64 26.6 44.5 23.4 42.7 67.2 47.3 Difference N Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 200 5.4 6.6 2.2 6.3 −5.5 7.0 (b)–(c), Medium vs. Large 153 −5.4 7.4 1.4 7.0 3.4 7.9 (a)–(c), Small vs. Large 179 0.0 7.3 3.6 6.8 −2.1 7.6 (d)–(e), Non-innovators vs. Innovators 266 −1.0 6.3 0.9 5.9 0.4 6.6 (e1)–(e2), NTM vs. NTF innovators 94 10.1 10.2 23.2 10.0 −20.5* 10.7 Note *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 56 allowed). 21 In the survey, we deﬁne “ability” as a person’s potential ability to perform their duties (e.g., willingness and attitude, cognitive ability, and interpersonal skills). We deﬁne “performance” as a person’s results achieved in performing their duties. The results evaluated include, for example, patent applications, prototype products, and commercialization. 1.6.2.2 Feedback These results are consistent with Manso’s (2011) theo- retical prediction that timely feedback on interim performance promotes explorative innovation and Azoulay et al.’s (2011) ﬁnding that detailed and high-quality feedback from experts leads researchers to produce higher-impact scientiﬁc articles. While we found a positive link between seeking feedback from external experts and new-to-market innovation, we note that there may be other factors that affect from whom a ﬁrm seeks feedback. For instance, ﬁrms that obtain funding from VC investors are more likely to have feedback from external experts, because VC investors usually provide “value-added services” such as strategic and operational guidance (Gompers et al. 2020). Another example is small ﬁrms, which make up a sizable share of new-to-market innovators (Table 1.4) and may seek advice from outside experts to compensate for a lack of experts within the ﬁrm. 1.7 Evaluation of R&D Personnel 57 1.7 Evaluation of R&D Personnel This section, based on our discussion in Sect. 1.2.3, examines how ﬁrms design compensation contracts and incentive schemes to increase their R&D personnel’s motivation for innovation. We begin with an overview of salary schemes employed for R&D personnel (Sect. 1.7.1). We then look more deeply into compensation contracts and incentive schemes for R&D personnel. First, we examine to what extent Japanese ﬁrms use performance-based evaluations (Sect. 1.7.2). Speciﬁcally, in Sect. 1.7.2.1, we focus on the relative weights assigned to ability and performance in the evaluation of a young R&D employee.21 In addition, in Sect. 1.7.2.2, we consider the criteria ﬁrms use for the evaluation of R&D personnel. We use ﬁrms’ responses to examine whether there is a negative association between performance-based evaluation and the likelihood of innovation (as suggested, e.g., by Holmström and Milgrom 1991) or a positive association (as suggested by Prendergast 2002 and Foss and Laursen 2005). Second, we look at what kinds of pecuniary and non-pecuniary incentives ﬁrms offer to R&D personnel and examine whether there is a link between these incentives and innovation (Sect. 1.7.3). Third, we examine long-term incentives for R&D employees from two different perspectives (Sect. 1.7.4). Speciﬁcally, in Sect. 1.7.4.1, we revisit some of the questions used in Sects. 1.7.2.2 and 1.7.3. and regard the following two items as incentives for long-term success: whether a ﬁrm employs the amount of sales generated by new products as an evaluation criterion for the long-term success of R&D personnel (see Table 1.21 in Sect. 1.7.2.2), and whether a ﬁrm employs invention reward schemes as incentives for long-term success (see Table 1.22 in Sect. 1.7.3). In Sect. 1.7.4.2, we focus on another potential long-term incentive for innovation: the possibility of promotion of R&D personnel to top management levels. Speciﬁcally, we ask ﬁrms whether a person that belonged to an R&D organization in the past became a director on the board of the ﬁrm. 1.7.1 Salary Schemes for R&D Personnel We conducted pre-interviews with several Japanese ﬁrms that actively engaged in R&D activities and found that the majority of Japanese ﬁrms employ the same salary schemes for their R&D personnel as for other employees. Based on this ﬁnding, we asked ﬁrms what kind of salary scheme they employed for R&D personnel and whether it was different from the salary schemes for other employees (multiple choices allowed). Table 1.19 reports the results for this question. We ﬁnd that the most frequen 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 58 58 1 R&D Management Practices and Innovation: E Table 1.19 Salary schemes for R&D personnel N A speciﬁc salary scheme for R&D personnel Starting salary varies depending on the educational background Salary scheme based on performance-based pay R&D personnel can choose from among various schemes None of the schemes listed are employed Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 608 8.1 27.2 72.5 44.7 14.6 35.4 2.1 14.5 18.1 38.5 By ﬁrm size (a) Small 313 10.2 30.3 66.5 47.3 14.1 34.8 1.6 12.6 21.7 41.3 (b) Medium 194 7.7 26.8 77.3 42.0 17.5 38.1 2.1 14.2 13.9 34.7 (c) Large 101 2.0 14.0 82.2 38.5 10.9 31.3 4.0 19.6 14.9 35.7 Non-innovators vs. Innovators (d) Non-innovators 278 8.6 28.1 70.1 45.8 11.5 32.0 1.4 11.9 19.8 39.9 (e) Innovators 329 7.6 26.5 74.5 43.7 17.3 37.9 2.7 16.3 16.7 37.4 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 8.8 28.5 63.2 48.7 14.0 35.0 1.8 13.2 22.8 42.3 (e2) New-to-ﬁrm (NTF) innovators 135 4.4 20.7 77.8 41.7 15.6 36.4 0.7 8.6 14.8 35.7 (continued) 59 1.7 Evaluation of R&D Personnel Table 1.19 Salary schemes for R&D personnel Difference N A speciﬁc salary scheme for R&D personnel Starting salary varies depending on the educational background Salary scheme based on performance- based pay R&D personnel can choose from among various schemes None of the schemes listed are employed Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 507 2.5 2.7 −10.9* 4.1 −3.5 3.3 −0.5 1.2 7.8 3.6 (b)–(c), Medium vs. Large 295 5.8 2.8 −4.9 5.0 6.6 4.4 −1.9 2.0 −0.9 4.3 (a)–(c), Small vs. Large 414 8.2* 3.1 −15.7* 5.2 3.2 3.9 −2.4 1.7 6.9 4.6 (d)–(e), Non-innovators vs. 1.7.1 Salary Schemes for R&D Personnel Innovators 607 1.0 2.2 −4.3 3.6 −5.8 2.9 −1.3 1.2 3.1 3.1 (e1)–(e2), NTM vs. NTF innovators 192 4.3 3.7 −14.6 6.9 −1.5 5.7 1.0 1.6 8.0 6.0 Note Respondents were asked to choose all that apply. , , and indicate signiﬁcance at the 1, 5, and 10% levels respectively 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 60 answer is that the “starting salary varies depending on the educational background” (72.5%). Because such education-based salary schemes are widely used among ﬁrms in Japan not only for R&D personnel but also for other employees, the result is consistent with the ﬁnding in our pre-interviews. The second most frequent answer is that the ﬁrm uses a “salary system based on performance-based pay” (14.6%), followed by “a speciﬁc salary scheme for R&D personnel that differs from that for other employees” (8.1%) and “R&D personnel can choose from among various salary schemes” (2.1%). Meanwhile, 18.1% of ﬁrms in the sample answered that “none of the salary schemes listed above are employed.” Because the percentage shares of ﬁrms that use a performance-based salary scheme and a speciﬁc salary scheme for R&D personnel are relatively low, we infer that the majority of Japanese ﬁrms do not provide pay-for-performance compensation schemes for R&D employees. Looking at the subsample results, we ﬁnd that larger ﬁrms are more likely to employ education-based salary schemes. The share is highest for large ﬁrms (82.2%) and lowest for small ﬁrms (66.5%). In contrast, the share of ﬁrms that provide a speciﬁc salary scheme for R&D personnel is lowest for large ﬁrms (2.0%) and highest for small ﬁrms (10.2%). The share of ﬁrms that employ a performance-based salary system is signiﬁcantly higher for product innovators (17.3%) than for non-innovators (11.5%). Although the share of ﬁrms that employ a performance-based salary scheme is less than 20% even among product innovators, this ﬁnding suggests that there is a positive association between performance-based pay and product innovations. Meanwhile, there is little difference between product innovators and non-innovators in terms of the other salary schemes that we listed. Among product innovators, the share of ﬁrms that employ an education-based salary scheme is signiﬁcantly lower for new-to-market innovators (63.2%) than for new-to-ﬁrm innovators (77.8%). Because education-based salary schemes are widely usedamongJapaneseﬁrms, thelower ratiofor new-to-market innovators maysuggest that they are more likely to employ other salary schemes including performance- based compensation schemes. Note Figures represent the weight that respondent ﬁrms put on performance as opposed to ability in the evaluation of an R&D employee in their early 30s. “Performance” refers to the level of achieve- ment met in performing the job, while “ability” refers to the abilities demonstrated in performing the job. *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 1.7.1 Salary Schemes for R&D Personnel However, the results in Table 1.19 do not bear this out: the differences between new-to-market and new-to-ﬁrm innovators with regard to the use of performance-based pay or R&D personnel-speciﬁc schemes are small and insigniﬁcant. Finally, the share of ﬁrms that answered “none of the schemes above are employed” is higher among new-to-market innovators (22.8%) than for new-to-ﬁrm innovators (14.8%), which suggests that new-to-market innovators are more likely to employ salary schemes not listed in our survey, but the difference between the two is again insigniﬁcant. Overall, while we ﬁnd that new-to-market innovators are less likely to employ education-based salary schemes, we cannot pin down which salary schemes they are more likely to employ for R&D personnel. 1.7 Evaluation of R&D Personnel 61 1.7.2.1 Weights on Performance and Ability in Evaluation In this subsection, we focus on the results for the question about the weight ﬁrms put on “ability” and “performance” when evaluating R&D personnel in their early 30s. Table 1.20 shows the results for the weight put on performance (the results for ability are omitted to save space since the two weights add up to 100%). The mean of the approximate weight ﬁrms put on performance is 46.4%, while the mean of the weight on ability is 53.6%. Looking at subsamples, we observe several differences across ﬁrm groups in terms of the weight they put on performance. First, while the means of the weight on performance are quite similar across small, medium, and large ﬁrms, large ﬁrms put a somewhat higher weight on performance (49.3%) than small ﬁrms (44.8%) and this difference is signiﬁcant. Second, product innovators put a signiﬁcantly higher weight on performance than non-innovators: the mean of the weight is 47.9% for product innovators and 44.6% for non-innovators. This suggests that there is a positive link Table 1.20 Weight put on performance in employee evaluation N Mean (%) S.D. Median Entire sample 599 46.4 18.1 50.0 By ﬁrm size (a) Small 308 44.8 18.5 50.0 (b) Medium 192 47.4 16.9 50.0 (c) Large 99 49.3 18.6 50.0 Non-innovators vs. Innovators (d) Non-innovators 274 44.6 18.1 50.0 (e) Innovators 325 47.9 17.9 50.0 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 46.0 18.8 50.0 (e2) New-to-ﬁrm (NTF) innovators 133 47.6 18.7 50.0 Difference N Mean (%) S.E. (a)–(b), Small vs. Medium 500 −2.6 1.6 (b)–(c), Medium vs. Large 291 −2.0 2.2 (a)–(c), Small vs. Large 407 −4.5 2.1 (d)–(e), Non-innovators vs. Innovators 599 −3.3 1.5 (e1)–(e2), NTM vs. NTF innovators 190 −1.6 3.0 Table 1.20 Weight put on performance in employee evaluation Note Figures represent the weight that respondent ﬁrms put on performance as opposed to ability in the evaluation of an R&D employee in their early 30s. “Performance” refers to the level of achieve- ment met in performing the job, while “ability” refers to the abilities demonstrated in performing the job. 1.7.2.1 Weights on Performance and Ability in Evaluation *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 62 between product innovation and performance-based evaluation.22 Together with the positive link between product innovation and performance-based salary schemes in Table 1.19, these results are consistent with Foss and Laursen’s (2005) ﬁnding of a positive link between innovation and the use of performance pay.23 Finally, among product innovators, we do not ﬁnd a signiﬁcant difference between new-to-market and new-to-ﬁrm innovators. 23 Note that while we ﬁnd evidence of a positive link between product innovation and the use of performance pay, in this monograph we do not examine the reasons for such a link. A possible expla- nation is provided by Prendergast (2002), who argues that there may be a positive link between the use of pay-for-performance and uncertainty (i.e., innovation, in our case) since in uncertain environ- ments, a principal (manager) delegates responsibility to agents (employees) and uses performance- based pay to compensate for agents’ unobservable effort (see Sect. 1.2.3). To check whether the positive link between the weight ﬁrms put on performance in their evaluation of employees and product innovation we ﬁnd in Table 1.20 is consistent with Prendergast’s (2002) argument, we examine the link between the delegation of authority in hiring R&D personnel to R&D organiza- tions (Table 1.12) and the weight put on performance (Table 1.20). However, we ﬁnd that the mean of the weight on performance is almost identical between ﬁrms that delegate the initiative in hiring R&D personnel to R&D organizations and ﬁrms that do not. Similarly, we do not ﬁnd a positive link between delegation and the likelihood of employing a performance-based salary scheme in Table 1.19. Investigating the mechanism through which product innovation is positively linked to performance is an issue we leave for future research. 22 We note that such a positive link between product innovation and performance-based evaluation could be due to the positive link between product innovation and ﬁrm size on the one hand and the positive link between the weight put on performance and ﬁrm size on the other hand. 1.7.2.2 Criteria for the Evaluation of R&D Personnel Table 1.21 presents results regarding the criteria that ﬁrms employ for the evaluation of R&D personnel (multiple choices allowed). The survey provided seven criteria as possible choices. Two of the criteria are related to the ability of R&D personnel, namely,“researchpapersandconferencepresentations”and“acquisitionofqualiﬁca- tions/degree.” The other four criteria are related to their performance, namely, “patent applications/registrations,” “commercialization (launch) of new products,” “amount of sales generated by new products to which the R&D employee contributed,” and “R&D progress, including compliance with a schedule.” We ﬁnd that the most widely used criterion is “R&D progress” (71.5%). The shares of ﬁrms using the other criteria are relatively similar. In descending order, they are “acquisition of qualiﬁcations/degrees” (18.6%), “amount of sales generated by new products” (14.7%), “commercialization of new products” (14.7%), “patent applications/registrations” (9.6%), and “research papers and presentations” (4.4%). The share of ﬁrms answering that “none of the criteria listed above are employed” is 8.1%. These ﬁgures indicate that the share of the most frequently cited performance- related criterion (R&D progress) is about four times as large as the share of the most frequently cited ability-related criterion (acquisition of qualiﬁcation/degree). This suggests that most ﬁrms in the sample are likely to employ performance- rather than ability-related criteria for the evaluation of R&D employees. That said, it should 22 We note that such a positive link between product innovation and performance-based evaluation could be due to the positive link between product innovation and ﬁrm size on the one hand and the positive link between the weight put on performance and ﬁrm size on the other hand. 63 1.7 Evaluation of R&D Personnel Table 1.21 Criteria for the evaluation of R&D personnel N Ability-related criteria Performance-related criteria None of the criteria listed are employed Research papers and conference presentations Acquisition of qualiﬁcations / degrees Patent applications / registrations Commercialization (Launch) of new products Amount of sales generated by new products to which the R&D employee contributed R&D progress, including compliance with schedule Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. 1.7.2.2 Criteria for the Evaluation of R&D Personnel NTF innovators 191 3.3 3.4 −7.9 6.1 9.1 4.6 6.3 5.3 3.4 5.7 −7.0 7.0 −0.4 4.1 Note Respondents were asked to choose all that apply * ** and * indicate signiﬁcance at the 1 5 and 10% levels respectively 1.7 Evaluation of R&D Personnel 65 be noted that R&D progress, including whether an R&D employee complies with a schedule, may not be a clear-cut criterion of employees’ performance since it does not reﬂect the outcome of their R&D activities. be noted that R&D progress, including whether an R&D employee complies with a schedule, may not be a clear-cut criterion of employees’ performance since it does not reﬂect the outcome of their R&D activities. Large ﬁrms tend to utilize a variety of performance- and ability-related criteria for evaluation. Speciﬁcally, the shares of large ﬁrms using ability-related criteria are signiﬁcantly higher than those of small ﬁrms: the percentage share of ﬁrms using “research papers/presentations” is 7.9% for large ﬁrms and 3.5% for small ﬁrms, while the percentage share of ﬁrms using “acquisition of qualiﬁcations/degrees” is 29.7% for large ﬁrms and 17.9% for small ﬁrms. Similarly, product innovators are more likely to employ both ability- and performance-related criteria, as the share of ﬁrms that replied “none of the - criteria listed are employed” is signiﬁcantly smaller among product innovators (5.5%) than non-innovators (11.2%). Speciﬁcally, the share of ﬁrms using “R&D progress” is signiﬁcantly higher among product innovators (75.3%) than non- innovators (66.9%). Among product innovators, the share of ﬁrms using “patent applications/registrations” is signiﬁcantly higher among new-to-market innovators (15.8%) than new-to-ﬁrm innovators (6.7%). Again, this result is consistent with the positive link between the use of performance pay and innovation found by Foss and Laursen (2005). A larger share of new-to-market innovators use “commercialization of new products” and “research papers/presentations” as criteria, although the differ- ences between new-to-ﬁrm innovators with regard to these criteria are statistically insigniﬁcant. 1.7.2.2 Criteria for the Evaluation of R&D Personnel Entire sample 607 4.4 20.6 18.6 39.0 9.6 29.4 14.7 35.4 14.7 35.4 71.5 45.2 8.1 27.3 By ﬁrm size (a) Small 313 3.5 18.4 17.9 38.4 8.9 28.6 17.9 38.4 15.3 36.1 67.1 47.1 10.2 30.3 (b) Medium 193 4.1 20.0 14.0 34.8 9.8 29.9 10.9 31.2 14.5 35.3 76.7 42.4 5.7 23.2 (c) Large 101 7.9 27.1 29.7 45.9 10.9 31.3 11.9 32.5 12.9 33.7 75.2 43.4 5.9 23.8 Non-innovators vs. Innovators (d) Non-innovators 278 4.3 20.4 15.8 36.6 9.0 28.7 15.8 36.6 15.1 35.9 66.9 47.1 11.2 31.5 (e) Innovators 328 4.6 20.9 20.7 40.6 10.1 30.1 13.7 34.5 14.3 35.1 75.3 43.2 5.5 22.8 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 7.0 25.8 12.3 33.1 15.8 36.8 17.5 38.4 17.5 38.4 68.4 46.9 7.0 25.8 (e2) New-to-ﬁrm (NTF) innovators 134 3.7 19.0 20.1 40.3 6.7 25.1 11.2 31.6 14.2 35.0 75.4 43.2 7.5 26.4 (continued) 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 64 Table 1.21 Criteria for the evaluation of R&D personnel Difference N Ability-related criteria Performance-related criteria None of the criteria listed are employed Research papers and conference presentations Acquisition of qualiﬁcations / degrees Patent applications / registrations Commercialization (Launch) of new products Amount of sales generated by new products to which the R&D employee contributed R&D progress, including compliance with schedule Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 506 −0.6 1.7 3.9 3.4 −0.9 2.7 7.0** 3.3 0.8 3.3 −9.6 4.1 4.5* 2.5 (b)–(c), Medium vs. Large 294 −3.8 2.8 −15.7*** 4.8 −1.0 3.7 −1.0 3.9 1.6 4.3 1.4 5.2 −0.2 2.9 (a)–(c), Small vs. Large 414 −4.4* 2.4 −11.8 4.6 −1.9 3.3 6.0 4.2 2.5 4.1 −8.2 5.3 4.3 3.3 (d)–(e), Non-innovators vs. Innovators 606 −0.3 1.7 −4.9 3.2 −1.1 2.4 2.1 2.9 0.8 2.9 −8.4 3.7 5.7** 2.2 (e1)–(e2), NTM vs. 1.7.3 Incentive Schemes for R&D Personnel Next, we turn to incentive schemes for R&D personnel. The survey provided a list of six different types of incentives—three pecuniary and three non-pecuniary incentives—and asked respondents to tick those that the ﬁrm employed (multiple choices allowed). The non-pecuniary incentives were “in-house research presenta- tions,” “dispatch to university and/or support for studying abroad,” and “open recruit- ment for R&D projects,” while the pecuniary incentives were “awards for outstanding R&D results,” “rewards based on the number of patent applications” (referred to as “patent-based rewards” hereinafter), and “rewards based on the amount of proﬁts from inventions and patents (invention reward schemes).” Table 1.22 shows that the two most widely used incentives are “patent-based rewards” (49.6%) and “invention reward schemes” (44.8%). In descending order, the percentage shares of the other incentives are “awards for outstanding R&D results” (16.4%), “in-house research presentations” (10.5%), “dispatch to university/abroad” (4.8%), and “open recruitment for R&D projects” (1.6%). Meanwhile, 21.7% of sampleﬁrmsrespondedthat“noneoftheschemesaboveareemployed.”Theseresults indicate that among ﬁrms that employ at least one of the listed incentive schemes, pecuniary incentive schemes such as patent-based rewards and invention reward schemes are more likely to be employed than non-pecuniary incentive schemes. 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 66 g Table 1.22 Incentive schemes for R&D personnel N Non-pecuniary (intrinsic) incentives Pecuniary (extrinsic) incentives None of the schemes listed are employed In-house research presentations Dispatch to university and/or support for studying abroad Open recruitment for R&D projects Awards for outstanding R&D results Rewards based on the number of patent applications Rewards based on the amount of proﬁts from inventions and patents (invention reward system) Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 609 10.5 30.7 4.8 21.3 1.6 12.7 16.4 37.1 49.6 50.0 44.8 49.8 21.7 41.2 By ﬁrm size (a) Small 313 9.3 29.0 3.5 18.4 2.2 14.8 16.9 37.6 37.4 48.5 31.3 46.4 33.9 47.4 (b) Medium 194 14.4 35.2 4.6 21.1 1.0 10.1 17.5 38.1 58.8 49.4 54.6 49.9 8.8 28.3 (c) Large 102 6.9 25.4 8.8 28.5 1.0 9.9 12.7 33.5 69.6 46.2 67.6 47.0 8.8 28.5 Non-innovators vs. Innovators (d) Non-innovators 278 10.8 31.1 4.3 20.4 1.4 11.9 15.1 35.9 41.7 49.4 37.4 48.5 28.8 45.4 (e) Innovators 329 10.3 30.5 5.2 22.2 1.8 13.4 17.6 38.2 56.2 49.7 50.8 50.1 15.8 36.5 New-to-market vs. 1.7.3 Incentive Schemes for R&D Personnel p y We ﬁnd that product innovators are more likely to employ at least one of the incentive schemes listed in the questionnaire: the share of product innovators that replied “none of the schemes above are employed” is 15.8%, while that of non- innovators is 28.8%. Consistent with this ﬁnding, the share of ﬁrms using a particular incentive scheme is larger for product innovators than innovators for all schemes except in-house research presentations. In particular, the share of product innovators that employ patent-based rewards (56.2%) is signiﬁcantly higher than that of non- innovators (41.7%). Similarly, the share of product innovators that employ invention reward schemes (50.8%) is signiﬁcantly higher than that of non-innovators (37.4%). Among product innovators, we ﬁnd that new-to-market innovators are less likely to employ non-pecuniary incentive schemes than new-to-ﬁrm innovators. The shares of new-to-market innovators that use “in-house research presentations” (1.8%) and “dispatch to university/studying abroad” (0.0%) are signiﬁcantly lower than those of new-to-ﬁrm innovators (10.4% and 6.7%, respectively). Turning to pecuniary incentives, the shares of new-to-market innovators that use patent-based rewards (47.4%) and invention reward schemes (45.6%) are also lower than those of new- to-ﬁrm innovators (55.2% and 50.7%, respectively), but the differences between new-to-market and new-to-ﬁrm innovators are insigniﬁcant. We ﬁnd that product innovators are more likely to employ at least one of the incentive schemes listed in the questionnaire: the share of product innovators that replied “none of the schemes above are employed” is 15.8%, while that of non- innovators is 28.8%. Consistent with this ﬁnding, the share of ﬁrms using a particular incentive scheme is larger for product innovators than innovators for all schemes except in-house research presentations. In particular, the share of product innovators that employ patent-based rewards (56.2%) is signiﬁcantly higher than that of non- innovators (41.7%). Similarly, the share of product innovators that employ invention reward schemes (50.8%) is signiﬁcantly higher than that of non-innovators (37.4%). Among product innovators, we ﬁnd that new-to-market innovators are less likely to employ non-pecuniary incentive schemes than new-to-ﬁrm innovators. The shares of new-to-market innovators that use “in-house research presentations” (1.8%) and “dispatch to university/studying abroad” (0.0%) are signiﬁcantly lower than those of new-to-ﬁrm innovators (10.4% and 6.7%, respectively). 1.7.3 Incentive Schemes for R&D Personnel New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 1.8 13.2 0.0 0.0 3.5 18.6 21.1 41.1 47.4 50.4 45.6 50.3 22.8 42.3 (e2) New-to-ﬁrm (NTF) innovators 134 10.4 30.7 6.7 25.1 1.5 12.2 17.2 37.8 55.2 49.9 50.7 50.2 20.1 40.3 (continued) 67 1.7 Evaluation of R&D Personnel 67 Table 1.22 Incentive schemes for R&D personnel Difference N Non-pecuniary (intrinsic) incentives Pecuniary (extrinsic) incentives None of the schemes listed are employed In-house research presentations Dispatch to university and/or support for studying abroad Open recruitment for R&D projects Awards for outstanding R&D results Rewards based on the number of patent applications Rewards based on the amount of proﬁts from inventions and patents (invention reward system) Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 507 −5.2* 2.9 −1.1 1.8 1.2 1.2 −0.6 3.5 −21.4* 4.5 −23.3* 4.4 25.1*** 3.8 (b)–(c), Medium vs. Large 296 7.6* 3.9 −4.2 2.9 0.1 1.2 4.8 4.5 −10.8* 5.9 −13.0 6.0 −0.1 3.5 (a)–(c), Small vs. Large 415 2.4 3.2 −5.3 2.4 1.3 1.6 4.2 4.2 −32.2* 5.5 −36.3* 5.3 25.0* 5.0 (d)–(e), Non-innovators vs. Innovators 607 0.5 2.5 −0.9 1.7 −0.4 1.0 −2.5 3.0 −14.5* 4.0 −13.3* 4.0 13.0* 3.3 (e1)–(e2), NTM vs. NTF innovators 191 −8.7 4.2 −6.7 3.3 2.0 2.3 3.9 6.1 −7.9 7.9 −5.1 7.9 2.7 6.5 Note Respondents were asked to choose all that apply. *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 68 1 R&D Management Practices and Innovation: Evidence from a Firm Survey Looking at subsamples, larger ﬁrms are more likely to employ patent-based rewards and invention reward schemes. For example, the share of large ﬁrms that employ patent-based rewards (69.6%) is signiﬁcantly higher than those of small and medium ﬁrms (37.4% and 58.8%, respectively). Among non-pecuniary incentives, the share of ﬁrms that use “in-house research presentations” is signiﬁcantly higher for medium-sized ﬁrms (14.4%) than for small ﬁrms (9.3%) and large ﬁrms (6.9%). Meanwhile, 33.9% of small ﬁrms used “none of the incentive schemes listed above,” suggesting that small ﬁrms employ other incentives not listed in the questionnaire or did not provide any incentives at all. 1.7.4.1 Rewards for Long-Term Success Toexaminetheroleofincentivesforlong-termsuccess,wefocusonthefollowingtwo items already discussed in Sects. 1.7.2.2 and 1.7.3: the “amount of sales generated by new products to which the R&D employee contributed,” which was one of the options in the question on criteria used for the evaluation of R&D personnel (Table 1.21), and “rewards based on the amount of proﬁts from inventions and patents (invention reward schemes),” which was one of the options in the question on incentive schemes for R&D personnel (Table 1.22). In the sections above, we used these items to examine whether a ﬁrm focused more on ability or performance in the evaluation of R&D personnel (Sect. 1.7.2.2) and whether a ﬁrm used pecuniary or non-pecuniary incentives (Sect. 1.7.3). In this subsection, we revisit these two items to examine whether ﬁrms employ rewards for long-term success since it takes time for product innovations to result in sales or proﬁts. Table 1.23 shows that 14.7% of sample ﬁrms employ “the sales amount generated by new products” for the evaluation of R&D personnel. As discussed with regard to Table 1.21, the share of ﬁrms using this criterion is not notably higher than the share of other criteria, presumably because it is difﬁcult to measure a single person’s contribution to the introduction of a new product. In terms of the share of ﬁrms using this criterion, there are no signiﬁcant differences among small, medium, and large ﬁrms, between product innovators and non-innovators, and between new-to-market and new-to-ﬁrm innovators. Next, turning to invention reward schemes, Table 1.23 shows that 44.8% of sample ﬁrms employ such schemes as part of their incentive for R&D personnel. Larger ﬁrms are more likely to employ invention reward schemes: the share is 67.6% for large ﬁrms, 54.6% for medium ﬁrms, and 31.3% for small ﬁrms. Similarly, product innova- tors(50.8%)aremorelikelytoemployinventionrewardschemesthannon-innovators (37.4%). Among product innovators, the share of ﬁrms that employ invention reward schemes is slightly lower for new-to-market innovators (45.6%) than for new-to-ﬁrm innovators (50.7%), but the difference between the two is insigniﬁcant. To summarize, we ﬁnd that product innovators are more likely to employ inven- tion reward schemes than non-innovators, but we do not ﬁnd a positive link between the “amount of sales generated by new products to which the R&D employee contributed” and product innovation. This suggests that whether rewards for long- term success contribute to making product innovation depends on the speciﬁc tool that a ﬁrm employs. 1.7.4 Incentives for Long-Term Success 1.7.4.1 Rewards for Long-Term Success 1.7.3 Incentive Schemes for R&D Personnel Turning to pecuniary incentives, the shares of new-to-market innovators that use patent-based rewards (47.4%) and invention reward schemes (45.6%) are also lower than those of new- to-ﬁrm innovators (55.2% and 50.7%, respectively), but the differences between new-to-market and new-to-ﬁrm innovators are insigniﬁcant. To sum up, we found that product innovators are more likely to employ pecuniary incentives than non-innovators. This result is consistent with studies such as Onishi (2013) and Sauermann and Cohen (2010) but inconsistent with studies that found a negative link between pecuniary incentives and innovation (e.g., Onishi et al. 2021). Regarding non-pecuniary incentives, we do not ﬁnd any associations between non- pecuniary incentive schemes and product innovation. This ﬁnding is inconsistent with Sauermann and Cohen (2010), who ﬁnd a positive link between non-pecuniary motives (i.e., preference for intellectual challenge and independence) and innovation (i.e., the number of patent applications). However, among product innovators, we ﬁnd that new-to-ﬁrm innovators are more likely to employ non-pecuniary incentives than new-to-market innovators. This ﬁnding indicates that the link between non-pecuniary incentive schemes and innovation outcomes depends on how innovation is measured. 1.7 Evaluation of R&D Personnel 69 1.7.4.1 Rewards for Long-Term Success We do not ﬁnd evidence that new-to-market innovators are more likely to use rewards for long-term success than new-to-ﬁrm innovators. 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 70 Table 1.23 Rewards for long-term success Amount of sales generated by new products Rewards based on the amount of proﬁts from inventions and patents (invention reward system) N Share (%) S.D. N Share (%) S.D. Entire sample 607 14.7 35.4 609 44.8 49.8 By ﬁrm size (a) Small 313 15.3 36.1 313 31.3 46.4 (b) Medium 193 14.5 35.3 194 54.6 49.9 (c) Large 101 12.9 33.7 102 67.6 47.0 Non-innovators vs. Innovators (d) Non-innovators 278 15.1 35.9 278 37.4 48.5 (e) Innovators 328 14.3 35.1 329 50.8 50.1 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 17.5 38.4 57 45.6 50.3 (e2) New-to-ﬁrm (NTF) innovators 134 14.2 35.0 134 50.7 50.2 Difference N Share (%) S.E. N Share (%) S.E. (a)–(b), Small vs. Medium 506 0.8 3.3 507 −23.3* 4.4 (b)–(c), Medium vs. Large 294 1.6 4.3 296 −13.0 6.0 (a)–(c), Small vs. Large 414 2.5 4.1 415 −36.3* 5.3 (d)–(e), Non-innovators vs. Innovators 606 0.8 2.9 607 −13.3* 4.0 (e1)–(e2), NTM vs. NTF innovators 191 3.4 5.7 191 −5.1 7.9 Note The results for “Amount of sales generated by new products” are reproduced from Table 1.21, while the results for “Rewards based on the amount of proﬁts from inventions and patents (invention reward system)” are reproduced from Table 1.22. *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 71 1.7 Evaluation of R&D Personnel Note The table shows the share of ﬁrms that have a director on the board that belonged to an R&D organization in the past. *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 1.7.4.2 Possibility of Promotion The survey examines another potential incentive for R&D personnel for long-term success: the possibility of promotion. As discussed in Sect. 1.7.3, promotion is a typical reward for long-term success. In the survey, we asked whether any of the direc- tors on the board (e.g., chairperson, president, vice president) belonged to an R&D organization in the past. For the majority of Japanese ﬁrms, some board members are internally promoted from among employees who have worked at the ﬁrm for a long time. Given this, the possibility that an R&D employee could potentially be nomi- nated as a director on the board may work as a long-term incentive for rank-and-ﬁle R&D personnel. Table 1.24 shows that 38.3% of survey ﬁrms have at least one director on the board who belonged to an R&D organization in the past. Looking at subsamples, there are no signiﬁcant differences among small (37.6%), medium (37.9%), and large (41.2%) ﬁrms. Meanwhile, product innovators are more likely to have directors on the board who belonged to an R&D organization than non-innovators. The share of ﬁrms that have a board member from an R&D organization is 42.6% among product innovators, while it is 33.1% among non-innovators, and the difference between the Table 1.24 Directors on the board from R&D organizations N Share (%) S.D. Entire sample 611 38.3 48.7 By ﬁrm size (a) Small 314 37.6 48.5 (b) Medium 195 37.9 48.7 (c) Large 102 41.2 49.5 Non-innovators vs. Innovators (d) Non-innovators 278 33.1 47.1 (e) Innovators 331 42.6 49.5 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 40.4 49.5 (e2) New-to-ﬁrm (NTF) innovators 136 41.9 49.5 Difference N Share (%) S.E. (a)–(b), Small vs. Medium 509 −0.4 4.4 (b)–(c), Medium vs. Large 297 −3.2 6.0 (a)–(c), Small vs. Large 416 −3.6 5.6 (d)–(e), Non-innovators vs. Innovators 609 −9.5 3.9 (e1)–(e2), NTM vs. NTF innovators 193 −1.6 7.8 Table 1.24 Directors on the board from R&D organizations Note The table shows the share of ﬁrms that have a director on the board that belonged to an R&D organization in the past. , , and indicate signiﬁcance at the 1, 5, and 10% levels respectively 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 72 two subsamples is signiﬁcant. Among product innovators, we ﬁnd no differences between the share for new-to-market innovators (40.4%) and new-to-ﬁrm innovators (41.9%). To summarize, the results in this subsection and Sect. 1.8 Risk Preferences and Corporate Culture Recent studies show that ﬁrms’ risk preferences and corporate culture are impor- tant determinants of their risk-taking, including investment in R&D, as discussed in Sect. 1.2.4. In this section, we examine how ﬁrms’ risk preferences and corporate culture are linked with innovation outcomes. In Sect. 1.8.1, we provide an overview of the results for three questions that try to capture ﬁrms’ risk preferences. First, the survey directly asked whether respondents felt that their ﬁrm is taking appropriate risk in R&D projects. Second, it set a hypo- thetical question about an R&D project and asked about the maximum amount that a ﬁrm would invest in this project to indirectly infer ﬁrms’ risk preferences. Third, it asked respondent ﬁrms to choose between two otherwise identical projects: one that has a greater NPV but negative cash ﬂow for the ﬁrst few years and another that has a smaller NPV but positive cash ﬂow throughout its duration. In Sect. 1.8.2, we measure ﬁrms’ corporate culture by employing the CVF (Cameron et al. 2014) introduced in Sect. 1.2.4, which categorizes corporate culture into the following quadrants: Collaborate, Control, Compete, and Create. 1.7.4.2 Possibility of Promotion 1.7.4.1 indicate that product innovators are more likely to employ invention reward schemes and have a director on the board who belonged to an R&D organization than non-innovators. This suggests that long-term incentives may be effective in promoting product innova- tions. However, we do not ﬁnd evidence that new-to-market innovators are more likely to employ incentive schemes for long-term success than new-to-market innova- tors. This result does not support Manso’s (2011) argument that reward for long-term success is essential for motivating exploration. 1.8.1 Risk Preferences (2021), who asked the same question in their survey of large North American ﬁrms and found that 60.2% respondents felt their ﬁrm took the “right amount of risk.” We ﬁnd that the share of ﬁrms where the respondent believes the ﬁrm takes too much risk is higher for larger ﬁrms: the shares for large, medium, and small ﬁrms are 12.1%, 10.2%, and 5.4%, respectively, and the difference between large and small ﬁrms and that between medium and small ﬁrms are both statistically signiﬁcant. By contrast, the share of ﬁrms where respondents felt their ﬁrm takes an appropriate level of risk is highest for small ﬁrms (73.7%). g The share of ﬁrms where respondents felt their ﬁrm takes the appropriate level of risk is signiﬁcantly higher for product innovators (72.0%) than non-innovators (63.9%), suggesting taking the right amount of risk is important for product inno- vation. Among product innovators, the share of new-to-market innovators where respondents thought their ﬁrm takes the appropriate level of risk is 82.4%, which is signiﬁcantly larger than the corresponding share for new-to-ﬁrm innovators (65.2%). In contrast, the share of new-to-market innovators where respondents thought that their ﬁrm does not take much risk is 13.7%, which is signiﬁcantly smaller than the corresponding share for new-to-ﬁrm innovators (27.0%). Again, this result suggests that appropriate risk taking is important for explorative innovations. Table 1.26 shows the results for the following question: “Suppose there is an R&D project that is expected to generate gross sales of 100 million yen immediately if it is successful but gross sales of 0 yen if it fails. Assume that the probability that the project is successful is 10%. How much would you invest in the project?” This question also tries to capture ﬁrms’ subjective risk preferences, again based on respondents’ own judgement. The expected return of this R&D project is 10 million yen (10%×100 million yen), meaning that a risk-neutral ﬁrm would invest 10 million yen.24 Accordingly, in Table 1.26, we classify ﬁrms as “risk-neutral” if they are willing to invest 10 million yen, as “risk-averse” if they are willing to invest less than 10 million yen, and as “risk-tolerant” if they are willing to invest more than 10 million yen. We ﬁnd that the share of risk-averse ﬁrms is 37.4%, that of risk-neutral ﬁrms 30.8%, and that of risk-tolerant ﬁrms 31.8%. 1.8.1 Risk Preferences Table 1.25 shows the results for the following question: “Do you think your R&D organization is taking appropriate risks in R&D projects to achieve its goals?” This question tries to capture ﬁrms’ subjective risk preferences based on respondents’ own judgement. Among the 606 ﬁrms that responded to this question, 60 ﬁrms (13.2%) chose “Do not know.” We exclude these ﬁrms, resulting in 526 observations used for Table 1.25. We ﬁnd that 68.4% of respondents in the sample believe that their ﬁrm takes an appropriate level of risk, 23.4% believe that their ﬁrm does not take much risk, and 8.2% believe that their ﬁrm takes too much risk. These results are similar to those 1.8 Risk Preferences and Corporate Culture 73 Table 1.25 Risk preferences: Respondents’ assessment of their ﬁrm’s taking of risk N Does not take much risk Takes appropriate level of risk Takes too much risk Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 526 23.4 42.4 68.4 46.5 8.2 27.4 By ﬁrm size (a) Small 259 20.8 40.7 73.7 44.1 5.4 22.7 (b) Medium 176 27.8 44.9 61.9 48.7 10.2 30.4 (c) Large 91 22.0 41.6 65.9 47.7 12.1 32.8 Non-innovators vs. Innovators (d) Non-innovators 233 27.0 44.5 63.9 48.1 9.0 28.7 (e) Innovators 293 20.5 40.4 72.0 45.0 7.5 26.4 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 51 13.7 34.8 82.4 38.5 3.9 19.6 (e2) New-to-ﬁrm (NTF) innovators 115 27.0 44.6 65.2 47.8 7.8 27.0 Difference N Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 267 −7.0* 4.1 11.8*** 4.5 −4.8* 2.5 (b)–(c), Medium vs. Large 267 5.9 5.7 −4.0 6.2 −1.9 4.0 (a)–(c), Small vs. Large 350 −1.1 5.0 7.8 5.5 −6.7** 3.1 (d)–(e), Non-innovators vs. Innovators 526 6.6* 3.7 −8.1** 4.1 1.5 2.4 (e1)–(e2), NTM vs. NTF innovators 166 −13.2* 7.0 17.1 7.6 −3.9 4.2 Note Figures represent the percentage share of ﬁrms where the survey respondent chose a particular answer in response to the following question: “Do you think your R&D organization is taking appropriate risks in R&D projects to achieve its goals?” Firms where the respondent answered with “Do not know” are excluded. , , and indicate signiﬁcance at the 1, 5, and 10% levels respectively 74 1 R&D Management Practices and Innovation: Evidence from a Firm Survey obtained by Graham et al. 24 We assume that respondents did not consider the costs involved in making the product when answering this question. 1.8.1 Risk Preferences The t-tests among the subsamples indicate that there are no signiﬁcant differences in these shares among small, medium, and large ﬁrms; between product innovators and non-innovators; and between new- to-market and new-to-ﬁrm innovators. Finally, in the survey we asked a hypothetical question about the choice between two otherwise identical R&D projects. Project 1 offers a larger expected cumulative proﬁt (net present value) but is expected to make losses for several years after the launch of the product. Project 2 has a smaller expected cumulative proﬁt (net present value) but is expected to generate stable proﬁts after the launch of the product. We assume all other conditions (e.g., initial investment costs, the probability of success of the project, project duration, etc.) are the same for both Projects 1 and 2. We expect 1.8 Risk Preferences and Corporate Culture 75 Table 1.26 Risk preferences: Risk aversion N Risk-averse Risk-neutral Risk-tolerant Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 588 37.4 48.4 30.8 46.2 31.8 46.6 By ﬁrm size (a) Small 306 39.9 49.0 30.7 46.2 29.4 45.6 (b) Medium 190 33.7 47.4 31.1 46.4 35.3 47.9 (c) Large 92 37.0 48.5 30.4 46.3 32.6 47.1 Non-innovators vs. Innovators (d) Non-innovators 269 36.4 48.2 30.5 46.1 33.1 47.1 (e) Innovators 319 38.2 48.7 31.0 46.3 30.7 46.2 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 43.9 50.1 24.6 43.4 31.6 46.9 (e2) New-to-ﬁrm (NTF) innovators 132 39.4 49.0 32.6 47.0 28.0 45.1 Difference N Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 496 6.2 4.5 −0.3 4.3 −5.9 4.3 (b)–(c), Medium vs. Large 282 −3.3 6.1 0.6 5.9 2.7 6.1 (a)–(c), Small vs. Large 398 2.9 5.8 0.3 5.5 −3.2 5.5 (d)–(e), Non-innovators vs. Innovators 588 −1.8 4.0 −0.6 3.8 2.4 3.9 (e1)–(e2), NTM vs. NTF innovators 189 4.5 7.8 −8.0 7.3 3.5 7.2 Note This table is constructed using responses to the following question: “Suppose there is an R&D project that is expected to generate gross sales of 100 million yen immediately if it is successful but gross sales of 0 yen if it fails. Assume that the probability that the project is successful is 10% (hence, the probability of failure is 90%). How much would you invest in the project? 1.8.1 Risk Preferences Please enter the approximate amount.“ Firms are classiﬁed as “risk-neutral” if the investment amount is 10 million yen, as “risk-averse” if the investment amount is less than 10 million yen, and as “risk-tolerant” if the investment amount exceeds 10 million yen. *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 76 that less risk-averse ﬁrm will choose Project 1 and more risk-averse ﬁrms will choose Project 2. Graham et al. (2021) ask a similar question in their corporate survey and argue that this question also measures the short-termism of a ﬁrm. In this regard, we expect ﬁrms that have a larger discount factor to choose the NPV-inferior Project 2 and ﬁrms that tolerate early losses to choose the NPV-superior Project 1. Table1.27presentstheresult.Amongthe606ﬁrmsthatrespondedtothisquestion, 174 ﬁrms (27.7%) replied with “do not know.” We exclude these ﬁrms, leaving us with 432 observations. Table 1.27 shows that 28.7% of ﬁrms would choose Project 1 and the remaining 71.3% would choose Project 2. Graham et al. (2021) report that 59.4% of the ﬁrms responding to their survey chose Project 1, which is completely different from our result. The result of our survey suggests that many Japanese ﬁrms tend to avoid short-term losses even if a project makes a long-term proﬁt, which Table 1.27 Risk preferences: Choice regarding an NPV-superior but initially unproﬁtable project N Share (%) S.D. Entire sample 432 28.7 45.3 By ﬁrm size (a) Small 225 22.7 42.0 (b) Medium 141 32.6 47.1 (c) Large 66 40.9 49.5 Non-innovators vs. Innovators (d) Non-innovators 182 23.6 42.6 (e) Innovators 250 32.4 46.9 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 38 42.1 50.0 (e2) New-to-ﬁrm (NTF) innovators 107 27.1 44.7 Difference N Share (%) S.E. (a)–(b), Small vs. Medium 366 −10.0 4.7 (b)–(c), Medium vs. Large 207 −8.3 7.1 (a)–(c), Small vs. Large 291 −18.2* 6.1 (d)–(e), Non-innovators vs. Innovators 432 −8.8** 4.4 (e1)–(e2), NTM vs. NTF innovators 145 15.0* 8.7 able 1.27 Risk preferences: Choice regarding an NPV-superior but initially unproﬁtable project Note This table is constructed using responses to the following question: “Suppose the following two R&D projects: Project 1: The expected cumulative proﬁt (net present value) is large, but the project is expected make losses for several years after the launch of the product. 1.8.1 Risk Preferences Project 2: The expected cumulative proﬁt (net present value) is small, but the project is expected to generate stable proﬁts after the launch of the product. Assume all other conditions (e.g., initial investment costs, the probability of success of the project, project duration, etc.) are the same for both Projects 1 and 2. Which project would you choose?” The table shows the share of ﬁrms that chose Project 1. *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 8 Risk Preferences and Corporate Culture 77 1.8 suggests that Japanese ﬁrms are more risk-averse or more myopic than U.S. ﬁrms regarding the choice of R&D projects. suggests that Japanese ﬁrms are more risk-averse or more myopic than U.S. ﬁrms regarding the choice of R&D projects. Next, comparing subsamples, we ﬁnd that larger ﬁrms are more likely to choose the NPV-superior Project 1: the share of ﬁrms that would choose Project 1 is 40.9% among large ﬁrms, 32.6% among medium ﬁrms, and 22.7% among small ﬁrms, and the differences between large and small ﬁrms and between medium and small ﬁrms are both signiﬁcant. While this suggests that larger ﬁrms are more risk-tolerant, it should be noted that there may be other factors that affect ﬁrms’ choice. For example, small ﬁrms are more likely to be ﬁnancially constrained, and this might make them choose Project 2, which generates a stable cash ﬂow after the launch of the product. In this case, small ﬁrms are not necessarily inherently more risk averse; instead, their choice might reﬂect ﬁnancial constraints, and if such constraints were controlled for, their risk aversion may not differ from larger ﬁrms. Further, we ﬁnd that product innovators (32.4%) are more likely to choose Project 1 than non-innovators (23.6%). This suggests that ﬁrms that succeeded in making product innovations are more risk tolerant. Among innovating ﬁrms, we ﬁnd that the share of new-to-market innovators (42.1%) that would choose Project 1 is signif- icantly larger than that of new-to-ﬁrm innovators (27.1%), indicating that new-to- market innovators are more risk tolerant and/or more tolerant of early losses than new-to-ﬁrm innovators. To sum up, the results for our ﬁrst question about risk preferences suggest that taking the appropriate level of risk is important for product innovation, especially for new-to-market innovation. 1.8.1 Risk Preferences The results for the second question about risk prefer- ences are similar for product innovators and non-innovators and for new-to-market and new-to-ﬁrm innovators, suggesting that there is no link between risk preferences and innovation. The results for our third question regarding risk preferences indicate that product innovators, especially new-to-market innovators, are more risk tolerant and immune to short-termism. The results for the ﬁrst and third questions are consis- tent with studies reporting a positive effect of risk tolerance on innovation and/or exploration (Ederer and Manso 2013; Krieger et al. 2022; Carson et al. 2020; Tian and Wang 2014). It should be noted, however, that how risk preferences are linked with innovation depends on the empirical proxy used for risk preferences. 1.8.2 Corporate Culture To measure a ﬁrm’s corporate culture in the Competing Values Framework (CVF) developed by Cameron et al. (2014) (see Sect. 1.2.4), we asked ﬁrms to choose up to three options out of eight to describe their corporate culture. The eight options that we provided were “teamwork,” “bottom-up approach,” “leadership,” “rule-based decision making,” “customer ﬁrst,” “proﬁtability,” “market impact,” and “creativity.” We chose these options based on studies employing the CVF (Cameron et al. 2014; 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 78 Fiordelisi and Ricci 2014; Thakor 2016; Hanaeda et al. 2020). The options corre- spond to the following quadrants: “teamwork” and “bottom-up approach” repre- sent a Collaborate-oriented culture, “leadership” and “rule-based decision making” represent a Control-oriented culture, “customer ﬁrst” and “proﬁtability” repre- sent a Compete-oriented culture, and “market impact” and “creativity” represent a Create-oriented culture. Table 1.28 shows the percentage shares of ﬁrms that chose each option. Note that these shares do not add up to 100 percent because ﬁrms were asked to choose up to three options. The option that ﬁrms chose most frequently is “customer ﬁrst” (72.4%), followed by “proﬁtability” (45.0%), “teamwork” (36.0%), and “creativity” (29.9%). Because the top two options, “customer ﬁrst” and “proﬁtability,” represent a Compete-oriented culture, it appears that many Japanese ﬁrms value competitiveness and prioritize customers and shareholders. Signiﬁcantly larger shares of large ﬁrms than small and/or medium ﬁrms chose “bottom-up approach,” “rule-based decision making,” and “customer ﬁrst.” On the other hand, signiﬁcantly larger shares of small ﬁrms than medium and/or large ﬁrms chose “market impact” and “creativity,” both of which correspond to a Create-oriented culture. The latter result suggests that many small ﬁrms value creativity. We fail to ﬁnd any correlations between corporate culture, including a Create- oriented culture, and the likelihood of making product innovations. The difference in the percentage shares of each of the options between product innovators and non- innovators are all insigniﬁcant. We ﬁnd, however, a signiﬁcant difference between new-to-market innovators and new-to-ﬁrm innovators: the share of new-to-market innovators that chose “market impact” (29.8%) and “creativity” (43.9%) is signiﬁ- cantly larger than that of new-to-ﬁrm innovators (market impact: 12.5%, creativity: 18.4%). This indicates that, among ﬁrms that succeeded in making product innova- tions, new-to-market innovators put high value on creativity as part of their corporate culture. As discussed in Sect. 1.8.2 Corporate Culture 1.2.4, as far as we are aware, there are no empirical studies that examine the relationship between corporate culture and innovation using the CVF. We note, however, that our ﬁnding that new-to-market innovation is positively associated with a Create-oriented culture is consistent, or at least not inconsistent, with the study by Graham et al. (2021), who do not use the CVF but ﬁnd that “adapt- ability,” which is one of the cultural values they measure, is positively correlated with creativity. Because “adaptability” corresponds to an orientation toward “indi- viduality and ﬂexibility” in the CVF (on the top of the Y axis in Fig. 1.1), ﬁrms with a culture that values adaptability can be classiﬁed as falling into either the Collabo- rate or Create quadrants. In addition, the ﬁnding that new-to-market innovators put higher value on creativity than new-to-ﬁrm innovators is consistent with Manso’s (2011) theoretical conjecture that a corporate culture that tolerates early failure and rewards long-term success motivates exploration. 1.8 Risk Preferences and Corporate Culture 79 Table 1.28 Corporate culture using the CVF N Collaborate-oriented Control-oriented Teamwork Bottom-up approach Leadership Rule-based decision making Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 609 36.0 48.0 8.9 28.5 12.6 33.3 20.2 40.2 By ﬁrm size (a) Small 313 34.8 47.7 6.7 25.1 13.7 34.5 16.3 37.0 (b) Medium 195 35.9 48.1 9.2 29.0 12.3 32.9 23.1 42.2 (c) Large 101 39.6 49.2 14.9 35.7 9.9 30.0 26.7 44.5 Non-innovators vs. Innovators (d) Non-innovators 277 35.4 47.9 6.9 25.3 12.6 33.3 17.7 38.2 (e) Innovators 330 36.4 48.2 10.6 30.8 12.4 33.0 22.4 41.8 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 29.8 46.2 5.3 22.5 8.8 28.5 21.1 41.1 (e2) New-to-ﬁrm (NTF) innovators 136 39.7 49.1 13.2 34.0 13.2 34.0 25.0 43.5 Difference N Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 508 −1.1 4.4 −2.5 2.4 1.4 3.1 −6.8* 3.6 (b)–(c), Medium vs. Large 296 −3.7 5.9 −5.6 3.9 2.4 3.9 −3.7 5.3 (a)–(c), Small vs. Large 414 −4.8 5.5 −8.1 3.2 3.8 3.8 −10.4 4.5 (d)–(e), Non-innovators vs. Innovators 607 −1.0 3.9 −3.7 2.3 0.2 2.7 −4.7 3.3 (e1)–(e2), NTM vs. 1.9 Conclusion This monograph provided a detailed account of the current R&D management prac- tices in Japanese ﬁrms based on a unique ﬁrm survey. Using data from the survey, we presented descriptive statistics and conducted t-tests to make inferences about the link between R&D management practices and (i) ﬁrms’ success in making product innovation and (ii) the choice between exploration (new-to-market product innova- tion) and exploitation (new-to-ﬁrm product innovation) among product innovators. We ﬁnd many interesting and instructive facts about the R&D management practices of Japanese ﬁrms. Speciﬁcally, we ﬁnd the following. • Organizational structure of R&D (Sect. 1.5): Product innovators are more likely to have a “hybrid” R&D organizational structure with both centralized and decen- tralized R&D activities than non-innovators. While some studies ﬁnd that ﬁrms with a centralized R&D structure are more likely to generate explorative innova- tions that have a higher level of impact (e.g., Argyres and Silverman 2004), we do not ﬁnd a link between new-to-market product innovation and the likelihood of having a centralized R&D structure. Regarding the delegation of authority to R&D organizations, we ﬁnd that the share of ﬁrms where both the R&D organization and the human resources depart- ment take the initiative in hiring R&D personnel is higher among product inno- vators than non-innovators. In contrast, the share of ﬁrms where the R&D organi- zation takes the initiative is signiﬁcantly lower for product innovators than non- innovators. These results suggest that there is no link, or a negative link, between the delegation of authority in hiring R&D personnel and the success of product innovation. However, among product innovators, the share of ﬁrms where the R&D organization takes the initiative is higher among new-to-market innovators than new-to-ﬁrm innovators. This result is consistent with studies ﬁnding that the delegation of authority to R&D organizations promotes exploration (Acemoglu et al. 2007; Kastl et al. 2013). We note, however, that in Sect. 1.6.1.3 we do not ﬁnd a link between new-to-market innovation and the delegation of authority regarding R&D project management. It seems safe to conclude that whether delegation of authority to R&D organizations motivates exploration is ambiguous and depends on the empirical proxy used. • Staged project management (Sect. 1.6): Product innovators are more likely to implementstagedprojectmanagement,setinterimgoals(milestones),andprovide feedback to R&D personnel than non-innovators. 1.8.2 Corporate Culture NTF innovators 193 −9.9 7.6 −8.0 4.9 −4.5 5.1 −3.9 6.8 (continued) 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 80 Table 1.28 Corporate culture using the CVF N Compete-oriented Create-oriented Customer ﬁrst Proﬁtability Market impact Creativity Share (%) S.D. Share (%) S.D. Share (%) S.D. Share (%) S.D. Entire sample 609 72.4 44.7 45.0 49.8 16.6 37.2 29.9 45.8 By ﬁrm size (a) Small 313 69.0 46.3 45.4 49.9 19.5 39.7 35.8 48.0 (b) Medium 195 73.3 44.3 47.2 50.0 14.9 35.7 23.1 42.2 (c) Large 101 81.2 39.3 39.6 49.2 10.9 31.3 24.8 43.4 Non-innovators vs. Innovators (d) Non-innovators 277 70.4 45.7 46.9 50.0 15.5 36.3 30.0 45.9 (e) Innovators 330 74.2 43.8 43.3 49.6 17.6 38.1 29.4 45.6 New-to-market vs. New-to-ﬁrm innovators (e1) New-to-market (NTM) innovators 57 68.4 46.9 42.1 49.8 29.8 46.2 43.9 50.1 (e2) New-to-ﬁrm (NTF) innovators 136 73.5 44.3 39.7 49.1 12.5 33.2 18.4 38.9 Difference N Share (%) S.E. Share (%) S.E. Share (%) S.E. Share (%) S.E. (a)–(b), Small vs. Medium 508 −4.3 4.2 −1.8 4.6 4.6 3.5 12.7* 4.2 (b)–(c), Medium vs. Large 296 −7.9 5.2 7.6 6.1 4.0 4.2 −1.7 5.2 (a)–(c), Small vs. Large 414 −12.2 5.1 5.8 5.7 8.6 4.3 11.0 5.4 (d)–(e), Non-innovators vs. Innovators 607 −3.8 3.6 3.6 4.1 −2.1 3.0 0.6 3.7 (e1)–(e2), NTM vs. NTF innovators 193 −5.1 7.1 2.4 7.8 17.3* 5.9 25.5* 6.7 Note Respondent ﬁrms were asked to choose up to three options that describe the ﬁrm’s corporate culture. *, , and * indicate signiﬁcance at the 1, 5, and 10% levels respectively 1.9 Conclusion 81 1.9 Conclusion Regarding the choice between exploration and exploitation, we ﬁnd that ﬁrms that introduced new-to-market product innovations are more likely to incorporate opinions from external experts into interim feedback. This result is consistent with theoretical studies predicting that timely feedback on performance promotes explorative innovation (Manso 2011) and empirical studies that ﬁnd a positive link between feedback and explo- ration (Azoulay et al. 2011). There are no signiﬁcant differences between product innovators and non-innovators and between new-to-market innovators and new- to-ﬁrm innovators regarding the extent to which ﬁrms consider the achievement of 1 R&D Management Practices and Innovation: Evidence from a Firm Survey 82 milestones as important when deciding whether to continue an R&D project. This result suggests that the effect of the threat of termination on innovation, in partic- ular explorative innovation, is ambiguous. Our result is consistent with Manso’s (2011) theoretical prediction but inconsistent with empirical ﬁndings by Ederer and Manso (2013) and Mao et al. (2014), who ﬁnd that the threat of termination is detrimental to exploration. • Compensation and incentive schemes for R&D personnel (Sect. 1.7): Product innovators are more likely to employ a salary scheme based on performance- based pay, put greater weight on performance than ability in the evaluation of R&D personnel, and employ pecuniary incentive schemes such as rewards based on the number of patent applications, than non-innovators. We also ﬁnd that, among product innovators, new-to-market innovators are more likely to employ patent applications/registrations as a criterion in evaluating R&D personnel than new-to-ﬁrm innovators. Our results are consistent with studies that ﬁnd a positive link between innovation and pay-for-performance (e.g., Foss and Laursen 2005). Further, they are also consistent with studies that ﬁnd a positive link between inno- vation and pecuniary incentives (Onishi 2013; Sauermann and Cohen 2010) but inconsistent with theoretical studies arguing that pecuniary incentives for R&D employees may adversely affect their intrinsic motivation (Bénabou and Tirole 2003; Kreps 1997) and empirical studies reporting a negative link between inno- vation and pecuniary incentives (e.g., Onishi et al. 2021). While some studies ﬁnd a positive link between innovation and non-pecuniary incentives (e.g., Sauermann and Cohen 2010), we do not ﬁnd such evidence. 1.9 Conclusion Finally, we ﬁnd that the share of ﬁrms that employ invention reward schemes, in which R&D employees receive rewards based on the amount of proﬁts the ﬁrm has made from the inventions and patents that the R&D employee was engaged in, are larger for product innovators than non-innovators. We also ﬁnd that product innovators are more likely to have a director on the board who belonged to an R&D organization in the past than non- innovators, which suggests that there is a positive link between innovation and reward for long-term success. However, we do not ﬁnd any evidence for a positive link between new-to-market innovation and reward for long-term success, which means that we do not ﬁnd support for Manso’s (2011) argument that reward for long-term success encourages exploration. • Risk preferences and corporate culture (Sect. 1.8): Using the three questions that try to capture a ﬁrm’s risk preferences, we ﬁnd that product innovators are more likely to take the appropriate level of risk than non-innovators. We also ﬁnd that product innovators are more risk-tolerant and immune to short-termism. The results that product innovators are taking the appropriate level of risk and are risk- tolerant are driven mainly by new-to-market rather than new-to-ﬁrm innovators. Our results are consistent with studies that ﬁnd a positive link between risk toler- ance and innovation and/or exploration (Ederer and Manso 2013; Krieger et al. 2022; Carson et al. 2020; Tian and Wang 2014). We note, however, that how risk preferences are associated with innovation depends on the way risk preferences are measured. • Risk preferences and corporate culture (Sect. 1.8): Using the three questions that try to capture a ﬁrm’s risk preferences, we ﬁnd that product innovators are more likely to take the appropriate level of risk than non-innovators. We also ﬁnd that product innovators are more risk-tolerant and immune to short-termism. The results that product innovators are taking the appropriate level of risk and are risk- tolerant are driven mainly by new-to-market rather than new-to-ﬁrm innovators. Our results are consistent with studies that ﬁnd a positive link between risk toler- ance and innovation and/or exploration (Ederer and Manso 2013; Krieger et al. 2022; Carson et al. 2020; Tian and Wang 2014). We note, however, that how risk preferences are associated with innovation depends on the way risk preferences are measured. © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5 1.9 Conclusion 1.9 Conclusion 83 We ﬁnd no link between product innovation and any of the corporate cultures (quadrants) identiﬁed by the CVF. We ﬁnd, however, that among product inno- vators, new-to-market innovators put higher value on a Create-oriented culture than new-to-ﬁrm innovators, which suggests that corporate culture does play an important role in motivating explorative innovation, as suggested by Manso (2011). In our discussion of the survey results, we linked the various ﬁndings with the theoretical and empirical literature on which our survey questions were based in order to outline possible mechanisms at work. However, the ﬁndings based on descriptive statistics and univariate analyses using t-tests represent only a ﬁrst step, and more rigorous statistical analyses employing multivariate regression models to examine the link between R&D organization and innovation while controlling for a range of factors are needed. Such factors include differences in ﬁrm size, industry, and ﬁrms’ ﬁnancial condition. Moreover, there were other ﬁndings which we did not investigate much in this monograph. For instance, we found that the empirical link between some of the R&D management practices (such as the delegation of authority to hire employees to R&D organization and the implementation of staged project management) and innovation outcomes depends on the proxy used for innovation outcomes—in our case, whether a ﬁrm has made any product innovations or new-to-market product innovations. This suggests that a management practice that is effective in increasing the likelihood of product innovation may not be effective in encouraging explorative innovation. We leave this issue as well as more rigorous analyses for future research. Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made. g The images or other third party material in this chapter are included in the chapter’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. Survey of R&D Management Practices • This survey is supported by Grants-in-Aid for Scientiﬁc Research (B) provided by the Japan Society for the Promotion of Science. • Your responses are strictly conﬁdential and no information provided by individual survey participants will be disclosed. Survey results will be statistically processed maintaining anonymity. • We recommend that you respond online. Access the following URL (https://res earch.surece.co.jp/rd-mgmt2020/) and log in using the ID and password provided in the attached “Request for online response.” • Please submit the completed questionnaire by February 17, 2020 (Monday). • We will send a summary report of the survey results to survey participants that have responded. • Unless otherwise stated, provide your answers as of FY2018. • Answer all questions on a non-consolidated basis, that is, on the basis of the activ- ities of your company only and exclude those of afﬁliated companies including the parent company and subsidiaries. • For a deﬁnition of “R&D” in this survey and detailed deﬁnitions of terms marked with * in the questionnaire, see the glossary of terms attached to this survey. • In what follows, we will ask about (1) R&D expenditure, R&D personnel, and R&D organizational structure, (2) R&D project management, (3) personnel eval- uation of researchers and engineers, and (4) R&D outputs. We would appreciate it if the person most qualiﬁed to respond to these issues answers the survey. © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5 8 85 Appendix 86 1. R&D expenditure, R&D personnel, and R&D organizational structure These questions are about the R&D expenditure1 of your company. These questions are about the R&D expenditure1 of your company. Q1: Please provide the approximate percentage shares of funding sources for R&D expenditure. Funding from headquarters (or the business unit to which R&D organization belongs) Commissions received from other business units within the company Funding from outside the company (commissions, subsidies, grants, etc.) Other Total % % % % 100% 1) “R&D expenditure” refers to the total amount of R&D expenditure spent inside and outside the company, irrespective of the source of the funds (your own funds, externally received funds, etc.) These questions are about the R&D expenditure of your company. Q1: Please provide the approximate percentage shares of funding sources for R&D expenditure. Funding from headquarters (or the business unit to which R&D organization belongs) Commissions received from other business units within the company Funding from outside the company (commissions, subsidies, grants, etc.) Other Total % % % % 100% 1) “R&D expenditure” refers to the total amount of R&D expenditure spent inside and outside the company, irrespective of the source of the funds (your own funds, externally received funds, etc.) Q2: To what extent do you take the following items (a)-(f) into account when determining total R&D expenditure? Choose one answer for each item. Fully taken into account To some extent taken into account Not very much taken into account Not taken into account at all (a) Gross sales in the previous year (b) Profits in the previous year (c) R&D expenditure in the previous year (d) Labor costs of R&D organization(s) (e) Cumulative costs of individual R&D projects (f) Annual sales goals for new products as a share of total sales Q2: To what extent do you take the following items (a)-(f) into account when determining total R&D expenditure? Choose one answer for each item. 87 Appendix The following questions are about your company’s R&D personnel. 2 Q3: Please provide the approximate age composition of your R&D personnel (percentage shares). © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5 8 Under 24 y/o From 25 to 34 y/o From 35 to 44 y/o From 45 to 54 y/o 55 y/o and older Total % % % % % 100% 2) “R&D personnel” refers to persons who have at least a bachelor’s degree (excluding junior college) or have equivalent or greater expertise, conduct R&D activities on their own specific topic, and engage in R&D activities for more than half of their working hours. The following questions are about your R&D organization(s). 3 Q6: How many R&D organizations does your company have? Number of R&D organizations (Number of R&D organizations overseas) 3) “R&D organization” refers to an organization (e.g., department, division) in which R&D personnel mainly conduct R&D. Even if the name of the organization does not include “research” or “development,” an organization that conducts R&D activities is regarded as an “R&D organization” for the purpose of this survey. Please provide the number of organizations that in your company’s organization chart can be regarded as “R&D organizations.” Q7: Among the R&D organizations in Q6, does your company have one or more R&D organizations that are (a) highly independent of business units and/or (b) directly controlled by business units? Choose all that apply. □ (a) My company has one or more R&D organizations that are highly independent of business units (e.g., central research laboratory, development department). □ (b) My company has one or more R&D organizations that are directly controlled by business units (e.g., pharmaceuticals development division). The following questions are about your R&D organization(s). 3 Q6: How many R&D organizations does your company have? Number of R&D organizations (Number of R&D organizations overseas) 3) “R&D organization” refers to an organization (e.g., department, division) in which R&D personnel mainly conduct R&D. Even if the name of the organization does not include “research” or “development,” an organization that conducts R&D activities is regarded as an “R&D organization” for the purpose of this survey. Please provide the number of organizations that in your company’s organization chart can be regarded as “R&D organizations ” The following questions are about your R&D organization(s). * Q6: How many R&D organizations does your company have? 3) “R&D organization” refers to an organization (e.g., department, division) in which R&D personnel mainly conduct R&D. © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5 8 Under 24 y/o From 25 to 34 y/o From 35 to 44 y/o From 45 to 54 y/o 55 y/o and older Total % % % % % 100% 2) “R&D personnel” refers to persons who have at least a bachelor’s degree (excluding junior college) or have equivalent or greater expertise, conduct R&D activities on their own specific topic, and engage in R&D activities for more than half of their working hours. Q4: Which department takes the initiative in hiring R&D personnel? Choose one answer. R&D organization takes the initiative Human resources department takes the initiative Both R&D organization and human resources department take the initiative Other Q5: To what extent are researchers’ own wishes taken into account when transferring R&D personnel? Choose one answer. Fully taken into account To some extent taken into account Not very much taken into account Not taken into account at all The following questions are about your R&D organization(s). 3 Q6: How many R&D organizations does your company have? Number of R&D organizations (Number of R&D organizations overseas) 3) “R&D organization” refers to an organization (e.g., department, division) in which R&D personnel mainly conduct R&D. Even if the name of the organization does not include “research” or “development,” an organization that conducts R&D activities is regarded as an “R&D organization” for the purpose of this survey. Please provide the number of organizations that in your company’s organization chart can be regarded as “R&D organizations.” Q7: Among the R&D organizations in Q6, does your company have one or more R&D organizations that are (a) highly independent of business units and/or (b) directly controlled by business units? Choose all that apply. □ (a) My company has one or more R&D organizations that are highly independent of business units (e.g., central research laboratory, development department). □ (b) My company has one or more R&D organizations that are directly controlled by business units (e.g., pharmaceuticals development division). The following questions are about your company’s R&D personnel. 2 The following questions are about your company s R&D personnel. Q3: Please provide the approximate age composition of your R&D personnel (percentage shares). © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5 8 (a) R&D organizations that are highly independent of business units (b) R&D organizations that are directly controlled by business units Total % % 100% 2. R&D project management Q11: Please specify the approximate number of R&D projects in progress. Number of R&D projects in progress Projects 4) “R&D project” refers to R&D activities for which the R&D personnel involved, a budget, a deadline, etc., have been set in order to achieve specific research outcomes. Q12: Please provide the approximate share of R&D projects that have continuously been ongoing for more than three years. Percentage of R&D projects that have continuously been ongoing for more than three years % The following questions are about the number of R&D projects. 4 Q9: Please provide the approximate R&D expenditure shares of R&D organizations falling under (a) and (b) in Q7, respectively. (a) R&D organizations that are highly independent of business units (b) R&D organizations that are directly controlled by business units Total % % 100% Q9: Please provide the approximate R&D expenditure shares of R&D organizations falling under (a) and (b) in Q7, respectively. (a) R&D organizations that are highly independent of business units (b) R&D organizations that are directly controlled by business units Total % % 100% ease provide the approximate R&D expenditure shares of R&D organizations falling under (a) and (b) in Q7, respectively. Q10: Please provide the approximate shares of R&D personnel belonging to R&D organizations falling under (a) and (b) in Q7, respectively. (a) R&D organizations that are highly independent of business units (b) R&D organizations that are directly controlled by business units Total % % 100% 2. R&D project management Q11: Please specify the approximate number of R&D projects in progress. Number of R&D projects in progress Projects 4) “R&D project” refers to R&D activities for which the R&D personnel involved, a budget, a deadline, etc., have been set in order to achieve specific research outcomes. Q12: Please provide the approximate share of R&D projects that have continuously been ongoing for more than three years. Percentage of R&D projects that have continuously been ongoing for more than three years % The following questions are about the number of R&D projects. 4 Q10: Please provide the approximate shares of R&D personnel belonging to R&D organizations falling under (a) and (b) in Q respectively. © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5 8 Even if the name of the organization does not include “research” or “development,” an organization that conducts R&D activities is regarded as an “R&D organization” for the purpose of this survey. Please provide the number of organizations that in your company’s organization chart can be regarded as “R&D organizations.” Q7: Among the R&D organizations in Q6, does your company have one or more R&D organizations that are (a) highly independent of business units and/or (b) directly controlled by business units? Choose all that apply. □ (a) My company has one or more R&D organizations that are highly independent of business units (e.g., central research laboratory, development department). □ (b) My company has one or more R&D organizations that are directly controlled by business units (e.g., pharmaceuticals development division). Q7: Among the R&D organizations in Q6, does your company have one or more R&D organizations that are (a) highly independent of business units and/or (b) directly controlled by business units? Choose all that apply. □ (a) My company has one or more R&D organizations that are highly independent of business units (e.g., central research laboratory, development department). □ (b) My company has one or more R&D organizations that are directly controlled by business units (e.g., pharmaceuticals development division). Appendix 88 Only answer Q8 to Q10 and (b) in Q7; otherwise, proceed to Q11. Q8: Specify the number of R&D organizations falling under (a) and (b) in Q7, respectively. (a) R&D organizations that are highly independent of business units (b) R&D organizations that are directly controlled by business units if you chose both (a) Only answer Q8 to Q10 and (b) in Q7; otherwise, proceed to Q11. Q8: Specify the number of R&D organizations falling under (a) and (b) in Q7, respectively. (a) R&D organizations that are highly independent of business units (b) R&D organizations that are directly controlled by business units if you chose both (a) Only answer Q8 to Q10 and (b) in Q7; otherwise, proceed to Q11. if you chose both (a) Q9: Please provide the approximate R&D expenditure shares of R&D organizations falling under (a) and (b) in Q7, respectively. (a) R&D organizations that are highly independent of business units (b) R&D organizations that are directly controlled by business units Total % % 100% Q10: Please provide the approximate shares of R&D personnel belonging to R&D organizations falling under (a) and (b) in Q7, respectively. © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5 8 (a) R&D organizations that are highly independent of business units (b) R&D organizations that are directly controlled by business units Total % % 100% 2. R&D project management 2. R&D project management The following questions are about the number of R&D projects. 4 Q11: Please specify the approximate number of R&D projects in progress. Projects Q12: Please provide the approximate share of R&D projects that have continuously been ongoing for more than three years. 89 Appendix Q13: Do you have any R&D projects that have been terminated or suspended within the past three years? Choose one answer. Q14: Please provide the approximate share of projects where R&D organizations have the authority to decide whether to terminate/suspend or continue the project. Moreover, if there is a pre-determined upper limit to the R&D expenditure up to which R&D organizations have the authority to decide whether to terminate/suspend or continue the project, please check “Pre-determined” and enter the approximate amount. Percentage of projects where the R&D organization has the authority to decide to terminate/suspend or continue the project Upper limit to R&D expenditure up to which R&D organization has authority to decide whether to terminate/suspend or continue the project. If there is a pre-determined limit, please enter the approximate amount. Pre-determined Not pre-determined ○ million yen % Q14: Please provide the approximate share of projects where R&D organizations have the authority to decide whether to terminate/suspend or continue the project. Moreover, if there is a pre-determined upper limit to the R&D expenditure up to which R&D organizations have the authority to decide whether to terminate/suspend or continue the project, please check “Pre-determined” and enter the approximate amount. Q14: Please provide the approximate share of projects where R&D organizations have the authority to decide whether to terminate/suspend or continue the project. Moreover, if there is a pre-determined upper limit to the R&D expenditure up to which R&D organizations have the authority to decide whether to terminate/suspend or continue the project, please check “Pre-determined” and enter the approximate amount. Percentage of projects where the R&D organization has the authority to decide to terminate/suspend or continue the project Upper limit to R&D expenditure up to which R&D organization has authority to decide whether to terminate/suspend or continue the project. If there is a pre-determined limit, please enter the approximate amount. Pre-determined Not pre-determined ○ million yen % The following questions are about your R&D project management. © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5 8 Whether intermediate goals were achieved Fully taken into account To some extent taken into account Not very much taken into account Not taken into account at all (a) Initial stages (e.g., idea/basic research) (b) Late stages (e.g., preparation for launching new goods/services) Q18: To what extent do you take into account whether intermediate goals (milestones) were achieved when assessing whether to terminate/suspend or continue the R&D project? Choose one answer for (a) initial stages and (b) late stages, respectively. Only answer Q19 if you chose “Yes” in Q16; otherwise, proceed to Q21. Q19: Do you provide feedback on the interim evaluation results to the R&D personnel in charge of the project? Choose one answer. Yes No proceed to Q19 Only answer Q20 if you chose “Yes” in Q19; otherwise, proceed to Q21. Only answer Q20 if you chose “Yes” in Q19; otherwise, proceed to Q21. Only answer Q20 if you chose “Yes” in Q19; otherwise, proceed to Q21. Q20: Do you incorporate the following items (a)–(c) when providing feedback on the interim evaluation results? Choose all applicable answers for each item. If you do not incorporate them, choose “Not incorporated.” Initial stages (e.g., ideas/basic research) Late stages (e.g., preparation for launching new goods/services) Not incorporated (a) Opinions from other research teams in the same or other R&D organizations □ □ □ (b) Opinions from non-R&D organizations (business units and head office) within your company □ □ □ (c) Opinions (including informal ones) from external experts outside your company □ □ □ Q20: Do you incorporate the following items (a)–(c) when providing feedback on the interim evaluation results? Choose all applicable answers for each item. If you do not incorporate them, choose “Not incorporated.” Q20: Do you incorporate the following items (a)–(c) when providing feedback on the interim evaluation results? Choose all applicable answers for each item. If you do not incorporate them, choose “Not incorporated.” The following questions are about your company’s preferences with regard to R&D projects and corporate culture. Note: Please answer Q21–Q24 based on your own judgement. Q21: Suppose there is an R&D project that is expected to generate gross sales of 100 million yen immediately if it is successful but gross sales of 0 yen if it fails. Assume that the probability that the project is successful is 10% (hence, the probability of failure is 90%). How much would you invest in the project? © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5 8 Q15: Please specify the average number of years from the commencement of an R&D project to the achievement of final results. Average number of years from the commencement of an R&D project to the achievement of final results Year(s) The following questions are about your R&D project management. Q16: Do you implement staged project management5 for your R&D projects? Choose one answer. If you do, specify the average number of stages. Yes Number of stages No Proceed to Q21 5) “Staged project management” refers to a method of R&D project management in which the project proceeds in consecutive multiple stages (phases). Staged project management is accompanied by an interim evaluation, which determines whether the project should be terminated/suspended/continued and reviews the project schedule. Q16: Do you implement staged project management5 for your R&D projects? Choose one answer. If you do, specify the average number of stages. Yes Number of stages No Proceed to Q21 5) “Staged project management” refers to a method of R&D project management in which the project proceeds in consecutive multiple stages (phases). Staged project management is accompanied by an interim evaluation, which determines whether the project should be terminated/suspended/continued and reviews the project schedule. 5) “Staged project management” refers to a method of R&D project management in which the project proceeds in consecutive multiple stages (phases). Staged project management is accompanied by an interim evaluation, which determines whether the project should be terminated/suspended/continued and reviews the project schedule. Only answer Q17 if you chose “Yes” in Q16; otherwise, proceed to Q21. Q17: Do you set intermediate goals (“milestones”) for the interim evaluation of projects? Choose one answer. Yes No proceed to Q19 Appendix 90 Only answer Q18 if you chose “Yes” in Q17; otherwise, proceed to Q19. Only answer Q18 if you chose “Yes” in Q17; otherwise, proceed to Q19. Only answer Q18 if you chose “Yes” in Q17; otherwise, proceed to Q19. Q18: To what extent do you take into account whether intermediate goals (milestones) were achieved when assessing whether terminate/suspend or continue the R&D project? Choose one answer for (a) initial stages and (b) late stages, respectively. © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5 8 Please enter the approximate amount. million yen 91 Q23: Do you think your R&D organization is taking appropriate risks in R&D projects to achieve its goals? Choose one answe Q24: Choose up to three words that describe your company’s corporate culture. □(a) Teamwork □(b) Leadership □(c) Customer first □(d) Market impact □(e) Creativity □(f) Profitability □(g) Rule-based decision making □(h) Bottom-up approach □(i) Other ( ) 3. Evaluation of R&D personnel 3. Evaluation of R&D personnel Appendix Q22: Suppose the following two R&D projects: ●Project 1: The expected cumulative profit (net present value) is large, but the project is expected make losses for several years after the launch of the product. ●Project 1: The expected cumulative profit (net present value) is large, but the project is expected make losses for several years after the launch of the product. ●Project 2: The expected cumulative profit (net present value) is small, but the project is expected to generate stable profits after the launch of the product. Assume all other conditions (e.g., initial investment costs, the probability of success of the project, project duration, etc.) are the same for both Projects 1 and 2. Which project do you think your company would choose? Choose one answer. Project 1 Project 2 Do not know ●Project 1: The expected cumulative profit (net present value) is large, but the project is expected make losses for several years after the launch of the product. ●Project 2: The expected cumulative profit (net present value) is small, but the project is expected to generate stable profits after the launch of the product. Assume all other conditions (e.g., initial investment costs, the probability of success of the project, project duration, etc.) are the same for both Projects 1 and 2. Which project do you think your company would choose? Choose one answer. Project 1 Project 2 Do not know Q23: Do you think your R&D organization is taking appropriate risks in R&D projects to achieve its goals? Choose one answer. Does not take much risk Takes appropriate level of risk Takes too much risk Do not know 3. Evaluation of R&D personnel 3. Evaluation of R&D personnel The following questions are about the evaluation of R&D personnel in your company. Q25: What salary scheme do you employ for R&D personnel? Choose all that apply. □ [a] A specific salary scheme for R&D personnel that differs from that for other employees □ [b] Starting salary varies depending on the educational background □ [c] Salary scheme based on performance-based pay □ [d] R&D personnel can choose from among various salary schemes □ [e] None of the salary schemes from [a]–[d] is employed. The following questions are about the evaluation of R&D personnel in your company. Q25: What salary scheme do you employ for R&D personnel? Choose all that apply. Q25: What salary scheme do you employ for R&D personnel? Choose all that apply. Appendix 92 Q26: Assume an R&D employee in their early 30s is being evaluated. Approximately what weights would be put on their ability and their performance in their evaluation?6 Q26: Assume an R&D employee in their early 30s is being evaluated. Approximately what weights would be put on their ability and their performance in their evaluation?6 Q26: Assume an R&D employee in their early 30s is being evaluated. Approximately what weights would be put on their ability and their performance in their evaluation?6 Ability Performance Total % % 100% * 6) “Ability” refers to the abilities demonstrated in performing the job. “Performance” refers to the level of achievement met in performing the job. Q26: Assume an R&D employee in their early 30s is being evaluated. Approximately what weights would be put on their ability and their performance in their evaluation?6 Ability Performance Total % % 100% 6) “Ability” refers to the abilities demonstrated in performing the job. “Performance” refers to the level of achievement met in performing the job. Q27: Do you employ the following criteria for the evaluation of R&D personnel? Choose all that apply. □ [a] Research papers and conference presentations □ [b] Patent applications/registrations □ [c] Commercialization (Launch) of new products □ [d] Amount of sales generated by new products to which the R&D employee contributed □ [e] R&D progress, including compliance with schedule □ [f] Acquisition of qualifications/degrees □ [g] None of the criteria from [a]–[f] are employed Q27: Do you employ the following criteria for the evaluation of R&D personnel? Choose all that apply. Q28: Do you employ the following incentive schemes for R&D personnel? 3. Evaluation of R&D personnel Choose all items that apply. □ [a] In-house research presentations □ [b] Dispatch to university and/or support for studying abroad □ [c] Open recruitment for R&D projects □ [d] Awards for outstanding R&D results □ [e] Rewards based on the number of patent applications □ [f] Rewards based on the amount of profits from inventions and patents (invention reward schemes) □ [g] None of the schemes from [a]–[f] are employed : Do you employ the following incentive schemes for R&D personnel? Choose all items that apply. Q29: Did any of the directors on the board (e.g., chairperson, president, vice president) belong to an R&D organization in the past? Choose one answer. Q29: Did any of the directors on the board (e.g., chairperson, president, vice president) belong to an R&D organization in the past? Choose one answer. 4. R&D results (This information will be used for sending you the survey results and for inquiries regarding your responses, if any.) Name Department Job title Number of years with the company No more than 5 years From 6 to 10 years From 11 to 15 years From 16 to 20 years More than 21 years Telephone number E-mail address Thank you for participating in our survey. Only answer Q32 if you chose “Yes” in Q31. Q32 concerns the novelty of, and turnover (revenue) from product innovations introduced in the market. Q32: Did you introduce the following types of product innovations in the market from FY2016 to FY2018? Choose all types of product innovations that apply and provide the percentage of total sales that such product innovations generated in FY2018. Also specify the approximate sales share of all products other than (X) and (Y) in FY2018. Introduced product innovations in the market (Over the 3 years from FY2016 to FY2018) Percentage of total sales in FY2018 (X) New or significantly improved goods/services that no competitor was offering (new goods and services in the market) □ % (Y) New or improved goods/services that were almost the same as or very similar to ones already offered by competitors (new goods and services for your company only) □ % (Z) All other goods/services other than (X) and (Y) above (including goods/services that remained unchanged or were only marginally modified and the resale of products purchased from other companies) (Z) = 100 − [(X) + (Y)] % Total sales in FY2018 (X) + (Y) + (Z) 1 0 0 % Only answer Q32 if you chose “Yes” in Q31. Q32 concerns the novelty of, and turnover (revenue) from product innovations introduced in the market. This is the end of the questions. Please write about yourself. (This information will be used for sending you the survey results and for inquiries regarding your responses, if any.) Name Department Job title Number of years with the company No more than 5 years From 6 to 10 years From 11 to 15 years From 16 to 20 years More than 21 years Telephone number E-mail address Thank you for participating in our survey. This is the end of the questions. Please write about yourself. 4. R&D results 4. R&D results The following questions are about the R&D results of your company. Q30: Did you make process innovations (introduce new or improved production processes and/or delivery methods, etc.)7 during the three years from FY2016 to FY2018? Choose one answer. 4. R&D results The following questions are about the R&D results of your company. Q30: Did you make process innovations (introduce new or improved production processes and/or delivery methods, etc.)7 during the three years from FY2016 to FY2018? Choose one answer. 93 Appendix Q31: Did you make product innovations (introduce new or improved goods or services in the market)8 during the three years from FY2016 to FY2018? Choose one answer. Q31: Did you make product innovations (introduce new or improved goods or services in the market)8 during the three years from FY2016 to FY2018? Choose one answer. Yes ○No Only answer Q32 if you chose “Yes” in Q31. Q32 concerns the novelty of, and turnover (revenue) from product innovations introduced in the market. Q32: Did you introduce the following types of product innovations in the market from FY2016 to FY2018? Choose all types of product innovations that apply and provide the percentage of total sales that such product innovations generated in FY2018. Also specify the approximate sales share of all products other than (X) and (Y) in FY2018. Introduced product innovations in the market (Over the 3 years from FY2016 to FY2018) Percentage of total sales in FY2018 (X) New or significantly improved goods/services that no competitor was offering (new goods and services in the market) □ % (Y) New or improved goods/services that were almost the same as or very similar to ones already offered by competitors (new goods and services for your company only) □ % (Z) All other goods/services other than (X) and (Y) above (including goods/services that remained unchanged or were only marginally modified and the resale of products purchased from other companies) (Z) = 100 − [(X) + (Y)] % Total sales in FY2018 (X) + (Y) + (Z) 1 0 0 % This is the end of the questions. Please write about yourself. Appendix Appendix Research and Development “Research and development (R&D)” refers to systematic research and creative efforts in science and technology undertaken for the acquisition of new knowledge on mate- rials, functions, and natural phenomena, and for new applications of the existing store of knowledge. R&D includes not only academic research but also activities related to the development of new products, the improvement of existing prod- ucts, and the development and technical improvement of products or production processes. For the purpose of this survey, R&D does not include activities for sales or management-related purposes. 2) R&D Personnel “R&D personnel” refers to individuals holding at least a bachelor’s degree (or having equivalent or greater expertise) and engaged in R&D activities in their area of exper- tise for more than half of their working hours. For the purpose of this survey, “R&D personnel” does not include the following persons: persons who mainly assist R&D personnel, persons who are engaged in technical services related to R&D activities under the guidance and supervision of R&D personnel, and persons who are engaged in clerical work, administration, accounting, etc. 1) R&D Expenditure The term “R&D expenditure” refers to the total amount of expenses on R&D, irre- spective of whether such expenses are funded internally (e.g., through retained earn- ings) or externally (e.g., through grants from the government). R&D expenditure includes expenditures spent within and outside your company. R&D expenditure spent within your company includes labor costs; expenditures on raw materials, tangible ﬁxed assets, intangible ﬁxed assets, and consumable supplies such as books; lease fees, etc. R&D expenditure spent outside your company refers to payments to outside vendors, irrespective of whether such payments are in the form of money in trust, subsides, allocations, etc. 4. R&D results (This information will be used for sending you the survey results and for inquiries regarding your responses, if any.) Name Department Job title Number of years with the company No more than 5 years From 6 to 10 years From 11 to 15 years From 16 to 20 years More than 21 years Telephone number E-mail address Thank you for participating in our survey. 94 3) R&D Organizations The term “R&D organizations” refers to organizations in which R&D personnel mainly conduct R&D. For the purpose of this survey, organizations that perform R&D activities are regarded as “R&D organizations” even if their name does not include the words “Research” or “Development.” For the number of R&D organizations in Q6, please answer with the number of R&D organizations in your company’s organization chart. For example, given the organization chart below, the company has four R&D organizations, which are boxed in bold. They consist of three R&D organizations that are highly independent of business units (i.e., the Development Department, the Central Research Laboratory, and the North American Research Laboratory under the “R&D Unit”) corresponding 95 95 Appendix 95 Fig. 1 Example organization chart Fig. 1 Example organization chart to Q7(a) and one R&D organization that is directly controlled by a business unit (i.e., the Pharmaceuticals Development Division under the “Pharmaceuticals Business Unit”) corresponding to Q7(b). Please do not include other R&D entities that are not listed in your company’s organization chart. For example, the hypothetical company in Fig. 1 also has a cosmetics development section/team under the “Cosmetics Division,” but the section/team is not shown in the company organization chart and therefore is not counted as an R&D organization. 4) R&D Projects The term “R&D projects” refers to projects that are conducted to accomplish speciﬁc research outcomes by a pre-determined deadline and with a designated budget and R&D personnel. Speciﬁc research outcomes include not only commercial- ized goods/services such as “low-power superconducting network devices” but also the development and improvement of production processes such as “an efﬁciency improvement in the manufacturing process by x%.” 7) Process Innovation “Process innovation” refers to new or signiﬁcantly improved production processes, methods of providing services and of delivering products or support activities. This includes signiﬁcant improvements in techniques, equipment, and/or software. Process innovation is deﬁned in this survey as something new to your company; your company does not have to be the ﬁrst to introduce this process. It does not matter whether the innovation was developed by your company or by other companies. 5) Staged Project Management “Staged project management” refers to the management of R&D projects in consec- utive stages, such as “ideation and concept development,” “preliminary assessment of commercialization,” “development,” “testing and validation,” and “production and marketing.” Staged project management also entails a phase-based interim evalua- tion that affects the decision about whether the project is continued, suspended, or abandoned, as well as a revision of the schedule. 96 Appendix 6) Ability and Performance in Employee Evaluation “Ability” in the evaluation of R&D employees refers to a person’s potential ability to perform their duties during the evaluation period. This includes “willingness and attitude (e.g., cooperativeness, discipline),” “cognitive ability (e.g., ability to assess and plan),” and “interpersonal skills (e.g., leadership, ability to arbitrate).” “Performance” in the evaluation of R&D employees refers to a person’s results achieved in performing their duties during the evaluation period. Usually, the results to be evaluated are set at the beginning of the evaluation period as “achievement goals,” taking account of the importance and difﬁculty of the person’s duties. For example, the results to be evaluated for R&D personnel include “patent applications,” “prototype products,” “commercialization,” etc. 8) Product Innovation “Product innovation” refers to new or signiﬁcantly improved goods or services with respect to their technical speciﬁcations, components and materials, software in the product, user friendliness, or other functional characteristics. This includes new combinations of existing technologies or technology upgrades of existing goods or services. Changes only in aesthetic characteristics or the resale of products invented and/or produced by other companies are not included. Product innovation in this survey is deﬁned as something new to your company; the good or service does not have to be new to the market. It does not matter whether your company or other companies developed the product innovation. References Acemoglu D, Aghion P, Lelarge C, Van Reenen J, Zilibotti F (2007) Technology, information, and the decentralization of the ﬁrm. Quart J Econ 122(4):1759–1799 Aghion P, Tirole J (1994) The management of innovation. Quart J Econ 109(4):1185–1209 Aghion P, Tirole J (1997) Formal and real authority in organizations. J Polit Econ 105(1):1–29 Andries P, Hünermund P (2020). Firm-level effects of staged investments in innovation: the moderating role of resource availability. Res Policy 49(7):103994 Argyres NS, Silverman BS (2004) R&D, organization structure, and the development of corporate technological knowledge. Strateg Manag J 25(8–9):929–958 Argyres N, Rios L, Silverman BS (2020) Organizational change and the dynamics of innovation formal R&D structure and intraﬁrm inventor networks. Strateg Manag J 41(11):2015–2049 Arora A, Belenzon S, Rios L (2011) The organization of R&D in American corporations: th determinants and consequences of decentralization. NBER Working Paper No. 17013 Arora A, Belenzon S, Rios L (2014) Make, buy, organize: the interplay between research, externa knowledge, and ﬁrm structure. Strateg Manag J 35(3):317–337 Azoulay P, Zivin JS, Manso G (2011) Incentives and creativity: evidence from the academic life sciences. RAND J Econ 42(3):527–554 Azoulay P, Lerner J (2012) Technological innovation and organizations. In: Gibbons R, Roberts J (eds) The handbook of organizational economics. Princeton, NJ: Princeton University Press. Azoulay P, Lerner J (2012) Technological innovation and organizations. In: Gibbons R, Roberts J ( d ) Th h db k f i i l i P i NJ P i U i i P g y Bénabou R, Tirole J (2003) Intrinsic and extrinsic motivation. Rev Econ Stud 70(3):489–520 Bloom N, Van Reenen J (2007) Measuring and explaining management practices across ﬁrms and countries. Quart J Econ 122(4):1351–1408 Bloom N, Van Reenen J (2010) Why do management practices differ across ﬁrms and countries? Econ Perspect 24(1):203–224 Bloom N, Brynjolfsson E, Foster L, Jarmin R, Patnaik M, Saporta-Ekstein I, Van Reenen J (2019) What drives differences in management practices? Am Econ Rev 109(5):1648–1683 Cameron KS, Quinn RE, DeGraff J, Thakor AV (2014) Competing values leadership, 2nd edn. Edward Elgar Publishing, Cheltenham, UK Carson R, Graff Zivin JS, Louviere J, Sadoff S, Shrader Jr JG (2020) The risk of caution: evidence from an R&D experiment. NBER Working Paper No. 26847 Cooper RG (1988) The new product process: a decision guide for management. J Mark Manag 3(3):238–255 Cooper RG (2017) Winning at new products: creating value through innovation, 5th edn. © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5 97 © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5 References Basic Books, New York © The Author(s) 2022 S. Haneda and A. Ono, R&D Management Practices and Innovation: Evidence from a Firm Survey, SpringerBriefs in Economics, https://doi.org/10.1007/978-981-16-9797-5 97 98 References Cornelli F, Yosha O (2003) Stage ﬁnancing and the role of convertible securities. Rev Econ Stud 70(1):1–32 Cramer JS, Hartog J, Jonker N, Van Praag CM (2002) Low risk aversion encourages the choice fo entrepreneurship: an empirical test of a truism. J Econ Behav Organ 48(1):29–36 Dahiya S, Ray K (2012) Staged investments in entrepreneurial ﬁnancing. J Corp Finan 18(5):1193– 1216 Dewett T (2007) Linking intrinsic motivation, risk taking, and employee creativity in an R&D environment. R&D Manag 37(3):197–208 g Ederer F, Manso G (2013) Is pay for performance detrimental to innovation? Manage Sci 59(7):1496–1513 Fiordelisi F, Ricci O (2014) Corporate culture and CEO turnover. J Corp Finan 28:66–82 Foss NJ, Laursen K (2005) Performance pay, delegation and multitasking under uncer innovativeness: an empirical investigation. J Econ Behav Organ 58(2):246–276 Gibbons R, Roberts J (eds) (2012) The handbook of organizational economics. Princeton University Press, Princeton, NJ Gompers PA (1995) Optimal investment, monitoring, and the staging of venture capital. Journal of Finance 50(5):1461–1489 Gompers PA, Gornall W, Kaplan SN, Strebulaev IA (2020) How do venture capitalists make decisions? J Financ Econ 135(1):169–190 Graham JR, Grennan J, Harvey CR, Rajgopal S (2021) Corporate culture: evidence from the ﬁeld. In: 27th Annual Conference on Financial Economics and Accounting Paper, Columbia Business School Research Paper No. 16-49, Duke I&E Research Paper No. 2016-33. Available at SSRN: https://ssrn.com/abstract=2805602 Guiso L, Sapienza P, Zingales L (2015) The value of corporate culture. J Financ Econ 117(1):60–76 Hanaeda H, Sasaki T, Suzuki K (2020). Corporate culture in Japanese ﬁrms: a survey analysis (in Japanese). In: Hanaeda H, Serita T, Peng X, Sasaki T, Suzuki K, Sasaki T. Nihon no Corporate Finance: Survey data niyoru bunseki (Corporate ﬁnance in Japan: Analyses based on survey data), HAKUTO-SHOBO Publishing Haneda S, Ito K (2018) Organizational and human resource management and innovation: which management practices are linked to product and/or process innovation? Res Policy 47(1):194–208 Hartmann M, Hassan A (2006) Application of real options analysis for pharmaceutical R&D project valuation: empirical results from a survey. Res Policy 35(3):343–354 Hartmann M, Hassan A (2006) Application of real options analysis for pha valuation: empirical results from a survey. Res Policy 35(3):343–354 Hellman T, Thiele V (2011) Incentives and innovation: a multitasking approach. References Am Econ Microecon 3(1):78–128 ( ) Holmström B (1989) Agency costs and innovation. J Econ Behav Organ 12(3):305–327 g y g Holmström B, Milgrom P (1991) Multitask principal-agent analyses: incentive contracts, asset ownership, and job design. J Law Econ Organ 7(Special Issue):24–52. p j g g p Hullova D, Simms CD, Trott P, Laczko P (2019) Critical capabilities for effective management of complementarity between product and process innovation: cases from the food and drink industry. Res Policy 48(1):339–354 y Kaplan SN, Strömberg P (2003) Financial contracting theory meets the real world: an empirica analysis of venture capital contracts. Rev Econ Stud 70(2):281–315 Kastl J, Martimort D, Piccolo S (2013) Delegation, ownership concentration and R&D spending evidence from Italy. J Ind Econ 61(1):84–107 y Kreps DM (1997) Intrinsic motivation and extrinsic motivation. Am Econ Rev 87(2):359–364 Krieger J, Li D, Papanikolaou D (2022) Missing novelty in drug development. Rev Financ Stud 35(2):636–679 Laursen KL, Foss NJ (2003) New human resource management practices, complementarities and the impact on innovation performance. Camb J Econ 27:243–263 Manso G (2011) Motivating innovation. J Finance 66(5):1823–1860 Mao Y, Tian X, Yu X (2014) Unleashing innovation. Mimeo March JG (1991) Exploration and exploitation in organizationa March JG (1991) Exploration and exploitation in organizational learning. Organ Sci 2(1):71–87 99 References Onishi K (2013) The effects of compensation plans for employee inventions on R&D productivity new evidence from Japanese panel data. Res Policy 42(2):367–378 Onishi K, Owan H, Nagaoka S (2021) How do inventors respond to ﬁnancial incentives? Evidence from the unanticipated court decision on employees’ inventions in Japan. J Law Econ 64(2):301– 339 Ono A, Haneda S, Ikeda Y, Inui T (2020) The current status of R&D management practices and innovations in Japan (in Japanese). NISTEP Discussion Paper No. 189 O’Reilly C, Chatman JA (1996) Culture as social control: corporations, cults, and commitment. In: Staw BM, Cummings LL (eds) Research in organizational behavior, vol 18. JAI Press, Greenwich, CT, pp 157–200 pp Prendergast C (1999) The provision of incentives in ﬁrms. J Econ Lit 37(1):7–63 Prendergast C (2002) The tenuous trade-off between risk and incentives. J Polit Econ 110(5):1071– 1102 Prendergast C (2011) What have we learnt about pay for performance? Economic and Social Review 42(2):113–134 Sahlman WA (1990) The structure and governance of venture-capital organizations. J Financ Econ 27(2):473–521 Sauermann H, Cohen WM (2010) What makes them tick? Employee motives and ﬁrm innovation. References Manage Sci 56(12):2134–2153 Smolnik T, Bergmann T (2020) Structuring and managing the new product development process: review on the evolution of the Stage-Gate process. J Bus Chem 17(1):41–57 Teece DJ (1996) Firm organization, industrial structure, and technological innovation. J Econ Behav Organ 31(2):193–224 Thakor AJ (2016) Corporate culture in banking. FRBNY Economic Policy Review Issue, August, 5–16 Tian X (2011) The causes and consequences of venture capital stage ﬁnancing. J Financ Econ 101(1):132–159 Tian X, Wang TY (2014) Tolerance for failure and corporate innovation. Rev Financ Stud 27(1):211– 255 Wang S, Zhou H (2004) Staged ﬁnancing in venture capital: moral hazard and risks. J Corp Finan 10(1):131–155 C (2019) i l i l i i h d ﬁ i k i 100 2 3 282 Yung C (2019) Entrepreneurial manipulation with staged ﬁnancing. J Bank Finance 100:273–282 Yung C (2019) Entrepreneurial manipulation with staged ﬁnancing. J Bank Finance 100:273–282
https://openalex.org/W4361305001	https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0282479&type=printable	English	null	COVID-19: Psychological distress, fear, and coping strategies among community members across the United Arab Emirates	PloS one	2,023	cc-by	8,367	Objectives We aimed to identify the factors associated with psychological distress, fear, and coping amongst community members across the UAE. Editor: Muhammad Arsyad Subu, School of Health Binawan: Universitas Binawan, INDONESIA COVID-19: Psychological distress, fear, and coping strategies among community members across the United Arab Emirates Rania Al DweikID1, Muhammad Aziz RahmanID2,3, Fathima Mohammed AhamedID4, Heba Ramada1, Yousef Al Sheble5, Sondos ElTaher6, Wendy Cross2, Deena Elsori7 1 College of Health Sciences, Abu Dhabi University, Abu Dhabi, United Arab Emirates, 2 Institute of Health and Wellbeing, Federation University Australia, Berwick, VIC, Australia, 3 Faculty of Public Health, Universitas Airlangga, Surabaya, Indonesia, 4 Al Ain Fertility Center, Abu Dhabi, United Arab Emirates, 5 School of Dentistry, Queen’s University Belfast, Belfast, United Kingdom, 6 Epidemiology Department, American University of Beirut, Beirut, Lebanon, 7 Faculty of Resilience, Rabdan Academy, Abu Dhabi, United Arab Emirates a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 raldw034@gmail.com OPEN ACCESS The COVID-19 pandemic impacted the psychosocial well-being of the United Arab Emirates [UAE] population like other communities internationally. Citation: Al Dweik R, Rahman MA, Ahamed FM, Ramada H, Al Sheble Y, ElTaher S, et al. (2023) COVID-19: Psychological distress, fear, and coping strategies among community members across the United Arab Emirates. PLoS ONE 18(3): e0282479. https://doi.org/10.1371/journal.pone.0282479 PLOS ONE PLOS ONE RESEARCH ARTICLE Methods Received: March 7, 2022 Accepted: February 15, 2023 Published: March 29, 2023 We conducted a cross-sectional online survey across the UAE during November 2020. Adults aged 18 years, living in the UAE who were able to respond to an online question- naire in English or Arabic were considered eligible to participate in the study. We used stan- dard validated tools to measure psychological distress, fear and coping. Kessler Psychological Distress Scale [K10] was used to assess psychological distress, Fear of COVID-19 Scale [FCV-19S] was used to assess the level of fear, and Brief Resilient Coping Scale [BRCS] was used to assess the coping strategies. Copyright: © 2023 Al Dweik et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper. Data Availability Statement: All relevant data are within the paper. A total of 417 individuals participated in this study with a mean age of 29 [± 10.7] years. More than half of the participants experienced high to very high levels of psychological dis- tress [55%] and a quarter experienced high levels of fear of COVID-19 [23.3%] with almost a third of them [36.2%] having low resilient coping. About 37.4% of the participants had work-related mental health impacts and 32.4% were perceived to have moderate to a great deal of distress due to a change of employment status during the pandemic. One in ten Competing interests: No competing interests. * raldw034@gmail.com Conclusion Community members in the UAE are at a higher risk of psychosocial distress and fear during the COVID-19 pandemic. Thus, healthcare providers and policymakers would need to be more alert to provide specific mental health support strategies for their wellbeing. Competing interests: No competing interests. Abbreviations: AOR, Adjusted Odds Ratio; BRCS, Brief Resilient Coping Scale; Cis, Confidence Interval; FCV-19S, Fear of COVID-19 Scale; K-10, Kessler Psychological Distress Scale; ORs, Odds Ratios; PCR, Polymerase Chain Reaction. PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 1 / 17 PLOS ONE COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates participants [9.4%] reported increased smoking. Increased smoking [AOR 8.66, 95% CIs 1.08–69.1,], increased alcohol drinking [AOR 2.39, 95% CIs 1.05–5.47] and higher levels of fear of COVID-19 [AOR 2.93, 95% CIs 1.83–4.67] were associated with moderate to very high levels of psychological distress. Being female [AOR 1.82, p = 0.030], having a pre- existing mental health condition [AOR 9.88, 95% p = 0.027], engaging in high-risk behaviors such as increased smoking [AOR 21.14, p = 0.003], increased alcohol drinking [AOR 1.48, p = 0.359] in the previous four weeks, and higher levels of fear of COVID-19 [AOR 4.18, p <0.001] were associated with moderate to very high levels of psychological distress. Also, being a smoker [AOR, 0.840, p = 0.011], and having a high level of fear [AOR 0.372, p = 0.001] were found to be associated with low resilient coping. PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 Introduction Since the emergence of COVID-19 it has surged exponentially across the world, with more than 276 million confirmed cases and over 5.3 million deaths as of 27th December 2021 [1]. The United States of America has reported the highest number of cases and deaths due to COVID-19 followed by India, Brazil, France, Russian Federation, the United Kingdom, Tur- key, Italy and Spain. In Bangladesh, the first three cases of COVID-19 were detected on 8th March 2020 [24]. The United Arab Emirates [UAE] was one of the first responders to the alerts of the COVID-19 outbreak; UAE’s government, in cooperation with the National Crisis and Emergency Management Authority [NCEMA], responded before WHO considered it a pan- demic [2]. In the UAE, the first cases of positive coronavirus were reported on January 23, 2020. To date 15 August 2022], there have been over 1,005,543 confirmed cases and a total of 2,339 deaths in the UAE [National Emergency Crisis and Disaster-UAE, 2022]. As of 15th August 2022, a total of 24,922,054 vaccine doses have been administered in the UAE [3]. g The COVID-19 pandemic brought unexpected challenges and the outbreak lead to com- promised physical and mental health [3–5]. Research on past infectious disease outbreaks, such as severe acute respiratory syndrome [SARS], swine flu, and influenza revealed a wide range of psychosocial impacts at individual, community, and international levels. However, there was limited information on the mental health impact of the COVID-19 pandemic on the UAE population [6]. More recently, studies investigating the psychological impacts of COVID-19 in China, Spain, Italy, India, and the UK reported moderate to severe stress, gener- alized anxiety, insomnia, and depression; which were associated with lockdowns, social isola- tion, changes in daily habits, public fear and feelings of uncertainty [7–11]. Factors such as longer quarantine duration, fear of infection and deaths, increase in anxiety, post-traumatic stress and depression, feelings of helplessness, guilt, panic, financial loss, deaths of family members and insufficient supplies of protective equipment and tests, affected public health globally [12–15]. Regulatory laws and sudden changes turned people’s lives upside down, leav- ing them in shock. Available literature suggested that a psychiatric epidemic was coexisting with the COVID-19 pandemic, increasing the strain of mental health issues [16]. Study design and settings This cross-sectional study was conducted across the United Arab Emirates [UAE] as part of a global study [22]. We followed the same method applied across the 17 countries in the global study. Participants were recruited using online platforms [social media and online groups] like what’s app, Facebook, Twitter, and Instagram] and were invited to answer an online survey. Introduction PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 2 / 17 PLOS ONE COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates Economic changes indirectly added psychological distress making people more prone to mental disorders [17]. Social isolation and virtual interactions have boosted the number of people suffering from mental illnesses such as anxiety and depression in the community set- tings of UAE [18]. The UAE has taken unprecedented precautionary measures including complete lockdowns against COVID-19 to control its spread and ensure the well-being of individuals [19]. The UAE government launched several initiatives to combat COVID-19 including surveillance and contact tracing, containment, mental health support, mass testing and treatment, govern- ment economic support, national vaccination program. Since March 2020, wearing a face mask was mandatory, and mobile apps like the "Al Hosn" app were a must to download to aid in quarantine and track the cases. Screening programs were routinely performed to help detect new COVID-19 cases as soon as possible for an immediate medical intervention to keep their lives [2]. Compared to developed countries, the UAE has the highest number of tests con- ducted per 1000 people [2]. To limit the spread of COVID-19, the UAE Government intro- duced physical distancing rules including restrictions on social gatherings, strict lockdowns [with a pause on all social, cultural, and sporting activities], quarantine, border closures, sus- pension of flights, and mandating public respiratory hygiene measures [20]. The UAE’s National Program for Happiness and Wellbeing launched an online national campaign to support the UAE’s community to overcome the possible psychological impact of the pandemic COVID-19 and to provide safe and confidential mental support to individuals who are impacted directly and indirectly by COVID-19. In addition, introduced another ini- tiative called ‘Hayat’ [Arabic word for life] is a psychological and moral support program to help federal government employees deal with the circumstances and anxiety associated with COVID-19. However, evidence-based evaluations on psychological distress, fear, and coping strategies were relatively scarce [21]. As the pandemic is still ongoing in the UAE, although it is under control, its psychological impact will not wane with its eradication. With limited research regarding the psychological distress and fear due to COVID-19 in UAE, this study aimed to investigate the factors leading to the psychological distress, the level of fear of COVID-19, and coping strategies among its population. Study population Adults aged 18 years, living in the UAE who were able to respond to an online questionnaire in English or Arabic were considered eligible to participate in the study. Participants included general community members, healthcare professionals, frontline or essential workers includ- ing COVID-19 hospitalized and none-hospitalized patients. Healthcare professionals and frontline or essential workers were individuals who self-identified themselves in the survey as being a doctor, nurse, or an individual in contact with patients/clients as part of their profes- sional responsibilities during the pandemic period. Patients were defined as individuals who utilized a health care service in the last six months at the time of data collection. PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 3 / 17 PLOS ONE COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates Study tool The structured survey questionnaire was derived from a prior study conducted globally across 17 countries including Australia, Malaysia, and Bangladesh by the same research group [23–25]. Psy- chological distress was measured using the Kessler Psychological Distress Scale [K-10] [26], fear was measured using the Fear of COVID-19 Scale [FCV-19S] [27], and coping was measured using Brief Resilient Coping Scale [BRCS] [28]. The reliability of those tools in the English version was examined in the Australian study, and it was found that they worked reliably for both migrants and non-migrants [26]. We also checked the differences in reliability of the tools in English and Arabic for our study; no differ- ence was observed. Having the multicultural population including migrants and non-migrants in the UAE, the study tools were also deemed suitable. We followed standard translation and back-translation process to create the Arabic version of the questionnaire, which was also pilot tested with necessary modifications [to clarify the languages further based on the feedback from the pilot study participants] were done as needed [24]. We used the same study tool which was used as part of the primary study in Australia [19] global study across 17 countries [24]. Study participants had the option to choose either English or Arabic version of the ques- tionnaire while responding to this study. Sampling The sample size was calculated using OpenEpi and the Snowball sampling technique was used for collecting the data. Assuming the population size is 10,069,862, with a 50% prevalence of stress globally [no national existing data available on the prevalence of stress in the UAE], at 95% confidence intervals and 80% power, the estimated minimum sample size was 384. Data collection An online survey was created using Google Forms. The survey link was made available in English with a separate link for participants who wished to take the survey in Arabic. Depend- ing on the survey link selected by the participant, the plain language information statement [PLIS] and a consent form showed on the first screen in English or Arabic. Participants who gave their full consent and met the survey’s requirements were allowed to proceed. Different social media channels, online community networks were used to distribute an invitation with the online survey URL and QR code, and the link to the survey was sent to staff, and students via the email databases of participating universities/hospitals, WhatsApp, and SMS text communications. The survey was open to anyone who had access to the survey URL, and no incentives were offered for taking part in the study. Participants had the option to review their responses before submitting the survey. All survey responses were anonymous. Data analyses The database was downloaded from the Google platform and IBM SPSS Statistics 28.0 statisti- cal software was used for data analyses. Descriptive statistics, such as means and standard devi- ations [SD], were generated for continuous variables; frequencies and percentages were generated for categorical variables. Fear of COVID-19 [based on the FCV-19S scoring] was categorized into low [score 7–21] and high [score 22–35], psychological distress [based on the K-10 scoring] was categorized into low [score 10–15] and moderate to very high [score 16– 50], and coping [based on the BRCS scoring] was categorized into low [score 4–13] and medium to high [score 14–20]. Logistic regression analyses were used to examine the associa- tion between variables. Multivariate analyses were carried out by adjusting age, gender, PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 4 / 17 PLOS ONE COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates smoking, alcohol intake, living status, place of birth, country, education, employment status, employment stress, healthcare worker, perceived financial impact, contact with COVID-19 case, experience due to COVID-19 [isolation/quarantine], and self-identification as a patient [who attended a clinic within last six months for any reasons]. Odds ratios [ORs] and 95% confidence intervals [CIs] were used to present the data. A cut-off of p<0.05 was considered statistically significant. For multivariate analyses, adjusted ORs [AOR] with 95% CIs were reported. Ethics approval and consent to participate Ethics approval was obtained from Abu Dhabi University Institutional Research Board [Ref: CoHS– 20-10-00024]. The survey was voluntary, and it was explained in plain English and Arabic, so that participants could make an informed decision about whether or not to partici- pate in the study. The survey was anonymous and the data were handled only by the investiga- tors listed in the study. Results A total of 417 individuals participated in this study and more than two-thirds of them [291, 70%] responded in English. The majority of them lived with family members [381, 91.4%] and belonged to the age group of 18–29 years [254, 60.9%]. Less than half of the participants [182, 43.6%] were born in the UAE and more than half of the study population [240, 57.6%] com- pleted at least a bachelor’s degree. Less than half of participants [156, 37.4%] lose/ reduced working hours/or were afraid to lose their jobs during this pandemic, and 32.4% [135] were perceived to have moderate to a great deal of distress due to a change of employment status due to pandemic. COVID-19 did not impact the financial situation of more than half of the participants [224,53.7%]. Only 13.4% [56] identified themselves as frontline or essential service workers, including doctors [1.7%], nurses [1%], and other healthcare professionals [6.2%] [Table 1]. More than half of the study participants [302,72.4%] did not report any comorbidity. However, 8.2% [34] reported having pre-existing psychiatric or mental health issues and 19.4% [81] reported other comorbidities conditions. More than half of the participants were never smoked or drink alcohol [71.9%, and 89% respectively]. However, of those who smoked, 9.4% [39] reported increased smoking in the previous four weeks [Table 1]. 22.7% [93] of the participants provided direct care to family members or patients with a known or suspected case of COVID-19. Almost half of the participants [205, 48.8%] used health services in the previous four weeks. 22.7% [93] of the participants provided direct care to family members or patients with a known or suspected case of COVID-19. Almost half of the participants [205, 48.8%] used health services in the previous four weeks. More than two-thirds of the participants [328, 78.8%] experienced moderate to very high levels of psychological distress, 23.3% [97] had high levels of fear of COVID-19, and more than half of the participants [262, 62.8%] had medium to high resilient coping [S1-S3 Tables in S1 File]. PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 Psychological distress Characteristics Total n [%] Total study participants 417 Age groups 405 18–29 years 254 [60.9] 30–59 years 149 [35.7] 60+ years 2 [0.48] Gender 414 Male 120 [28.8] Female 294 [70.5] Living Status 412 Live without family members [on your own/shared house/others] 31 [7.4] Live with family members [partner and/or children] 381 [91.4] Born in UAE 415 No 233 [55.9] Yes 182 [43.6] Completed level of education 414 Grade 1–12 125 [30.5] Trade/Certificate/Diploma 49 [11.8] Bachelor and above 240 [57.6] Current employment condition 399 Jobs affected by COVID-19 [lost job/working hours reduced/afraid of job loss] 156 [37.4] Have an income source [employed/Government benefits] 243 [58.3] Perceived distress due to change of employment status 367 A little to none 232 [55.6] Moderate to a great deal 135 [32.4] Self-identification as frontline or essential service worker 417 Yes 56 [13.4] No 361 [86.6] COVID-19 Impacted financial situation 417 Yes, impacted positively 36 [8.6] Yes, impacted negatively 157 [37.6] No impact 224 [53.7] Co-morbidities 417 No comorbid conditions 302[72.4] Psychiatric/ Mental health issues 34 [8.2] Other comorbid conditions 81 [19.4] Smoking 417 Never smoker 300[71.9] Ex-smoker 40 [9.6] Occasional smoker 16 [3.8] Less than weekly, but at least once a month 10 [2.4] Less than daily, but at least once a week 11 [2.6] Daily 40 [9.6] Increase smoking over the last 6 months 77 Yes 39 [9.4] No 38 [9.1] Current alcohol drinking 371 Yes 39 [9.4] (Continued) PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 6 / 17 PLOS ONE COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates Table 1. Psychological distress (Continued) Characteristics Total n [%] No 371 [89.0] Increased alcohol drinking over the last 6 months 39 Yes 16 [3.8] No 23 [5.5] Provide care to a family member/patient with a known/suspected case of COVID-19 410 No 285 [68.3] Unsure 32 [7.7] Yes 93[22.7] Experience related to COVID-19 pandemic [multiple responses possible] 405 No known exposure to COVID-19 336 [80.6] Tested positive for COVID-19 16 [3.8] Tested negative for COVID -19 but self-isolating 42 [10.1] Recent overseas travel history and was in quarantine 11 [2.6] Health Service use in the last 4 weeks 405 Yes 205 [48.8] No 200 [49.2] Type of health service users to overcome COVID-19 related stress in the last 4 weeks 30 Consult a GP 12 [2.9] Consulted a psychologist 3 [0.7] Consulted a psychiatrist 1 [0.2] Used mental health resources available in media 9 [2.2] Used mental health support services 1 [0.2] Used combination of services 4 [1.0] https://doi.org/10.1371/journal.pone.0282479.t001 adjusted, being female [AOR 1.82, p = 0.030], having a pre-existing mental health condition [AOR 9.88, 95% p = 0.027], and engaging in high-risk behaviors such as increased smoking [AOR 21.14, p = 0.003] in the previous four weeks, and higher levels of fear of COVID-19 [AOR 4.18, p <0.001] were associated with moderate to very high levels of psychological dis- tress. However, living with family members [AOR 0.487, p = 0.184] and having medium to high resilient coping [AOR 0.386, p = 0.002] were associated with low levels of psychological distress [Table 2]. Fear of COVID-19 The levels of fear of COVID-19 were associated with several factors following the adjustment of potential confounders. Higher levels of fear from COVID -19 were associated with factors such as: having a Bachelors and Masters level of education or above [AOR 4.14, p = 0.011], per- ceived distress due to changes in employment status [AOR 2.06, p = 0.009], impacted financial situation due to COVID-19 [AOR 1.38, p = 0.013], having pre-existing mental health issues or existing psychiatric problems [AOR 2.37, p = 0.030], and having moderate to a very high level of psychological distress [AOR 4.09, p = 0.001] [Table 3]. Psychological distress The results in Table 2 show the factors associated with moderate to very high psychological distress among the participants. Younger participants, females, participants living family members, having graduated, employed, those with pre-existing mental health conditions, did not identify themselves as frontline workers, increased smoking and alcohol drinking in the last four weeks, self-isolating, used health service in general, or used health service to overcome COVID-19 related stress in the last four weeks, and those with a higher level of fear of COVID-19 were more likely to develop moderate to very high levels of psychological distress compared to their counterparts [Table 4]. However, when potential confounders were PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 5 / 17 PLOS ONE COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates Table 1. Characteristic of the study population in UAE [November-2020]. PLOS ONE Table 2. Factors related to high psychological distress among the study population [based on K10 score] in UAE [November-2020]. Coping strategies The multivariate analyses showed that there was no significant difference between those with medium to high resilient coping when compared to those with low resilient coping based on the BRCS scale. On the other hand, being a smoker [AOR, 0.840, p = 0.011], and having a high PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 7 / 17 COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates PLOS ONE Table 2. (Continued) Characteristics Moderate to Very High [Score 16–50], n [%] Low [Score 10–15], n [%] p OR 95% Cl p AOR 95%Cl Yes 30 9 0.766 0.887 0.404–1.947 0.345 1.509 0.642–3.546 Increases alcohol drinking over the last 6 months 30 10 No 15 9 1 1 Yes 15 1 0.064 8.00 0.888–72.099 0.044 1.487 0.637–88.545 Provide care to a family member/patient with a known/ suspected case of COVID-19 322 88 No 218 67 1 1 Yes 104 21 0.194 1.193 0.914–1.558 0.543 0.680 0.197–2.351 Experience related COVID-19 pandemic 316 89 No known exposure to COVID-19 257 79 1 1 I had been tested negative for COVID-19 but self-isolating 36 6 0.192 1.821 0.740–4.482 0.212 1.8.23 0.710–4.684 I had been tested positive for COVID-19 16 0 0.323 0.531 0.152–1.862 0.125 0.286 0.058–1.413 Self-identification as a patient/Use of health service in the last 4 weeks 319 86 No 151 49 1 1 Yes 168 37 0.114 1.473 0.912–2.381 0.091 1.588 0.929–2.713 Healthcare service use in the last 6 months 162 36 Telehealth consultation/Used helpline 14 1 1 1 Visited a healthcare provider in person 129 30 0.263 3.256 0.412–25.732 0.906 1.149 0.114–11.544 Used both services 19 5 0.820 0.884 0.305–2.556 0.330 0.518 0.138–1.945 Level of fear of COVID-19 [FCV-19S categories] 328 89 Low [score 7–21] 240 80 1 1 High [score 22–35] 88 9 0.002 3.259 1.569–6.771 0.001 4.148 1.766–9.741 Level of coping [BRCS categories] 324 89 Low resilient coping [score 4–13] 130 21 1 1 Medium to high resilient coping [score 14–20] 194 68 0.005 0.461 0.269–0.789 0.002 0.383 0.208–0.704 Healthcare service use to overcome COVID-19 related stress in the last 4 weeks 316 89 No 281 88 1 1 Yes 35 1 0.019 10.961 1.48–81.16 0.016 12.134 1.596–92.249 Adjusted for: age, gender, living status, residence location, born in UAE, education and employment. level of fear [AOR 0.372, p = 0.001] were found to be associated with low resilient coping [Table 4]. level of fear [AOR 0.372, p = 0.001] were found to be associated with low resilient coping [Table 4]. level of fear [AOR 0.372, p = 0.001] were found to be associated with low resilient coping [Table 4]. PLOS ONE Characteristics Moderate to Very High [Score 16–50], n [%] Low [Score 10–15], n [%] p OR 95% Cl p AOR 95%Cl Total study participants 319 86 Age groups 319 86 18–29 years 222 32 1 1 30–59 years 96 53 <0.001 0.261 0.158–0.430 <0.001 0.305 0.152–0.614 60+ years 1 1 0.175 0.144 0.009–2.362 0.284 0.200 0.011–3.799 Gender 326 88 Male 82 38 1 1 Female 244 50 0.001 2.260 1.385–3.696 0.030 1.828 1.060–3.152 Living Status 324 88 Live without family members [on your own/shared house/ others] 26 5 1 1 Live with family members [partner and/or children] 298 83 0.462 0.690 0.257–1.854 0.242 0.529 0.182–1.536 Born in UAE 327 88 No 175 58 1 1 Yes 152 30 0.039 1.679 1.027–2.745 0.889 1.046 0.558–1.959 Completed level of education 326 88 Grade 1 to 6 Primary 1 0 1 1 Grade 7 to 12 106 18 0.388 2.944 0.254–34.186 0.332 3.374 0.289–39.348 Certificate/Diploma 46 3 0.135 7.667 0.531–110.650 0.077 11.308 0.770–166.070 Degree [Bachelor/Masters or above] 173 67 0.836 1.291 0.115–14.475 0.416 2.767 0.238–32.179 Current employment condition 328 89 Unemployed/Home duties 16 2 1 1 Jobs affected by COVID-19 [lost job/working hours reduced/afraid of job loss] 110 46 0.240 0.399 0.086–1.852 0.637 0.594 0.068–5.161 Have an income source [employed/Government benefits] 202 41 0.801 0.821 0.177–3.808 0.645 0.609 0.074–5.000 Perceived distress due to change of employment status 293 74 A little to None 169 63 1 1 Moderate to a great deal 124 11 <0.001 4.202 2.127–8.304 <0.001 4.379 2.018–9.582 Self-identification as frontline or essential service worker 328 89 No 286 75 1 1 Yes 42 14 0.474 0.787 0.408–1.516 0.863 1.081 0.500–2.338 COVID-19 impacted the financial situation 328 89 No 160 64 1 1 Yes 168 25 <0.001 4.086 2.196–7.600 <0.001 5.099 1.966–13.224 Co-morbidities 328 89 No 232 70 1 1 Psychiatric/Mental health problem 33 1 0.025 9.957 1.338–74.112 0.038 8.673 1.129–66.623 Other co-morbidities 63 18 0.856 1.056 0.587–1.901 0.038 2.094 1.042–4.207 Smoking 328 89 Never smoker 239 61 1 1 Ever smoker 89 28 0.873 0.988 0.856–1.141 0.732 0.906 0.515–1.594 Increased smoking in the last 4 weeks 62 15 No 25 13 1 1 Yes 37 2 0.005 9.620 1.996–46.370 0.003 61.006 4.063–915.988 Current alcohol drinking [last 4 weeks] 323 87 No 293 78 1 1 (Continued) y g , , p g g g y PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 8 / 17 PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 8 COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates Discussion This study is one of the first studies in identifying factors related to psychological distress, fear, and coping among the UAE residents during the COVID-19 pandemic. During an outbreak of infection, previous research has demonstrated a wide range of psy- chosocial effects on people at the individual and communal levels [29]. The results of this study showed that around 78.7% of respondents experienced moderate to very high levels of psychological distress which was higher than the findings from similar studies on psychologi- cal distress during the pandemic in Australia [62.6%], Bangladesh [69%], Malaysia [62.1%] and Saudi Arabia [72%] and China [53.8%] [28, 38–42] [24] PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 9 / 17 COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates PLOS ONE PLOS ONE Table 3. Factors associated with levels of fear of COVID-19 among the study population [based on FCV-19S score] in UAE [November-2020]. PLOS ONE Table 3. (Continued) Characteristics High [Score 22– 35], n [%] Low [Score 7–21], n [%] p OR 95% Cl p AOR 95%Cl No 63 222 1 1 Yes 32 93 0.747 1.040 0.819–1.321 0.201 0.369 0.080–1.701 Experience related COVID-19 pandemic 94 311 No known exposure to COVID-19 86 250 1 1 I had been tested positive for COVID-19 1 15 0.100 0.181 0.024–1.385 0.584 1.280 0.529–3.095 I had been tested negative for COVID-19 but self-isolating 7 35 0.207 0.579 0.248–1.352 0.209 1.476 0.805–2.706 I had recent overseas travel history and was in self-quarantine 0 11 0.999 0.201 0.369 0.080–1.701 Self-identification as a patient/Use of health service in the last 4 weeks 94 311 No 42 158 1 1 Yes 52 153 0.299 1.279 0.804–2.032 0.404 1.232 0.754–2.014 Healthcare service use in the last 4 weeks 49 149 Telehealth consultation/Used helpline 3 12 1 1 Visited a healthcare provider in person 22 126 0.945 0.955 0.255–3.580 0.287 0.316 0.038–2.635 Used both services 13 11 <0.001 4.512 1.854–10.985 0.014 3.383 1.284–8.914 Level of psychological distress [K10 categories] 97 320 Low [score 10–15] 9 80 1 1 Moderate to Very high [score 16–50] 88 240 0.002 3.259 1.569–6.771 0.001 4.093 1.754–9.548 Level of coping [BRCS categories] 93 320 Low resilient coping [score 4–13] 37 114 1 1 Medium to high resilient coping [score 14–20] 56 206 0.464 0.838 0.521–1.346 0.568 0.862 0.518–1.434 Healthcare service use to overcome COVID-19 related stress in the last 6 months 92 313 No 76 293 1 1 Yes 16 20 0.002 3.084 1.525–6.237 <0.001 3.582 1.707–7.516 https://doi.org/10.1371/journal.pone.0282479.t003 According to a previous study, patients with a history of smoking are more likely to develop severe COVID disease and be admitted to critical care [30]. This study found a significant association between increased alcohol consumption and smoking in the previous six months and higher psychological distress. However, we are not able to establish temporal relationship due to the nature of cross-sectional study design. Evidence suggests that practicing social distancing poses a challenge to the mental health of all family members of multiple generations living together because there was decreased social support through family interaction and cultural activities, which led to feelings of loneliness, negative emotions, and psychological distress [31]. In addition, members experienced fear of infections for themselves or their family, anxiety, fear of death, and other mental health con- cerns [32]. PLOS ONE Characteristics High [Score 22– 35], n [%] Low [Score 7–21], n [%] p OR 95% Cl p AOR 95%Cl Total study participants 94 311 Age groups 94 311 18–29 years 63 191 1 1 30–59 years 30 119 0.284 0.764 0.468–1.249 0.996 0.986 0.508–1.913 60+ years 1 1 0.435 3.032 0.187–49.182 0.333 4.175 0.231–75.333 Gender 95 319 Male 24 96 1 1 Female 71 223 0.363 1.274 0.756–2.144 0.470 1.229 0.702–2.150 Living Status 96 316 Live without family members [on your own/shared house/others] 8 23 1 1 Live with family members [partner and/or children] 88 293 0.732 0.863 0.373–1.998 0.644 0.812 0.335–1.965 Born in UAE 97 318 No 52 181 1 1 Yes 45 137 0.565 1.143 0.724–1.805 0.961 0.987 0.569–1.710 Current employment condition 97 320 Unemployed/Home duties 4 14 0.935 0.946 0.249–3.594 0.621 1.516 0.291–7.899 Jobs affected by COVID-19 [lost job/working hours reduced/afraid of job loss] 32 124 0.758 1.229 0.332–4.551 0.623 1.483 0.308–7.137 Have an income source [employed/Government benefits] 61 182 0.880 1.222 0.091–16.429 0.886 1.270 0.080–20.217 Perceived distress due to change of employment status 80 287 A little to None 41 191 1 1 Moderate to a great deal 39 96 0.013 1.893 1.145–3.127 0.009 2.062 1.203–3.536 Self-identification as frontline or essential service worker 97 320 No 86 275 1 1 Yes 11 45 0.492 0.782 0.387–1.578 0.930 0.967 0.456–2.049 COVID-19 impacted the financial situation 97 320 No 43 181 1 1 Yes 54 139 0.017 1.340 1.055–1.703 0.012 1.912 1.151–3.175 Co-morbidities 97 320 No 64 238 1 1 Psychiatric/Mental health problem 14 20 0.011 2.603 1.246–5.438 0.030 2.375 1.089–5.176 Other co-morbidities 19 62 0.661 1.140 0.636–2.042 0.994 1.002 0.521–1.930 Smoking 97 320 Never smoker 73 227 1 1 Ever smoker 24 93 0.693 0.971 0.841–1.122 0.470 0.814 0.465–1.423 Increased smoking in the last 4 weeks 15 62 No 2 36 1 1 Yes 13 26 0.006 9.00 1.869–43.34 0.031 6.50 1.191–35.486 Current alcohol drinking [last 4 weeks] 94 316 No 85 286 1 1 Yes 9 30 0.981 1.009 0.461–2.209 0.929 1.039 0.451–2.390 Increases alcohol drinking over the last 4 weeks 9 30 No 5 19 1 1 Yes 4 11 0.812 1.200 0.267–5.400 0.194 4.213 0.480–36.978 Provide care to a family member/patient with a known/suspected case of COVID-19 95 315 ar of COVID-19 among the study population [based on FCV-19S score] in UAE [November-2020]. PLOS ONE PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 10 / 17 COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 PLOS ONE Studies demonstrated a significant negative impact of fear of COVID-19 on mental health leading to depression, anxiety, or stress [33]. Fear of COVID-19 has also been found to be associated with decreased life satisfaction with increased mental health challenges in the context of the current pandemic [34–36]. The results of this study were consistent with the results from other countries including the US, Australia, and Bangladesh where females and those with pre-existing mental health conditions reported higher psychiatric distress [23, 24, 37]. There are a variety of reasons for this, including the fact that women disproportionately share the majority of caregiving tasks in both the formal and informal sectors. They are also more frequently the primary PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 11 / 17 COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates PLOS ONE Table 4. (Continued) Characteristics Medium to High [Score 14–20], n [%] Low [Score 4–13], n [%] p OR 95% Cl p AOR 95%Cl No 176 109 1 1 Yes 82 29 0.737 1.041 0.825–1.312 Experience related COVID-19 pandemic 254 147 No known exposure to COVID-19 217 118 1 1 Tested positive for COVID-19 10 6 0.873 0.919 0.326–2.592 0.469 0.646 0.198–2.109 Tested negative for COVID-19 but self-isolating 20 22 0.036 0.501 0.263–0.956 0.024 0.458 0.233–0.901 Recent overseas travel history and was in self-quarantine 10 1 0.106 5.514 0.697–43.605 0.455 2.280 0.262–19.842 Self-identification as a patient [visited a healthcare provider in the last 4 weeks] 257 144 No 122 78 1 1 Yes 135 66 0.199 1.308 0.869–1.969 0.300 1.260 0.813–1.953 Healthcare service use in the last 4 weeks 131 63 Telehealth consultation/Used helpline 7 8 1 1 Visited a healthcare provider in person 108 51 0.105 0.413 0.142–1.202 0.733 0.812 0.245–2.690 Used both services 16 4 0.276 1.899 0.601–5.937 0.130 2.794 0.738–10.585 Level of fear of COVID-19 [K10 categories] 262 151 Low [score 16–15] 68 21 1 1 Moderate to Very High [score 16–50] 194 130 0.005 0.461 0.269–0.789 0.001 0.372 0.202–0.684 Level of fear of COVID-19 [FCV-19S categories] 262 151 Low [score 10–15] 206 114 1 1 High [score 16–50] 56 37 0.464 0.838 0.521–1.346 0.576 0.865 0.519–1.440 Healthcare service use to overcome COVID-19 related stress in the last 4 weeks 257 148 No 230 139 1 1 Yes 27 9 0.136 1.813 0.828–3.968 0.193 1.716 0.761–3.872 https://doi.org/10.1371/journal.pone.0282479.t004 https://doi.org/10.1371/journal.pone.0282479.t004 caregivers in a household, which may exacerbate their anxiety and stress in a pandemic situ- ation [38] Higher levels of fear in this study were significantly associated with distress due to employ- ment status, impact on the financial situation, co-morbidity including psychiatric and mental health problems, and moderate to very high psychological distress. Similar results were reported in both Australian and Bangladesh results [23, 24]. Fear of contracting the infection, quarantine measures for infected individuals, and self-isolation/social distancing for the gen- eral population may have played a major role in influencing their mental health during the critical phases of the pandemic and could explain this association [39, 40]. A study reported that patients affected with COVID-19 had a high level of post-traumatic stress symptoms and a significantly higher level of depressive symptoms [41]. This study found that 62.8% of the participants had high resilient coping. PLOS ONE PLOS ONE Table 4. Factors associated with medium to high resilience coping among the study participants [based on BRCS score] in UAE [November-2020]. PLOS ONE Characteristics Medium to High [Score 14–20], n [%] Low [Score 4–13], n [%] p OR 95% Cl p AOR 95%Cl Total study participants 257 144 Age groups 257 144 18–29 years 149 101 1 1 30–59 years 106 43 0.021 1.671 1.081–2.582 0.275 1.386 0.771–2.490 60+ years 2 0 0.999 Gender 260 150 Male 75 45 1 1 Female 185 105 0.805 1.057 0.681–1.642 0.290 1.297 0.801–2.099 Living Status 257 151 Live without family members [on your own/shared house/others] 23 8 1 1 Live with family members [partner and/or children] 234 143 0.184 0.569 0.248–1.307 0.520 0.751 0.314–1.795 Born in UAE 262 149 No 163 70 1 1 Yes 99 79 0.003 0.538 0.358–0.809 0.090 0.655 0.402–1.068 Current employment condition 262 151 Unemployed/Home duties 16 2 1 1 Jobs affected by COVID-19 [lost job/working hours reduced/afraid of job loss] 106 50 0.184 0.353 0.076–1.639 0.091 0.161 0.019–1.339 Have an income source [employed/Government benefits] 140 99 0.062 0.236 0.052–1.076 0.46 0.120 0.015–0.960 Perceived distress due to change of employment status 227 140 A little to None 145 87 1 1 Moderate to a great deal 82 53 0.738 0.928 0.600–1.435 0.858 1.044 0.648–1.682 Self-identification as frontline or essential service worker 262 151 No 227 130 1 1 Yes 35 21 0.875 0.954 0.533–1.709 0.707 0.882 0.459–1.695 COVID-19 impacted the financial situation 262 151 No 140 80 1 1 Yes 122 71 0.831 1.023 0.827–1.266 0.410 1.216 0.763–1.939 Co-morbidities 262 151 No 197 105 1 1 Psychiatric/Mental health problem 15 15 0.102 0.533 0.251–1.133 0.084 0.494 0.222–1.100 Other co-morbidities 50 31 0.559 0.860 0.518–1.427 0.992 0.997 0.570–1.745 Smoking 262 151 Never smoker 201 95 1 1 Ever smoker 61 56 0.022 0.869 0.770–0.980 0.001 0.432 0.267–0.701 Increased smoking in the last 6 months 36 41 No 25 13 1 1 Yes 16 23 0.031 0.362 0.143–0.913 0.222 0.504 0.168–1.513 Current alcohol drinking [last 4 weeks] 255 151 No 228 139 1 1 Yes 27 12 0.384 1.372 0.673–1.123 0.967 0.984 0.460–2.103 Increases alcohol drinking over the last 6 months 27 12 No 15 9 1 1 Yes 12 3 0.249 2.40 0.530–10.877 0.768 0.731 0.091–5.895 Provide care to a family member/patient with a known/suspected case of COVID-19 258 148 PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 12 / 17 COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 Strengths The use of validated measures to evaluate the elements associated with psychological distress, fear, and coping methods among a large number of UAE residents during the COVID-19 pan- demic and the achievement of the target sample size was considered as strengths of the study. Another strength was achieving the target sample size within all COVID-19 restrictions. PLOS ONE These findings were similar to the results from the global study [57%] and Bangladesh study [57.1%] [22, 33]. This disparity could be explained by people learning from previous success experiences, allow- ing them to cope better [42]. The UAE Ministry of Health introduced psychological and moral support program called Hayat’ [Arabic word for life]to help federal government employees deal with the circumstances and anxiety associated with COVID-19 [23]. 13 / 17 PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 PLOS ONE COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates Limitations Since this was an online survey, the majority of the responses were from the age group 18 to 29 years, implying that they were more active on social media and had more online access. The use of the snowball sampling technique may have resulted in selection bias, and the survey’s self-reporting nature may have resulted in reporting bias. However, due to the restrictions on movement at the time of the pandemic, such sampling was deemed viable. Data were collected in late 2020, so the habituation effects could not be ruled out. Acknowledgments The authors would like to thank the Abu Dhabi University, patients, frontline health and other essential service workers, and general community members who participated in the study. Conclusion The COVID-19 pandemic had an adverse impact on mental health in the UAE community. The UAE government has taken unprecedented precautionary measures including complete lockdowns against COVID-19 to control its spread and ensure the well-being of individuals. The findings of this study showed that people with mental illnesses, females, and frontline workers were at high risk of fear and psychological distress. In general, the COVID-19 pan- demic had a major negative impact on public mental health. Healthcare authorities in collabo- ration with various sectors are recommended to apply for psychological help and design appropriate educational programs to improve the mental health of the public. PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 References 1. Organization WH. COVID-19 weekly epidemiological update, edition 58, 21 September 2021. 2021. 2. Al Hosany F, Ganesan S, Al Memari S, Al Mazrouei S, Ahamed F, Koshy A, et al. Response to COVID- 19 pandemic in the UAE: A public health perspective. Journal of Global Health. 2021;11. https://doi.org/ 10.7189/jogh.11.03050 PMID: 33828834 3. Suliman DM, Nawaz FA, Mohanan P, Modber MAKA, Musa MK, Musa MB, et al. UAE efforts in promot- ing COVID-19 vaccination and building vaccine confidence. Vaccine. 2021; 39[43]:6341–5. https://doi. org/10.1016/j.vaccine.2021.09.015 PMID: 34561138 4. Peeri NC, Shrestha N, Rahman MS, Zaki R, Tan Z, Bibi S, et al. The SARS, MERS and novel coronavi- rus [COVID-19] epidemics, the newest and biggest global health threats: what lessons have we learned? International journal of epidemiology. 2020; 49[3]:717–26. https://doi.org/10.1093/ije/dyaa033 PMID: 32086938 5. Thomas J, Barbato M, Verlinden M, Gaspar C, Moussa M, Ghorayeb J, et al. Psychosocial correlates of depression and anxiety in the United Arab Emirates during the COVID-19 pandemic. Frontiers in Psy- chiatry. 2020; 11:1248. https://doi.org/10.3389/fpsyt.2020.564172 PMID: 33240122 6. Saddik B, Hussein A, Albanna A, Elbarazi I, Al-Shujairi A, Temsah M-H, et al. The psychological impact of the COVID-19 pandemic on adults and children in the United Arab Emirates: a nationwide cross-sec- tional study. BMC psychiatry. 2021; 21[1]:1–18. 7. Huang Y, Zhao N. Generalized anxiety disorder, depressive symptoms and sleep quality during COVID-19 outbreak in China: a web-based cross-sectional survey. Psychiatry research. 2020; 288:112954. https://doi.org/10.1016/j.psychres.2020.112954 PMID: 32325383 8. Wang C, Pan R, Wan X, Tan Y, Xu L, Ho CS, et al. Immediate psychological responses and associated factors during the initial stage of the 2019 coronavirus disease [COVID-19] epidemic among the general population in China. International journal of environmental research and public health. 2020; 17 [5]:1729. https://doi.org/10.3390/ijerph17051729 PMID: 32155789 9. Shevlin M, McBride O, Murphy J, Miller JG, Hartman TK, Levita L, et al. Anxiety, depression, traumatic stress and COVID-19-related anxiety in the UK general population during the COVID-19 pandemic. BJPsych Open. 2020; 6[6]. https://doi.org/10.1192/bjo.2020.109 PMID: 33070797 10. Verma S, Mishra A. Depression, anxiety, and stress and socio-demographic correlates among general Indian public during COVID-19. International Journal of Social Psychiatry. 2020; 66[8]:756–62. https:// doi.org/10.1177/0020764020934508 PMID: 32567466 11. Lakhan R, Agrawal A, Sharma M. Prevalence of depression, anxiety, and stress during COVID-19 pan- demic. Journal of neurosciences in rural practice. 2020. https://doi.org/10.1055/s-0040-1716442 PMID: 33144785 12. Rubin GJ, Wessely S. The psychological effects of quarantining a city. Bmj. 2020;368. Author Contributions Conceptualization: Rania Al Dweik. Data curation: Rania Al Dweik, Heba Ramada, Yousef Al Sheble, Sondos ElTaher. Investigation: Rania Al Dweik. Methodology: Rania Al Dweik, Muhammad Aziz Rahman, Yousef Al Sheble. Methodology: Rania Al Dweik, Muhammad Aziz Rahman, Yousef Al Sheb Methodology: Rania Al Dweik, Muhammad Aziz Rahman, Yousef Al Sheble. Resources: Rania Al Dweik. Resources: Rania Al Dweik. Software: Rania Al Dweik. Software: Rania Al Dweik. Supervision: Rania Al Dweik. Validation: Rania Al Dweik. 14 / 17 14 / 17 PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 PLOS ONE COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates Visualization: Rania Al Dweik. Writing – original draft: Rania Al Dweik, Fathima Mohammed Ahamed. Writing – review & editing: Rania Al Dweik, Muhammad Aziz Rahman, Wendy Cross, Deena Elsori. Writing – review & editing: Rania Al Dweik, Muhammad Aziz Rahman, Wendy Cross, Deena Elsori. References https://doi.org/ 10.1136/bmj.m313 PMID: 31992552 13. Balkhy HH, Abolfotouh MA, Al-Hathlool RH, Al-Jumah MA. Awareness, attitudes, and practices related to the swine influenza pandemic among the Saudi public. BMC infectious diseases. 2010; 10[1]:1–7. 14. Sim K, Chan YH, Chong PN, Chua HC, Soon SW. Psychosocial and coping responses within the com- munity health care setting towards a national outbreak of an infectious disease. Journal of psychoso- matic research. 2010; 68[2]:195–202. https://doi.org/10.1016/j.jpsychores.2009.04.004 PMID: 20105703 15. Leung GM, Lam TH, Ho LM, Ho S, Chan B, Wong I, et al. The impact of community psychological responses on outbreak control for severe acute respiratory syndrome in Hong Kong. Journal of Epide- miology & Community Health. 2003; 57[11]:857–63. https://doi.org/10.1136/jech.57.11.857 PMID: 14600110 16. Hossain MM, Tasnim S, Sultana A, Faizah F, Mazumder H, Zou L, et al. Epidemiology of mental health problems in COVID-19: a review. F1000Research. 2020;9. 17. Frasquilho D, Matos MG, Salonna F, Guerreiro D, Storti CC, Gaspar T, et al. Mental health outcomes in times of economic recession: a systematic literature review. BMC public health. 2015; 16[1]:115. PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 15 / 17 PLOS ONE COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates 18. Hwang T-J, Rabheru K, Peisah C, Reichman W, Ikeda M. Loneliness and social isolation during the COVID-19 pandemic. Int Psychogeriatr. 2020; 32[10]:1217–20. https://doi.org/10.1017/ S1041610220000988 PMID: 32450943 18. Hwang T-J, Rabheru K, Peisah C, Reichman W, Ikeda M. Loneliness and social isolation during the COVID-19 pandemic. Int Psychogeriatr. 2020; 32[10]:1217–20. https://doi.org/10.1017/ S1041610220000988 PMID: 32450943 19. Cheikh Ismail L, Osaili TM, Mohamad MN, Al Marzouqi A, Jarrar AH, Abu Jamous DO, et al. Eating hab- its and lifestyle during COVID-19 lockdown in the United Arab Emirates: a cross-sectional study. Nutri- ents. 2020; 12[11]:3314. https://doi.org/10.3390/nu12113314 PMID: 33137947 20. Bedford J, Enria D, Giesecke J, Heymann DL, Ihekweazu C, Kobinger G, et al. COVID-19: towards con- trolling of a pandemic. The lancet. 2020; 395[10229]:1015–8. https://doi.org/10.1016/S0140-6736(20) 30673-5 PMID: 32197103 21. Furlong Y, Finnie T. Culture counts: the diverse effects of culture and society on mental health amidst COVID-19 outbreak in Australia. Irish journal of psychological medicine. 2020; 37[3]:237–42. https:// doi.org/10.1017/ipm.2020.37 PMID: 32406358 22. Rahman MA, Islam SMS, Tungpunkom P, Sultana F, Alif SM, Banik B, et al. COVID-19: Factors associ- ated with psychological distress, fear, and coping strategies among community members across 17 countries. Globalization and health. 2021; 17[1]:1–19. 23. References Rahman MA, Rahman S, Wazib A, Arafat S, Chowdhury ZZ, Uddin BMM, et al. COVID-19 related psy- chological distress, fear and coping: identification of high-risk groups in Bangladesh. Frontiers in psychi- atry. 2021:1399. https://doi.org/10.3389/fpsyt.2021.718654 PMID: 34484005 24. Rahman MA, Hoque N, Alif SM, Salehin M, Islam SMS, Banik B, et al. Factors associated with psycho- logical distress, fear and coping strategies during the COVID-19 pandemic in Australia. Globalization and Health. 2020; 16[1]:1–15. 25. Bahar Moni AS, Abdullah S, Bin Abdullah MFIL, Kabir MS, Alif SM, Sultana F, et al. Psychological dis- tress, fear and coping among Malaysians during the COVID-19 pandemic. PloS one. 2021; 16[9]: e0257304. https://doi.org/10.1371/journal.pone.0257304 PMID: 34506576 26. Enticott JC, Lin E, Shawyer F, Russell G, Inder B, Patten S, et al. Prevalence of psychological distress: How do Australia and Canada compare? Australian & New Zealand Journal of Psychiatry. 2018; 52 [3]:227–38. https://doi.org/10.1177/0004867417708612 PMID: 28523939 27. Ahorsu DK, Lin C-Y, Imani V, Saffari M, Griffiths MD, Pakpour AH. The fear of COVID-19 scale: devel- opment and initial validation. International journal of mental health and addiction. 2020:1–9. 28. Sinclair VG, Wallston KA. The development and psychometric evaluation of the Brief Resilient Coping Scale. Assessment. 2004; 11[1]:94–101. https://doi.org/10.1177/1073191103258144 PMID: 14994958 29. Hall RC, Hall RC, Chapman MJ. The 1995 Kikwit Ebola outbreak: lessons hospitals and physicians can apply to future viral epidemics. General hospital psychiatry. 2008; 30[5]:446–52. https://doi.org/10. 1016/j.genhosppsych.2008.05.003 PMID: 18774428 30. Guan W-j, Ni Z-y, Hu Y, Liang W-h, Ou C-q, He J-x, et al. Clinical characteristics of coronavirus disease 2019 in China. New England journal of medicine. 2020; 382[18]:1708–20. https://doi.org/10.1056/ NEJMoa2002032 PMID: 32109013 31. Mukhtar S. Pakistanis’ mental health during the COVID-19. Asian Journal of Psychiatry. 2020; 51:102127. https://doi.org/10.1016/j.ajp.2020.102127 PMID: 32344330 32. Rana W, Mukhtar S, Mukhtar S. Mental health of medical workers in Pakistan during the pandemic COVID-19 outbreak. Asian journal of psychiatry. 2020; 51:102080. https://doi.org/10.1016/j.ajp.2020. 102080 PMID: 32283512 33. Yıldırım M, Arslan G, Alkahtani AM. Do fear of COVID-19 and religious coping predict depression, anxi- ety, and stress among the Arab population during health crisis? Death studies. 2021:1–7. https://doi. org/10.1080/07481187.2021.1882617 PMID: 33555985 34. Taylor S, Landry CA, Paluszek MM, Fergus TA, McKay D, Asmundson GJ. COVID stress syndrome: Concept, structure, and correlates. Depression and anxiety. 2020; 37[8]:706–14. https://doi.org/10. 1002/da.23071 PMID: 32627255 35. Satici B, Gocet-Tekin E, Deniz ME, Satici SA. PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 References Adaptation of the Fear of COVID-19 Scale: Its association with psychological distress and life satisfaction in Turkey. International journal of mental health and addiction. 2020:1–9. https://doi.org/10.1007/s11469-020-00294-0 PMID: 32395095 36. Yıldırım M, Arslan G, O¨ zaslan A. Perceived risk and mental health problems among healthcare profes- sionals during COVID-19 pandemic: exploring the mediating effects of resilience and coronavirus fear. International Journal of Mental Health and Addiction. 2020:1–11. https://doi.org/10.1007/s11469-020- 00424-8 PMID: 33223977 37. French MT, Mortensen K, Timming AR. Psychological Distress and Coronavirus Fears During the Initial Phase of the COVID-19 Pandemic in the United States. The Journal of Mental Health Policy and Eco- nomics. 2020; 23[3]:93–100. PMID: 32853158 PLOS ONE \| https://doi.org/10.1371/journal.pone.0282479 March 29, 2023 16 / 17 PLOS ONE COVID-19: Psychological distress, fear, and coping strategies among community across the United Arab Emirates 38. Langer A, Meleis A, Knaul FM, Atun R, Aran M, Arreola-Ornelas H, et al. Women and health: the key for sustainable development. The Lancet. 2015; 386[9999]:1165–210. https://doi.org/10.1016/S0140-6736 (15)60497-4 PMID: 26051370 39. Brooks SK, Webster RK, Smith LE, Woodland L, Wessely S, Greenberg N, et al. The psychological impact of quarantine and how to reduce it: rapid review of the evidence. The lancet. 2020; 395 [10227]:912–20. https://doi.org/10.1016/S0140-6736(20)30460-8 PMID: 32112714 40. Yıldırım M, Arslan G. A Moderated Mediation Effect of Stress-Related Growth and Meaning in Life in the Association Between Coronavirus Suffering and Satisfaction With Life: Development of the Stress- Related Growth Measure. Frontiers in psychology. 2021; 12:529. https://doi.org/10.3389/fpsyg.2021. 648236 PMID: 33796058 41. Vindegaard N, Benros ME. COVID-19 pandemic and mental health consequences: Systematic review of the current evidence. Brain, behavior, and immunity. 2020; 89:531–42. https://doi.org/10.1016/j.bbi. 2020.05.048 PMID: 32485289 42. Lazarus RS, Folkman S. Stress, appraisal, and coping: Springer publishing company; 1984. 17 / 17
https://openalex.org/W4285695019	https://hal.inria.fr/hal-03232345/document	English	null	A categorical approach to secure compilation	arXiv (Cornell University)	2,020	cc-by	12,709	To cite this version: Stelios Tsampas, Andreas Nuyts, Dominique Devriese, Frank Piessens. A Categorical Approach to Secure Compilation. 15th International Workshop on Coalgebraic Methods in Computer Science (CMCS), Apr 2020, Dublin, Ireland. pp.155-179, 10.1007/978-3-030-57201-3_9. hal-03232345 A Categorical Approach to Secure Compilation Stelios Tsampas, Andreas Nuyts, Dominique Devriese, Frank Piessens To cite this version: Stelios Tsampas, Andreas Nuyts, Dominique Devriese, Frank Piessens. A Categorical Approach to Secure Compilation. 15th International Workshop on Coalgebraic Methods in Computer Science (CMCS), Apr 2020, Dublin, Ireland. pp.155-179, 10.1007/978-3-030-57201-3_9. hal-03232345 A Categorical Approach to Secure Compilation Stelios Tsampas, Andreas Nuyts, Dominique Devriese, Frank Piessens Distributed under a Creative Commons Attribution 4.0 International License A categorical approach to secure compilation Stelios Tsampas1 , Andreas Nuyts1 , Dominique Devriese (Corresponding)2 , and Frank Piessens1 1 KU Leuven, Leuven, Belgium name.surname@cs.kuleuven.be 2 Vrije Universiteit Brussel, Brussels, Belgium dominique.devriese@vub.be 1 KU Leuven, Leuven, Belgium name.surname@cs.kuleuven.be 2 Vrije Universiteit Brussel, Brussels, Belgium dominique.devriese@vub.be Abstract. We introduce a novel approach to secure compilation based on maps of distributive laws. We demonstrate through four examples that the coherence criterion for maps of distributive laws can potentially be a viable alternative for compiler security instead of full abstraction, which is the preservation and reflection of contextual equivalence. To that end, we also make use of the well-behavedness properties of distributive laws to construct a categorical argument for the contextual connotations of bisimilarity. Keywords: Secure compilation · Distributive laws · Structural Oper- ational Semantics HAL Id: hal-03232345 https://inria.hal.science/hal-03232345v1 Submitted on 21 May 2021 L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. Distributed under a Creative Commons Attribution 4.0 International License Tsampas et al. 2 In this paper we introduce a novel, categorical approach to secure compi- lation. The approach has two main components: the elegant representation of Structural Operational Semantics (SOS) [38] using category-theoretic distribu- tive laws [48] 3 and also maps of distributive laws [40, 50, 27] as secure compilers that preserve bisimilarity. Our method aims to be unifying, in that there is a general, shared formalism for operational semantics, and simplifying, in that the formal criterion for compiler security, the coherence criterion for maps of distributive laws, is straightforward and relatively easy to prove. The starting point of our contributions is an abstract proof on how coalge- braic bisimilarity under distributive laws holds contextual meaning in a manner similar to contextual equivalence (Section 4.3). We argue that this justifies the use of the coherence criterion for testing compiler security as long as bisimilarity adequately captures the underlying threat model. We then demonstrate the ef- fectiveness of our approach by appeal to four examples of compiler (in)security. The examples model classic, non-trivial problems in secure compilation: – An example of an extra processor register in the target language that conveys additional information about computations (Section 5). – An example of an extra processor register in the target language that conveys additional information about computations (Section 5). ( ) – A datatype mismatch between the type of variable (Section 6).l – The introduction of illicit control flow in the target language (Section 7) l e of incorrect local state encapsulation (Sectio – A case of incorrect local state encapsulation (Section 8). For each of these examples we present an insecure compiler that fails the co- herence criterion, then introduce security primitives in the target language and construct a secure compiler that respects it. We also examine how bisimilarity can be both a blessing and a curse as its strictness and rigidity sometimes lead to relatively contrived solutions. Finally, in Section 9, we discuss related work and point out potential avenues for further development of the underlying theory. On the structure and style of the paper This work is presented mainly in the style of programming language semantics but its ideas are deeply rooted in category theory. We follow an “on-demand” approach when it comes to im- portant categorical concepts: we begin the first example by introducing the base language used throughout the paper, While, and gradually present distributive laws when required. Tsampas et al. From the second example in Section 6 and on, we relax the categorical notation and mostly remain within the style of PL semantics. 1 Introduction As a field, secure compilation is the study of compilers that formally preserve abstractions across languages. Its roots can be tracked back to the seminal obser- vation made by Abadi [1], namely that compilers which do not protect high-level abstractions against low-level contexts might introduce security vulnerabilities. But it was the advent of secure architectures like the Intel SGX [15] and an ever- increasing need for computer security that motivated researchers to eventually work on formally proving compiler security. The most prominent [16, 18, 49, 35, 37, 32, 45] formal criterion for compiler security is full abstraction: A compiler is fully abstract if it preserves and reflects Morris-style contextual equivalence [31], i.e. indistinguishability under all pro- gram contexts, which are usually defined as programs with a hole. The intuition is that contexts represent the ways an attacker can interact with programs and so full abstraction ensures that such interactions are consistent between languages. Full abstraction is arguably a strong and useful property but it is also notori- ously hard to prove for realistic compilers, mainly due to the inherent challenge of having to reason directly about program contexts [37, 18, 9, 24]. There is thus a need for better formal methods, a view shared in the scientific commu- nity [10, 33]. While recent work has proposed generalizing from full abstraction towards the so-called robust properties [36, 2], the main challenge of quantifying over program contexts remains, which manifests when directly translating target contexts to the source (back-translation). Other techniques, such as trace seman- tics [35] or logical relations [17], require complex correctness and completeness proofs w.r.t. contextual equivalence in order to be applicable. A categorical approach to secure compilation 3 A categorical approach to secure compilation The constructors are respectively literals, a dereference operator var, binary arithmetic operations as well as unary operations. We let S be the set of lists of natural numbers. The role of S is that of a run-time store whose entries are referred by their index on the list using constructor var. We define function eval : S × E →N inductively on the structure of expressions. Definition 1 (Evaluation of expressions in While). eval store (lit n) = n eval store (var l) = get store l eval store (e1 b e2) = (eval store e1) [[b]] (eval store e2) eval store (u e) = [[u]] (eval store e) Programs in While language are generated by the following grammar: ⟨prog⟩::= skip \| N := ⟨expr⟩\| ⟨prog⟩; ⟨prog⟩\| while ⟨expr⟩⟨prog⟩ The operational semantics of our While language are introduced in Figure 1. We are using the notation s, x ⇓s′ to denote that program x, when supplied with s : S, terminates producing store s′. Similarly, s, x →s′, x′ means that program x, supplied with s, evaluates to x′ and produces new store s′. s, skip ⇓s s, l := e ⇓update s l (eval s e) s, p ⇓s′ s, p;q →s′, q s, p →s′, p′ s, p;q →s′, p′;q eval s e = 0 s, while e p →s, skip eval s e ̸= 0 s, while e p →s, p; while e p Fig. 1. Semantics of the While language. Fig. 1. Semantics of the While language. 2.1 Syntax and operational semantics We begin by defining the set of arithmetic expressions. We begin by defining the set of arithmetic expressions. ⟨expr⟩::= lit N \| var N \| ⟨expr⟩⟨bin⟩⟨expr⟩\| ⟨un⟩⟨expr⟩ 3 The authors use the term “Mathematical Operational Semantics”. The term “Bial- gebraic Semantics” is also used in the literature. 3 The authors use the term “Mathematical Operational Semantics”. The term “Bial- gebraic Semantics” is also used in the literature. 4 A copointed endofunctor is an endofunctor F equipped with a natural transformation F =⇒Id. 5i 2.2 While, categorically The categorical representation of operational semantics has various forms of incremental complexity but for our purposes we only need to use the most im- portant one, that of GSOS laws [48]. Definition 2. Given a syntax functor Σ and a behavior functor B, a GSOS law of Σ over B is a natural transformation ρ : Σ(Id×B) =⇒BΣ∗, where (Σ∗, η, µ) is the monad freely generated by Σ. Example 1. Let E be the set of expressions of the While-language. Then the syntax functor Σ : Set →Set for While is given by ΣX = ⊤⊎(N × E) ⊎(X × X) ⊎(E × X) where ⊎denotes a disjoint (tagged) union. The elements could be denoted as skip, l := e, x1; x2 and while e x respectively. The free monad Tsampas et al. 4 Σ∗satisfies Σ∗X ∼= X ⊎ΣΣ∗X, i.e. its elements are programs featuring program variables from X. Since While-programs run in interaction with a store and can terminate, the behavior functor is BX = S →(S×Maybe X), where S is the set of lists of natural numbers and X →Y denotes the exponential object (internal Hom) Y X.i ) The GSOS specification of While determines ρ. A premise s, p →s′, p′ denotes an element (p, b) ∈(Id × B)X where b(s) = (s′, just p′), and a premise s, p ⇓s′ denotes an element (p, b) where b(s) = (s′, nothing). A conclusion s, p →s′, p′ (where p ∈ΣX is further decorated above the line to ¯p ∈Σ(Id × B)X) specifies that ρ(¯p) ∈BΣ∗X sends s to (s′, just p′), whereas a conclusion s, p ⇓s′ specifies that s is sent to (s′, nothing). Concretely, ρX : Σ(X × BX) →BΣ∗X is the function (partially from [47]): skip 7→λ s.(s, nothing) l := e 7→λ s.(update s l (eval s e), nothing) while e (x, f) 7→λ s. { (s, just (x ; while e x)) if eval s e ̸= 0 (s, just (skip)) if eval s e = 0 (x, f) ; (y, g) 7→λ s. { (s′, just (x′ ; y)) if f(s) = (s′, just x′) (s′, just y) if f(s) = (s′, nothing) skip 7→λ s.(s, nothing) l := e 7→λ s.(update s l (eval s e), nothing) while e (x, f) 7→λ s. 5 We write B∞for the cofree comonad over B, which satisfies B∞X ∼= X × BB∞X. F =⇒Id. 2.2 While, categorically Note also that a : ΣA ∼= A and z : Z ∼= BZ are invertible. Example 2. Continuing Example 1, the initial bialgebra A is just the set of all While-programs. Meanwhile, the final bialgebra Z, which has the meaning of the set of behaviors, satisfies Z ∼= (S →S × Maybe Z). In other words, our attacker model is that of an attacker who can count execution steps and moreover, between any two steps, read out and modify the state. In Section 9, we discuss how we hope to consider weaker attackers in the future. 2.2 While, categorically { (s, just (x ; while e x)) if eval s e ̸= 0 (s, just (skip)) if eval s e = 0 (x, f) ; (y, g) 7→λ s. { (s′, just (x′ ; y)) if f(s) = (s′, just x′) (s′, just y) if f(s) = (s′, nothing) It has been shown by Lenisa et al. [28] that there is a one-to-one correspon- dence between GSOS laws of Σ over B and distributive laws of the free monad Σ∗over the cofree copointed endofunctor [28] Id × B.4 Definition 3 (In [26]). A distributive law of a monad (T, η, µ) over a copointed functor (H, ϵ) is a natural transformation λ : TH =⇒HT subject to the following laws: λ ◦η = Hη, ϵ ◦λ = Tϵ and λ ◦µ = Hµ ◦λ ◦Tλ. Given any GSOS law, it is straightforward to obtain the corresponding dis- tributive law via structural induction (In [50], prop. 2.7 and 2.8). By convention, we shall be using the notation ρ for GSOS laws and ρ∗for the equivalent dis- tributive laws unless stated otherwise. A distributive law λ based on a GSOS law ρ gives a category λ-Bialg of λ- bialgebras [48], which are pairs ΣX h−→X k−→BX subject to the pentagonal law k◦h = Bh∗◦ρX ◦Σ[id, k], where h∗is the inductive extension of h. Morphisms in λ-Bialg are arrows X →Y that are both algebra and coalgebra homomorphisms at the same time. The trivial initial B-coalgebra ⊥→B⊥lifts uniquely to the initial λ-bialgebra ΣΣ∗⊥ a−→Σ∗⊥ hλ −−→BΣ∗⊥, while the trivial final Σ-algebra Σ⊤→⊤lifts uniquely to the final λ-bialgebra ΣB∞⊤ gλ −→B∞⊤ z−→BB∞⊤5. Since Σ∗⊥is the set of programs generated by Σ and B∞⊤the set of behaviors cofreely generated by B, the unique bialgebra morphism f : Σ∗⊥→B∞⊤is the interpretation function induced by ρ. A categorical approach to secure compilation 5 Remark 1. We write A for Σ∗⊥and Z for B∞⊤, and refer to hλ : A →BA as the operational model for λ and to gλ : ΣZ →Z as the denotational model [48]. Note also that a : ΣA ∼= A and z : Z ∼= BZ are invertible. Remark 1. We write A for Σ∗⊥and Z for B∞⊤, and refer to hλ : A →BA as the operational model for λ and to gλ : ΣZ →Z as the denotational model [48]. 3 An extra register (Part I) For our threat model, where the attacker can directly observe labels, it makes sense to define contextual equivalence in Whileas: Definition 4. p ∼=q ⇐⇒∀c : C. c JpK ⇓⇐⇒c JqK ⇓ Where C is the set of one-hole contexts, J_K : C× A→Adenotes the plugging function and we write p ⇓when p eventually terminates. Contextual equivalence for While is defined analogously. It is easy to show that the simple “embedding” compiler from While to Whileis not fully abstract by examining terms a ≜while (var[0]) (0 := 0) and b ≜while (var[0] ∗2) (0 := 0), for which a ∼= b but a≇b. A context c ≜(obs 1 _); while (var[1] −1) skip will log the result of the while condition in aand bin var[1] and then either diverge or terminate depending on the value of var[1]. An initial var[0] value of 1 will cause c JaK to terminate but c JbK to diverge. Securely extending WhileTo deter malicious contexts from exploiting the extra information, we introduce sandboxing primitives to help hide it. We add an additional constructor in While, ⟨progr⟩+, and the following inference rules to form the secure version Whileof While. s, p ⇓v s′ s, p+ ⇓0 s′ s, p →v s′, p′ s, p+ →0 s′, p′+ s, p ⇓v s′ s, p+ ⇓0 s′ s, p →v s′, p′ s, p+ →0 s′, p′+ We now consider the compiler from While to Whilewhich, along with the obvious embedding, wraps the the translated terms in sandboxes. This looks to be effective as programs a and b are now contextually equivalent and the extra information is adequately hidden. We will show that this compiler is indeed a map of distributive laws between While and Whilebut to do so we need a brief introduction on the underlying theory. 3 An extra register (Part I) Let us consider the scenario where a malicious party can observe more informa- tion about the execution state of a program, either because information is being leaked to the environment or the programs are run by a more powerful machine. A typical example is the presence of an extra flags register that logs the result of a computation [34, 8, 35]. This is the intuition behind the augmented version of While with additional observational capabilities, While.f  The main difference is in the behavior so the notation for transitions has to slightly change. The two main transition types, s, x ⇓v s′ and s, x →v s′, x′ work similarly to While except for the label v : N produced when evaluating expres- sions. We also allow language terms to interact with the labels by introducing the constructor obs N ⟨prog⟩. When terms evaluate inside an obs block, the labels are sequentially placed in the run-time store. The rest of the constructors are identical but the distinction between the two languages should be clear. While the expressions are the same as before, the syntax functor is now ΣX = ΣX ⊎N × X, and the behavior functor is B= S →N × S × Maybe X. The full semantics can be found in Figure 2. As for While, they specify a GSOS law ρ: Σ(Id × B) =⇒BΣ∗ . s, skip ⇓0 s v = eval s e s, l := e ⇓v update s l v v = eval s e v ̸= 0 s, while e p →v s, skip s, p ⇓v s′ s, p; q →v s′, q s, p ⇓v s′ s′′ = update s′ n v s, obs n p →v s′′, skip s, p →v s′, p′ s, p; q →v s′, p′; q s, p →v s′, p′ s′′ = update s′ n v s, obs n p →v s′′, obs (n + 1) p′ v = eval s e v = 0 s, while e p →v s, p; while e p Fig. 2. Semantics of While. Fig. 2. Semantics of While. Traditionally, the (in)security of a compiler has been a matter of full ab- straction; a compiler is fully abstract if it preserves and reflects Morris-style [31] Tsampas et al. 6 contextual equivalence. Remark 3. We shall be writing Bc for the cofree copointed endofunctor Id × B over B and bc : Bc 1 =⇒Bc 2 for id × b. 4.1 Maps of distributive laws Assume two GSOS laws ρ1 : Σ1(Id × B1) =⇒B1Σ∗ 1 and ρ2 : Σ2(Id × B2) =⇒ B2Σ∗ 2, where (Σ∗ 1, η1, µ1) and (Σ∗ 2, η2, µ2) are the monads freely generated by Σ1 and Σ2 respectively. We shall regard pairs of natural transformations (σ : Σ∗ 1 =⇒ Σ∗ 2, b : B1 =⇒B2) as compilers between the two semantics, where σ acts as a syntactic translation and b as a translation between behaviors. Remark 2. If A1 and A2 are the sets of terms freely generated by Σ1 and Σ2, we can get the compiler c : A1 →A2 from σ. On the other hand, b generates a function d : Z1 →Z2 between behaviors via finality. Remark 3. We shall be writing Bc for the cofree copointed endofunctor Id × B over B and bc : Bc 1 =⇒Bc 2 for id × b. A categorical approach to secure compilation 7 Definition 5 (Adapted from [50]). A map of GSOS laws from ρ1 to ρ2 consists of a natural transformation σ : Σ∗ 1 =⇒Σ∗ 2 subject to the monad laws σ ◦η1 = η2 and σ ◦µ1 = µ2 ◦Σ∗ 2σ ◦σ paired with a natural transformation b : B1 =⇒B2 that satisfies the following coherence criterion: Definition 5 (Adapted from [50]). A map of GSOS laws from ρ1 to ρ2 consists of a natural transformation σ : Σ∗ 1 =⇒Σ∗ 2 subject to the monad laws σ ◦η1 = η2 and σ ◦µ1 = µ2 ◦Σ∗ 2σ ◦σ paired with a natural transformation b : B1 =⇒B2 that satisfies the following coherence criterion: Σ∗ 1Bc 1 Bc 1Σ∗ 1 Σ∗ 2Bc 2 Bc 2Σ∗ 2 ρ∗ 1 σ ◦Σ∗ 1bc bc ◦Bc 1σ ρ∗ 2 ⌟ Remark 4. A natural transformation σ : Σ∗ 1 =⇒Σ∗ 2 subject to the monad laws is equivalent to a natural transformation t : Σ1 =⇒Σ∗ 2. Theorem 1. If σ and b constitute a map of GSOS laws, then we get a compiler c : A1 →A2 and behavior transformation d : Z1 →Z2 satisfying d ◦f1 = f2 ◦c : A1 →Z2. As bisimilarity is exactly equality in the final coalgebra (i.e. equality under fi : Ai →Zi), c preserves bisimilarity [50]. 4.1 Maps of distributive laws If d is a monomorphism (which, under mild conditions, is the case in Set if every component of b is a monomorphism), then c also reflects bisimilarity. What is very important though, is that the well-behavedness properties of the two GSOS laws bestow contextual meaning to bisimilarity. Recall that the gold standard for secure compilation is contextual equivalence (Definition 4), which is precisely what is observable through program contexts. Bisimilarity is generally not the same as contextual equivalence, but we can instead show that in the case of GSOS laws or other forms of distributive laws, bisimilarity defines the upper bound (most fine-grained distinction) of observability up to program contexts. We shall do so abstractly in the next subsections. 4.2 Abstract program contexts The informal notion of a context in a programming language is that of a program with a hole [31]. Thus contexts are a syntactic construct that models external interactions with a program: a single context is an experiment whose outcome is the evaluation of the subject program plugged in the context. j p g p gg Naïvely, one may hope to represent contexts by a functor H sending a set of variables X to the set HX of terms in ΣX that may have holes in them. A complication is that contexts may have holes at any depth (i.e. any number of operators may have been applied to a hole), whereas ΣX is the set of terms that have exactly one operator in them, immediately applied to variables. One solution is to think of Y in HY as a set of variables that do not stand for terms, but for contexts. This approach is fine for multi-hole contexts, but if we also want to consider single-hole contexts and a given single-hole context c is not the hole itself, then precisely one variable in c should stand for a single-hole context, and all other variables should stand for terms. Thus, in order to support both single- and multi-hole contexts, we make H a two-argument functor, where HXY is the set of contexts with term variables from X and context variables from Y . Tsampas et al. 8 Definition 6. Let C be a distributive category [14] with products ×, coproducts ⊎, initial object ⊥and terminal object ⊤, as is the case for Set. A context functor for a syntax functor Σ : C →C, is a functor H : C × C →C (with application to (X, Y ) denoted as HXY ) such that there exist natural transformations hole : ∀(X, Y ).⊤→HXY and con : ∀X.X × HXX →X ⊎ΣX making the following diagram commute for all X: X × ⊤ X X × HXX X ⊎ΣX π1 ∼ = idX×hole(X,X) i1 conX The idea of the transformation con is the following: it takes as input a variable x ∈X to be plugged into the hole, and a context c ∈HXX with one layer of syntax. The functor HX is applied again to X rather than Y because x is assumed to have been recursively plugged into the context placeholders y ∈Y already. 4.2 Abstract program contexts We then make a case distinction: if c is the hole itself, then i1 x is returned. Otherwise, i2 c is returned. Definition 7. Let C be a category as in Definition 6 and assume a syntax functor Σ with context functor H. If Σ has an initial algebra (A, qA) (the set of programs) and HA has a strong initial algebra (CA, qCA) [23] (the set of contexts), then we define the plugging function [[ ]] : A × CA →A as the “strong inductive extension” [23] of the algebra structure [idA, qA] ◦conA : A × HAA →A on A, i.e. as the unique morphism that makes the following diagram commute: A × HACA A × CA A × HA(A × CA) A × HAA A ⊎ΣA A id×qCA ∼ = (π,st) [[ ]] id×HA[[ ]] conA [id,qA] A × HACA A × CA A × HA(A × CA) A × HAA A ⊎ΣA A id×qCA ∼ = (π,st) [[ ]] id×HA[[ ]] conA [id,qA] ⌟ The above definition of contextual functors is satisfied by both single-hole and multi-hole contexts, the construction of which we discuss below. Multi-hole contexts Given a syntax functor Σ, its multi-hole context functor is simply HXY = ⊤⊎ΣY . The contextual natural transformation con is the obvious map that returns the pluggee if the given context is a hole, and otherwise the context itself (which is then a program): con : ∀X.X × (⊤⊎ΣX) →X ⊎ΣX con ◦(id × i1) = i1 ◦π1 : ∀X.X × ⊤→X ⊎ΣX con ◦(id × i2) = i2 ◦π2 : ∀X.X × ΣX →X ⊎ΣX 9 A categorical approach to secure compilation The ‘pattern matching’ is justified by distributivity of C. For hole = i1 : ⊤→ ⊤⊎ΣX, we can see that con ◦(id × hole) = i1 ◦π1 as required by the definition of a context functor. The ‘pattern matching’ is justified by distributivity of C. For hole = i1 : ⊤→ ⊤⊎ΣX, we can see that con ◦(id × hole) = i1 ◦π1 as required by the definition of a context functor. Single-hole contexts It was observed by McBride [29] that for inductive types, i.e. least fixpoints / initial algebras µF of certain endofunctors F called contain- ers [4] or simply polynomials, their single-hole contexts are lists of ∂F(µF) where ∂F is the derivative of F 6. 6 The list operator itself arises from the derivative of the free monad operator. 4.2 Abstract program contexts Derivatives for containers, which were developed by Abbott et al. in [5], enable us to give a categorical interpretation of single-hole contexts as long as the syntax functor Σ is a container. It would be cumbersome to lay down the entire theory of containers and their derivatives, so we shall instead focus on the more restricted set of Simple Polyno- mial Functors [22] (or SPF), used to model both syntax and behavior. Crucially, SPF’s are differentiable and hence compatible with McBride’s construction. Definition 8 (Simple Polynomial Functors). The collection of SPF is the least set of functors C →C satisfying the following rules: Tsampas et al. Using derivatives we can define the context functor HXY = ⊤⊎((∂Σ X)×Y ) for syntax functor Σ. Then the initial algebra CA of HA is indeed List ((∂Σ) A), the set of single-hole contexts for A ∼= ΣA. Plugging Before defining con, we define an auxiliary function conStep : ∂Σ × Id =⇒Σ. We defer the reader to [29] for the full definition of conStep, which is inductive on the SPF Σ, and shall instead only define the case for coproducts. So, for ∂(F ⊎G) = ∂F ⊎∂G we have: conStepF ⊎G : (∂F ⊎∂G) × Id =⇒F ⊎G conStepF ⊎G ◦(i1 × id) = i1 ◦conStepF : ∂F × Id =⇒F ⊎G conStepF ⊎G ◦(i2 × id) = i2 ◦conStepG : ∂G × Id =⇒F ⊎G We may now define con : X × HXX →X ⊎ΣX as follows: con : ∀X.X × (⊤⊎(∂Σ X × X)) →X ⊎ΣX con ◦(id × i1) = i1 ◦π1 : ∀X.X × ⊤→X ⊎ΣX con ◦(id × i2) = i2 ◦conStepΣ ◦π2 : ∀X.X × (∂Σ X × X) →X ⊎ΣX con ◦(id × i2) = i2 ◦conStepΣ ◦π2 : ∀X.X × (∂Σ X × X) →X ⊎ΣX By setting hole = i1 : ⊤→⊤⊎(∂Σ X × X) we can see that con ◦(id × hole) = i1 ◦π1 as required by Definition 6. Definition 8 (Simple Polynomial Functors). The collection of SPF is the least set of functors C →C satisfying the following rules: id Id ∈SPF const J ∈Obj(C) KJ ∈SPF prod F, G ∈SPF F × G ∈SPF coprod F, G ∈SPF F ⊎G ∈SPF comp F, G ∈SPF F ◦G ∈SPF We can now define the differentiation action ∂: SPF →SPF by structural induction. Interestingly, it resembles simple derivatives for polynomial functions. Definition 9 (SPF derivation rules). ∂Id = ⊤, ∂KJ = ⊥, ∂(G ⊎H) = ∂G ⊎∂H, ∂(G × H) = (∂G × H) ⊎(G × ∂H), ∂(G ◦H) = (∂G ◦H) × ∂H. Definition 9 (SPF derivation rules). Definition 9 (SPF derivation rules). Definition 9 (SPF derivation rules). ∂Id = ⊤, ∂KJ = ⊥, ∂(G ⊎H) = ∂G ⊎∂H, ∂(G × H) = (∂G × H) ⊎(G × ∂H), ∂(G ◦H) = (∂G ◦H) × ∂H. ∂Id = ⊤, ∂KJ = ⊥, ∂(G ⊎H) = ∂G ⊎∂H Example 3. The definition of con for single-hole contexts might look a bit cryptic at first sight so we shall use a small example from [29] to shed some light. In the case of binary trees, locating a hole in a context can be thought of as traversing through a series of nodes, choosing left or right according to the placement of the hole until it is found. At the same time a record of the trees at the non-chosen branches must be kept so that in the end the entire structure can be reproduced.i Now, considering that the set of binary trees is the least fixed point of functor ⊤⊎(Id × Id), then the type of “abstract” choice at each intersection is the functor KBool × Id, where KBool stands for a choice of left or right and the Id part represents the passed structure. Lists of (KBool × Id) BinTree are exactly the sort of record we need to keep, i.e. they contain the same information as a tree with a single hole. And indeed KBool × Id is (up to natural isomorphism) the derivative of ⊤⊎(Id × Id)! ⌟ 10 4.3 Contextual coclosure Having established a categorical notion of contexts, we can now move towards formulating contextual categorical arguments about bisimilarity. We assume a context functor H for Σ such that HA has strong initial algebra (CA, qCA) (the object containing all contexts). First, since we prefer to work in more general categories than just Set, we will encode relations R ⊆X × Y as spans X r1 ←−R r2 −→Y . One may wish to consider only spans for which (r1, r2) : R →X × Y is a monomorphism, though this is not necessary for our purposes. We want to reason about contextually closed relations on the set of terms A, which are relations such that a1 R a2 implies (c Ja1K) R (c Ja2K) for all contexts c ∈CA. Contextual equivalence will typically be defined as the co- closure of equitermination: the greatest contextually closed relation that implies equitermination. For spans, this becomes: Definition 10. In a category as in Definition 6, a span A r1 ←−R r2 −→A is called contextually closed if there is a morphism J K : CA×R →R making the following diagram commute: CA × A CA × R CA × A A R A J K id×r1 id×r2 J K J K r1 r2 11 A categorical approach to secure compilation The contextual co-closure A ¯r1 ←−¯R ¯r2 −→A of an arbitrary span A r1 ←−R r2 −→A is the final contextually closed span on A with a span morphism ¯R →R. ⌟ We call terms bisimilar if the operational semantics f : A →Z assigns them equal behaviors: Definition 11. We define (strong) bisimilarity ∼bis as the pullback of the equal- ity span (idZ, idZ) : Z →Z × Z along f × f : A × A →Z × Z (if existent). Definition 11. We define (strong) bisimilarity ∼bis as the pullback of the equal- ity span (idZ, idZ) : Z →Z × Z along f × f : A × A →Z × Z (if existent). Theorem 2. Under the assumptions of 7, bisimilarity (if existent) is contextu- ally closed. Proof. We need to give a morphism of spans from CA × (∼bis) to (∼bis): CA × A CA × (∼bis) CA × A A (∼bis) A Z Z Z. 4.3 Contextual coclosure J K id×r1 id×r2 J K f r1 r2 w f idZ idZ By definition of (∼bis), it suﬀices to give a morphism of spans to the equality span on Z, i.e. to prove that f ◦J K ◦(id × r1) = f ◦J K ◦(id × r2). To this end, consider the following diagram (parameterized by i ∈{1, 2}), in which every polygon is easily seen to commute: (∼bis) × HACA (∼bis) × CA A × HACA A × CA (∼bis) × HA(A × CA) A × HA(A × CA) (∼bis) × HAA A × HAA A ⊎ΣA A (∼bis) × HAZ (∼bis) × HZZ Z × HZZ Z ⊎ΣZ Z id×qCA ∼ = ri×id (π1,HA(ri×id)◦st) ri×id id×qCA ∼ = (π1,st) J K ri×id id×HAJ K id×HAJ K ri×id id×HAf conA f×Hf f [id,qA] f⊎Σf f id×Hf idZ w×id conZ [id,qZ] The bottom-right square stems from the underlying GSOS law: it is the algebra homomorphism part of the bialgebra morphism between the initial and the final bialgebras. Commutativity of the outer diagram reveals that f ◦J K ◦(ri × id) is, regardless of i, the strong inductive extension of [idZ, qZ] ◦conZ ◦(w × HfidZ) : (∼bis) × HAZ →Z. Thus, it is independent of i. ⊓⊔ (∼bis) × HACA (∼bis) × CA A × HACA A × CA (∼bis) × HA(A × CA) A × HA(A × CA) (∼bis) × HAA A × HAA A ⊎ΣA A (∼bis) × HAZ (∼bis) × HZZ Z × HZZ Z ⊎ΣZ Z id×qCA ∼ = ri×id (π1,HA(ri×id)◦st) ri×id id×qCA ∼ = (π1,st) J K ri×id id×HAJ K id×HAJ K ri×id id×HAf conA f×Hf f [id,qA] f⊎Σf f id×Hf idZ w×id conZ [id,qZ] The bottom-right square stems from the underlying GSOS law: it is the algebra homomorphism part of the bialgebra morphism between the initial and the final bialgebras. Commutativity of the outer diagram reveals that f ◦J K ◦(ri × id) is, regardless of i, the strong inductive extension of [idZ, qZ] ◦conZ ◦(w × HfidZ) : (∼bis) × HAZ →Z. Thus, it is independent of i. ⊓⊔ 12 Tsampas et al. Tsampas et al. 12 Corollary 2. In Set, bisimilarity implies contextual equivalence.7 Proof. Bisimilarity implies equitermination. This yields an implication between their coclosures. ⊓⊔ Proof. Bisimilarity implies equitermination. This yields an implication between their coclosures. 4.3 Contextual coclosure ⊓⊔ ⊓⊔ Comparing Corollary 1 to contextual equivalence in Definition 4 reveals their key difference. Contextual equivalence makes minimal assumptions on the un- derlying observables, which are simply divergence and termination. On the other hand, the contextual coclosure of bisimilarity assumes maximum observability (as dictated by the behavior functor) and in that sense it represents the upper bound of what can be observed through contexts. Consequently, this criterion is useful if the observables adequately capture the threat model, which is true for the examples that follow. This theorem echoes similar results in the broader study of coalgebraic bisim- ulation [11, 41]. There are, however, two differences. The first is that our theorem allows for extra flexibility in the definition of contexts as the theorem is paramet- ric on the context functor. Second, by making the context construction explicit we can directly connect (the contextual coclosure of) bisimilarity to contextual equivalence (Corollary 2) and so have a more convincing argument for using maps of distributive laws as secure compilers. 7 Note that we can not conclude that preservation of bisimilarity would imply preser- vation of contextual equivalence. 5 An extra register (Part II) The next step is to define the syntax and behavior natural transformations. The first compiler, σ: Σ =⇒Σ, is a very simple mapping of constructors in While to their Whilecounterparts. The second natural transformation, σ: Σ =⇒ Σ∗ , is more complex as it involves an additional layer of syntax in While. Definition 12 (Sandboxing natural transformation). Consider the nat- ural transformation e : Σ =⇒Σwhich embeds Σ in Σ. Using PL notation, we define σ: ΣX →Σ∗ X : p 7→e(p)+. This yields a monad morphism σ∗ : Σ∗→Σ∗ (Remark 4). Definition 12 (Sandboxing natural transformation). Consider the nat- ural transformation e : Σ =⇒Σwhich embeds Σ in Σ. Using PL notation, we define σ: ΣX →Σ∗ X : p 7→e(p)+. This yields a monad morphism σ∗ : Σ∗→Σ∗ (Remark 4). Defining the natural translation between behaviors is a matter of choosing a designated value for the added observable label. The only constraint is that the chosen value has to coincide with the label that the sandbox produces. Band Bare identical so we need a single natural transformation b : B =⇒B/: b : ∀X. (S →S × Maybe X) →S →N × S × Maybe X b f = λ(s : S) →(0, f(s)) 7 Note that we can not conclude that preservation of bisimilarity would imply preser- vation of contextual equivalence. 13 A categorical approach to secure compilation While to WhileWe now have the natural translation pairs (σ, b) and (σ, b), which allows us to check the coherence criterion from Section 4.1. We shall be using a graphical notation that provides for a good intuition as to what failure or success of the criterion means. For example, Fig. 3 shows failure of the coherence criterion for the first pair. l := e v = eval s e s ⇓update s l v l := e v = eval s e s ⇓v/0 update s l v ρ∗ σ∗ ◦Σ∗bc bc ◦Bcσ∗  ρ∗  Fig. 3. Failure of the criterion for (σ, b). l := e v = eval s e s ⇓update s l v l := e v = eval s e s ⇓v/0 update s l v ρ∗ σ∗ ◦Σ∗bc bc ◦Bcσ∗  ρ∗  Fig. 3. Failure of the criterion for (σ, b). 5 An extra register (Part II) The horizontal arrows in the di- agram represent the two semantics, ρ∗and ρ∗ , while the vertical ar- rows are the two horizontal com- positions of the natural translation pair. The top-left node holds an el- ement of Σ∗(Id × B), which in this case is an assignment operation. The two rightmost nodes represent be- haviors, so the syntactic element is missing from the left side of the transition arrows.i Fig. 3. Failure of the criterion for (σ, b). In the upper path, the term is first applied to the GSOS law ρ∗and the result is then passed to the translation pair, thus producing the designated label 0, typeset in blue for convenience. In the lower path, the term is first applied to the translation and then goes through the target semantics, ρ∗ , where the label v is produced. It is easy to find such an s so that v ̸= 0. l := e v = eval s e s ⇓update s l v l := e+ v = eval s e s, l := e ⇓v update s l v s ⇓0 update s l v ρ∗ σ∗ ◦Σ∗bc bc ◦Bcσ∗  ρ∗  Fig. 4. The coherence criterion for (σ, b). While to WhileThe same ex- ample is investigated for the sec- ond translation pair (σ, b). Fig- ure 4 shows what happens when we test the same case as before. Applying ρ∗ to l := e+ is simi- lar to ρacting twice. The inner- most transition is the intermedi- ate step and as it only appears in the bottom path it is typeset in red. This time the diagram commutes as the label produced in the inner layer, v, is effectively erased by the sandboxing rules of While. l := e v = eval s e s ⇓update s l v l := e+ v = eval s e s, l := e ⇓v update s l v s ⇓0 update s l v ρ∗ σ∗ ◦Σ∗bc bc ◦Bcσ∗  ρ∗  Fig. 4. The coherence criterion for (σ, b). l := e v = eval s e s ⇓update s l v l := e+ v = eval s e s, l := e ⇓v update s l v s ⇓0 update s l v ρ∗ σ∗ ◦Σ∗bc bc ◦Bcσ∗  ρ∗  Fig. 4. The coherence criterion for (σ, b). Fig. 4. 6 State mismatch Having established our categorical foundations, we shall henceforth focus on examples. The first one involves a compiler where the target machine is not nec- essarily more powerful than the source machine, but the target value primitives are not isomorphic to the ones used in the source. This is a well-documented problem [37], which has led to failure of full abstraction before [8, 18, 25]. For example, we can repeat the development of While except we substitute natural numbers with integers. We call this new version WhileZ. ⟨ ⟩ \| \| ⟨ ⟩⟨b ⟩⟨ ⟩\| ⟨ ⟩⟨ ⟩ For example, we can repeat the development of While except we substitute natural numbers with integers. We call this new version WhileZ. ⟨expr⟩::= lit Z \| var N \| ⟨expr⟩⟨bin⟩⟨expr⟩\| ⟨un⟩⟨expr⟩ The behavior functor also differs in that the store type S is substituted with SZ, the set of lists of integers. We can define the behavioral natural transformation bZ : B =⇒BZ as the best “approximation” between the two behaviors. In Set: bZ : ∀X. (S →S × (⊤⊎X)) →SZ →SZ × (⊤⊎X) bZ f = [toZ, id] ◦f ◦toN 0 := min(var[0],0) [n] ⇓[0] 0 := min(var[0],0) [−1] ⇓[−1/0] ρ∗ Σ∗bc Z bc Z ρ∗ Z Fig. 6. Failure of the criterion for (id, bZ). Where toN replaces all negative numbers in the store with 0 and toZ typecasts S to SZ. It is easy to see that the identity compiler from While to WhileZ is not fully abstract. For example, the expressions 0 and min(var[0], 0) are identical in While but can be distinguished in WhileZ (if var[0] is negative). This is reflected in the coherence criterion diagram for the identity compiler in Figure 6, when initiating the store with a negative integer. 0 := min(var[0],0) [n] ⇓[0] 0 := min(var[0],0) [−1] ⇓[−1/0] ρ∗ Σ∗bc Z bc Z ρ∗ Z Fig. 6. Failure of the criterion for (id, bZ). 0 := min(var[0],0) [n] ⇓[0] 0 := min(var[0],0) [−1] ⇓[−1/0] ρ∗ Σ∗bc Z bc Z ρ∗ Z Fig. 6. Failure of the criterion for (id, bZ). Fig. 6. Failure of the criterion for (id, bZ). The solution is to create a special environment where WhileZ forgets about negative integers, in essence copying what bZ does on the store. 5 An extra register (Part II) The coherence criterion for (σ, b). An endo-compiler for WhileIf Ais the set of closed terms for While, the compiler u : A→A, which “escapes” Whileterms from their sandboxes can be elegantly modeled using category theory. As before, it is not possible to express it using a simple natural transformation Σ=⇒Σ. We can, however, use the free pointed endofunctor [28] over Σ, Id ⊎Σ. What we want is to map non-sandboxed terms to themselves and lift the extra layer of syntax from sandboxed terms. Intuitively, for a set of variables X, ΣX is one layer of syntax “populated” with elements of X. If X ⊎ΣX is the union of ΣX with the set of variables X, lifting the sandboxing layer is mapping the X in X+ to the left of X ⊎ΣX and the rest to themselves at the right. 14 Tsampas et al. s, p →v s′, q p+ s →0 s′, q+ s, p →v s′, q p s →v/0 s′, q ρ∗  σ∗ u Bcσ∗ u ρ∗  Fig. 5. Failure of the criterion for σu. Fig. 5. Failure of the criterion for σu. This is obviously not a secure compiler as it allows discerning previously indistinguishable programs. As we can see in Figure 5, the coherence criterion fails in the expected manner. 6 State mismatch This is a special kind of sandbox, written ⟨_⟩, for which we introduce the following rules: toN(s), p ⇓s′ s, ⟨p⟩⇓s′ toN(s), p →s′, p′ s, ⟨p⟩→s′, ⟨p′⟩ A categorical approach to secure compilation 15 0 := min(var[0],0) [n] ⇓[0] ⟨0 := min(var[0],0)⟩ [−1] ⇓[0] ρ∗ σ∗ Z ◦Σ∗bc Z bc Z ◦Bcσ∗ Z ρ∗ Z Fig. 7. The coherence criterion for (σZ, bZ). 0 := min(var[0],0) [n] ⇓[0] ⟨0 := min(var[0],0)⟩ [−1] ⇓[0] ρ∗ σ∗ Z ◦Σ∗bc Z bc Z ◦Bcσ∗ Z ρ∗ Z Fig. 7. The coherence criterion for (σZ, bZ). Fig. 7. The coherence criterion for (σZ, bZ). We may now repeat the construction from Definition 12 to define the compiler σZ. We can easily verify that the pair (σZ, bZ) constitutes a map of distributive laws. For instance, Figure 7 demonstrates how the previous failing case now works under (σZ, bZ). Tsampas et al. Instruction nop is the no-operation, stop halts execution and assign is anal- ogous to the assignment operation in While. The br instruction is what really defines Low, as it stands for bidirectional relative branching. Semantics of Low Figure 8 shows the operational semantics of Low. The exe- cution state of a running program consists of a run-time store and the program counter register PC ∈Z that points at the instruction being processed. If the pro- gram counter is zero, the leftmost instruction is executed. If the program counter is greater than zero, then the current instruction is further to the right. Other- wise, the program counter is out-of-bounds and execution stops. The categorical interpretation suggests a GSOS law ρL of syntax functor ΣLX = inst⊎(inst×X) over behavior functor BLX = S × Z →S × Z × Maybe X. An insecure compiler This time we start with the behavioral translation, which is less obvious as we have to go from BX = S →S × Maybe X to BLX = S × Z →S × Z × Maybe X. The increased arity in BL poses an interesting question as to what the program counter should mean in While. It makes sense to consider the program counter in While as zero since a program in While is treated uniformly as a single statement. bL : ∀X. (S →S × Maybe X) →S × Z →S × Z × Maybe X bL f (s, 0) = { (s′, 1, nothing) if f s = (s′, nothing) (s′, 0, just y) if f s = (s′, just y) bL f (s, n ̸= 0) = (s, n, nothing) When it comes to translating terms, a typical compiler from While to Low would untangle the tree-like structure of While and convert it to a list of Low in- structions. For while statements, the compiler would use branching to simulate looping in the low-level. Example 4. Let us look at a simple case of a loop. The While program while (var 0 < 2) (1 := var 1 + 1) is compiled to br !(var 0 < 2) 3 ;; assign 1 (var 1 + 1) ;; br (lit 1) -2 ⌟ while (lit 0) (0 := lit 0) s →s, skip br !(lit 0) 3 ;; assign 0 lit 0 ;; br (lit 1) −2 s, 1 ⇓s, 1 s, 1 →s[07→0], 2... 7 Control Flow Many low-level languages support unrestricted control flow in the form of jump- ing or branching to an address. On the other hand, control flow in high-level languages is usually restricted (think if-statements or function calls). A compiler from the high-level to the low-level might be insecure as it exposes source-level programs to illicit control flow. This is another important and well-documented example of failure of full abstraction [8, 37, 3, 33]. s, 0, stop [;; x] ⇓s, 0 v = eval s e s′ = update s n v s, 0, assign n v [;; x] →s′, 1, assign n v [;; x] PC ≥0 s, PC, x ⇓s′, PC′ s, PC + 1, i ;; x ⇓s′, PC′ + 1 v = eval s e v = 0 s, 0, br e z [;; x] →s, 1, br e z [;; x] v = eval s e v ̸= 0 s, 0, br e z [;; x] →s, z, br e z [;; x] PC < 0 s, PC, i ;; x ⇓s, PC p = nop [;; x] s, 0, p →s, 1, p PC ≥0 s, PC, x →s′, PC′, x′ s, PC + 1, i ;; x →s′, PC′ + 1, i ;; x′ PC ̸= 0 s, PC, i ⇓s, PC Fig. 8. Semantics of the Low language. Elements in square brackets are optional. Fig. 8. Semantics of the Low language. Elements in square brackets are optional. We introduce low-level language Low, the programs of which are non-empty lists of instructions. Low differs significantly from While and its derivatives in both syntax and semantics. For the syntax, we define the set of instructions ⟨inst⟩and set of programs ⟨asm⟩. ⟨inst⟩::= nop \| stop \| assign N ⟨expr⟩\| br ⟨expr⟩Z ⟨asm⟩::= ⟨inst⟩\| ⟨inst⟩;; ⟨asm⟩ ⟨inst⟩::= nop \| stop \| assign N ⟨expr⟩\| br ⟨expr⟩Z ⟨asm⟩::= ⟨inst⟩\| ⟨inst⟩;; ⟨asm⟩ 16 Tsampas et al. h cL BLcL ◦bA hL Fig. 9. cL is not a coalgebra homomorphism. This compiler, called cL, cannot be defined in terms of a natural transformation Σ =⇒ Σ∗ L as per Remark 4, but it is inductive on the terms of the source language. In this case we can directly compare the two op- erational models bA ◦h : A → BLA (where h : A →BA) and hL : AL →BLAL and notice that cL : A →AL is not a coalgebra homomorphism (Figure 9). The key is while (lit 0) (0 := lit 0) s →s, skip br !(lit 0) 3 ;; assign 0 lit 0 ;; br (lit 1) −2 s, 1 ⇓s, 1 s, 1 →s[07→0], 2... h cL BLcL ◦bA hL Fig. 9. cL is not a coalgebra homomorphism. while (lit 0) (0 := lit 0) s →s, skip br !(lit 0) 3 ;; assign 0 lit 0 ;; br (lit 1) −2 s, 1 ⇓s, 1 s, 1 →s[07→0], 2... h cL BLcL ◦bA hL Fig. 9. cL is not a coalgebra homomorphism. A categorical approach to secure compilation 17 that the program counter in Low allows for finer observations on programs. Take for example the case for while (lit 0) (0 := lit 0), where the loop is always skipped. In Low, we can still access the loop body by simply pointing the program counter to it. This is a realistic attack scenario because Low allows manipulation of the program counter via the br instruction. Solution By comparing the semantics between While in Figure 1 and Low in Figure 8 we find major differences. The first one is the reliance of Low to a program counter which keeps track of execution, whereas While executes state- ments from left to right. Second, the sequencing rule in While dictates that statements are removed from the program state8 upon completion. On the other hand, Low keeps the program state intact at all times. Finally, there is a stark contrast between the two languages in the way they handle while loops. To address the above issues we introduce a new sequencing primitive ;;c and a new looping primitive loop for Low, which prohibit illicit control flow and properly propagate the internal state. Furthermore, we change the semantics of the singleton assign instruction so that it mirrors the peculiarity of its While counterpart. 8 We are not referring to the store, but to the internal, algebraic state. Tsampas et al. The additions can be found in Figure 10. PC ̸= 0 s, PC, x ;;c y ⇓s′, PC s, 0, x ⇓s′, z s, 0, x ;;c y →s′, 0, y s, 0, x →s′, z, x′ s, 0, x ;;c y →s′, 0, x′ ;;c y PC ̸= 0 s, PC, loop e x ⇓s, PC v = eval s e v = 0 s, 0, loop e x →s, 0, stop v = eval s e v ̸= 0 s, 0, loop e x →s, 0, x ;;c loop e x v = eval s e s′ = update s n v s, 0, assign n v ⇓s′, 0 Fig. 10. Secure primitives for the Low language. Fig. 10. Secure primitives for the Low language. We may now define the simple “embedding” natural transformation σE : Σ =⇒ΣL, which maps skip to stop, assignments to assign, sequencing to ;;c and while to loop. Figure 11 shows success of the coherence criterion for the while case. Since the diagram commutes for all cases, (σE, bE) is a map of GSOS laws between While and the secure version of Low. This guarantees that, remarkably, de- spite the presence of branching, a low-level attacker cannot illicitly access code that is unreachable on while (lit 0) p s →s, skip loop (lit 0) p s, 1 ⇓s, 1 ρ∗ σ∗ E ◦Σ∗bc L bc L ◦Bcσ∗ E ρ∗ L Fig. 11. The coherence criterion for (σE, bL). Fig. 11. The coherence criterion for (σE, bL). Tsampas et al. 18 the high-level. Regardless, the solution is a bit contrived in that the new Low primitives essentially copy what While does. This is partly because the above are complex issues involving radically different languages but also due to the current limitations of the underlying theory. We elaborate further on said limitations, as well as advantages and future improvements, at Section 9. 9 Examples of this are enclaves in Intel SGX [15] and object capabilities in CHERI [51]. 8 Local state encapsulation A categorical approach to secure compilation 19 m′ = update m (l + L ∗sp) (evalSP m sp e) (m, sp), l := e ⇓(m′, sp) sp > 0 (m, sp), return ⇓(m, sp −1) (m, sp), p ⇓(m′, sp) (m, sp), p; q →(m′, sp), q (m, sp), p →(m′, sp), p′ (m, sp), p; q →(m′, sp), p′; q (m, sp), skip ⇓(m, sp) evalSP m sp e = 0 (m, sp), while e p →(m, sp), skip evalSP m sp e ̸= 0 (m, sp), while e p →(m, sp), p; while e p (m, sp), frame ⇓(m, sp + 1) Fig. 13. Semantics of the Stack language. m′ = update m (l + L ∗sp) (evalSP m sp e) (m, sp), l := e ⇓(m′, sp) sp > 0 (m, sp), return ⇓(m, sp −1) (m, sp), p ⇓(m′, sp) (m, sp), p; q →(m′, sp), q (m, sp), p →(m′, sp), p′ (m, sp), p; q →(m′, sp), p′; q (m, sp), skip ⇓(m, sp) evalSP m sp e = 0 (m, sp), while e p →(m, sp), skip evalSP m sp e ̸= 0 (m, sp), while e p →(m, sp), p; while e p (m, sp), frame ⇓(m, sp + 1) Fig. 13. Semantics of the Stack language. m′ = update m (l + L ∗sp) (evalSP m sp e) (m, sp), l := e ⇓(m′, sp) sp > 0 (m, sp), return ⇓(m, sp −1) (m, sp), p ⇓(m′, sp) (m, sp), p; q →(m′, sp), q (m, sp), p →(m′, sp), p′ (m, sp), p; q →(m′, sp), p′; q (m, sp), skip ⇓(m, sp) evalSP m sp e = 0 (m, sp), while e p →(m, sp), skip evalSP m sp e ̸= 0 (m, sp), while e p →(m, sp), p; while e p (m, sp), frame ⇓(m, sp + 1) Fig. 13. Semantics of the Stack language. Fig. 13. Semantics of the Stack language. Low-level stack In typical low-level instruction sets like the Intel x86 [21] or MIPS [30] there is a single, continuous memory which is partitioned in frames via processor registers. Figure 13 shows the semantics of Stack, a variant of WhileB with the same syntax that incorporates a simple low-level stack. 8 Local state encapsulation High-level programming language abstractions often involve some sort of private state space that is protected from other objects. Basic examples include func- tions with local variables and objects with private members. Low-level languages do not offer such abstractions but when it comes to secure architectures, there is some type of hardware sandboxing 9 to facilitate the need for local state encap- sulation. Compilation schemes that respect confidentiality properties have been a central subject in secure compilation work [37, 8, 18, 46], dating all the way back to Abadi’s seminal paper [1]. In this example we will explore how local state encapsulation fails due to lack of stack clearing [46, 44]. We begin by extending While to support blocks which have their own private state, thus introducing WhileB. More precisely, we add the frame and return commands that denote the beginning and end of a new block. We also have to modify the original behavior functor B to act on a stack of stores by simply specifying BBX = [S] →[S] × Maybe X, where [S] denotes a list of stores. For reasons that will become apparent later on, we shall henceforth consider stores of a certain length, say L. m, skip ⇓m v = eval' m e m′ = update' m l v m, l := e ⇓m′ m, p ⇓m′ m, p; q →m′, q m, p →m′, p′ m, p; q →m′, p′; q eval' m e = 0 m, while e p →m, skip eval' m e ̸= 0 m, while e p →m, p; while e p s0 = [0, 0, . . . , 0] m, frame ⇓s0 :: m s :: m, return ⇓m Fig. 12. Semantics of the WhileB language. Fig. 12. Semantics of the WhileB language. The semantics for WhileB can be found in Figure 12. Command frame allo- cates a new private store by appending one to the stack of stores while return pops the top frame from the stack. This built-in, automatic (de)allocation of frames guarantees that there are no traces of activity, in the form of stored values, of past blocks. The rest of the semantics are similar to While, only now evaluating an expression and updating the state acts on a stack of stores instead of a single, infinite store and var expressions act on the active, topmost frame. 8 Local state encapsulation Failure of the criterion for (id, bB). Solution It is clear that the lack of stack frame initialization in Stack is the lead cause of failure so we introduce the following fix in the frame rule. m′ = (take (L ∗sp) m) ++ s0 ++ (drop ((L + 1)‘ ∗sp) m) (m, sp), frame ⇓(m′, sp + 1) The idea behind the new frame rule is that the L-sized block in position sp, which is going to be the new stack frame, has all its values replaced by zeroes. As we can see in Figure 15, the coherence criterion is now satisfied and the example described earlier no longer works. frame [] ⇓[s0] frame s′ = s0 ++ (drop L s) (s, 0) ⇓(s′, 1) ρ∗ B Σ∗ Bbc B bc B ρ∗ St Fig. 15. The coherence criterion for (id, bB) under the new frame rule. Fig. 15. The coherence criterion for (id, bB) under the new frame rule. 8 Local state encapsulation The difference is that the stack frames are all sized L, the same size as each individual store in WhileB, so at each frame and return we need only increment and decrement the stack pointer. The presence of the stack pointer, which is essentially a natural number, means that the behavior of Stack is BSX = S ×N →S ×N×Maybe X. The new evaluation function, evalSP, works similarly to eval in Definition 1, except for var l expressions that dereference values at offset l + L ∗sp. An insecure compiler WhileB and Stack share the same syntax so we only need a behavioral translation, which is all about relating the two different notions of stack. We thus define natural transformation bB : BB =⇒BS: bB : ∀X. ([SL] →[SL] × Maybe X) →S →N →S × N × Maybe X bB f s sp = (override (join m) s, len m, y) where (m, y) = f (div s sp) div s sp = (take L s) :: (div (drop L s) (sp −1)) override s′ s = s′ ++ drop (len s′) s We “divide” an infinite list by the number of stack frames, feed the result to the behavior function f and join (“flatten”) it back together while keeping the original part of the infinite list which extends beyond the active stack intact. Note that in the case of the frame command f adds a new frame to the list of stores. The problem is that in WhileB the new frame is initialized to 0 in contrast to Stack where frame does not initialize new frames. This leads to a failure of the coherence criterion for (id, bB) as we can see in Figure 14. Failure of the criterion is meaningful in that it underlines key problems of this compiler which can be exploited by a low-level attacker. First, the low-level calling convention indirectly allows terms to access expired stack frames. Second, violating the assumption in WhileB that new frames are properly initialized breaks behavioral equivalence. For example, programs a ≜frame ; 0 := var[0]+1 and b ≜frame ; 0 := 1 behave identically in WhileB but not in Stack. Tsampas et al. 20 frame [] ⇓[s0] frame s′ = override s0 s (s, 0) ⇓(s/s′, 1) ρ∗ B Σ∗ Bbc B bc B ρ∗ St Fig. 14. Failure of the criterion for (id, bB). Fig. 14. 9 Discussion and future work On Mathematical Operational Semantics The cases we covered in this paper are presented using Plotkin’s Structural Operational Semantics [38], yet their foundations are deeply categorical [48]. Consequently, for one to use the methods presented in this paper, the semantics involved must fall within the framework of distributive laws, the generality of which has been explored in the past [47, 50], albeit not exhaustively. To the best of our knowledge, Section 7 and Section 8 show the first instances of distributive laws as low-level machines. Bialgebraic semantics are well-behaved in that bisimilarity is a congruence [19]. We used that to show that two bisimilar programs will remain bisimilar irrespec- tive of the context they are plugged into, which is not the same as contextual equivalence. However, full abstraction is but one of a set of proposed characteri- zations of secure compilation [36, 2] and the key intuition is that our framework A categorical approach to secure compilation 21 is suitable as long as bisimilarity adequately captures the threat model. While this is the case in the examples, we can imagine situations where the threat model is weaker than the one implied by bisimilarity. For example, language Whilein Section 3 includes labels in its transition structure and the underlying model is accurate in that Whileterms can ma- nipulate said labels. However, if we were to remove obs statements from the syntax, the threat model becomes weaker than the one implied by bisimilarity. Similarly in Section 7 and Low, we could remove the implicit assumption that the program counter can be manipulated by a low-level attacker.if This issue can be classified as part of the broader effort towards coalgebraic weak bisimilarity, a hard problem which has been an object of intense, ongoing scientific research [42, 39, 20, 13, 43, 42, 12]. Of particular interest is the work by Abou-Saleh and Pattinson [7, 6] about bialgebraic semantics, where they use techniques introduced in [20] to obtain a more appropriate semantic domain for effectful languages as a final coalgebra in the Kleisli category of a suitable monad. This method is thus a promising avenue towards exploring weaker equivalences in bialgebraic semantics, as long as these can be described by a monad. 9 Discussion and future work On Maps of Distributive Laws Maps of distributive laws were first men- tioned by Power and Watanabe [40], then elaborated as Well-behaved transla- tions by Watanabe [50] and more recently by Klin and Nachyla [27]. Despite the few examples presented in [50, 27], this paper is the first major attempt towards applying the theory behind maps of distributive laws in a concrete problem, let alone in secure compilation. p From a theoretical standpoint, maps of distributive laws have remained largely the same since their introduction. This comes despite the interesting develop- ments discussed in Section 9 regarding distributive laws, which of course are the subjects of maps of distributive laws. We speculate the existence of Kleisli maps of distributive laws that guarantee preservation of equivalences weaker than bisimilarity. We plan to develop this notion and explore its applicability in future work. Conclusion It is evident that the systematic approach presented in this work may markedly streamline proofs for compiler security as it involves a single, sim- ple coherence criterion. Explicit reasoning about program contexts is no longer necessary, but that does not mean that contexts are irrelevant. On the contrary, the guarantees are implicitly contextual due to the well-behavedness of the se- mantics. Finally, while the overall usability and eventual success of our method remains a question mark as it depends on the expressiveness of the threat model, the body of work in coalgebraic weak bisimilarity and distributive laws in Kleisli categories suggests that there are many promising avenues for further progress. Acknowledgements. This work was partially supported by the Research Fund KU Leuven. Andreas Nuyts holds a PhD fellowship from the Research Founda- tion - Flanders (FWO). Tsampas et al. 22 References [1] Martín Abadi. “Protection in Programming-Language Translations”. In: Secure Internet Programming, Security Issues for Mobile and Distributed Objects. 1999, pp. 19–34. doi: 10.1007/3-540-48749-2\_2. url: https: [2] Carmine Abate et al. Journey Beyond Full Abstraction: Exploring Robust Property Preservation for Secure Compilation. 2018. arXiv: 1807.04603 [cs.PL]. [3] Carmine Abate et al. “When Good Components Go Bad: Formally Secure Compilation Despite Dynamic Compromise”. In: Proceedings of the 2018 ACM SIGSAC Conference on Computer and Communications Security, CCS 2018, Toronto, ON, Canada, October 15-19, 2018. Ed. by David Lie et al. ACM, 2018, pp. 1351–1368. isbn: 978-1-4503-5693-0. doi: 10.1145/ 3243734.3243745. url: https://doi.org/10.1145/3243734.3243745. [4] Michael Gordon Abbott, Thorsten Altenkirch, and Neil Ghani. “Contain- ers: Constructing strictly positive types”. In: Theor. Comput. Sci. 342.1 (2005), pp. 3–27. doi: 10.1016/j.tcs.2005.06.002. url: https://doi.org/10. 1016/j.tcs.2005.06.002.f [5] Michael Gordon Abbott et al. “for Data: Differentiating Data Structures”. In: Fundam. Inform. 65.1-2 (2005), pp. 1–28. url: http://content.iospress. com/articles/fundamenta-informaticae/fi65-1-2-02. [6] Faris Abou-Saleh. “A coalgebraic semantics for imperative programming languages”. PhD thesis. Imperial College London, UK, 2014. url: http: //hdl.handle.net/10044/1/13693.f [7] Faris Abou-Saleh and Dirk Pattinson. “Towards Effects in Mathematical Operational Semantics”. In: Electr. Notes Theor. Comput. Sci. 276 (2011), pp. 81–104. doi: 10.1016/j.entcs.2011.09.016. url: https://doi.org/10. 1016/j.entcs.2011.09.016. [8] Pieter Agten et al. “Secure Compilation to Modern Processors”. In: 25th IEEE Computer Security Foundations Symposium, CSF 2012, Cambridge, MA, USA, June 25-27, 2012. Ed. by Stephen Chong. IEEE Computer Society, 2012, pp. 171–185. isbn: 978-1-4673-1918-8. doi: 10.1109/CSF. 2012.12. url: https://doi.org/10.1109/CSF.2012.12. p // g/ / [9] Amal Ahmed and Matthias Blume. “An equivalence-preserving CPS trans- lation via multi-language semantics”. In: Proceeding of the 16th ACM SIG- PLAN international conference on Functional Programming, ICFP 2011, Tokyo, Japan, September 19-21, 2011. Ed. by Manuel M. T. Chakravarty, Zhenjiang Hu, and Olivier Danvy. ACM, 2011, pp. 431–444. isbn: 978-1- 4503-0865-6. doi: 10.1145/2034773.2034830. url: https://doi.org/10. 1145/2034773.2034830. [10] Amal Ahmed et al. “Secure Compilation (Dagstuhl Seminar 18201)”. In: Dagstuhl Reports 8.5 (2018). Ed. by Amal Ahmed et al., pp. 1–30. issn: 2192-5283. doi: 10.4230/DagRep.8.5.1. url: http://drops.dagstuhl.de/ opus/volltexte/2018/9891. A categorical approach to secure compilation 23 [11] Falk Bartels. On Generalised Coinduction and Probabilistic Specification Formats: Distributive Laws in Coalgebraic Modelling. 2004. [12] Filippo Bonchi et al. “Lax Bialgebras and Up-To Techniques for Weak Bisimulations”. In: 26th International Conference on Concurrency Theory, CONCUR 2015, Madrid, Spain, September 1.4, 2015. Ed. by Luca Aceto and David de Frutos-Escrig. Vol. 42. LIPIcs. Schloss Dagstuhl - Leibniz- Zentrum fuer Informatik, 2015, pp. 240–253. References isbn: 978-3-939897-91-0. doi: 10.4230/LIPIcs.CONCUR.2015.240. url: https://doi.org/10.4230/LIPIcs. CONCUR.2015.240. [13] Tomasz Brengos. “Weak bisimulation for coalgebras over order enriched monads”. In: Logical Methods in Computer Science 11.2 (2015). doi: 10. 2168/LMCS-11(2:14)2015. url: https://doi.org/10.2168/LMCS-11(2: 14)2015. [14] J. Robin B. Cockett. “Introduction to Distributive Categories”. In: Math- ematical Structures in Computer Science 3.3 (1993), pp. 277–307. doi: 10. 1017/S0960129500000232. url: https://doi.org/10.1017/S0960129500000232 [15] Victor Costan and Srinivas Devadas. “Intel SGX Explained”. In: IACR Cryptology ePrint Archive 2016 (2016), p. 86. url: http://eprint.iacr.org/ 2016/086. [16] Dominique Devriese, Marco Patrignani, and Frank Piessens. “Fully-abstract compilation by approximate back-translation”. In: Proceedings of the 43rd Annual ACM SIGPLAN-SIGACT Symposium on Principles of Program- ming Languages, POPL 2016, St. Petersburg, FL, USA, January 20 - 22, 2016. 2016, pp. 164–177. doi: 10.1145/2837614.2837618. url: https: //doi.org/10.1145/2837614.2837618. [17] Derek Dreyer, Amal Ahmed, and Lars Birkedal. “Logical Step-Indexed Logical Relations”. In: Logical Methods in Computer Science 7.2 (2011). doi: 10.2168/LMCS-7(2:16)2011. url: https://doi.org/10.2168/LMCS- 7(2:16)2011. [18] Cédric Fournet et al. “Fully abstract compilation to JavaScript”. In: The 40th Annual ACM SIGPLAN-SIGACT Symposium on Principles of Pro- gramming Languages, POPL ’13, Rome, Italy - January 23 - 25, 2013. 2013, pp. 371–384. doi: 10.1145/2429069.2429114. url: https://doi.org/ 10.1145/2429069.2429114. / [19] Jan Friso Groote and Frits W. Vaandrager. “Structured Operational Se- mantics and Bisimulation as a Congruence”. In: Inf. Comput. 100.2 (1992), pp. 202–260. doi: 10.1016/0890-5401(92)90013-6. url: https://doi.org/ 10.1016/0890-5401(92)90013-6. [20] Ichiro Hasuo, Bart Jacobs, and Ana Sokolova. “Generic Trace Semantics via Coinduction”. In: Logical Methods in Computer Science 3.4 (2007). doi: 10.2168/LMCS-3(4:11)2007. url: https://doi.org/10.2168/LMCS-3(4: 11)2007. ) [21] Intel 64 and IA-32 Architectures Software Developer’s Manual. Intel Cor- poration. 2016. url: https://www.intel.com/content/dam/www/public/ Tsampas et al. 24 us/en/documents/manuals/64-ia-32-architectures-software-developer- instruction-set-reference-manual-325383.pdf. us/en/documents/manuals/64-ia-32-architectures-software-developer- instruction-set-reference-manual-325383.pdf. [22] Bart Jacobs. Introduction to Coalgebra: Towards Mathematics of States and Observation. Vol. 59. Cambridge Tracts in Theoretical Computer Science. Cambridge University Press, 2016. isbn: 9781316823187. doi: 10.1017/ CBO9781316823187. url: https://doi.org/10.1017/CBO9781316823187.i [23] Bart Jacobs. “Parameters and Parametrization in Specification, Using Dis- tributive Categories”. In: Fundam. Inform. 24.3 (1995), pp. 209–250. doi: 10.3233/FI-1995-2431. url: https://doi.org/10.3233/FI-1995-2431. / // / / [24] Radha Jagadeesan et al. “Local Memory via Layout Randomization”. In: Proceedings of the 24th IEEE Computer Security Foundations Symposium, CSF 2011, Cernay-la-Ville, France, 27-29 June, 2011. IEEE Computer Society, 2011, pp. 161–174. isbn: 978-1-61284-644-6. doi: 10.1109/CSF. 2011.18. url: https://doi.org/10.1109/CSF.2011.18. // / / [25] Andrew Kennedy. “Securing the .NET programming model”. In: Theor. Comput. Sci. 364.3 (2006), pp. 311–317. doi: 10.1016/j.tcs.2006.08.014. url: https://doi.org/10.1016/j.tcs.2006.08.014. [26] Bartek Klin. References “Bialgebras for structural operational semantics: An intro- duction”. In: Theor. Comput. Sci. 412.38 (2011), pp. 5043–5069. doi: 10. 1016/j.tcs.2011.03.023. url: https://doi.org/10.1016/j.tcs.2011.03.023. [27] Bartek Klin and Beata Nachyla. “Presenting Morphisms of Distributive Laws”. In: 6th Conference on Algebra and Coalgebra in Computer Sci- ence, CALCO 2015, June 24-26, 2015, Nijmegen, The Netherlands. 2015, pp. 190–204. doi: 10.4230/LIPIcs.CALCO.2015.190. url: https://doi. org/10.4230/LIPIcs.CALCO.2015.190. [28] Marina Lenisa, John Power, and Hiroshi Watanabe. “Distributivity for endofunctors, pointed and co-pointed endofunctors, monads and comon- ads”. In: Electr. Notes Theor. Comput. Sci. 33 (2000), pp. 230–260. doi: 10.1016/S1571-0661(05)80350-0. url: https://doi.org/10.1016/S1571- 0661(05)80350-0. ( ) [29] Conor Mcbride. The Derivative of a Regular Type is its Type of One-Hole Contexts (Extended Abstract). 2001. [30] MIPS Architecture for Programmers Volume II-A: The MIPS32 Instruc- tion Set Manual. MIPS Technologies. 2016. url: https://s3-eu-west-1. amazonaws.com/downloads-mips/documents/MD00086-2B-MIPS32BIS- AFP-6.06.pdf. [31] James H. Morris. “Lambda-Calculus Models of Programming Languages”. PhD thesis. Massachusetts Institute of Technology, 1968. [32] Max S. New, William J. Bowman, and Amal Ahmed. “Fully abstract com- pilation via universal embedding”. In: Proceedings of the 21st ACM SIG- PLAN International Conference on Functional Programming, ICFP 2016, Nara, Japan, September 18-22, 2016. Ed. by Jacques Garrigue, Gabriele Keller, and Eijiro Sumii. ACM, 2016, pp. 103–116. isbn: 978-1-4503-4219- 3. doi: 10.1145/2951913.2951941. url: https://doi.org/10.1145/2951913. 2951941. A categorical approach to secure compilation 25 [33] Marco Patrignani, Amal Ahmed, and Dave Clarke. “Formal Approaches to Secure Compilation: A Survey of Fully Abstract Compilation and Related Work”. In: ACM Comput. Surv. 51.6 (Feb. 2019), 125:1–125:36. issn: 0360- 0300. doi: 10.1145/3280984. [34] Marco Patrignani, Dave Clarke, and Frank Piessens. “Secure Compilation of Object-Oriented Components to Protected Module Architectures”. In: Programming Languages and Systems - 11th Asian Symposium, APLAS 2013, Melbourne, VIC, Australia, December 9-11, 2013. Proceedings. Ed. by Chung-chieh Shan. Vol. 8301. Lecture Notes in Computer Science. Springer, 2013, pp. 176–191. isbn: 978-3-319-03541-3. doi: 10.1007/978-3- 319-03542-0\ 13. url: https://doi.org/10.1007/978-3-319-03542-0\ 13. [35] Marco Patrignani, Dominique Devriese, and Frank Piessens. “On Modular and Fully-Abstract Compilation”. In: IEEE 29th Computer Security Foun- dations Symposium, CSF 2016, Lisbon, Portugal, June 27 - July 1, 2016. IEEE Computer Society, 2016, pp. 17–30. isbn: 978-1-5090-2607-4. doi: 10.1109/CSF.2016.9. url: https://doi.org/10.1109/CSF.2016.9. [36] Marco Patrignani and Deepak Garg. “Robustly Safe Compilation”. In: Pro- gramming Languages and Systems - 28th European Symposium on Pro- gramming, ESOP 2019, Held as Part of the European Joint Conferences on Theory and Practice of Software, ETAPS 2019, Prague, Czech Repub- lic, April 6-11, 2019, Proceedings. References 2019, pp. 469–498. doi: 10.1007/978-3- 030-17184-1\_17. url: https://doi.org/10.1007/978-3-030-17184-1\_17. [37] Marco Patrignani et al. “Secure Compilation to Protected Module Archi- tectures”. In: ACM Trans. Program. Lang. Syst. 37.2 (2015), 6:1–6:50. doi: 10.1145/2699503. url: https://doi.org/10.1145/2699503. [38] Gordon D. Plotkin. “A structural approach to operational semantics”. In: J. Log. Algebr. Program. 60-61 (2004), pp. 17–139. [39] Andrei Popescu. “Weak Bisimilarity Coalgebraically”. In: Algebra and Coal- gebra in Computer Science, Third International Conference, CALCO 2009, Udine, Italy, September 7-10, 2009. Proceedings. 2009, pp. 157–172. doi: 10.1007/978-3-642-03741-2\_12. url: https://doi.org/10.1007/978-3- 642-03741-2\_12. [40] John Power and Hiroshi Watanabe. “Distributivity for a monad and a comonad”. In: Electr. Notes Theor. Comput. Sci. 19 (1999), p. 102. doi: 10.1016/S1571-0661(05)80271-3. url: https://doi.org/10.1016/S1571- 0661(05)80271-3. [41] Jurriaan Rot et al. “Enhanced coalgebraic bisimulation”. In: Mathematical Structures in Computer Science 27.7 (2017), pp. 1236–1264. doi: 10.1017/ S0960129515000523. url: https://doi.org/10.1017/S0960129515000523. [42] Jan Rothe and Dragan Masulovic. “Towards Weak Bisimulation For Coal- gebras”. In: Electr. Notes Theor. Comput. Sci. 68.1 (2002), pp. 32–46. doi: 10.1016/S1571-0661(04)80499-7. url: https://doi.org/10.1016/S1571- 0661(04)80499-7. Tsampas et al. 26 [43] Jan J. M. M. Rutten. “A note on coinduction and weak bisimilarity for while programs”. In: ITA 33.4/5 (1999), pp. 393–400. doi: 10.1051/ita: 1999125. url: https://doi.org/10.1051/ita:1999125. [44] Lau Skorstengaard, Dominique Devriese, and Lars Birkedal. “Reasoning About a Machine with Local Capabilities - Provably Safe Stack and Return Pointer Management”. In: Programming Languages and Systems - 27th European Symposium on Programming, ESOP 2018, Held as Part of the European Joint Conferences on Theory and Practice of Software, ETAPS 2018, Thessaloniki, Greece, April 14-20, 2018, Proceedings. Ed. by Amal Ahmed. Vol. 10801. Lecture Notes in Computer Science. Springer, 2018, pp. 475–501. isbn: 978-3-319-89883-4. doi: 10.1007/978- 3- 319- 89884- 1\_17. url: https://doi.org/10.1007/978-3-319-89884-1\_17. [45] Lau Skorstengaard, Dominique Devriese, and Lars Birkedal. “StkTokens: Enforcing Well-Bracketed Control Flow and Stack Encapsulation Using Linear Capabilities”. In: Proc. ACM Program. Lang. 3.POPL (Jan. 2019), 19:1–19:28. issn: 2475-1421. doi: 10.1145/3290332. [46] Stelios Tsampas, Dominique Devriese, and Frank Piessens. “Temporal Safety for Stack Allocated Memory on Capability Machines”. In: 32nd IEEE Com- puter Security Foundations Symposium, CSF 2019, Hoboken, NJ, USA, June 25-28, 2019. IEEE, 2019, pp. 243–255. isbn: 978-1-7281-1407-1. doi: 10.1109/CSF.2019.00024. url: https://doi.org/10.1109/CSF.2019.00024. [47] Daniele Turi. “Categorical Modelling of Structural Operational Rules: Case Studies”. In: Category Theory and Computer Science, 7th International Conference, CTCS ’97, Santa Margherita Ligure, Italy, September 4-6, 1997, Proceedings. 1997, pp. 127–146. doi: 10.1007/BFb0026985. url: https://doi.org/10.1007/BFb0026985. [48] Daniele Turi and Gordon D. Plotkin. References “Towards a Mathematical Opera- tional Semantics”. In: Proceedings, 12th Annual IEEE Symposium on Logic in Computer Science, Warsaw, Poland, June 29 - July 2, 1997. 1997, pp. 280–291. doi: 10.1109/LICS.1997.614955. url: https://doi.org/10. 1109/LICS.1997.614955. / [49] Thomas Van Strydonck, Frank Piessens, and Dominique Devriese. “Linear Capabilities for Fully Abstract Compilation of Separation-Logic-Verified Code”. In: Proc. ACM Program. Lang. ICFP (2019). accepted. ( ) [50] Hiroshi Watanabe. “Well-behaved Translations between Structural Oper- ational Semantics”. In: Electr. Notes Theor. Comput. Sci. 65.1 (2002), pp. 337–357. doi: 10.1016/S1571-0661(04)80372-4. url: https://doi.org/ 10.1016/S1571-0661(04)80372-4. [51] Robert N. M. Watson et al. “CHERI: A Hybrid Capability-System Ar- chitecture for Scalable Software Compartmentalization”. In: 2015 IEEE Symposium on Security and Privacy, SP 2015, San Jose, CA, USA, May 17-21, 2015. IEEE Computer Society, 2015, pp. 20–37. isbn: 978-1-4673- 6949-7. doi: 10.1109/SP.2015.9. url: https://doi.org/10.1109/SP.2015.9.
https://openalex.org/W4220994106	https://upcommons.upc.edu/bitstream/2117/367228/1/pathogens-11-00366%20%281%29.pdf	English	null	The Origin and Maintenance of Tuberculosis Is Explained by the Induction of Smear-Negative Disease in the Paleolithic	Pathogens	2,022	cc-by	15,828	pathogens pathogens pathogens Citation: Cardona, P.-J.; Català, M.; Prats, C. The Origin and Maintenance of Tuberculosis Is Explained by the Induction of Smear-Negative Disease in the Paleolithic. Pathogens 2022, 11, 366. https://doi.org/10.3390/ pathogens11030366 Keywords: Homo sapiens; Mycobacterium tuberculosis; disease spectrum; Paleolithic; Neolithic; demography; resistance; tolerance; chronicity; SEIR model; coinfection; coevolution; eco-immunology; mutualism; inequality; poverty Academic Editors: Delphi Chatterjee and Jordi B. Torrelles Article Pere-Joan Cardona 1,2,3,4,, Martí Català 5,6 and Clara Prats 5,6 Pere-Joan Cardona 1,2,3,4,, Martí Català 5,6 and Clara Prats 5,6 Pere-Joan Cardona 1,2,3,4,, Martí Català 5,6 and Clara Prats 5,6 1 Unitat de Tuberculosi Experimental, Germans Trias i Pujol Research Institute (IGTP), 08916 Badalona, Spain 2 Microbiology Department, North Metropolitan Clinical Laboratory, ‘Germans Trias i Pujol’ University 1 Unitat de Tuberculosi Experimental, Germans Trias i Pujol Research Institute (IGTP), 08916 Badalona, Spain 2 1 Unitat de Tuberculosi Experimental, Germans Trias i Pujol Research Institute (IGTP), 08916 Badalona, Spain 2 Microbiology Department, North Metropolitan Clinical Laboratory, ‘Germans Trias i Pujol’ University Hospital, 08916 Badalona, Spain 3 Genetics and Microbiology Department, Universitat Autònoma de Barcelona, 08916 Barcelona, Spain 4 Centro de Investigación Biomédica en Red de Enfermedades Respiratorias (CIBERES), 28029 Madrid, Spain 5 Comparative Medicine and Bioimage Centre of Catalonia (CMCiB), Germans Trias i Pujol Research Institute (IGTP), 08916 Badalona, Spain; mcatala@igtp.cat (M.C.); clara.prats@upc.edu (C.P.) 6 Departament de Física, Escola d’Enginyeria Agroalimentària i de Biosistemes de Barcelona, Universitat Politècnica de Catalunya (UPC)-BarcelonaTech, 08916 Badalona, Spain Correspondence: pcardonai.germanstrias@gencat.cat Abstract: Is it possible that the origin of Mycobacterium tuberculosis (Mtb) infection was around 70,000 years before the common era? At that time Homo sapiens was just another primate species with discrete growth and a very low-density geographic occupation. Therefore, it is difﬁcult to understand the origin of a highly virulent obligate human pathogen. We have designed a new SEIR model (TBSpectr) that allows the differentiation of smear-positive and -negative tuberculosis. The model reconciles currently accepted growth rates for the Middle Paleolithic (0.003%/year) and Neolithic (0.1%/year). The obtained data link the origin of Mtb infection in the Middle Paleolithic to the induction of smear-negative TB, and reveal that its persistence required interrelations among hunter–gatherer groups, while the risk of human extinction was negligible. It also highlights the number of people infected per case and the fast progression to disease for Mtb infection maintenance, as well as the link between poor health in the Neolithic with the increased incidence of more severe forms of TB (smear-positive). In conclusion, our data support the origin of TB as a well-tolerated, highly persistent disease, even in low-density populations, showing the difﬁculty of its eradication and highlighting the necessity for providing better health conditions to humans to reduce its severity. pathogens pathogens 1. Introduction Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- iations. Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- iations. There is a current hypothesis that establishes the origin of Mycobacterium tuberculosis (Mtb) infection in the middle Palaeolithic age, 70,000 years before the common era (BCE), as determined by molecular timing bases [1]. This hypothesis presents a great challenge, as it is difﬁcult to reconcile how a pathogen with such extraordinary virulence could coevolve with a host, Homo sapiens, which at that time represented a fragile animal species [2,3]. In fact, tuberculosis (TB) is the greatest killer of humankind. It has been estimated to have caused 1,000,000,000 deaths in the last 200 years [4]. This impact appears to be the ﬁnal phase of a formidable incidence, records of which ﬁrst started in Europe in the eighteenth century, and coinciding with the industrial revolution and the compilation of the ﬁrst consistent epidemiological records, with mortality incidences in big cities such as Stockholm, Hamburg and London peaking at around 900 deaths/100,000 inhabitants [5]. When the origin of this epidemic seemed to be uncertain [6], and it was thought to be the consequence of the sudden emergence of crowded cities [7], growing evidence began Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). https://www.mdpi.com/journal/pathogens Pathogens 2022, 11, 366. https://doi.org/10.3390/pathogens11030366 Pathogens 2022, 11, 366 2 of 16 to emerge that in reality it had always been with us [8]. The reason for the perception of this sudden gap in the European TB incidence may lie in the lack of availability of a precise means of diagnosis until the studies by René Laennec. His work was instrumental in correlating the pathology with its clinical or physical symptoms at the beginning of the nineteenth century [9]. Due to its condition as an obligate pathogen of Homo sapiens, Mtb has been evolving with humanity since its origin, and when we examine the historical references it appears to have adapted to all kinds of cultural changes, which have allowed its sustainable evolution despite its incredibly mortal capacity [8]. 1. Introduction ‘Ancient’ Mtb lineages appeared by 70,000 BCE, at a time when humans were organized into small tribes of hunter–gatherers of around 50 individuals living in Africa, with an effective population limited to a million people, who then started to expand towards the east in the process known as the second out-of- Africa [3,10,11]. This represents a very low population density [12], of around 30 people per 100 km2, which challenges the permanence of many infectious diseases, especially those that are human obligate parasites [10]. This low density and sustained ‘non-growing’ condition for hundreds of thousands of years, around 0.003% annually [13], was the consequence of a nomadic lifestyle where child-rearing carried a high cost, for both transportation and breeding, so it is estimated that each woman could only raise one child every 4 years [14]. Even when child mortality was high (27%), it was estimated that 46% of children died before the age of 15 [15]; these people had good quality of life, which resulted in a life expectancy of around 33 years [14,16], because once they reached the age of 15, 67% of them lived to an age of 45 or older [17]. Interestingly, modern Mtb lineages (2 to 4) appeared by 46,000 BCE, at the time of increasing expansion, progressively displacing the ancients ones (1, 5 and 6). In fact, nowadays ancient strains are detected in limited regional areas, i.e., West (lineages 5 and 6) and East Africa, the Philippines and Indian Ocean Rim (lineage 1) [1,18]. Modern lineages were less aggressive, as they were less proinﬂammatory, but having a greater capacity to disseminate through aerosols [19], and it is the one that ﬁnally became predominant in humans [20]. Mtb expansion was fuelled by the Neolithic revolution and the resulting explosive population growth, of around 0.1% annually, thanks to the progressive change towards sedentary life and faming-based activities, which led to higher birth rates (about one child raised every two years), but with a lower quality of life due to the impact of social inequalities [21,22], harder work duties and a less varied diet, which caused a reduction in life expectancy [16,23]. The greater dissemination capacity of modern lineages was markedly favored in the regions with a higher population growth. This can explain why the ancient ones can be. 1. Introduction Several models have been set up to try to improve understanding of the origin of Mtb infection in such a low-density population. One of the initial ones is based on the hypothesis that the mechanism of infection of Mtb was originally based on a late progression towards active disease (i.e., prolonged latency) of more than one generation, with the possibility that younger and more susceptible individuals became infected [24]. Adapting this criterion, Zheng et al. [25] built a model of TB transmission [26,27] using a population of 100 individuals. They concluded that to sustain TB, Mtb would have had to have a progression to disease of up to 50%, which clearly exceeds the value of 5–10% accepted nowadays [28]. Recently, we developed a susceptible–exposed–infectious–recovered (SEIR) model, which demonstrated the extraordinary impact of Mtb, causing the extinction of infected groups [19]. This could only be overcome by an unprecedented population increase attributable to annual population growth rates of 1% and 2.6% instead of the accepted 0.003% and 0.1% for the Paleolithic and Neolithic ages, respectively [13]. The study had major drawbacks, as it required a dramatic re-evaluation of the population growth parameters in prehistory. Recent data based on the precise determination of mortality and self-cure in the natural history of TB in the prechemotherapy era have obliged us to modify this model. In this study, the authors were able to better distinguish the prognosis of smear-positive (SP) and Pathogens 2022, 11, 366 3 of 16 3 of 16 smear-negative (SN) patients as a sign of TB severity [29]. This study shows a dramatic difference between the two forms and allows us to better explore the trade-offs that made this coevolution possible. Severe forms (SP) had an annual mortality ratio of 0.389 and self-cure ratio of 0.250, while mild forms (SN) had values of 0.025 and 0.125, respectively, reducing the mortality rate by 15-fold. Taking the TB disease spectrum into account, the impact of mild forms is paramount, as it drastically reduces mortality but maintains the possibility of disseminating the infection, albeit at a lower ratio. It is known that patients with SP have ten times higher levels of bacilli in their sputum than those with SN [30]. From this study, we hypothesized that SN patients might be the clue to reconciling the scenario of a lo- density ‘non-growing’ population with the origin and maintenance of TB. 1. Introduction The main objective of the present article was to evaluate the SN proportion (p) needed to maintain the consensual annual population growth rates established in Paleolithic and Neolithic societies, and also to determine the impacts of ancient and modern Mtb lineages. Thus, we built a new SEIR model (TBSpectr), in which we distinguished the two clinical forms and allocated them according to a smear-negative proportion ‘p’, depending on the health status of the host, which determined the induction of one clinical form or the other. Our work suggests that Mtb and modern humans were able to coevolve thanks to the presence of SN lesions, due to the better health status present before the Neolithic revolution. It was precisely the deteriorating health in that period that led to the increase in SP forms, even when the predominant Mtb lineages (modern ones) were less virulent. This highlights the character of TB as a poverty-related disease, and its greater impact on socially depressed sectors of the population. Additionally, the benign nature of its origin makes Mtb a highly human-adapted pathogen, which will be difﬁcult to eradicate using the current diagnostic methods. 2.1. Mtb Infection in the Paleolithic Was Possible Thanks to High Smear-Negative Proportion (p) Values The latent TB infection can become active through endogenous reactivation or exogenous reinfection. Patients with active TB can naturally recover (R), becoming non-infectious. We identiﬁed two categories in I and R, according to the spectrum of the disease, and we distinguished smear-negative (SN) and smear-positive (SP) forms of active TB. The driver for the evolution towards both forms of the disease depends on the value of a smear-negative proportion (p), which in turn depends on the health status of the host. By the same token, we have included a factor depending on the bacillary load (k) linked to smear-negative cases. Latent infected persons (I) can drain bacilli, lose immunity and become susceptible (S). Recovered persons can relapse to active TB through endogenous reactivation or exogenous reinfection. Factors included are: birth rate (π), transmission rate (β), drainage of infection (δ), fast progression (f), immunity (i), endogenous reactivation (a), exogenous reinfection (r), natural mortality (µ), smear-negative proportion (p), bacillary charge (k), reactivation factor in recovered (w), cure in smear-positive (csp) and smear-negative (csn), mortality caused by TB in smear- positive (µTBsp) and smear-negative (µTBsn), infectious smear-negative (Isn), infectious smear-positive (Isp), recovered smear-negative (Rsn) and recovered smear-positive (Rsp) compartments. Parameter Values Sources Paleolithic Neolithic Smear-negative proportion (p) 0.488 (A) 0.677 (M) 0.096 (A) 0.268 (M) explored Mortality/year caused by TB (µTB) 0.389 (SP)/0.025 (SN) [29] Infected people per case/year (e) A = 10 (SP)/1 (SN); M = 20 (SP)/2 (SN) [30,35] Bacillary charge (k) 0.1 [30] Fast progression (f) 0.099 (A)/0.0825 (M) 0.1238 (A)/0.1031 (M) [36] Reactivation from infection (a) f 0.3 Bacillary drainage and immunity reduction (δ) 0.1-a-r [37,38] Reduced progression due to immunity (i) 0.1 [39] TB natural cure (c) 0.231 (SP)/0.130 (SN) [29] Increased progression in recovered (w) 7 [40] SP: Smear positive; SN: Smear negative; A: Ancient strain; M: Modern strain Infected people per case/year (e) A = 10 (SP)/1 (SN); M = 20 (SP)/2 (SN) [30,35] Bacillary charge (k) 0.1 [30] Fast progression (f) 0.099 (A)/0.0825 (M) 0.1238 (A)/0.1031 (M) [36] Reactivation from infection (a) f 0.3 Bacillary drainage and immunity reduction (𝛿) 0.1-a-r [37,38] Reduced progression due to immunity (i) 0.1 [39] TB natural cure (c) 0.231 (SP)/0.130 (SN) [29] Increased progression in recovered (w) 7 [40] SP: Smear-positive; SN: Smear-negative; A: Ancient strain; M: Modern strain. Figure 1. TBSpectr model. Each compartment refers to the set of individuals by disease status: sus- ceptible, exposed, infected, recovered. 2.1. Mtb Infection in the Paleolithic Was Possible Thanks to High Smear-Negative Proportion (p) Values by TB (𝜇TB) 0.389 (SP)/0.025 (SN) [29] Infected people per case/year (e) A = 10 (SP)/1 (SN); M = 20 (SP)/2 (SN) [30,35] Bacillary charge (k) 0.1 [30] Fast progression (f) 0.099 (A)/0.0825 (M) 0.1238 (A)/0.1031 (M) [36] Reactivation from infection (a) f 0.3 Bacillary drainage and immunity reduction (𝛿) 0.1-a-r [37,38] Reduced progression due to immunity (i) 0.1 [39] TB natural cure (c) 0.231 (SP)/0.130 (SN) [29] Increased progression in recovered (w) 7 [40] SP: Smear-positive; SN: Smear-negative; A: Ancient strain; M: Modern strain. Figure 1. TBSpectr model. Each compartment refers to the set of individuals by disease status: sus- ceptible, exposed, infected, recovered. New-born individuals are assumed to be susceptible. A TB infection can remain latent (E) or can directly develop into infectious active TB (I). The latent TB Figure 1. TBSpectr model. Each compartment refers to the set of individuals by disease status: susceptible, exposed, infected, recovered. New-born individuals are assumed to be susceptible. A TB infection can remain latent (E) or can directly develop into infectious active TB (I). The latent Parameter Values Sources Paleolithic Neolithic Smear-negative proportion (p) 0.488 (A) 0.677 (M) 0.096 (A) 0.268 (M) explored Mortality/year caused by TB (µTB) 0.389 (SP)/0.025 (SN) [29] Infected people per case/year (e) A = 10 (SP)/1 (SN); M = 20 (SP)/2 (SN) [30,35] Bacillary charge (k) 0.1 [30] Fast progression (f) 0.099 (A)/0.0825 (M) 0.1238 (A)/0.1031 (M) [36] Reactivation from infection (a) f 0.3 Bacillary drainage and immunity reduction (δ) 0.1-a-r [37,38] Reduced progression due to immunity (i) 0.1 [39] TB natural cure (c) 0.231 (SP)/0.130 (SN) [29] Increased progression in recovered (w) 7 [40] SP: Smear-positive; SN: Smear-negative; A: Ancient strain; M: Modern strain. by TB (𝜇TB) Infected people per case/year (e) A = 10 (SP)/1 (SN); M = 20 (SP)/2 (SN) [30,35] Bacillary charge (k) 0.1 [30] Fast progression (f) 0.099 (A)/0.0825 (M) 0.1238 (A)/0.1031 (M) [36] Reactivation from infection (a) f 0.3 Bacillary drainage and immunity reduction (𝛿) 0.1-a-r [37,38] Reduced progression due to immunity (i) 0.1 [39] TB natural cure (c) 0.231 (SP)/0.130 (SN) [29] Increased progression in recovered (w) 7 [40] SP: Smear-positive; SN: Smear-negative; A: Ancient strain; M: Modern strain. 2.1. Mtb Infection in the Paleolithic Was Possible Thanks to High Smear-Negative Proportion (p) Values Parameter Values Sources Paleolithic Neolithic Smear-negative proportion (p) 0.488 (A) 0.677 (M) 0.096 (A) 0.268 (M) explored Mortality/year caused by TB (µTB) 0.389 (SP)/0.025 (SN) [29] Infected people per case/year (e) A = 10 (SP)/1 (SN); M = 20 (SP)/2 (SN) [30,35] Bacillary charge (k) 0.1 [30] Fast progression (f) 0.099 (A)/0.0825 (M) 0.1238 (A)/0.1031 (M) [36] Reactivation from infection (a) f 0.3 Bacillary drainage and immunity reduction (δ) 0.1-a-r [37,38] Reduced progression due to immunity (i) 0.1 [39] TB natural cure (c) 0.231 (SP)/0.130 (SN) [29] Increased progression in recovered (w) 7 [40] SP: Smear-positive; SN: Smear-negative; A: Ancient strain; M: Modern strain. y (𝜇) Infected people per case/year (e) A = 10 (SP)/1 (SN); M = 20 (SP)/2 (SN) [30,35] Bacillary charge (k) 0.1 [30] Fast progression (f) 0.099 (A)/0.0825 (M) 0.1238 (A)/0.1031 (M) [36] Reactivation from infection (a) f 0.3 Bacillary drainage and immunity reduction (𝛿) 0.1-a-r [37,38] Reduced progression due to immunity (i) 0.1 [39] TB natural cure (c) 0.231 (SP)/0.130 (SN) [29] Increased progression in recovered (w) 7 [40] SP: Smear-positive; SN: Smear-negative; A: Ancient strain; M: Modern strain. Figure 1. TBSpectr model. Each compartment refers to the set of individuals by disease status: sus- ceptible, exposed, infected, recovered. New-born individuals are assumed to be susceptible. A TB infection can remain latent (E) or can directly develop into infectious active TB (I). The latent TB infection can become active through endogenous reactivation or exogenous reinfection. Patients with active TB can naturally recover (R), becoming non-infectious. We identified two categories in I and R, according to the spectrum of the disease, and we distinguished smear-negative (SN) and smear-positive (SP) forms of active TB. The driver for the evolution towards both forms of the dis- ease depends on the value of a smear-negative proportion (p), which in turn depends on the health status of the host. By the same token, we have included a factor depending on the bacillary load (k) linked to smear-negative cases. Latent infected persons (I) can drain bacilli, lose immunity and be- come susceptible (S). Recovered persons can relapse to active TB through endogenous reactivation Figure 1. TBSpectr model. Each compartment refers to the set of individuals by disease status: susceptible, exposed, infected, recovered. New-born individuals are assumed to be susceptible. A TB infection can remain latent (E) or can directly develop into infectious active TB (I). 2.1. Mtb Infection in the Paleolithic Was Possible Thanks to High Smear-Negative Proportion (p) Values 2.1. Mtb Infection in the Paleolithic Was Possible Thanks to High Smear-Negative Proportion (p) Values Adjustments of the continuous TBSpectr model (Figure 1) to the currently accepted human population growth rates in the Paleolithic (0.0003%/year) and Neolithic (0.1%/year) (Figure 2) gave us the relation between natality (λ) and the smear-negative proportion (p), which determined the allocation towards SN TB forms. We considered the demographic parameters for each period of time and Mtb lineage (Table 1). Once obtained, we established the natality value (λ) according to the annual population growth rate that was able to ﬁt both Mtb lineages in the Paleolithic (0.032) and Neolithic (0.044) periods. This allowed us to determine the value for the smear-negative proportion (p) corresponding to each Mtb lineage for each cultural period. Thus, the p values were higher in the Paleolithic (0.488), corresponding to the better health status, which was then increased by the emergence of the modern lineages (0.677), which also corresponded to its lower virulence. The Neolithic period, due to the higher population growth, allowed the clinical forms to worsen dramatically according to the p values, corresponding to a poorer health status, which decreased in both modern and ancient lineages to 0.263 and 0.096, respectively. Table 1. Parameters and references. Parameter Values Sources Paleolithic Neolithic Annual population growth rate (gr) 0.003% 0.1% [13,31,32] Natality (λ) 0.032 0.044 15 years old surviving children/women 2.11 2.33 Natural mortality/year (µ) 1/33 1/26.5 [33,34] Table 1. Parameters and references. Pathogens 2022, 11, 366 4 of 16 Table 1. Cont. Mortality/yea by TB (𝜇 Table 1. Cont. Mortality/yea by TB (𝜇 Table 1. Cont. Parameter Values Sources Paleolithic Neolithic Smear-negative proportion (p) 0.488 (A) 0.677 (M) 0.096 (A) 0.268 (M) explored Mortality/year caused by TB (µTB) 0.389 (SP)/0.025 (SN) [29] Infected people per case/year (e) A = 10 (SP)/1 (SN); M = 20 (SP)/2 (SN) [30,35] Bacillary charge (k) 0.1 [30] Fast progression (f) 0.099 (A)/0.0825 (M) 0.1238 (A)/0.1031 (M) [36] Reactivation from infection (a) f 0.3 Bacillary drainage and immunity reduction (δ) 0.1-a-r [37,38] Reduced progression due to immunity (i) 0.1 [39] TB natural cure (c) 0.231 (SP)/0.130 (SN) [29] Increased progression in recovered (w) 7 [40] SP: Smear-positive; SN: Smear-negative; A: Ancient strain; M: Modern strain. 2.1. Mtb Infection in the Paleolithic Was Possible Thanks to High Smear-Negative Proportion (p) Values New-born individuals are assumed to be susceptible. A TB infection can remain latent (E) or can directly develop into infectious active TB (I). The latent TB infection can become active through endogenous reactivation or exogenous reinfection. Patients with active TB can naturally recover (R), becoming non-infectious. We identified two categories in I and R, according to the spectrum of the disease, and we distinguished smear-negative (SN) and smear-positive (SP) forms of active TB. The driver for the evolution towards both forms of the dis- ease depends on the value of a smear-negative proportion (p), which in turn depends on the health status of the host. By the same token, we have included a factor depending on the bacillary load (k) linked to smear-negative cases. Latent infected persons (I) can drain bacilli, lose immunity and be- come susceptible (S). Recovered persons can relapse to active TB through endogenous reactivation Figure 1. TBSpectr model. Each compartment refers to the set of individuals by disease status: susceptible, exposed, infected, recovered. New-born individuals are assumed to be susceptible. A TB infection can remain latent (E) or can directly develop into infectious active TB (I). The latent TB infection can become active through endogenous reactivation or exogenous reinfection. Patients with active TB can naturally recover (R), becoming non-infectious. We identiﬁed two categories in I and R, according to the spectrum of the disease, and we distinguished smear-negative (SN) and smear-positive (SP) forms of active TB. The driver for the evolution towards both forms of the disease depends on the value of a smear-negative proportion (p), which in turn depends on the health status of the host. By the same token, we have included a factor depending on the bacillary load (k) linked to smear-negative cases. Latent infected persons (I) can drain bacilli, lose immunity and become susceptible (S). Recovered persons can relapse to active TB through endogenous reactivation or exogenous reinfection. Factors included are: birth rate (π), transmission rate (β), drainage of infection (δ), fast progression (f), immunity (i), endogenous reactivation (a), exogenous reinfection (r), natural mortality (µ), smear-negative proportion (p), bacillary charge (k), reactivation factor in recovered (w), cure in smear-positive (csp) and smear-negative (csn), mortality caused by TB in smear- positive (µTBsp) and smear-negative (µTBsn), infectious smear-negative (Isn), infectious smear-positive (Isp), recovered smear-negative (Rsn) and recovered smear-positive (Rsp) compartments. Pathogens 2022, 11, 366 5 of 16 mear- ositive Figure 2. 2.1. Mtb Infection in the Paleolithic Was Possible Thanks to High Smear-Negative Proportion (p) Values Relation between the smear-negative proportion (p) and natality. Adjustment of natality (A) and effective natality is understood as the number of surviving 15-year-old children per woman (B) using the parameters of the TBSpectr model and the accepted growth population rates for the Paleolithic (0.0003%/year) and Neolithic (0.1%/year). Colored lines represent the values for the Paleolithic period and the infection with ancient (orange) and modern (violet) variants of Mtb, as well as the Neolithic period and the infection with ancient (blue) and Modern (yellow) variants of Mtb. Dots of corresponding colors show the values for the smear-negative proportion (p) chosen in each case. Figure 2. Relation between the smear-negative proportion (p) and natality. Adjustment of natality (A) and effective natality is understood as the number of surviving 15-year-old children per woman (B) using the parameters of the TBSpectr model and the accepted growth population rates for the Paleolithic (0.0003%/year) and Neolithic (0.1%/year). Colored lines represent the values for the Paleolithic period and the infection with ancient (orange) and modern (violet) variants of Mtb, as well as the Neolithic period and the infection with ancient (blue) and Modern (yellow) variants of Mtb. Dots of corresponding colors show the values for the smear-negative proportion (p) chosen in each case. Figure 2. Relation between the smear-negative proportion (p) and natality. Adjustment of natality (A) and effective natality is understood as the number of surviving 15-year-old children per woman (B) i h f h TBS d l d h d h l i f h Figure 2. Relation between the smear-negative proportion (p) and natality. Adjustment of natality (A) and effective natality is understood as the number of surviving 15-year-old children per woman Figure 2. Relation between the smear-negative proportion (p) and natality. Adjustment of natality (A) and effective natality is understood as the number of surviving 15-year-old children per woman (B) using the parameters of the TBSpectr model and the accepted growth population rates for the Paleolithic (0.0003%/year) and Neolithic (0.1%/year). Colored lines represent the values for the Paleolithic period and the infection with ancient (orange) and modern (violet) variants of Mtb, as well as the Neolithic period and the infection with ancient (blue) and Modern (yellow) variants of Mtb. Dots of corresponding colors show the values for the smear-negative proportion (p) chosen in each case. Figure 2. Relation between the smear-negative proportion (p) and natality. 2.1. Mtb Infection in the Paleolithic Was Possible Thanks to High Smear-Negative Proportion (p) Values Adjustment of natality (A) and effective natality is understood as the number of surviving 15-year-old children per woman (B) using the parameters of the TBSpectr model and the accepted growth population rates for the Paleolithic (0.0003%/year) and Neolithic (0.1%/year). Colored lines represent the values for the Paleolithic period and the infection with ancient (orange) and modern (violet) variants of Mtb, as well as the Neolithic period and the infection with ancient (blue) and Modern (yellow) variants of Mtb. Dots of corresponding colors show the values for the smear-negative proportion (p) chosen in each case. Paleolithic period and the infection with ancient (orange) and modern (violet) variants of Mtb, as well as the Neolithic period and the infection with ancient (blue) and Modern (yellow) variants of Mtb. Dots of corresponding colors show the values for the smear-negative proportion (p) chosen in each case. ( y ) ( y ) p Paleolithic period and the infection with ancient (orange) and modern (violet) variants of Mtb, as well as the Neolithic period and the infection with ancient (blue) and Modern (yellow) variants of Mtb. Dots of corresponding colors show the values for the smear-negative proportion (p) chosen in each case. 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate Figure 3 shows the dynamics of the different compartments studied according to the smear-negative proportion (p). It is important to note the parabolic evolution vs. the ex- We also looked at the number of children of 15 or older (Figure 2B), as the number of fertile individuals available is an interesting factor that contributes to population growth. In this case, the difference between the two periods of time was not great (2.11 vs. 2.33), indicating a higher mortality of children less than 15 years old in the Neolithic. ponential decline in SN and SP infectious cases, respectively. This indicates the need for the initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate ponential decline in SN and SP infectious cases, respectively. 2.1. Mtb Infection in the Paleolithic Was Possible Thanks to High Smear-Negative Proportion (p) Values This indicates the need for the initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate modern ones; otherwise they would contribute to decrease the p value, which would lead to a dramatic reduction in SN lesions, leading to the clearance of Mtb. Another inter- esting point is that the percentage of SP infectious cases relies mainly on the historical period studied, meaning that the lowering of health status in the Neolithic was responsi- ble for the marked increase in the severity of TB. The better health status is important in the Paleolithic when looking at the percentages of recovered and infectious SN cases, as it shows a wider gap in the Paleolithic than in the Neolithic. This means that even when ancient lineages were more virulent, because they had originated in the Paleolithic, they were well tolerated. On the other hand, modern lineages increased the percentages of peo- ple exposed, regardless of the historical period studied, confirming the greater capacity for dissemination. Figure 3 shows the dynamics of the different compartments studied according to the smear-negative proportion (p). It is important to note the parabolic evolution vs. the exponential decline in SN and SP infectious cases, respectively. This indicates the need for the initial Mtb strains, belonging to ancient lineages, to have greater virulence than the modern ones; otherwise they would contribute to decrease the ‘p’ value, which would lead to a dramatic reduction in SN lesions, leading to the clearance of Mtb. Another interesting point is that the percentage of SP infectious cases relies mainly on the historical period studied, meaning that the lowering of health status in the Neolithic was responsible for the marked increase in the severity of TB. The better health status is important in the Paleolithic when looking at the percentages of recovered and infectious SN cases, as it shows a wider gap in the Paleolithic than in the Neolithic. This means that even when ancient lineages were more virulent, because they had originated in the Paleolithic, they were well tolerated. On the other hand, modern lineages increased the percentages of people exposed, regardless of the historical period studied, conﬁrming the greater capacity for dissemination. Pathogens 2022, 11, 366 hogens 2022, 11, x FOR P 6 of 16 Figure 3. Evolution of the percentages of individuals in each SEIR compartment in relatio smear-negative proportion (p) according to the continuous TBSPectr model. ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate Pictures show t lution rates obtained for susceptible (A), exposed (B), infected SN (C), infected SP (D), re SN (E) and recovered SP (F) cases. Fractions are independent of initial conditions. Colore represent the values for the Paleolithic period and the infection with a ancient (orange) and (violet) variants of Mtb, as well as the Neolithic period and the infection with ancient (bl modern (yellow) variants of Mtb. Dots of each corresponding color show the value of the negative proportion (p) chosen in each case. Figure 3. Evolution of the percentages of individuals in each SEIR compartment in relation to the smear-negative proportion (p) according to the continuous TBSPectr model. Pictures show the evolution rates obtained for susceptible (A), exposed (B), infected SN (C), infected SP (D), recovered SN (E) and recovered SP (F) cases. Fractions are independent of initial conditions. Colored lines represent the values for the Paleolithic period and the infection with a ancient (orange) and modern (violet) variants of Mtb, as well as the Neolithic period and the infection with ancient (blue) and modern (yellow) variants of Mtb. Dots of each corresponding color show the value of the smear- negative proportion (p) chosen in each case. Figure 3. Evolution of the percentages of individuals in each SEIR compartment in relation smear-negative proportion (p) according to the continuous TBSPectr model. Pictures show th lution rates obtained for susceptible (A), exposed (B), infected SN (C), infected SP (D), rec SN (E) and recovered SP (F) cases. Fractions are independent of initial conditions. Colore represent the values for the Paleolithic period and the infection with a ancient (orange) and m (violet) variants of Mtb, as well as the Neolithic period and the infection with ancient (blu modern (yellow) variants of Mtb. Dots of each corresponding color show the value of the negative proportion (p) chosen in each case Figure 3. Evolution of the percentages of individuals in each SEIR compartment in relation to the smear-negative proportion (p) according to the continuous TBSPectr model. Pictures show the evolution rates obtained for susceptible (A), exposed (B), infected SN (C), infected SP (D), recovered SN (E) and recovered SP (F) cases. Fractions are independent of initial conditions. ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate For reference, there is a grey dotted horizontal line marking the presence of 1 person (100log10). Figure 4. Evolution of the population in the continuous TBSpectr model towards the stationary state. Pictures show the projections of 1000 simulations in a group of 100 people with a random initial distribution among compartments (S0 ∈[0, 100], E0 ∈[0, 40], Ic0, In0, Rc0, Rn0 ∈[0, 100−S0−E0], the sum of all compartments is equal to 100) during 1000 years of evolution. The thick lines correspond to the central scenario with S0 = 80, E0 = 20 and other compartments starting at zero. Evolution rates are drawn for the Paleolithic period and the infection with ancient (A) and modern (B) variants of Mtb, as well as the Neolithic period and the infection with ancient (C) and modern (D) variants of Mtb. Colour lines represent the different compartments of susceptible (blue), exposed (orange), infected SN (yellow), infected SP (violet), recovered SN (green) and recovered SP (cerulean blue). For reference, there is a grey dotted horizontal line marking the presence of 1 person (100log10). This scenario changes radically in the Neolithic period when SP infectious cases emerge dramatically, as shown in Figure 3. This is remarkable in the case of infections caused by ancient lineages. In the case of the modern ones, this emergence is influenced by the higher dissemination capacity that increases both SP and SN infectious cases to- gether with the exposed ones, causing for the first time a clear dip in the susceptible pop- ulation. Interestingly, at stationary equilibrium, both SN and SP infectious cases become similar. However, in the end the impact of Mtb infection only stops after 100 years, during the faster growth of the global population shown in the Neolithic, which can be seen by looking at the larger number of people in all compartments after the 1000-year period and explains the predominance of modern lineages (Figure S1). This scenario changes radically in the Neolithic period when SP infectious cases emerge dramatically, as shown in Figure 3. This is remarkable in the case of infections caused by ancient lineages. In the case of the modern ones, this emergence is inﬂuenced by the higher dissemination capacity that increases both SP and SN infectious cases together with the exposed ones, causing for the ﬁrst time a clear dip in the susceptible population. ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate The thick lines correspon the central scenario with S0 = 80, E0 = 20 and other compartments starting at zero. Evolution r are drawn for the Paleolithic period and the infection with ancient (A) and modern (B) variant Mtb, as well as the Neolithic period and the infection with ancient (C) and modern (D) variant Mtb. Colour lines represent the different compartments of susceptible (blue), exposed (orange) fected SN (yellow), infected SP (violet), recovered SN (green) and recovered SP (cerulean blue). reference there is a grey dotted horizontal line marking the presence of 1 person (100log10) Figure 4. Evolution of the population in the continuous TBSpectr model towards the stationary Pictures show the projections of 1000 simulations in a group of 100 people with a random i distribution among compartments (S0 ∈[0, 100], E0 ∈[0, 40], Ic0, In0, Rc0, Rn0 ∈[0, 100−S0−E0 sum of all compartments is equal to 100) during 1000 years of evolution. The thick lines corres to the central scenario with S0 = 80, E0 = 20 and other compartments starting at zero. Evolution are drawn for the Paleolithic period and the infection with ancient (A) and modern (B) varian Mtb, as well as the Neolithic period and the infection with ancient (C) and modern (D) var of Mtb. Colour lines represent the different compartments of susceptible (blue), exposed (ora infected SN (yellow), infected SP (violet), recovered SN (green) and recovered SP (cerulean blue 0 Figure 4. Evolution of the population in the continuous TBSpectr model towards the stationary state. Pictures show the projections of 1000 simulations in a group of 100 people with a random initial distribution among compartments (S0 ∈ [0, 100], E0 ∈ [0, 40], Ic0, In0, Rc0, Rn0 ∈ [0, 100−S0−E0], the sum of all compartments is equal to 100) during 1000 years of evolution. The thick lines correspond to the central scenario with S0 = 80, E0 = 20 and other compartments starting at zero. Evolution rates are drawn for the Paleolithic period and the infection with ancient (A) and modern (B) variants of Mtb, as well as the Neolithic period and the infection with ancient (C) and modern (D) variants of Mtb. Colour lines represent the different compartments of susceptible (blue), exposed (orange), in- fected SN (yellow), infected SP (violet), recovered SN (green) and recovered SP (cerulean blue). ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate Colored lines represent the values for the Paleolithic period and the infection with a ancient (orange) and modern (violet) variants of Mtb, as well as the Neolithic period and the infection with ancient (blue) and modern (yellow) variants of Mtb. Dots of each corresponding color show the value of the smear- negative proportion (p) chosen in each case. 2.3. The SN Recovered Cases Were Crucial for Maintaining Mtb Infection in the Paleolithi 2.3. The SN Recovered Cases Were Crucial for Maintaining Mtb Infection in the Paleolithic While the Neolithic Era Was Marked by the Arrival of SP Lesions 2.3. The SN Recovered Cases Were Crucial for Maintaining Mtb Infection in the Paleolithi 2.3. The SN Recovered Cases Were Crucial for Maintaining Mtb Infection in the Paleolithic While the Neolithic Era Was Marked by the Arrival of SP Lesions While the Neolithic Era Was Marked by the Arrival of SP Lesions The evolution of the people in the different compartments over a thousand yea ing the continuous TBSpectr model to find the stationary distribution (Figure 4) s that the SN recovered compartment was the most important one for ensuring the p tence of Mtb infection in the Paleolithic, while also maintaining a reduced percenta infectious people with SN lesions. The impact of SP infectious cases was minimal i period. The emergence of modern lineages led to a slight increase in these values, m The evolution of the people in the different compartments over a thousand years using the continuous TBSpectr model to ﬁnd the stationary distribution (Figure 4) shows that the SN recovered compartment was the most important one for ensuring the persistence of Mtb infection in the Paleolithic, while also maintaining a reduced percentage of infectious people with SN lesions. The impact of SP infectious cases was minimal in this period. The emergence of modern lineages led to a slight increase in these values, making the SP infec- tious values even more negligible, but notably increasing the number of people exposed, which increased to reach a similar percentage to that of the group of susceptible people. 7 of 16 o Pathogens 2022, 11, 366 g , , 7 of 16 Figure 4. Evolution of the population in the continuous TBSpectr model towards the stationary state. ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate Pictures show the projections of 1000 simulations in a group of 100 people with a random initial distribution among compartments (S0 ∈ [0, 100], E0 ∈ [0, 40], Ic0, In0, Rc0, Rn0 ∈ [0, 100−S0−E0], the sum of all compartments is equal to 100) during 1000 years of evolution. The thick lines correspond to the central scenario with S0 = 80, E0 = 20 and other compartments starting at zero. Evolution rates are drawn for the Paleolithic period and the infection with ancient (A) and modern (B) variants of Mtb, as well as the Neolithic period and the infection with ancient (C) and modern (D) variants of Mtb. Colour lines represent the different compartments of susceptible (blue), exposed (orange), in- fected SN (yellow), infected SP (violet), recovered SN (green) and recovered SP (cerulean blue). For reference, there is a grey dotted horizontal line marking the presence of 1 person (100log10). This scenario changes radically in the Neolithic period when SP infectious cases Figure 4. Evolution of the population in the continuous TBSpectr model towards the stationary state Pictures show the projections of 1000 simulations in a group of 100 people with a random initial distribution among compartments (S0 ∈[0, 100], E0 ∈[0, 40], Ic0, In0, Rc0, Rn0 ∈[0, 100−S0−E0], the sum of all compartments is equal to 100) during 1000 years of evolution. The thick lines correspond to the central scenario with S0 = 80, E0 = 20 and other compartments starting at zero. Evolution rates are drawn for the Paleolithic period and the infection with ancient (A) and modern (B) variants of Mtb, as well as the Neolithic period and the infection with ancient (C) and modern (D) variants of Mtb. Colour lines represent the different compartments of susceptible (blue), exposed (orange) infected SN (yellow), infected SP (violet), recovered SN (green) and recovered SP (cerulean blue). For reference, there is a grey dotted horizontal line marking the presence of 1 person (100log10). This scenario changes radically in the Neolithic period when SP infectious cases Figure 4. Evolution of the population in the continuous TBSpectr model towards the stationary st Pictures show the projections of 1000 simulations in a group of 100 people with a random in distribution among compartments (S0 ∈ [0, 100], E0 ∈ [0, 40], Ic0, In0, Rc0, Rn0 ∈ [0, 100−S0−E0], the s of all compartments is equal to 100) during 1000 years of evolution. ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate Interestingly, at stationary equilibrium, both SN and SP infectious cases become similar. However, in the end the impact of Mtb infection only stops after 100 years, during the faster growth of the global population shown in the Neolithic, which can be seen by looking at the larger number of people in all compartments after the 1000-year period and explains the predominance of modern lineages (Figure S1). Pathogens 2022, 11, 366 Pathogens 2022, 11, x FOR 8 of 16 8 of 17 2.4. Persistence of Mtb Infection Required Interrelations among Hunter–Gatherer Groups in the Paleolithic While the Risk of Human Extinction Was Negligible 2.4. Persistence of Mtb Infection Required Interrelations among Hunter–Gatherer Groups in the Paleolithic While the Risk of Human Extinction Was Negligible When examining the discrete TBSpectr model to evaluate the probability of Mtb infection clearance (Figure 5), it is clear that regardless of the lineage, its survival would be impossible in reduced human groups. Data show that a minimum of a 1000-person community was necessary to maintain this during the Paleolithic, and slightly less under the infection with modern lineages. This means that the groups of 50 hunter–gatherers had to interrelate, otherwise the infection would have disappeared. This factor was less important in the Neolithic due to the higher population growth rate. On the other hand, data on the capacity of Mtb infection to cause the extinction of humankind (Figure 6) show that this was very low, at roughly around 0.2% after 500 years of coevolution in a limited group of 50 people. When examining the discrete TBSpectr model to evaluate the probability of Mtb in- fection clearance (Figure 5), it is clear that regardless of the lineage, its survival would be impossible in reduced human groups. Data show that a minimum of a 1000-person com- munity was necessary to maintain this during the Paleolithic, and slightly less under the infection with modern lineages. This means that the groups of 50 hunter–gatherers had to interrelate, otherwise the infection would have disappeared. This factor was less im- portant in the Neolithic due to the higher population growth rate. On the other hand, data on the capacity of Mtb infection to cause the extinction of humankind (Figure 6) show that this was very low, at roughly around 0.2% after 500 years of coevolution in a limited group of 50 people. Figure 5. ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate We used 1000 simulations with random initial conditions, using parameters from Table 1. Figure 5. Relation between the size of the human group and the clearance of the Mtb infection. Heatmap of the end values for 1000 year’s evolution on the TBSpectr discrete model using the initial conditions in each compartment found in the equilibrium phase shown previously (Figure 4). For reference, we have included a vertical dotted red line at the time where there is a 100% clearance for the smallest population, a horizontal dotted red line to reference the evolution in a population size of 100 people and a white line at the population size where there is 0% clearance in the Paleolithic period under the infection of Mtb ancient lineages. Clearance means the lack of population in the Exposed, Infectious, and Recovered compartments. We used 1000 simulations with random initial conditions, using parameters from Table 1. Figure 5. Relation between the size of the human group and the clearance of the Mtb infection. Heatmap of the end values for 1000 year’s evolution on the TBSpectr discrete model using the initial conditions in each compartment found in the equilibrium phase shown previously (Figure 4). For reference, we have included a vertical dotted red line at the time where there is a 100% clearance for the smallest population, a horizontal dotted red line to reference the evolution in a population size of 100 people and a white line at the population size where there is 0% clearance in the Paleolithic period under the infection of Mtb ancient lineages. Clearance means the lack of population in the Exposed, Infectious, and Recovered compartments. We used 1000 simulations with random initial conditions, using parameters from Table 1. Figure 5. Relation between the size of the human group and the clearance of the Mtb infection. Heatmap of the end values for 1000 year’s evolution on the TBSpectr discrete model using the initial conditions in each compartment found in the equilibrium phase shown previously (Figure 4). For reference, we have included a vertical dotted red line at the time where there is a 100% clearance for the smallest population, a horizontal dotted red line to reference the evolution in a population size of 100 people and a white line at the population size where there is 0% clearance in the Paleolithic period under the infection of Mtb ancient lineages. ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate Relation between the size of the human group and the clearance of the Mtb infection. Heatmap of the end values for 1000 year’s evolution on the TBSpectr discrete model using the initial conditions in each compartment found in the equilibrium phase shown previously (Figure 4). For reference, we have included a vertical dotted red line at the time where there is a 100% clearance for the smallest population, a horizontal dotted red line to reference the evolution in a population size of 100 people and a white line at the population size where there is 0% clearance in the Paleolithic period under the infection of Mtb ancient lineages. Clearance means the lack of population in the Exposed, Infectious, and Recovered compartments. We used 1000 simulations with random initial conditions, using parameters from Table 1. Figure 5. Relation between the size of the human group and the clearance of the Mtb infection Heatmap of the end values for 1000 year’s evolution on the TBSpectr discrete model using the initial conditions in each compartment found in the equilibrium phase shown previously (Figure 4). For reference, we have included a vertical dotted red line at the time where there is a 100% clearance for the smallest population, a horizontal dotted red line to reference the evolution in a population size of 100 people and a white line at the population size where there is 0% clearance in the Paleolithic period under the infection of Mtb ancient lineages. Clearance means the lack of population in the Exposed, Infectious, and Recovered compartments. We used 1000 simulations with random initial conditions, using parameters from Table 1. Figure 5. Relation between the size of the human group and the clearance of the Mtb infection. Heatmap of the end values for 1000 year’s evolution on the TBSpectr discrete model using the initial conditions in each compartment found in the equilibrium phase shown previously (Figure 4). For reference, we have included a vertical dotted red line at the time where there is a 100% clearance for the smallest population, a horizontal dotted red line to reference the evolution in a population size of 100 people and a white line at the population size where there is 0% clearance in the Paleolithic period under the infection of Mtb ancient lineages. Clearance means the lack of population in the Exposed, Infectious, and Recovered compartments. ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate Clearance means the lack of population in the Exposed, Infectious, and Recovered compartments. We used 1000 simulations with random initial conditions, using parameters from Table 1. Pathogens 2022, 11, 366 Pathogens 2022, 11, x FOR 9 of 16 9 of 17 Figure 6. Relation of the size of the human group to the extinction of humankind. Heatmap of the end values for 1000 year’s evolution on the TBSpectr discrete model using the initial conditions in each compartment found in the equilibrium phase shown previously (Figure 4). Extinction means the disappearance of humankind in the group explored. Figure 6. Relation of the size of the human group to the extinction of humankind. Heatmap of the end values for 1000 year’s evolution on the TBSpectr discrete model using the initial conditions in each compartment found in the equilibrium phase shown previously (Figure 4). Extinction means the disappearance of humankind in the group explored. 2 5 Th N b f P l I f d P C d F P i Di A h M Figure 6. Relation of the size of the human group to the extinction of humankind. Heatmap of the end values for 1000 year’s evolution on the TBSpectr discrete model using the initial conditions in each compartment found in the equilibrium phase shown previously (Figure 4). Extinction means the disappearance of humankind in the group explored. Figure 6. Relation of the size of the human group to the extinction of humankind. Heatmap of the end values for 1000 year’s evolution on the TBSpectr discrete model using the initial conditions in each compartment found in the equilibrium phase shown previously (Figure 4). Extinction means the disappearance of humankind in the group explored. 2.5. The Number of People Infected Per Case and Fast Progression to Disease Are the Most Important Factors in Maintaining TB in Existence 2.5. The Number of People Infected Per Case and Fast Progression to Disease Are the Most Important Factors in Maintaining TB in Existence Important Factors in Maintaining TB in Existence A sensitivity analysis (Figure 7) using a range of values per parameter shown in Table 2 was used to analyse the influence of the different factors that define the TBSpectr model. ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate The most important ones are precisely those that we have seen as differentiating ancient and modern Mtb lineages, i.e., the people infected per case/year (e) and rapid progression towards active TB (f), as these are the determinants for increasing the numbers in all com- partments of infected people. In this regard, the immunity factor (i) seems to have the same influence, but in this case has to be read inversely, as the lower the value, the higher protection. It was of interest to confirm that the increase in the smear-negative proportion (p) increased the number of SN cases and reduced the SP ones, as expected. The reduction in infected cases by the higher mortality caused by TB (μTB,sp) was also expected, as it re- duces the chances of infection, thereby causing an increase in the susceptible compart- ment. The increase in natality (𝜆) linked to the increase in natural mortality (μ) was also expected, as it is precisely what happened in the Neolithic. It is interesting to note that the increase in the annual population growth rate (gr) reduced the number of exposed com- partments while decreasing the SP infectious one (Isp). Also noteworthy is the positive correlation between the cure of SP infectious people (csp) and the number of SN infectious people, showing an obvious interrelation between the two compartments. Finally, it is also interesting to observe the corroboration of how the increase in the charge value (k) A sensitivity analysis (Figure 7) using a range of values per parameter shown in Table 2 was used to analyse the inﬂuence of the different factors that deﬁne the TBSpectr model. The most important ones are precisely those that we have seen as differentiating ancient and modern Mtb lineages, i.e., the people infected per case/year (e) and rapid progression towards active TB (f), as these are the determinants for increasing the numbers in all compartments of infected people. In this regard, the immunity factor (i) seems to have the same inﬂuence, but in this case has to be read inversely, as the lower the value, the higher protection. It was of interest to conﬁrm that the increase in the smear-negative proportion (p) increased the number of SN cases and reduced the SP ones, as expected. ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate Parameter Paleolithic Value Neolithic Value Sensitivity Analysis Range % gr 0.003% 0.1% (0.003, 0.1) µ 0.03030 0.03846 (0.0286, 0.04) p 0.488 (A) 0.677 (M) 0.096 (A) 0.268 (M) (0, 1) µTB 0.389 (SP) 0.025 (SN) (0.02, 0.4) e A = 10 (SP) 1 (SN); M = 20 (SP) 2 (SN) (1, 20) f 0.099 (A) 0.0825 (M) 0.1238 (A) 0.1031 (M) (0, 0.13) i 0.1 (0.05, 0.5) c 0.231 (SP) 0.130 (SN) (0.1, 0.4) w 7 (1, 7) k 0.1 (0.05, 0.2) SP: Smear-positive; SN: Smear-negative; A: Ancient strain; M: Modern strain. We used a heatmap to show the Partial Rank Correlation Coefﬁcient on the TBSpectr discrete model after analyzing 1000 simulations. The inﬂuence of the evolution was shown in the relationships between the infected people per case/year (e), fast progression (f), immunity (i), bacillary charge (k), natural mortality (µ), reactivation factor in recovered Figure 7. Sensitivity analysis. We used a heatmap to show the Partial Rank Correlation Coefficient on the TBS discrete model after analyzing 1000 simulations. The influence of the evolution was sh in the relationships between the infected people per case/year (e), fast progression (f) it (i) b ill h (k) t l t lit ( ) ti ti f t i d Figure 7. Sensitivity analysis. Table 2. Sensitivity analysis. Parameter Paleolithic Value Neolithic Value Sensitivity Analysis Range % igure 7. Sensitivity analysis. Figure 7. Sensitivity analysis. We used a heatmap to Table 2. Sensitivity analysis. p p discrete model after analyzing 1000 simulations. The influence of the evolution was sh n the relationships between the infected people per case/year (e), fast progression (f) munity (i), bacillary charge (k), natural mortality (𝜇), reactivation factor in recovered nnual population growth rate (gr), cure in smear-positive (csp), mortality caused by 𝜇TBsp) and smear-negative proportion (p) in the evolution of clearance in groups of CL100) and 1000 (CL1000) people, extinction in groups of 100 (EX100) and 1000 (EX1 people and the equilibrium fractions of susceptible (S), exposed (E), infectious smear- tive (Isn), infectious smear-positive (Isp), recovered smear-negative (Rsn) and recov mear-positive (Rsp) compartments. Table 2. Sensitivity analysis. ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate Parameter Paleolithic Value Neolithic Value Sensitivity Analysis Range % gr 0.003% 0.1% (0.003, 0.1) µ 0.03030 0.03846 (0.0286, 0.04) p 0.488 (A) 0.677 (M) 0.096 (A) 0.268 (M) (0, 1) µTB 0.389 (SP) 0.025 (SN) (0.02, 0.4) e A = 10 (SP) 1 (SN); M = 20 (SP) 2 (SN) (1, 20) f 0.099 (A) 0.0825 (M) 0.1238 (A) 0.1031 (M) (0, 0.13) i 0.1 (0.05, 0.5) c 0.231 (SP) 0.130 (SN) (0.1, 0.4) w 7 (1, 7) k 0.1 (0.05, 0.2) SP: Smear-positive; SN: Smear-negative; A: Ancient strain; M: Modern strain. iscrete model after analyzing 1000 simulations. The influence of the evolution was sh n the relationships between the infected people per case/year (e), fast progression (f) Parameter Paleolithic Value Neolithic Value Sensitivity Analysis Range % Parameter Paleolithic Value Neolithic Value Sensitivity Analysis Range gr 0.003% 0.1% (0.003, 0.1) 𝜇 0.03030 0.03846 (0.0286, 0.04) p 0.488 (A) 0.677 (M) 0.096 (A) 0.268 (M) (0, 1) 𝜇TB 0.389 (SP) 0.025 (SN) (0.02, 0.4) e A = 10 (SP) 1 (SN); M = 20 (SP) 2 (SN) (1, 20) 𝑓 0.099 (A) 0.0825 (M) 0.1238 (A) 0.1031 (M) (0, 0.13) i 0.1 (0.05, 0.5) We used a heatmap to show the Partial Rank Correlation Coefﬁcient on the TBSpectr discrete model after analyzing 1000 simulations. The inﬂuence of the evolution was shown in the relationships between the infected people per case/year (e), fast progression (f), immunity (i), bacillary charge (k), natural mortality (µ), reactivation factor in recovered (w), annual population growth rate (gr), cure in smear-positive (csp), mortality caused by TB (µTBsp) and smear-negative proportion (p) in the evolution of clearance in groups of 100 (CL100) and 1000 (CL1000) people, extinction in groups of 100 (EX100) and 1000 (EX1000) people and the equilibrium fractions of susceptible (S), exposed (E), infectious smear-negative (Isn), infectious smear-positive (Isp), recovered smear-negative (Rsn) and recovered smear-positive (Rsp) compartments. ponential decline in SN and SP infectious cases, respectively. This indicates the need for he initial Mtb strains, belonging to ancient lineages, to have greater virulence than the d th i th ld t ib t t d th ‘ ’ l hi h ld 2.2. Ancient Lineages Had to Be More Virulent in the Paleolithic in Order to Sustain Mtb Infection but They Offered a Better Recovery Rate The reduction in infected cases by the higher mortality caused by TB (µTB,sp) was also expected, as it reduces the chances of infection, thereby causing an increase in the susceptible compartment. The increase in natality (λ) linked to the increase in natural mortality (µ) was also expected, as it is precisely what happened in the Neolithic. It is interesting to note that the increase in the annual population growth rate (gr) reduced the number of exposed compartments while decreasing the SP infectious one (Isp). Also noteworthy is the positive correlation between the cure of SP infectious people (csp) and the number of SN infectious people, showing an obvious interrelation between the two compartments. Finally, it is also interesting to observe the corroboration of how the increase in the charge value (k) raises the likelihood of SN infectious cases developing from recovered cases. 10 of 16 10 10 of 16 10 Pathogens 2022, 11, 366 gens 2022, 11, x FOR PE Figure 7. Sensitivity analysis. We used a heatmap to show the Partial Rank Correlation Coefficient on the TBSp discrete model after analyzing 1000 simulations. The influence of the evolution was sh in the relationships between the infected people per case/year (e), fast progression (f) munity (i), bacillary charge (k), natural mortality (𝜇), reactivation factor in recovered annual population growth rate (gr), cure in smear-positive (csp), mortality caused b (𝜇TBsp) and smear-negative proportion (p) in the evolution of clearance in groups o (CL100) and 1000 (CL1000) people, extinction in groups of 100 (EX100) and 1000 (EX1 people and the equilibrium fractions of susceptible (S), exposed (E), infectious smear- ative (Isn), infectious smear-positive (Isp), recovered smear-negative (Rsn) and recov smear-positive (Rsp) compartments. Table 2. Sensitivity analysis. Parameter Paleolithic Value Neolithic Value Sensitivity Analysis Rang gr 0.003% 0.1% (0.003, 0.1) 𝜇 0.03030 0.03846 (0.0286, 0.04) Figure 7. Sensitivity analysis. Table 2. Sensitivity analysis. c w 3. Discussion have theorized the origin of Mtb as a progressive increase in the hydrophobicity of the cell wall. Thus, the origin of Mtb complex would be M. kansasii, with a smooth morphology—a hydrophilic, environmental mycobacterium that acts as an opportunistic pathogen of several mammals, including humans. With the progressive loss of the polar sugars together with the acquisition of apolar ones, there was an evolution towards M. canettii and canetti/tuberculosis species to ﬁnally become the Mtb complex MRCA and a human obligate pathogen [47]. g p g Overall, this means that Mtb is the result of thousands of years of evolution of en- vironmental mycobacteria, living in cell-free media (water and dust) or colonizing free living amoebas [48,49], and moving to colonize the ‘pulmonary amoebas’, which we can consider the alveolar macrophages, to ﬁnally becoming an obligate parasite thanks to its capacity to disseminate through aerosols [47]. However, originally this parasitization had to be sustainable in the context of a low-density human population, based in small tribal groups with a necessary interrelation between them, as demonstrated by other authors [50], to avoid clearance of the infection. Our data indicate that even by developing mild SN TB forms with low dissemination capacity, the obligate parasitization was possible. In fact, recent data on molecular epidemiology support a greater impact of subclinical TB cases than expected [51], challenging the status of the current diagnostic methodology if we are to ﬁnally eradicate this infection. In fact, looking at the countries with the lowest TB incidence rates, it appears that there is a long-lasting persistent low incidence of the infection [52], which is usually attributed to imported cases, but that we might also attribute to the original nature of Mtb infection, which tends to generate mild SN TB forms. The persistence of Mtb in our tissues would also provide some evolutive advantage to humans, for instance by increasing trained immunity and allowing people to better control acute respiratory viral infections [53]. This is a sort of mutualism, an ‘old friends’ relationship that has also been linked to several colonizing microorganisms, including environmental mycobacteria, through the balance of our inﬂammatory responses, which has led to the evolution of the human immune system [54,55]. c w 3. Discussion w 7 (1, 7) k 0.1 (0.05, 0.2) P: Smear-positive; SN: Smear-negative; A: Ancient strain; M: Modern strain. Discussion The TBSpectr model is a very significant correction of our previous model [19], w nked the origin of TB to an unprecedent population increase (×20 times in 100 years The TBSpectr model is a very signiﬁcant correction of our previous model [19], which linked the origin of TB to an unprecedent population increase (×20 times in 100 years). In that case we based our assumptions on the data from a systematic review of the evolution of TB cases from the prechemotherapy era [41]. Recent data from Ragonnet et al. [29] led us to reconsider our previous work, as the authors were able to enrich that review and differentiate the evolution of SP and SN-TB separately. In our case, the model has a major ﬂaw, as it does not consider the interrelation between the two forms of TB, but it Pathogens 2022, 11, 366 11 of 16 11 of 16 better represents the TB spectrum to the point that it drastically changes our perception of its origin. better represents the TB spectrum to the point that it drastically changes our perception of its origin. g Following this change, our model makes an even stronger case for how coevolution between Mtb and humanity emerged around 70,000 years BCE [1]. This hypothesis is based on molecular clock calibrations that are constantly revisited [42]. Furthermore, other methodologies led to the hypothesis that its origin is linked with the control of ﬁre, at around 300,000 to 400,000 BCE [43]. In fact, there is one report that claims the presence of Leptomeningitis tuberculosa in the endocranial surface of a hominid fossil with an esti- mated age of around 500,000 BCE in Kocaba¸s (Turkey) [44]. Although the interpretation of this ﬁnding has been somewhat controversial [45], it may be hypothesized that before a complete settlement as a human obligate parasite, the Mycobacterium species ‘attempted’ several times to become the most recent common ancestor (MRCA) of the Mtb complex that we identify nowadays. In this sense, Gutierrez et al. identiﬁed a very ancient an- cestor (3,000,000 BCE) linked to the smooth tubercle bacilli (M. canettii) strains, which are still isolated from human TB today [46]. This position is supported by looking at the evolution of the lipid composition of the mycobacterial cell wall. In this regard, Jankute et al. c w 3. Discussion In this regard, our hypothesis on the origin of the expansion of SP TB forms is supported by the reduction in standard of living, which came with the Neolithic revolution and led to increased TB severity and mortality. This took place despite the lower virulence of the newly evolved modern lineages, which had a lower inﬂammatory capacity but also had higher dissemination capability [19], and would not take place if the humankind would keep the healthier Paleolithic standard of living. Our work sustains the hypothesis or idea that Mtb not only evolved to become a competent aerosol-disseminated pathogen among humans, but also gained the ability to remain sustainable among low-density populations. It was the human cultural changes in lifestyle with the Neolithic revolution that broke this balance and generated TB, the greatest killer of humankind. Pathogens 2022, 11, 366 12 of 16 12 of 16 Overall, our data support the concept that Mtb infection is highly adapted to persist and coevolve with humans thanks to its ability to cause long-lasting SN-TB. From a practical point of view, this means that its eradication would be very difﬁcult because of its capacity to remain among us discretely, challenging the current diagnostic tools. It also emphasizes the poverty-related nature of the disease and the need to provide better global health status to prevent severe forms of the disease in order to be able to reduce its terrible morbidity and mortality. The greatest strength of the model is the differentiation between SN and SP so that it is able to incorporate recent data on the determination of mortality and self-cure in the natural history of TB in the prechemotherapy era [29]. Nevertheless, the transition between SN and SP was not explored because of a lack of information about its rate, which is a limitation of the model that should be addressed in future revisions. Another limitation is that children under 15 years of age are not included in the model. Their inclusion could modify the outcome dynamics. Further work should also address the interactions between communities, which is a factor that could reveal important information concerning the maintenance of the infection. 4.1. TBSpectr Model 4.1. TBSpectr Model We designed a compartmental mathematical model based on a set of differential equations to describe the dynamics of the evolution of Mycobacterium tuberculosis complex p p p y The population is divided into several compartments according to their status with regards to the infection cycle: S, susceptible; E, non-infectious exposed; ISP, smear-positive infectious; ISN, smear-negative infectious; RSP, recovered from a smear-positive TB course; RSN, recovered from a smear-negative TB course. Newborn individuals appear at a rate π in the susceptible compartment. Mortality is given by µ when it is not caused by TB, µTB,SP when it is caused by SP tuberculosis and µTB,SN when it is caused by SN tuberculosis. Flows between compartments are shown in Figure 1 and given by this set of equations: dS dt = π + δE −µS −βIS (1) dE dt = (1 −f ) βSI −(µ + δ + i(a + rI))E (2) (1) dE dt = (1 −f ) βSI −(µ + δ + i(a + rI))E (2) dISP dt = f (1 −p)βSI + i(1 −p)(a + rI)E + (1 −p)wi(a + rI)R −(µ + µTB,SP + cSP)ISP (3) f pβSI + ip(a + rI)E + pwi(a + rI)R −(µ + µTB,SN + cSN)ISN dISN dt = f pβSI + ip(a + rI)E + pwi(a + rI)R −(µ + µTB,SN + cSN)ISN dRSP dt = cSPISP −(µ + wi(a + rI))RSP (5) dRSN dt = cSNISN −(µ + wi(a + rI))RSN (6) dN dt = dS dt + dE dt + dI dt + dR dt (7) (6) (7) where I = ISP + ISN and R = RSP + RSN. The different parameters that appear in the equations are the following: birth rate (π); percentage of bacillary drainage of the infection on exposed individuals (δ); non-TB-caused mortality rate per year (µ); TB transmission rate (β); percentage of fast progression of the active disease (f); reduced progression due to immunity (i); reactivation of an infection (a); risk of disease caused by a reinfection (r); TB-caused mortality rate (µTB,SP); increased progression in recovered in comparison to naïve (w); smear-negative proportion (p); natural cure rate of smear-negative (cSN); and Pathogens 2022, 11, 366 13 of 16 natural cure rate of smear-positive (cSP). Table 1 presents a summary of the parameters, the ranges of values explored and the sources, when available. The following paragraphs describe the relationship among some of them. 4.1. TBSpectr Model p g The transmission rate β depends on the number of new infections caused by a particu- lar SP or SN case (eSP and eSN), as well as on the proportion of each infectious compartment: β = 1 N eSPISP+eSN ISN ISP+ISN [27]. The probability of showing rapid progression of the disease dur- ing the ﬁrst year post-infection or reinfection is given by f [36,56], while the progression to active TB once TB infection remains latent (E) or active TB naturally recovers (R) is included with the parameter a, which represents 30% of the fast progression according to the data from Sloot et al. [36]. The parameter r provides the risk of disease caused by reinfection as r = f β + a(1 −f )β. This relationship considers that given a reinfection, its progression to active TB can be due both to fast progression (f) and reactivation (a) of the new infection [19]. We consider a protective ratio i of 0.1 (i.e., those infected people that do not develop the disease have at least 90% protection against the onset of disease) [39]. The drainage of the infection on exposed individuals is given by δ [37,38], which is assumed to be reduced by the possibility of endogenous or exogenous reactivation of the infection, deﬁned as a and r, respectively. Thus, bacillary drainage is deﬁned as: δ = 0.1 −a −r. Uys et al. [40] determined that recovered subjects have a seven-fold higher chance of developing disease, which is given by w in our model. The birth rate (π) and natality rate (λ) determine the newborn individuals π = λ(S + E + I + R). The natality parameter is adjusted using all other input parameters to ﬁt the annual population growth rate (gr). The number of births per fertile woman and year are computed as 2·µ−1·λ. We also included the k factor (0.1) to take into account the fact that SN-TB patients have ten times lower levels of bacilli in their sputum than SP patients, and that this has an impact on reducing the reactivation cases in recovered SN-TB [30]. The assumptions behind each of the equations and values have been discussed at length previously [19]. The main novelty of this update of the model is the distinction between SP and SN-TB. 4.1. TBSpectr Model In this regard, we introduced the smear-negative proportion p, which determines the percentage of active cases that show a smear-negative course. The annual spontaneous cure (cSP and cSN) and dying (µTB,SP and µTB,SN) rates depend on the SP or SN nature of the disease and are given by Ragonnet et al. [29]. We did not include a transition between SN and SP, as we considered only the ﬁnal clinical status and could not establish the rate of transition between them. 4.2. Assessment of Uncertainty and Sensitivity in the System The uncertainty and sensitivity analysis of the TBSpectr model was performed as described in [57]. We used a Latin hypercube sampling (LHS) technique to generate 1000 different parameter sets representative of the parameter space (sampling-based method). Parameters between the values shown in Table 2 were explored. The partial rank correlation coefﬁcient (PRCC) was computed at each time step for each of the parameters and suscepti- ble, exposed, infected, recovered and total populations, as were the annual incidence and death rates. We also computed the ﬁnal PRCC rates between input parameters and TB clearance and community extinction, using discrete resolution (see below). This analysis allowed for the assessment of the individual effect of each parameter on each outcome, with a linear ruling out of the effects of the uncertainty on the rest of the parameters. 4.3. Continuous and Discrete Resolution of the Models The continuous TBSpectr model was numerically integrated with MATLAB using the Euler method, with an integration step of 1/10 years. As a result of the integration, we obtained the dynamics of each of the model’s variables. y As we discussed in our previous paper [23], the limited size of some of the communities studied suggests the suitability of exploring a discrete resolution of the model, using natural numbers to describe the variable dynamics. With such discrete resolution, which was also based on the Euler integration method, we converted each of the ﬂows at each integration Pathogens 2022, 11, 366 14 of 16 14 of 16 step into a natural number using Poisson random distribution. This use of randomness on the ﬂow rounding makes it possible that two communities with the same initial conditions and model parameters diverge in their evolution. In particular, there is a chance that a certain community achieves clearance of the infection, while other communities remain affected by TB or can even become extinct because of the pathogen. step into a natural number using Poisson random distribution. This use of randomness on the ﬂow rounding makes it possible that two communities with the same initial conditions and model parameters diverge in their evolution. In particular, there is a chance that a certain community achieves clearance of the infection, while other communities remain affected by TB or can even become extinct because of the pathogen. Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/pathogens11030366/s1, Figure S1: Hypothetical evolution of the Mtb lineages in relation to the protective factor and natality. Author Contributions: Conceptualization, P.-J.C.; methodology, P.-J.C. and C.P.; software, P.-J.C., C.P. and M.C.; validation, P.-J.C., C.P. and M.C.; formal analysis, P.-J.C., C.P. and M.C.; investigation, P.-J.C., C.P. and M.C.; resources, P.-J.C., C.P. and M.C.; data curation, P.-J.C., C.P. and M.C.; writing—original draft preparation, P.-J.C.; writing—review and editing, P.-J.C.; visualization, P.-J.C.; supervision, P.-J.C.; project administration, P.-J.C.; funding acquisition, P.-J.C. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by “La Caixa” Foundation (ID 100010434), under agreement LCF/PR/GN16/10290002. Institutional Review Board Statement: Not applicable. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Informed Consent Statement: Not applicable. Acknowledgments: The project leading to these results received funding from “La Caixa” Foundation (ID 100010434), under agreement LCF/PR/GN16/10290002. 4.3. Continuous and Discrete Resolution of the Models We are grateful to Harvey Evans and Tom Yohanna for their precise corrections of the English text. Conﬂicts of Interest: The authors declare no potential conﬂict of interest. Conﬂicts of Interest: The authors declare no potential conﬂict of interest. References 1. Comas, I.; Coscolla, M.; Luo, T.; Borrell, S.; Holt, K.E.; Kato-Maeda, M.; Parkhill, J.; Malla, B.; Berg, S.; Thwaites, G.; et al. Out-of-Africa migration and Neolithic coexpansion of Mycobacterium tuberculosis with modern humans. Nat. Genet. 2013, 45, 1176–1182. [CrossRef] [PubMed] ] [ ] y, R.D.; Drummond, A.J. mtDNA variation predicts population size in humans and reveals a major Southern man prehistory. Mol. Biol. Evol. 2008, 25, 468–474. [CrossRef] [PubMed] 2. Atkinson, Q.D.; Gray, R.D.; Drummond, A.J. mtDNA variation predicts population size in humans a Asian chapter in human prehistory. Mol. Biol. Evol. 2008, 25, 468–474. [CrossRef] [PubMed] Asian chapter in human prehistory. Mol. Biol. Evol. 2008, 25, 468 474. [CrossRef] [PubMed] 3. Miller, E.F.; Manica, A.; Amos, W. Global demographic history of human populations inferred from whole mitochondrial genomes. R. Soc. Open Sci. 2018, 5, 180543. [CrossRef] [PubMed] 3. Miller, E.F.; Manica, A.; Amos, W. Global demographic history of human populations inferred from w R. Soc. Open Sci. 2018, 5, 180543. [CrossRef] [PubMed] 4. Paulson, T. Epidemiology: A mortal foe. Nature 2013, 502, S2–S3. [CrossRef] [PubMed] 5 St d W V i ti i l bilit t t b l i i A i t d R d l i f diff t t l id i ? I t J 4. Paulson, T. Epidemiology: A mortal foe. Nature 2013, 502, S2–S3. [CrossRef] [PubMed] , p gy , , [ ] [ ] 5. Stead, W. Variation in vulnerability to tuberculosis in America today: Random, or legacies of different ancestral epidemics? Int. J. Tuberc. Lung Dis. 2001, 5, 807–814. [PubMed] h h l b l d p gy 5. Stead, W. Variation in vulnerability to tuberculosis in America today: Random, or legacies of different ancestral epidemics? Int. J. Tuberc. Lung Dis. 2001, 5, 807–814. [PubMed] g 6. Dubos, R.; Dubos, J. The White Plague: Tuberculosis, Man, and Society; Rutgers University Press: New Brunswick, NJ, USA, 1987; 277p. Available online: https://books.google.es/books?id=DcgZbgSOPBAC&printsec=frontcover&hl=ca&source=gbs_ ViewAPI&redir_esc=y#v=onepage&q&f=false (accessed on 12 September 2019). 6. Dubos, R.; Dubos, J. The White Plague: Tuberculosis, Man, and Society; Rutgers University Press: New Brunswick, NJ, USA, 1987; 277p. Available online: https://books.google.es/books?id=DcgZbgSOPBAC&printsec=frontcover&hl=ca&source=gbs_ ViewAPI&redir_esc=y#v=onepage&q&f=false (accessed on 12 September 2019). y p g q p 7. Stead, W.W. The origin and erratic global spread of tuberculosis. How the past explains the present and is the key to the future. Clin. Chest Med. 1997, 18, 65–77. [CrossRef] 7. Stead, W.W. References Re 20. Brites, D.; Gagneux, S. Co-evolution of Mycobacterium tuberculosis and Homo sapiens. Immunol. Rev. 2015, 264, 6–24. [CrossRef] 21. Bentley, R.A.; Bickle, P.; Fibiger, L.; Nowell, G.M.; Dale, C.W.; Hedges, R.E.M.; Hamilton, J.; Wahl, J.; Francken, M.; Grupe, G.; et al. Community differentiation and kinship among Europe’s ﬁrst farmers. Proc. Natl. Acad. Sci. USA 2012, 109, 9326–9330. [CrossRef] 21. Bentley, R.A.; Bickle, P.; Fibiger, L.; Nowell, G.M.; Dale, C.W.; Hedges, R.E.M.; Hamilton, J.; Wahl, J.; Francken, M.; Grupe, G.; et al. Community differentiation and kinship among Europe’s ﬁrst farmers. Proc. Natl. Acad. Sci. USA 2012, 109, 9326–9330. [CrossRef] 22. Scheidel, W. The Great Leveler: Violence and the History of Inequality from the Stone Age to the Twenty-First Century; Princeton University Press: Princeton, NJ, USA, 2018; 504p. Available online: https://press.princeton.edu/books/paperback/9780691183 251/the-great-leveler (accessed on 25 January 2021). g y 23. Guzmán, R.A.; Weisdorf, J. The Neolithic Revolution from a price-theoretic perspective. J. Dev. Econ. 23. Guzmán, R.A.; Weisdorf, J. The Neolithic Revolution from a price-theoretic perspective. J. Dev. Econ. 2011, 96, 209–219. [CrossRef] 23. Guzmán, R.A.; Weisdorf, J. The Neolithic Revolution from a price-theoretic perspective. J. Dev. Econ. 2011, 96, 209–219. [CrossRef] 24. Blaser, M.J.; Kirschner, D. The equilibria that allow bacterial persistence in human hosts. Nature 2007, 449, 843–849. [CrossRef] h h l d l d l h l f h l l h l f b l i 23. Guzmán, R.A.; Weisdorf, J. The Neolithic Revolution from a price-theoretic perspective. J. Dev. Econ. 2011, 96, 209–219. [CrossRef] 24. Blaser, M.J.; Kirschner, D. The equilibria that allow bacterial persistence in human hosts. Nature 2007, 449, 843–849. [CrossRef] 25. Zheng, N.; Whalen, C.C.; Handel, A. Modeling the potential impact of host population survival on the evolution of M. tuberculosis latency. PLoS ONE 2014, 9, e105721. [CrossRef] [PubMed] , ; , J p p p J 24. Blaser, M.J.; Kirschner, D. The equilibria that allow bacterial persistence in human hosts. Nature 2007, 4 q p 25. Zheng, N.; Whalen, C.C.; Handel, A. Modeling the potential impact of host population survival on the evolution of M. tuberculosis latency. PLoS ONE 2014, 9, e105721. [CrossRef] [PubMed] 25. Zheng, N.; Whalen, C.C.; Handel, A. Modeling the potential impact of host population survival on th latency. PLoS ONE 2014, 9, e105721. [CrossRef] [PubMed] y 26. Blower, S.M.; McLean, A.R.; Porco, T.C.; Small, P.M.; Hopewell, P.C.; Sanchez, M.A.; Moss, A.R. The intrinsic transmission dynamics of tuberculosis epidemics. References Nat. Med. 1995, 1, 815–821. [CrossRef] y p 27. Porco, T.C.; Blower, S.M. Quantifying the intrinsic transmission dynamics of tuberculosis. Theor. P [CrossRef] [PubMed] 27. Porco, T.C.; Blower, S.M. Quantifying the intrinsic transmission dynamics of tuberculosis. Theor. Popul. Biol. 1998, 54, 117–132. [CrossRef] [PubMed] 28. Menzies, N.A.; Wolf, E.; Connors, D.; Bellerose, M.; Sbarra, A.N.; Cohen, T.; Hill, A.N.; Yaesoubi, R.; Galer, K.; White, P.J.; et al. Progression from latent infection to active disease in dynamic tuberculosis transmission models: A systematic review of the validity of modelling assumptions. Lancet Infect. Dis. 2018, 18, e228–e238. [CrossRef] y g p f 29. Ragonnet, R.; Flegg, J.A.; Brilleman, S.L.; Tiemersma, E.W.; Melsew, Y.A.; McBryde, E.S.; Trauer, J.M. Revisiting the Natural History of Pulmonary Tuberculosis: A Bayesian Estimation of Natural Recovery and Mortality Rates. Clin. Infect. Dis. 2021, 73, e88–e96. [CrossRef] [PubMed] 30. Palaci, M.; Dietze, R.; Hadad, D.J.; Ribeiro, F.K.C.; Peres, R.L.; Vinhas, S.A.; Maciel, E.L.N.; do Valle Dettoni, V.; Horter, L.; Boom, W.H.; et al. Cavitary disease and quantitative sputum bacillary load in cases of pulmonary tuberculosis. J. Clin. Microbiol. 2007, 45, 4064–4066. [CrossRef] [PubMed] 31. Gilligan, I. Climate, Clothing, and Agriculture in Prehistory; Cambridge University Press: Cambridge, UK, 2018. Available online: https://www.cambridge.org/core/product/identiﬁer/9781108555883/type/book (accessed on 12 September 2019). 32. Chamberlain, A.T. Demography in Archaeology; Cambridge University Press: Cambridge, UK, 2006. Available online: http: //ebooks.cambridge.org/ref/id/CBO9780511607165 (accessed on 15 September 2019). 33. Kaplan, H.; Hill, K.; Lancaster, J.; Hurtado, A.M. A theory of human life history evolution: Diet, intelligence, and longevity. Evol. Anthropol. Issues News Rev. 2000, 9, 156–185. [CrossRef] 34. Galor, O.; Moav, O. The Neolithic Revolution and Contemporary Variations in Life Expectancy. Working Papers Brown University, Department of Economics. 2007, Volume 14, pp. 1–39. Available online: http://www.ssrn.com/abstract=1012650 (accessed on 15 September 2019). [CrossRef] p 35. Stýblo, K.; Meijer, J.; Sutherland, I. Tuberculosis Surveillance Research Unit Report No. 1: The transmission of tubercle bacilli; its trend in a human population. Bull. Int. Union Tuberc. 1969, 42, 1–104. Available online: http://www.ncbi.nlm.nih.gov/pubmed/ 5004883 (accessed on 15 September 2019). 5004883 (accessed on 15 September 2019). 36. Sloot, R.; Schim van der Loeff, M.F.; Kouw, P.M.; Borgdorff, M.W. Risk of Tuberculosis after Recent Exposure. A 10-Year Follow-up Study of Contacts in Amsterdam. Am. J. Respir. Crit. Care Med. 2014, 190, 1044–1052. [CrossRef] 37. Cardona, P.-J. A dynamic reinfection hypothesis of latent tuberculosis infection. Infection 2009, 37, 80–8 37. Cardona, P.-J. References The origin and erratic global spread of tuberculosis. How the past explains the present and is the key to the future. Clin. Chest Med. 1997, 18, 65–77. [CrossRef] 8. Barberis, I.; Bragazzi, N.L.; Galluzzo, L.; Martini, M. The history of tuberculosis: From the ﬁrst histo of Koch’s bacillus. J. Prev. Med. Hyg. 2017, 58, E9–E12. [PubMed] 9. Roguin, A. Rene Theophile Hyacinthe Laënnec (1781–1826): The man behind the stethoscope. Clin. Med. Res. 2006, 4, 230–235. [CrossRef] [PubMed] 9. Roguin, A. Rene Theophile Hyacinthe Laënnec (1781–1826): The man behind the stethoscope. Clin. Med. Res. 2006, 4, 230–235. [CrossRef] [PubMed] 10. Henn, B.M.; Cavalli-Sforza, L.L.; Feldman, M.W. The great human expansion. Proc. Natl. Acad. Sci. USA 2012, 109, 17758–17764. [CrossRef] 11. Oppenheimer, S. Out-of-Africa, the peopling of continents and islands: Tracing uniparental gene trees across the map. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 2012, 367, 770–784. [CrossRef] 12. Tryon, C.A.; Faith, J.T. A demographic perspective on the Middle to Later Stone Age transition from Nasera rockshelter, Tanzania. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 2016, 371, 20150238. [CrossRef] [PubMed] 13. Cohen, M.N. The Food Crisis in Prehistory: Overpopulation and the Origins of Agriculture; Yale University Press: New Haven, CT, USA, 1977; 341p. Available online: https://yalebooks.yale.edu/book/9780300023510/food-crisis-prehistory (accessed on 12 September 2019). 15 of 16 15 of 16 Pathogens 2022, 11, 366 14. Johnson, A.W.; Earle, T.K. The Evolution of Human Societies: From Foraging Group to Agrarian State; Stanford University Press: Redwood, CA, USA, 1987; 360p. Available online: https://books.google.es/books/about/The_Evolution_of_Human_Societies. html?id=m7eGx61SSioC&redir_esc=y (accessed on 12 September 2019). p p g g html?id=m7eGx61SSioC&redir_esc=y (accessed on 12 September 2019). html?id=m7eGx61SSioC&redir_esc=y (accessed on 15. Volk, A.A.; Atkinson, J.A. Infant and child death in the human environment of evolutionary adaptation 34, 182–192. [CrossRef] [ ] 16. Armelagos, G.J.; Goodman, A.H.; Jacobs, K.H. The origins of agriculture: Population growth during a period of declining health. Popul. Environ. 1991, 13, 9–22. [CrossRef] p 17. Gurven, M.; Kaplan, H. Longevity Among Hunter- Gatherers: A Cross-Cultural Examination. Popul. Dev. Rev. 2007, 33, 321–365. [CrossRef] [ ] 18. Coscolla, M.; Gagneux, S. Consequences of genomic diversity in Mycobacterium tuberculosis. Semin. Immunol. 2014, 26, 431–444. [CrossRef] [PubMed] [ ] [ ] 19. Cardona, P.-J.; Català, M.; Prats, C. Origin of tuberculosis in the Paleolithic predicts unprecedented population growth and female resistance. Sci. Rep. 2020, 10, 42. [CrossRef] [PubMed] p 20. Brites, D.; Gagneux, S. Co-evolution of Mycobacterium tuberculosis and Homo sapiens. Immunol. References C.; Wu, H.H.T.; Ridell, M.; Garton, N.J.; Barer, M.R.; Minnikin, D.E.; Bhatt, A.; Besra, G.S. The role of hydrophobicity in tuberculosis evolution and pathogenicity. Sci. Rep. 2017, 7, 1315. [CrossRef] [PubMed] Jankute, M.; Nataraj, V.; Lee, O.Y. C.; Wu, H.H.T.; Ridell, M.; Garton, N.J.; Barer, M.R.; Minnikin, D.E.; Bha role of hydrophobicity in tuberculosis evolution and pathogenicity. Sci. Rep. 2017, 7, 1315. [CrossRef] [Pub en Salah, S.; Khlif, M.; Raoult, D.; Drancourt, M. Survival of environmental mycobacteria in Acanthamo 48. Adékambi, T.; Ben Salah, S.; Khlif, M.; Raoult, D.; Drancourt, M. Survival of environm polyphaga. Appl. Environ. Microbiol. 2006, 72, 5974–5981. [CrossRef] [PubMed] p yp g pp 49. Mba Medie, F.; Ben Salah, I.; Henrissat, B.; Raoult, D.; Drancourt, M. Mycobacterium tuberculosis Complex Mycobacteria as Amoeba-Resistant Organisms. PLoS ONE 2011, 6, e20499. [CrossRef] [PubMed] g 50. McGrath, J.W. Social networks of disease spread in the lower illinois valley: A simulation approach. Am. J. Phys. Anthropol. 1988, 77, 483–496. [CrossRef] [PubMed] -Muñoz, I.; Torres-Puente, M.; Villamayor, L.M.; Borrás, R.; Borrás-Máñez, M.; Bosque, M.; Camarena, 51. Xu, Y.; Cancino-Muñoz, I.; Torres-Puente, M.; Villamayor, L.M.; Borrás, R.; Borrás-Máñez, M.; Bosque, M.; Camarena, J.J.; Colomer-Roig, E.; Colomina, J.; et al. High-resolution mapping of tuberculosis transmission: Whole genome sequencing and phylogenetic modelling of a cohort from Valencia Region, Spain. PLoS Med. 2019, 16, e1002961. [CrossRef] [PubMed] p y g g g p 52. Geiter, L. Ending Neglect: The Elimination of Tuberculosis in the United States; National Academies Press: Washington, DC, USA, 2000. Available online: http://www.ncbi.nlm.nih.gov/pubmed/25077271 (accessed on 1 April 2021). p g p p 53. Netea, M.G.; van der Meer, J.W.M. Trained Immunity: An Ancient Way of Remembering. Cell Host Microbe 2017, 21, 297–300. [CrossRef] [PubMed] 54. Jackson, J.A.; Friberg, I.M.; Little, S.; Bradley, J.E. Review series on helminths, immune modulation and the hygiene hypoth- esis: Immunity against helminths and immunological phenomena in modern human populations: Coevolutionary legacies? Immunology 2009, 126, 18–27. [CrossRef] [PubMed] 55. Rook, G.A. Regulation of the immune system by biodiversity from the natural environment: An ecosystem service essential to health. Proc. Natl. Acad. Sci. USA 2013, 110, 18360–18367. [CrossRef] 56. Trauer, J.M.; Moyo, N.; Tay, E.-L.; Dale, K.; Ragonnet, R.; McBryde, E.S.; Denholm, J.T. Risk of Active Tuberculosis in the Five Years Following Infection . . . 15%? Chest 2016, 149, 516–525. [CrossRef] [PubMed] 57. Marino, S.; Hogue, I.B.; Ray, C.J.; Kirschner, D.E. References A dynamic reinfection hypothesis of latent tuberculosis infection. Infection 2009, 37, 80–86. [CrossRef] [PubMed] 38. Menzies, D. What Does Tuberculin Reactivity after Bacille Calmette-Guérin Vaccination Tell Us? Clin. Infect. Dis. 2000, 31, S71–S74. [CrossRef] [PubMed] 38. Menzies, D. What Does Tuberculin Reactivity after Bacille Calmette-Guérin Vaccination Tell Us? Clin. Infect. Dis. 2000, 31, S71–S74. [CrossRef] [PubMed] 39. Heimbeck, J. Incidence of tuberculosis in young adult women, with special reference to employment. Br. J. Tuberc. 1938, 32, 154–166. [CrossRef] 40. Uys, P.; Brand, H.; Warren, R.; van der Spuy, G.; Hoal, E.G.; van Helden, P.D. The Risk of Tuberculosis Reinfection Soon after Cure of a First Disease Episode Is Extremely High in a Hyperendemic Community. PLoS ONE 2015, 10, e0144487. [CrossRef] [PubMed] Pathogens 2022, 11, 366 16 of 16 41. Tiemersma, E.W.; van der Werf, M.J.; Borgdorff, M.W.; Williams, B.G.; Nagelkerke, N.J.D. Natural History of Tuberculosis: Duration and Fatality of Untreated Pulmonary Tuberculosis in HIV Negative Patients: A Systematic Review. PLoS ONE 2011, 6, e17601. [CrossRef] [PubMed] ] [ ] rdo, F.; Duchêne, S.; Brites, D.; Gagneux, S. The molecular clock of Mycobacterium tuberculosis. PLOS Pathog. 201 sRef] [PubMed] 43. Chisholm, R.H.; Trauer, J.M.; Curnoe, D.; Tanaka, M.M. Controlled ﬁre use in early humans might have triggered the evolutionary emergence of tuberculosis. Proc. Natl. Acad. Sci. USA 2016, 113, 9051–9056. [CrossRef] g 44. Kappelman, J.; Alçiçek, M.C.; Kazancı, N.; Schultz, M.; Özkul, M.; ¸Sen, ¸S. FirstHomo erectus from Turkey and implications for migrations into temperate Eurasia. Am. J. Phys. Anthropol. 2008, 135, 110–116. [CrossRef] g p y p 45. Roberts, C.A.; Pﬁster, L.-A.; Mays, S. Letter to the editor: Was tuberculosis present in Homo erectus in Turkey? Am. J. Phys. Anthropol. 2009, 139, 442–444. [CrossRef] p 46. Gutierrez, M.C.; Brisse, S.; Brosch, R.; Fabre, M.; Omaïs, B.; Marmiesse, M.; Supply, P.; Vincent, V. Ancient origin and gene mosaicism of the progenitor of Mycobacterium tuberculosis. PLoS Pathog. 2005, 1, e5. [CrossRef] p g y g 47. Jankute, M.; Nataraj, V.; Lee, O.Y.-C.; Wu, H.H.T.; Ridell, M.; Garton, N.J.; Barer, M.R.; Minnikin, D.E.; Bhatt, A.; Besra, G.S. The role of hydrophobicity in tuberculosis evolution and pathogenicity. Sci. Rep. 2017, 7, 1315. [CrossRef] [PubMed] 47. Jankute, M.; Nataraj, V.; Lee, O.Y.-C.; Wu, H.H.T.; Ridell, M.; Garton, N.J.; Barer, M.R.; Minnikin, D.E.; role of hydrophobicity in tuberculosis evolution and pathogenicity. Sci. Rep. 2017, 7, 1315. [CrossRef] [ 47. Jankute, M.; Nataraj, V.; Lee, O.Y. References A methodology for performing global uncertainty and sensitivity analysis in systems biology. J. Theor. Biol. 2008, 254, 178–196. [CrossRef]
https://openalex.org/W4291028696	https://zenodo.org/records/6984116/files/4314ijist13.pdf	English	null	ENHANCED GREEN FIREWALL FOR EFFICIENT DETECTION AND PREVENTION OF MOBILE INTRUDER USING GREYLISTING METHOD	Zenodo (CERN European Organization for Nuclear Research)	2,014	cc-by	4,640	International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 ABSTRACT: Wireless sensor networks are nowadays widely popular and has become an integral part in the military applications for human monitoring, thermal detection etc. Security of Wireless sensor network (WSN) becomes a very important issue with the rapid development of WSN that is vulnerable to a wide range of attacks such as sinkhole attacks due to deployment in the hostile environment and having limited resources. Intrusion detection system is one of the major and efficient defensive methods against attacks in WSN. One such detection technique is black listing technology. But using only Black listing technology is not suitable for a mobile intruder since it was designed considering only a static intruding node in a WSN. So it is necessary to build an energy efficient Intrusion detection system for sinkhole attack by a mobile intruder in WSN. We are intended to design an energy efficient system for detection of sinkhole and elimination of a mobile intruder from WSN nodes using a technology called greylisting. This technology uses pre alarm packets to warn the neighboring nodes about the intruder and the energy consumed by the pre alarm packets for making an alarm is much lesser than that of the packets used in black listing technology. Thus this method will serve as the solution for the dilemma in providing the security for WSN in sinkhole attack. G.Pradeep Kumar 1, R.Chakkaravarthy2,S.Arun kishorre3, L.S.Sathiyamurthy4 1- Assistant Professor, ECE dept., Velammal College of Engineering & Technology, Madurai. 2 Final year students, ECE dept., Velammal College of Engineering & Technology, Madurai. 2 Final year students, ECE dept., Velammal College of Engineering & Technology, Madurai. ENHANCED GREEN FIREWALL FOR EFFICIENT DETECTION AND PREVENTION OF MOBILE INTRUDER USING GREYLISTING METHOD G.Pradeep Kumar 1, R.Chakkaravarthy2,S.Arun kishorre3, L.S.Sathiyamurth 2.1 sinkhole attack definition and detection From the survey of Security Attacks in Wireless Sensor Networks done by Mr.ManishMPatel and Dr.AkshaiAggarwal Research Scholars in Gujarat Technological University, Sinkhole attacks typically work by making a compromised node look especially attractive to surrounding nodes with respect to the routing algorithm. Sinkhole attacks are difficult to counter because routing information supplied by a node is difficult to verify. As an example, a laptop-class adversary has a strong power radio transmitter that allows it to provide a high-quality route by transmitting with enough power to reach a wide area of the network. An approach to detect sinkhole attack using data consistency and network flow information is proposed in. It finds a list of suspected nodes and estimating the attacked area. Then using network flow graph, it effectively identifies the intruder in the list. Hop-count monitoring mechanism for detecting sinkhole attack is discussed in. Author has proposed Anomaly Detection System (ADS), which analyses the magnitude of hop- counts stored in a node’s routing table. Whenever any sensor node sends its message, all of four EM (Extra Monitor) nodes with high gain antenna will receive the message and RSSI value. If the destination of receive message is BS, then all of EM nodes will send RSSI value to the RSSI Based Sinkhole Detector to localize the position of the sender node. After that the visual geographic map will be updated. If the flow of receive message does not correspond with normal flow of visual geographic map, then sinkhole attack will be detected. By monitoring the CPU usage of each node in fixed time interval, the base station calculates the difference of CPU usage of each node. After comparing the difference with a threshold, the base station would identify whether a node is malicious or not. Proposed routing algorithm uses mobile agents to collect information of all mobile sensor nodes to make every node aware of the entire network so that a valid node will not listen the cheating information from malicious or compromised node. It does not need any encryption or decryption mechanism to detect the sinkhole attack. Whenever a node advertises, it finds the digest of the message using the MD5/SHA1 algorithm, and sends it along the original path [30]. 2. RELATED WORK 2. RELATED WORK 1. INTRODUCTION: Wireless Sensor Networks are highly distributed networks of small, lightweight wireless nodes, deployed in large numbers, monitors the environment or system by measuring physical parameters such as temperature, pressure, humidity. Any disturbance caused by a node that is non-member to a particular cluster of nodes is said to be an intrusion.An intrusion detection system (IDS) monitors network traffic and monitors for suspicious activity and alerts the cluster head about the intruder. There are many intrusion detection systems at present appealing 100% efficiency but the hard truth is that none of them could reach it. One of those steps towards a secure WSN is blacklisting technology. Blacklisting is the technology of maintaining a list of unauthorised nodes by the existing nodes in order to stop communication with those nodes. A node refers the blacklist when there is a need to communicate with a node or when there is a data transfer between those nodes. Though blacklisting is an efficient way of prevention of mobile intruder for maintaining the blacklist a lot of energy is consumed. This became a major drawback in the blacklisting technology. 99 DOI : 10.5121/ijist.2014.4313 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 In this paper we propose an enhanced green firewall using the greylisting technology for the energy efficient prevention of an intrusion by the mobile intruder. • Green firewall uses greylisting technology which has less energy consumption compared to blacklisting. • Enhanced green firewall checks and identifies the sinkhole attack and then blocks it using greylisting technology in an energy efficient manner. • Pre-alarm packets are used instead of alarm packets to intimate the nodes in a cluster about the intruder in the network • Pre-alarm packets are used instead of alarm packets to intimate the nodes in a cluster about the intruder in the network 2.1 sinkhole attack definition and detection At the same time, send the message to the advertising node, which will either keep the message as it is if it is a trustable node, or alter the message if it is a sinkhole. This advertising node should then generate the digest for the message it is going to transmit and send it forward. The destination detects the attack only when the digest obtained from both the paths are different. 100 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 3. ENHANCED GREEN FIREWALL Before getting into the concept of enhanced green firewall it is necessary to define the following terms which are necessary to explain about green firewall and the way to enhance it. The following terms are coined in terms of wireless sensor networks which may differ from the original meaning of the term. 2.2 Blacklisting From the paper Spectrum-Aware Wireless Sensor Networks done by Mr.Peng Du and Prof.George Roussos Dept. of Computer Science and Information Systems Birkbeck, University of London a blacklist consists of 16 bits that represent individual frequencies and channels are blacklisted by setting corresponding bits to 1. The sizing of blacklist controls the maximum number of channels allowed to exclude. This parameter can be either static or, alternatively, dynamic to cover any channel falling short of certain threshold, provided that at least one channel remains usable. The communication goes ahead if the prospective channel is not blacklisted; otherwise an alternative must be generated with Equation and checked again. This iterative process terminates when an admissible channel is found. Blacklists are periodically updated at intervals of Tu to reﬂect latest spectral condition. The synchronization of blacklists is crucial to maintaining communication between peers. In ADV slots, nodes insert their local blacklists to ADV payload and propagation is achieved simultaneously with standard TSCH timing synchronization. An important detail is that blacklisting is deactivated in ADV slots so that common hopping sequence can be easily recovered in case of desynchronized blacklists since ADV packets are always exchanged using default hopping sequence. Figure 1: Blacklisting the entire region Figure 1: Blacklisting the entire region 3.1 Blacklist Every node contains a blacklist, which come from decision node. After receiving the alarm packet from its decision node, the node transfers the node in the alarm packet into the blacklist if it already exists in the greylist or adds it into its blacklist. As a result, it is guaranteed that every node would be informed before the intruder moves to it. Every node would defend against the intruder before it begins to attack. To the nodes listed in the blacklist, every node should block and isolate the node in blacklist. 101 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 Figure2 : blacklisting the nearby nodes alone Figure2 : blacklisting the nearby nodes alone 3.3 Pre-alarm Packet It includes node ID and is only broad- casted by normal node. The node broadcast a pre-alarm packet when it put one node into blacklist. A pre-alarm packet cannot be forwarded and it is not propagated for more than one hop. Neighbor nodes can receive the pre-alarm packet and put the intruder into greylist. 3.4 Alarm Packet It include node ID and is only sent by decision node. Decision node regulates the pre-alarm list, which is used to count the pre-alarm packet from his cluster members. This list consists of two properties: node id and pre-alarm count. Once the decision node receives a pre-alarm packet from its cluster member, it adds the pre-alarm count of the corresponding node. When the pre-alarm count reaches the threshold, which is in proportion with the cluster size, it shows that an intruder has entered the cluster. The decision node conducts an alarm packet to its cluster members. Alarm packet is only sent by decision node and can not be forward by other node. 3.5 Model of Green Firewall 3.5 Model of Green Firewall 3.5 Model of Green Firewall When intrusion detection system ﬁnds an intruder in a WSN, it will inform every node in the WSN to isolate the intruder if the intruder is mobile. The method is to ﬂood the alarm packet to all nodes in WSNs. It leads every node in the WSN to receive and forward the alarm packet, which consumes a lot of energy in network. Green ﬁrewall is used to protect WSNs against attacks in the networks with less energy consumption. The features of WSNs are multi-hop and wireless communication. A node receives and sends packets through its neighbor nodes. A node does not have connection with the other nodes in the network, if it’s all neighbors cut off their links to it. The attacker has actually been isolated from the network even if it is still physically located in the network. The green ﬁrewall isolates the intruder by using the above principle. Green ﬁrewall need not broadcast the alarm packet to all nodes. Instead, it only broadcasts the alarm packet to the nodes which enclose the intruder. The method effectively reduces redundant alarm packets transmissions and meanwhile it decreases the energy consumption in WSNs When intrusion detection system ﬁnds an intruder in a WSN, it will inform every node in the WSN to isolate the intruder if the intruder is mobile. The method is to ﬂood the alarm packet to all nodes in WSNs. It leads every node in the WSN to receive and forward the alarm packet, which consumes a lot of energy in network. Green ﬁrewall is used to protect WSNs against attacks in the networks with less energy consumption. The features of WSNs are multi-hop and wireless communication. A node receives and sends packets through its neighbor nodes. A node does not have connection with the other nodes in the network, if it’s all neighbors cut off their links to it. The attacker has actually been isolated from the network even if it is still physically located in the network. The green ﬁrewall isolates the intruder by using the above principle. Green ﬁrewall need not broadcast the alarm packet to all nodes. Instead, it only broadcasts the alarm packet to the nodes which enclose the intruder. 3.1 Greylist Every node contains a greylist. Nodes in greylist come from pre-alarm packet broadcasted by other neighbors. A node put some node into its greylist when it receives node ID in pre-alarm packet from its neighbors. The node knows that the intruder in the greylist is close to itself, but the intruder may not enter its communication coverage. When the intruders enter its communication coverage, the node will transfer the intruder into blacklist and defence the intruder. In the meantime, the node broadcast pre-alarm packets include the intruder to its decision node and its neighbors. 102 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 Figure 3 : Greylisting by sending pre alarm packets Figure 3 : Greylisting by sending pre alarm packets Figure 3 : Greylisting by sending pre alarm packets Figure 3 : Greylisting by sending pre alarm packets Figure 3 : Greylisting by sending pre alarm packets Figure 3 : Greylisting by sending pre alarm packets 3.5 Model of Green Firewall 3.5 Model of Green Firewall 3.5 Model of Green Firewall The method effectively reduces redundant alarm packets transmissions and meanwhile it decreases the energy consumption in WSNs When intrusion detection system ﬁnds an intruder in a WSN, it will inform every node in the WSN to isolate the intruder if the intruder is mobile. The method is to ﬂood the alarm packet to all nodes in WSNs. It leads every node in the WSN to receive and forward the alarm packet, which consumes a lot of energy in network. Green ﬁrewall is used to protect WSNs against attacks in the networks with less energy consumption. The features of WSNs are multi-hop and wireless communication. A node receives and sends packets through its neighbor nodes. A node does not have connection with the other nodes in the network, if it’s all neighbors cut off their links to it. The attacker has actually been isolated from the network even if it is still physically located in the network. The green ﬁrewall isolates the intruder by using the above principle. Green ﬁrewall need not broadcast the alarm packet to all nodes. Instead, it only broadcasts the alarm packet to the nodes which enclose the intruder. The method effectively reduces redundant alarm packets transmissions and meanwhile it decreases the energy consumption in WSNs 4. IMPLEMENTATION 4. IMPLEMENTATION 4. IMPLEMENTATION We have implemented enhanced green firewall using JNS (Java Network Simulator). For a sample we have created 16 nodes and we have shown the sinkhole attack by a mobile intruder and the way it is detected and formation of a blacklist and also briefed about the formation of greylist and have compared the performance of greylist and blacklist. We have implemented enhanced green firewall using JNS (Java Network Simulator). For a sample we have created 16 nodes and we have shown the sinkhole attack by a mobile intruder and the way it is detected and formation of a blacklist and also briefed about the formation of greylist and have compared the performance of greylist and blacklist. We have implemented enhanced green firewall using JNS (Java Network Simulator). For a sample we have created 16 nodes and we have shown the sinkhole attack by a mobile intruder and the way it is detected and formation of a blacklist and also briefed about the formation of greylist and have compared the performance of greylist and blacklist. 4.2 Blacklisting and greylisting After the detection of sinkhole the node is blacklisted only by the nearby nodes and greylisted by all the other nodes where there is a possibility of the intruder may move next. Thus continuous monitoring of all nodes is avoided and energy consumption is saved. 4.1 Sinkhole detection 4.1 Sinkhole detection 4.1 Sinkhole detection We use the concept of overhearing for sinkhole attack detection. Initially when a mobile intruder comes in as a sinkhole it transmits a packet to the nearby node in order to get the packets naming itself as one of the nodes in that cluster. In such cases the node will inform this to the cluster head, the cluster head in reply sends a “hello” packet to every node in that cluster. The original nodes of the cluster will reply for the message to the cluster head but the sinkhole will drop that packet which is its normal behaviour. Thus the cluster head comes to know that node is malicious and sends information about that node to all other nodes about the intruder in form of sinkhole. We use the concept of overhearing for sinkhole attack detection. Initially when a mobile intruder comes in as a sinkhole it transmits a packet to the nearby node in order to get the packets naming itself as one of the nodes in that cluster. In such cases the node will inform this to the cluster head, the cluster head in reply sends a “hello” packet to every node in that cluster. The original nodes of the cluster will reply for the message to the cluster head but the sinkhole will drop that packet which is its normal behaviour. Thus the cluster head comes to know that node is malicious and sends information about that node to all other nodes about the intruder in form of sinkhole. We use the concept of overhearing for sinkhole attack detection. Initially when a mobile intruder comes in as a sinkhole it transmits a packet to the nearby node in order to get the packets naming itself as one of the nodes in that cluster. In such cases the node will inform this to the cluster head, the cluster head in reply sends a “hello” packet to every node in that cluster. The original nodes of the cluster will reply for the message to the cluster head but the sinkhole will drop that packet which is its normal behaviour. Thus the cluster head comes to know that node is malicious and sends information about that node to all other nodes about the intruder in form of sinkhole. 103 103 103 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 5. SIMULATION RESULTS The major check here is the energy that is being used to implement green firewall and existing blacklisting technology. In order to check the energy being used in the blacklisting and green firewall we have to have some assumptions as follows. Let us take the energy used for blacklisting as Eb = + {∑ } + ∑ + The energy used for green firewall as Eg, Where, Ei- energy used for initial blacklisting Ei- energy used for initial blacklisting Ean- energy used by all nodes Ean- energy used by all nodes Ebl- energy used after blacklisting Em- energy used for maintaining and checking nodes with blacklist Em- energy used for maintaining and checking nodes with blacklist Similarly we can calculate the energy used for Green firewall implementation, Similarly we can calculate the energy used for Green firewall implementation, CONCLUSION CONCLUSION CONCLUSION In this paper, we proposed an energy-efﬁcient intrusion prevention mechanism in WSNs called green ﬁrewall to isolate the intruder. It includes two kinds of list: greylist and blacklist, and two kinds of packets: pre-alarm packet and alarm packet. We design their local propagating mechanism of these information to reduce energy consumption. We conducted theoretical analysis and compare it with the traditional ﬂooding broadcast blacklist. To reveal the performance of green ﬁrewall, we also performed extensive simulation with ﬁve representative scenarios: static, crossing movement, short- distance movement, long-distance movement, and multipleintruders. The results show that the green ﬁrewall can provide low control overhead by reducing the number of alarm packet transmissions, green ﬁrewall can effectively reduce the energy consumption and make it an energy-efﬁcient intrusion prevention mechanism in WSNs. In this paper, we proposed an energy-efﬁcient intrusion prevention mechanism in WSNs called green ﬁrewall to isolate the intruder. It includes two kinds of list: greylist and blacklist, and two kinds of packets: pre-alarm packet and alarm packet. We design their local propagating mechanism of these information to reduce energy consumption. We conducted theoretical analysis and compare it with the traditional ﬂooding broadcast blacklist. To reveal the performance of green ﬁrewall, we also performed extensive simulation with ﬁve representative scenarios: static, crossing movement, short- distance movement, long-distance movement, and multipleintruders. The results show that the green ﬁrewall can provide low control overhead by reducing the number of alarm packet transmissions, green ﬁrewall can effectively reduce the energy consumption and make it an energy-efﬁcient intrusion prevention mechanism in WSNs. In this paper, we proposed an energy-efﬁcient intrusion prevention mechanism in WSNs called green ﬁrewall to isolate the intruder. It includes two kinds of list: greylist and blacklist, and two kinds of packets: pre-alarm packet and alarm packet. We design their local propagating mechanism of these information to reduce energy consumption. We conducted theoretical analysis and compare it with the traditional ﬂooding broadcast blacklist. To reveal the performance of green ﬁrewall, we also performed extensive simulation with ﬁve representative scenarios: static, crossing movement, short- distance movement, long-distance movement, and multipleintruders. The results show that the green ﬁrewall can provide low control overhead by reducing the number of alarm packet transmissions, green ﬁrewall can effectively reduce the energy consumption and make it an energy-efﬁcient intrusion prevention mechanism in WSNs. Eg= Ei+Esn+Egl Where, Figure4 :Step 1- sending hi packets to check for sinkhole Figure 5 :Step 2- sinkhole detection Figure 5 :Step 2- sinkhole Figure 5 :Step 2- sinkhole Figure4 :Step 1- sending hi packets to check for sinkhole detection Figure 5 :Step 2- sinkhole Figure4 :Step 1- sending hi packets to check for sinkhole detection 104 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 q y Figure 6: blacklisting by sending alarm packets Fig 7: greylisting by sending pre- alarm packets Energy comparison between enhanced green firewall and blacklisting q y Figure 6: blacklisting by sending alarm packets Fig 7: greylisting by sending pre- alarm packets Energy comparison between enhanced green firewall and blacklisting q y Figure 6: blacklisting by sending alarm packets Fig 7: greylisting by sending pre- alarm packets Energy comparison between enhanced green firewall and blacklisting Figure 6: blacklisting by sending alarm packets alarm packets Figure 6: blacklisting by sending alarm packets alarm packets Figure 6: blacklisting by sending alarm packets alarm packets Fig 7: greylisting by sending pre- Fig 7: greylisting by sending pre- Fig 7: greylisting by sending pre- Figure 6: blacklisting by sending alarm packets alarm packets Figure 6: blacklisting by sending alarm packets alarm packets Figure 6: blacklisting by sending alarm packets alarm packets Figure 6: blacklisting by sending alarm packets alarm packets Figure 6: blacklisting by sending alarm packets alarm packets Figure 6: blacklisting by sending alarm packets alarm packets Fig 7: greylisting by sending pre- Fig 7: greylisting by sending pre- Fig 7: greylisting by sending pre- Energy comparison between enhanced green firewall and blacklisting Energy comparison between enhanced green firewall and blacklisting Energy comparison between enhanced green firewall and blacklisting CONCLUSION 0 1000 2000 3000 4000 5000 CONCLUSION blacklisting greylisting for static greylisting for mobile CONCLUSION Eg= Ei+Esn+Egl Where, Esn- energy used by specific nodes for monitoring sn- energy used by specific nodes for monitoring Egl- energy used for maintain the greylist Thus from the above equations we can give dummy values and make an analysis by a chart between time and energy consumed as shown below. [3] A Non cryptographic method of sink hole attack detection in wireless sensor networks, D.Sheela, Naveen kumar. C and Dr. G.Mahadevan, 978-1-4577-0590-8/11/$26.00 ©2011 IEEE [3] A Non cryptographic method of sink hole attack detection in wireless sensor networks, D.Sheela, Naveen kumar. C and Dr. G.Mahadevan, 978-1-4577-0590-8/11/$26.00 ©2011 IEEE [3] A Non cryptographic method of sink hole attack detection in wireless sensor networks, D.Sheela, Naveen kumar. C and Dr. G.Mahadevan, 978-1-4577-0590-8/11/$26.00 ©2011 IEEE REFERENCES REFERENCES REFERENCES [1] Detection of sinkhole Attack in Wireless Sensor Networks using Message Digest Algorithms, S.Sharmila and DrGUmamaheswari978-1-61284-764-1/11/$26.00 ©2011 IEEE [1] Detection of sinkhole Attack in Wireless Sensor Networks using Message Digest Algorithms, S.Sharmila and DrGUmamaheswari978-1-61284-764-1/11/$26.00 ©2011 IEEE [1] Detection of sinkhole Attack in Wireless Sensor Networks using Message Digest Algorithms, S.Sharmila and DrGUmamaheswari978-1-61284-764-1/11/$26.00 ©2011 IEEE [1] Detection of sinkhole Attack in Wireless Sensor Networks using Message Digest Algorithms, S.Sharmila and DrGUmamaheswari978-1-61284-764-1/11/$26.00 ©2011 IEEE [1] Detection of sinkhole Attack in Wireless Sensor Networks using Message Digest Algorithms, S.Sharmila and DrGUmamaheswari978-1-61284-764-1/11/$26.00 ©2011 IEEE [1] Detection of sinkhole Attack in Wireless Sensor Networks using Message Digest Algorithms, S.Sharmila and DrGUmamaheswari978-1-61284-764-1/11/$26.00 ©2011 IEEE [2] A Sinkhole Attack Detection Scheme in Mintroute Wireless Sensor Networks, Murad A. Rassam, AnazidaZainal, Mohd. AizainiMaarof, and Mohammed Al-Shaboti, 2978-1-4673-4786-0/12/$31.00 ©2012 IEEE [2] A Sinkhole Attack Detection Scheme in Mintroute Wireless Sensor Networks, Murad A. Rassam, AnazidaZainal, Mohd. AizainiMaarof, and Mohammed Al-Shaboti, 2978-1-4673-4786-0/12/$31.00 ©2012 IEEE [2] A Sinkhole Attack Detection Scheme in Mintroute Wireless Sensor Networks, Murad A. Rassam, AnazidaZainal, Mohd. AizainiMaarof, and Mohammed Al-Shaboti, 2978-1-4673-4786-0/12/$31.00 ©2012 IEEE [2] A Sinkhole Attack Detection Scheme in Mintroute Wireless Sensor Networks, Murad A. Rassam, AnazidaZainal, Mohd. AizainiMaarof, and Mohammed Al-Shaboti, 2978-1-4673-4786-0/12/$31.00 ©2012 IEEE [2] A Sinkhole Attack Detection Scheme in Mintroute Wireless Sensor Networks, Murad A. Rassam, AnazidaZainal, Mohd. AizainiMaarof, and Mohammed Al-Shaboti, 2978-1-4673-4786-0/12/$31.00 ©2012 IEEE [2] A Sinkhole Attack Detection Scheme in Mintroute Wireless Sensor Networks, Murad A. Rassam, AnazidaZainal, Mohd. AizainiMaarof, and Mohammed Al-Shaboti, 2978-1-4673-4786-0/12/$31.00 ©2012 IEEE [3] A Non cryptographic method of sink hole attack detection in wireless sensor networks, D.Sheela, Naveen kumar. C and Dr. G.Mahadevan, 978-1-4577-0590-8/11/$26.00 ©2011 IEEE [3] A Non cryptographic method of sink hole attack detection in wireless sensor networks, D.Sheela, Naveen kumar. C and Dr. G.Mahadevan, 978-1-4577-0590-8/11/$26.00 ©2011 IEEE [3] A Non cryptographic method of sink hole attack detection in wireless sensor networks, D.Sheela, Naveen kumar. C and Dr. G.Mahadevan, 978-1-4577-0590-8/11/$26.00 ©2011 IEEE 105 105 105 International Journal of Information Sciences and Techniques (IJIST) Vol.4, No.3, May 2014 [4] Improved Two-factor User Authentication in Wireless Sensor Networks, BinodVaidya, DimitriosMakrakis, Hussein T. Mouftah, 978-1-4244-7742-5/10/$26.00 ©2010 IEEE [5] Node Localization in WSN Based on Weighted Vectors Centroid Algorithm,Xiaoqin Su, Zhaoming Lei, 978-0-7695-4543-1/11 $25.00 © 2011 IEEE [6] Security Attacks in Wireless Sensor Networks: A Survey, Mr. REFERENCES REFERENCES REFERENCES Manish M Patel , Dr.AkshaiAggarwal, 978-1-4799-0317-7/13/$31.00©2013 IEEE Spectrum-Aware Wireless Sensor Networks, Peng Du, prof.Georgebroussous, 978-1-4577-1348- 4/13/$31.00 ©2013 IEEE [8] AshfaqHussainFarooqi, FarrukhAslamKhan ,Jin Wang Sungyoung Lee (2012) A novel intrusion detection framework for wireless sensor networks, Pervasive Ubiquitous Computing DOI 10.1007/s00779- 012-0529-y, 2012. y [9] Y. Zhou, Y.G Fang, Y.C. Zhang, Securing Wireless Sensor Networks: a Survey, IEE Communications Surveys & Tutorials, 2008, Vol.10, No.8, pp.6-28 [10] Jinsong Wu, Honggang Hu, Murat Uysal, High-Rate Distributed Space- Time-Frequency Coding for Wireless Cooperative Networks, IEEE Trans- actions on Wireless Communications, 2011, Vol.10, No.2, pp.614-625 [11] K.Ren, W.Lou, K.Zeng, P.J.Moran, On Broadcast Authentication in Wire- less Sensor Networks,IEEE Transactions on Wireless Communications, 2007, Vol.6, No.11, pp.4136-4144 106
https://openalex.org/W2065794691	https://europepmc.org/articles/pmc4398432?pdf=render	English	null	Insect-Inspired Navigation Algorithm for an Aerial Agent Using Satellite Imagery	PloS one	2,015	cc-by	8,542	Introduction Natural selection has sculpted sensory systems and brains such that they can coordinate navi- gation tasks quickly, repeatedly, and with little error. Of particular interest are small-brained insects, which by virtue of their inherent parsimony may offer insights for understanding the first principles of navigational algorithms [1,2]. Deciphering the algorithms inside a navigating insect’s head might also bring societal benefits such as the development of smart autonomous vehicles and technology to augment the sensory and navigational abilities of visually impaired individuals. Data Availability Statement: All relevant data are within the paper. Insect-Inspired Navigation Algorithm for an Aerial Agent Using Satellite Imagery Douglas D. Gaffin1, Alexander Dewar2, Paul Graham2, Andrew Philippides3 1 Department of Biology, University of Oklahoma, Norman, Oklahoma, United States of America, 2 School of Life Sciences, University of Sussex, Brighton, United Kingdom, 3 Department of Informatics, University of Sussex, Brighton, United Kingdom ddgaffin@ou.edu * ddgaffin@ou.edu * ddgaffin@ou.edu Academic Editor: Eric James Warrant, Lund University, SWEDEN Academic Editor: Eric James Warrant, Lund University, SWEDEN , University, SWEDEN Received: August 8, 2014 Accepted: February 18, 2015 Published: April 15, 2015 Copyright: © 2015 Gaffin et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Received: August 8, 2014 Copyright: © 2015 Gaffin et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. RESEARCH ARTICLE OPEN ACCESS Citation: Gaffin DD, Dewar A, Graham P, Philippides A (2015) Insect-Inspired Navigation Algorithm for an Aerial Agent Using Satellite Imagery. PLoS ONE 10(4): e0122077. doi:10.1371/journal.pone.0122077 Academic Editor: Eric James Warrant, Lund University, SWEDEN Abstract Humans have long marveled at the ability of animals to navigate swiftly, accurately, and across long distances. Many mechanisms have been proposed for how animals acquire, store, and retrace learned routes, yet many of these hypotheses appear incongruent with behavioral observations and the animals’ neural constraints. The “Navigation by Scene Fa- miliarity Hypothesis” proposed originally for insect navigation offers an elegantly simple so- lution for retracing previously experienced routes without the need for complex neural architectures and memory retrieval mechanisms. This hypothesis proposes that an animal can return to a target location by simply moving toward the most familiar scene at any given point. Proof of concept simulations have used computer-generated ant’s-eye views of the world, but here we test the ability of scene familiarity algorithms to navigate training routes across satellite images extracted from Google Maps. We find that Google satellite images are so rich in visual information that familiarity algorithms can be used to retrace even tortu- ous routes with low-resolution sensors. We discuss the implications of these findings not only for animal navigation but also for the potential development of visual augmentation sys- tems and robot guidance algorithms. OPEN ACCESS Citation: Gaffin DD, Dewar A, Graham P, Philippides A (2015) Insect-Inspired Navigation Algorithm for an Aerial Agent Using Satellite Imagery. PLoS ONE 10(4): e0122077. doi:10.1371/journal.pone.0122077 Insect-Inspired Navigation Using Satellite Images visual images [5] to a human-like map sense [6,7]. The storage of long, accurately ordered se- quences of views, or of the complex relationships between them (allocentric representation), however, would seem to require complex circuitry and has proven difficult to implement in closed loop models [8]. What’s more, such schemes seem to rely on anthropocentric ideas of how humans might design a navigation system. Competing Interests: Co-author Paul Graham is a PLOS ONE Editorial Board member. The authors verify that this is the case. This does not alter the authors' adherence to PLOS ONE Editorial policies and criteria. In general, the constraints imposed by the limited neural resources of insects seem to pro- mote the use of efficient task-dependent solutions to sensorimotor challenges [9,10,11]. In this spirit, the Navigation by Scene Familiarity Hypothesis (NSFH) was developed as a model of route navigation in ants [12,13,14]. NSFH is an egocentric algorithm that does not require stor- age of complex relational memories among visual scenes. The visual information required to navigate a route is stored in a single neural network, which after route learning can output a fa- miliarity score for any view presented to it. Subsequently, the visual information of a perceived scene addresses the memory directly and navigation is achieved implicitly by using a trained network to assess the familiarity of visual scenes, thus enabling a search for familiar scenes. Baddeley et al. [13] produced a virtual agent that used an NSFH-inspired algorithm to retrace visual routes consisting of computer-simulated silhouettes of bushes, trees, and tussocks. The agent generated routes that shared many characteristics with those documented in field studies of navigating ants [15,16,17]. Such parsimonious algorithms can work because of the structured learning made possible by redundancy within navigational systems. In insects and many animals, for instance, path in- tegration [18,17] can guide direct routes through unfamiliar terrain, providing an opportunity to learn visual information [19]. Similarly, insects have dedicated learning behaviors, called learning flights or learning walks, which also provide ideal opportunities to learn visual infor- mation [20,21,17]. In these cases, some of the computational load associated with navigational learning is outsourced to innate behaviors that position the agent at the correct place to learn simple views. The aim of this work was to test the NSFH approach for a novel visual ecology, namely that produced by an aerial platform with a downward facing, low-resolution visual sensor. To pro- vide a simple and flexible simulation we used satellite images taken from Google Maps. Google Maps is a rich, readily accessible source of satellite images that can be scaled to various altitudes and used to test an artificial agent’s tracking proficiency across a range of landscapes and sen- sor morphologies. While other studies have used satellite images, maps, and panoramic street views for localization and navigation [22,23,24], we use an insect-inspired approach to assess scene familiarity by computing pixel-by-pixel scene differences from a fixed sensor matrix. A premise of the NSFH is that natural scenes are sufficiently rich with information that even an agent with a low-resolution sensor will not get confused and follow an incorrect path. We developed MATLAB scripts to access Google satellite images, analyze their information content given sensors of various complexities, and test the ability of hypothetical autonomous agents to retrace various training routes. We emphasize that the point of this work is not to op- timize the navigational algorithms. Rather our aim is to show that downward facing views of natural scenes contain sufficient information for route navigation via NSFH-style algorithms, and to investigate how navigation performance depends on the interplay between environment and sensor resolution. Data Availability Statement: All relevant data are within the paper. Funding: DG was supported by a grant from the Research Council of the University of Oklahoma Norman Campus. AD and PG are both funded by the Biotechnology and Biological Sciences Research Council (grant codes: BB/F015925/1 and BB/ H013644). AP is funded by the Engineering and Physical Sciences Research Council (EP/I031758/1) and an EU FP7 grant (ICT 308943). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. For example, a foraging ant or a worker honeybee precisely navigates long distances from nest to food and back [3,4]. How can circuitry capable of such a complex task be contained in less than a cubic millimeter of brain tissue? Many hypotheses have been proposed to suggest how these insects navigate, ranging from sequential recall and template matching of stored PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 1 / 14 Visual System Our visual system acted as a downward facing “beam” that detected a 640 x 640 pixel scene from the larger 3600 x 3600 satellite image and spanned approximately 100 m of land surface (a 100 m patch of land from 250 m in height represents a beam of 22.6° in width). We applied the MATLAB function “histeq” to enhance contrast using histogram equalization; analogous processes occur in many animals that use temporal and/or spatial neural summation to en- hance contrast [25]. We also pixelated each scene to a specific matrix size and pixel depth (number of gray lev- els). For example, in the landscape shown in Fig 1A, we reduced the resolution of each scene to a 50 x 50 pixel matrix, with each pixel scaled to 10 levels of gray. However, to compare sensor resolutions we also varied the pixel densities (10 x 10, 20 x 20, 40 x 40, 80 x 80) and pixel depth (black/white, 10 gray levels, 100 gray levels) during scene processing. The corners of these ma- trices were cropped to form circles so that we could more easily compare directional informa- tion between scenes (Fig 1B). In the example of Fig 1A and 1B, circularization reduced pixel coverage by 21.5% (from 2,500 to 1,963 pixels). The result is a low-resolution sensor with limit- ed field of view and no distortion due to perspective. Accessing Google Maps Accessing Google Maps We obtained an application programming interface key by registering with Google Maps at https://code.google.com/apis/console, which then allows 25,000 map requests per day. We PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 2 / 14 Insect-Inspired Navigation Using Satellite Images wrote a MATLAB script that allows the user to choose the latitude and longitude of desired scenes and the level of zoom of the image, from 0 (whole world) to 21 (very close up), the map type (roadmap, satellite, terrain, hybrid), and the image format (png/png8, png32, gif, jpg, jpg- baseline). In the examples described here, we chose the satellite map type and a zoom level of 18 (~ 250 m above the ground); we saved our images in the png32 format. wrote a MATLAB script that allows the user to choose the latitude and longitude of desired scenes and the level of zoom of the image, from 0 (whole world) to 21 (very close up), the map type (roadmap, satellite, terrain, hybrid), and the image format (png/png8, png32, gif, jpg, jpg- baseline). In the examples described here, we chose the satellite map type and a zoom level of 18 (~ 250 m above the ground); we saved our images in the png32 format. We centered our main study landscape on the Meeting House on the University of Sussex campus, UK, which is located at longitude 50.8649 and latitude -0.0880. We stitched together nine contiguous scenes (3 x 3 arrangement), after cropping out the Google watermark from the bottom of each scene. The final product was a 3600 x 3600 pixel satellite image of the Sussex campus and some of the surrounding forest and fields. In addition to the Sussex landscape, we chose and constructed additional landscapes repre- senting a range of visual complexities. Our simple landscape was a patch of open water in the middle of Lake Derwentwater, south of Keswick, UK (longitude 54.5739, latitude -3.1476); our medium landscape was a sandy region dotted with vegetated hummocks near Kelso Dunes in the Mojave National Preserve, USA (longitude 34.9410, latitude -115.8234); our complex land- scape was centered on the Anne Frank House in Amsterdam, NL (long PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 Landscape Information Analysis To assess the familiarity of different parts of the surveyed region, we first extracted 10,000 cir- cularized scenes in a 100 x 100 grid (Fig 1A). The centers of adjacent scenes were displaced by about 5 m, resulting in 95% overlap between neighbors. The similarity or familiarity of each scene is then calculated by comparing it to all other scenes on a pixel-by-pixel basis with the ab- solute difference between pixels summed to generate a relative scene-to-scene difference score, the image difference score, which is summed across all 10,000 scenes to give a 100 x 100 image difference topography map of the surveyed region (Fig 1C). In this map the greater the differ- ence value at a given point, the greater the difference of the 1,963 pixels in the scene at that point to the same pixels of all of the other 9,999 scenes. A high difference value therefore repre- sents a scene that is distinct relative to other scenes in that environment. The image difference between the scene viewed currently and a stored scene can be used to guide spatial behavior in two ways [26,27]. Because image difference between scenes oriented in the same direction increases with distance between them, a single scene can be an attractor to the PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 3 / 14 Insect-Inspired Navigation Using Satellite Images Fig 1. Scene familiarity landscapes from satellite images. A Satellite images (zoom level = 18; ~250 m camera altitude) of an University of Sussex campus in Falmer, England were stitched together in MATLAB to form a single 3,600 x 3,600 pixel grayscale The focal scene (highlighted by the right-most central white circle) is centered on the Meeting House on the Sussex campus (long -0.0880). Ten thousand 640x640 scenes were sampled at equally spaced points in a 100x100 grid (white dots) across the larger la processed to 50 x 50 resolution with 10 levels of gray. These sample scenes were circularized prior to similarity processing. B Exp i S f i i i f i i S Fig 1. Scene familiarity landscapes from satellite images. A Satellite images (zoom level = 18; ~250 m camera altitude) of an area encompassing the University of Sussex campus in Falmer, England were stitched together in MATLAB to form a single 3,600 x 3,600 pixel grayscale image (scale bar = 100 m). Insect-Inspired Navigation Using Satellite Images scene and a scene shifted two sampling points to the left. These locations are indicated by white circles in A. C All 10,000 scenes in A were compared to all other scenes. The contour plot indicates average image difference values by scene position. The plot in C is superimposed atop the campus satellite image for reference. Dotted rectangles d and e show regions of relatively high and relatively low distinctness. These two positions are used to generate the volcano plots shown in D and E. These plots show the absolute pixel-by-pixel image differences of the focal scene compared to the surrounding 225 scenes (15 x 15 square). The red arrows below the volcanoes indicate the orientation of the best-matched surrounding scene rotated through 360 degrees and compared to the static focal scene; arrow length varies directly with the goodness of the match. Satellite images used under a CC BY license, with permission from Esri, original copyright 2015. doi:10.1371/journal.pone.0122077.g001 location it was stored at. To visualize the use of a stored view in this way, in Fig 1D and 1E we show the image difference between focal and surrounding views which create surface “volcanoes.” If an agent maintains a constant heading and moves to reduce the difference between current and stored scenes, it will navigate down the surface in a form of gradient descent. From a given surface plot we can thus gauge how useful a stored image would be as an attractor by looking for smooth gradients in image differences in the region around the image. Alternatively, a scene can be used as a “visual compass;” that is, the orientation of a stored focal scene can be recovered by rotating the current scene until it best matches the stored scene. We can visualize this information by find- ing the minima in image differences between a focal scene and systematically rotated scenes (the rotational image difference function, RIDF) from locations surrounding the focal location and plotting the resultant directions as arrows below the surface volcanoes (Fig 1D and 1E). To assess scene information among various landscapes, we used volcano plots and RIDF analyses for the Derwentwater, Mojave, and Amsterdam satellite regions. In each case, we used 100 x 100 scene spacing and compared 400 adjacent scenes (in a 20 x 20 square) to five focal scenes. To calculate comparator measures of landscape information, we first made perpendicu- lar slices through the volcanoes (Fig 2A) and averaged the absolute scene difference values by distance from the focal scene for the four directions (Fig 2B). We then determined the absolute scene difference value and the distance from the focal point (in meters) at the 50% maximal dif- ference point (p50) on the resultant curve. To assess visual compass information, we calculated RIDFs for a location by rotating the scene 360 degrees in one-degree increments, comparing each to the focal scene. Volcanoes and RIDFs are averaged from values obtained from five evenly spaced focal scenes from across the region (specifically, by row and column: 25,25; 25,75; 50,50; 75,25; 75,75). Landscape Information Analysis The focal scene (highlighted by the right-most central white circle) is centered on the Meeting House on the Sussex campus (longitude: 50.8649; latitude: -0.0880). Ten thousand 640x640 scenes were sampled at equally spaced points in a 100x100 grid (white dots) across the larger landscape. Each scene was processed to 50 x 50 resolution with 10 levels of gray. These sample scenes were circularized prior to similarity processing. B Expanded views of the central PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 4 / 14 PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 Route following Algorithm We developed and tested several algorithms during the course of these studies. We describe here our best prototype based on speed and accuracy of route retracing. We note that the pur- pose of this paper is not to develop an algorithm with arbitrarily good performance; rather we use the performance of route following algorithms as proof that the information within this vi- sual ecology is sufficient for navigation using a familiarity algorithm. Creating the training path. The program presents the selected satellite region and queries the user to select the number of points to be marked in the training path. Cross hairs appear over the satellite image to guide the user’s placement of training points. Once all of the training points are placed, the program interpolates and plots the contiguous points between each suc- cessive point to create a continuous line for the main path and then increases the width of this path by a scene on either side (yellow dashes, Fig 3). The bearing of travel between successive user-placed points is then determined based on a 360-degree compass system with north being zero. To simulate what the agent would “see” as it made its way along the path, each scene in the training path is then rotated by this bearing and stored in memory. Recapitulating the training path. To initiate recapitulation, the user places the agent near the beginning of the learned route, which is akin to an ant or a bee either leaving its hive for a 5 / 14 PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 Insect-Inspired Navigation Using Satellite Images Fig 2. Comparator measures of landscape information. A Slices are made through the familiarity volcano in the four cardinal directions and the values are averaged to produce the curve shown in B. The absolute scene difference value and the distance from the focal point (in meters) are derived from this curve at the 50% maximal difference point (p50). doi:10.1371/journal.pone.0122077.g002 Fig 2. Comparator measures of landscape information. A Slices are made through the familiarity volcano in the four cardinal directions and the values are averaged to produce the curve shown in B. The absolute scene difference value and the distance from the focal point (in meters) are derived from this curve at the 50% maximal difference point (p50). Fig 2. Comparator measures of landscape information. Route following Algorithm A Slices are made through the familiarity volcano in the four cardinal directions and the values are averaged to produce the curve shown in B. The absolute scene difference value and the distance from the focal point (in meters) are derived from this curve at the 50% maximal difference point (p50). doi:10.1371/journal.pone.0122077.g002 journey to a learned food source, or leaving a food source, to return home. Ants are known to scan the world when in an unfamiliar location [28] and this behavior inspires our agent’s strat- egy. The agent’s first scan is based on the direction of the most familiar scene at that point. The agent produces a scan in an arc (blue lines, Fig 3) whose size varies depending on a user-set value (here we used a value that produced an arc that spanned about 40 m of ground surface). The initial point sampled on the arc is in the direction of the current path with subsequent journey to a learned food source, or leaving a food source, to return home. Ants are known to scan the world when in an unfamiliar location [28] and this behavior inspires our agent’s strat- egy. The agent’s first scan is based on the direction of the most familiar scene at that point. The agent produces a scan in an arc (blue lines, Fig 3) whose size varies depending on a user-set value (here we used a value that produced an arc that spanned about 40 m of ground surface). The initial point sampled on the arc is in the direction of the current path with subsequent 6 / 14 PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 Insect-Inspired Navigation Using Satellite Images Fig 3. Schematic of route tracking algorithm. The yellow dotted arrows indicate a user-selected route. The blue arc scans are based on the direction of the most familiar scene at that point (dotted black lines). Initial samples are made in the direction of the current path with subsequent samples taken at progressively wider angles on the arc, alternating left and right views (green points). At each point sampled, the agent rotates the focal current scene through 360 degrees and compares each rotation to all stored memory scenes (yellow path). Once a familiarity threshold is met, no further points on the arc are sampled. Scene Information Analysis We explored the potential quality of various scenes for visual navigation by examining the scene difference information as shown by volcano and RIDF plots of the region. Fig 4 shows these analyses for three selected satellite regions of varying complexity. Fig 4B and 4C show sample volcano and RIDF plots averaged from the five points shown in Fig 4A. This simple analysis gives us some sense of how useful a stored view would be for navigation. Volcano plots (Fig 4B) show a central low point whose depth relates to how distinct a view is within that envi- ronment. As we would expect the more visually rich scene of Amsterdam provides scenes that are more distinct than the surroundings. The volcano plots show that even the homogenous landscapes of deserts and lakes do provide distinct views, however the minima are shallow. RIDF analysis (Fig 4C) indicates whether usable directional information remains in the views of a world as produced by a particular sensor. We see that our low-resolution, small beam visual sensor provides navigationally relevant information from our world; information is available even within the relatively featureless Derwentwater landscape using the 80 x 80 sen- sor (though it should be noted that in a live situation, changing wave patterns, light glint, and shadows should negate visual information present in static satellite images). Volcano plots also highlight a catchment area whereby one could descend a gradient of the image difference function to return to the location of the stored scene. An analysis of the p50 point of these curves (Fig 4D and 4E) for various sensor designs highlights two key points: (1) lower resolution sensors produce a larger catchment area. And, (2) the number of gray levels does not impact on the catchment area in complex scenes. Taken together, these scene analyses show that directional and positional information can be derived from views with a range of resolutions and in natural environments. However, high- resolution views show better performance when used as a “visual compass” and as expected, the visually rich environment of Amsterdam also allows for better performance. Route following Algorithm The agent moves forward to this best-matched point (red line) and casts its next scan based on an extension of this red line segment (see Methods for further description). doi:10 1371/journal pone 0122077 g003 Fig 3. Schematic of route tracking algorithm. The yellow dotted arrows indicate a user-selected route. The blue arc scans are based on the direction of the most familiar scene at that point (dotted black lines). Initial samples are made in the direction of the current path with subsequent samples taken at progressively wider angles on the arc, alternating left and right views (green points). At each point sampled, the agent rotates the focal current scene through 360 degrees and compares each rotation to all stored memory scenes (yellow path). Once a familiarity threshold is met, no further points on the arc are sampled. The agent moves forward to this best-matched point (red line) and casts its next scan based on an extension of this red line segment (see Methods for further description). doi:10.1371/journal.pone.0122077.g003 doi:10.1371/journal.pone.0122077.g003 samples taken at progressively wider angles from this direction, alternating the left and the right sides (green points, Fig 3). At each point sampled, the agent rotates its current focal scene through 360 degrees at user- defined increments, comparing each incremental rotation to all stored memory scenes. If a user-set threshold of familiarity is met (here we used a threshold of 20% of the mean of differ- ence values obtained by comparing all rotations of the current surveyed point to all scenes in the training path), then no further points on the current arc are sampled and the agent moves forward to this best-matched point (red line, Fig 3). It casts its next scan based on the forward extension of the red line segment (dotted black lines, Fig 3). The program continues until the PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 7 / 14 Insect-Inspired Navigation Using Satellite Images re-tracked path is within a designated distance of the end of the training path, runs off the sat- ellite area, or is terminated by the user. re-tracked path is within a designated distance of the end of the training path, runs off the sat- ellite area, or is terminated by the user. PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 Tests of route recapitulation algorithm Having shown that the information for navigation exists, we now turn our attention to whether sensors of this type could be used to form a simple visual route navigation algorithm. However, we reiterate that our goal for this analysis is to show that this information can be used for navi- gation and not to develop an optimal navigational algorithm. We first tested the performance of the tracking algorithm to navigate a simple S-shaped training path across the three experi- mental satellite regions. Fig 5A shows sample paths and re-tracings for two sensor resolutions (10 x 10 and 40 x 40; 10 levels of gray). The insets show examples of perceived scenes. Fig 5B summarizes the performance of all four resolutions and three pixel depths across the three re- gions. The only re-tracings that failed to reach the target were for the 10 x 10 (BW, 10 gray), 20 x 20 (BW), and 40 x 40 (BW) sensors in the Derwentwater region. The 10 x 10 (100 gray) and 20 x 20 (10 gray) sensors meandered a bit, but still reached the target. The Derwentwater 20 x 20 (100 gray), 40 x 40 (10 & 100 gray), and 80 x 80 (BW, 10, & 100 gray) sensors had no diffi- culty traversing the path correctly. All of the sensors tracked the training path accurately for the Mojave and Amsterdam regions. The histeq transformation improved performance of the auto-tracking algorithm. For example, in these trials the pre-transformation algorithm recapit- ulated the path correctly in 4 of the 12 pixel density/depth combinations for Derwentwater and PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 8 / 14 Insect-Inspired Navigation Using Satellite Images Fig 4. The spatial information available in natural scenes. A Five sampled focal scenes are shown as white dots superimpos Derwentwater, Mojave, and Amsterdam (scale bar = 100 m). B The averaged volcano plots for these five scenes, each comparin surrounding 400 scenes (20 x 20 grid) are shown for a 20 x 20 pixel-resolution with 100 gray level pixel depth. A common y-axis i differences among regions. C RIDF curves averaged across the five landscape sample points for the four sensor resolutions indi 100 gray level pixel depth). Absolute scene difference information (D) and distance from focal scene (E) are plotted for each regio Insect-Inspired Navig Fig 4. The spatial information available in natural scenes. A Five sampled focal scenes are shown as white dots superimposed on satellite regions for Derwentwater, Mojave, and Amsterdam (scale bar = 100 m). B The averaged volcano plots for these five scenes, each comparing focal scenes to the surrounding 400 scenes (20 x 20 grid) are shown for a 20 x 20 pixel-resolution with 100 gray level pixel depth. A common y-axis is used to show informatio differences among regions. C RIDF curves averaged across the five landscape sample points for the four sensor resolutions indicated in the legend (all at 100 gray level pixel depth). Absolute scene difference information (D) and distance from focal scene (E) are plotted for each region for each of the sensor Fig 4. The spatial information available in natural scenes. A Five sampled focal scenes are shown as white dots superimposed on satellite regions for Derwentwater, Mojave, and Amsterdam (scale bar = 100 m). B The averaged volcano plots for these five scenes, each comparing focal scenes to the surrounding 400 scenes (20 x 20 grid) are shown for a 20 x 20 pixel-resolution with 100 gray level pixel depth. A common y-axis is used to show information differences among regions. C RIDF curves averaged across the five landscape sample points for the four sensor resolutions indicated in the legend (all at 100 gray level pixel depth). Absolute scene difference information (D) and distance from focal scene (E) are plotted for each region for each of the sensor PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 9 / 14 We tested the ability of the auto-tracking algorithm to retrace simple, S-shaped t i i t ( ll d t ) d th D t t M j d A t d t llit i ( l b 100 ) A S l t d th Fig 5. Route recapitulation performance across different landscapes. We tested the ability of the auto-tracking Fig 5. Route recapitulation performance across different landscapes. We tested the ability of the auto-tracking algorithm to retrace simple, S-shaped training routes (yellow dots) drawn across the Derwentwater, Mojave, and Amsterdam satellite regions (scale bar = 100 m). A Sample retraced paths are shown for two sensor resolutions (10 x 10 and 40 x 40) and 10 gray level pixel depth (insets show examples of sensor images). B Cumulative distance of departure of the retraced route from the training route is shown for the three regions at the four sensor resolutions and three pixel depths. Plotted are the summed geometric distances of each best-matched point (red) to the appropriate training path point (in sequence) based on a 100 x 100 point sampling grid imposed on each region (actual values given above bars). Satellite images used under a CC BY license, with permission from Esri, original copyright 2015. n performance across different landscapes. We tested Fig 5. Route recapitulation performance across different landscapes. We tested the ability of the auto-tracking algorithm to retrace simple, S-shaped training routes (yellow dots) drawn across the Derwentwater, Mojave, and Amsterdam satellite regions (scale bar = 100 m). A Sample retraced paths are shown for two sensor resolutions (10 x 10 and 40 x 40) and 10 gray level pixel depth (insets show examples of sensor images). B Cumulative distance of departure of the retraced route from the training route is shown for the three regions at the four sensor resolutions and three pixel depths. Plotted are the summed geometric distances of each best-matched point (red) to the appropriate training path point (in sequence) based on a 100 x 100 point sampling grid imposed on each region (actual values given above bars). Satellite images used under a CC BY license, with permission from Esri, original copyright 2015. Insect-Inspired Navigation Using Satellite Images resolutions and pixel depths as calculated at the p50 position on the information volcanoes (as described in Fig 2). Satellite images used under a CC BY license, with permission from Esri, original copyright 2015. resolutions and pixel depths as calculated at the p50 position on the information volcanoes (as described in Fig 2). Satellite images used under a CC BY license, with permission from Esri, original copyright 2015. resolutions and pixel depths as calculated at the p50 position on the information volcanoes (as described in Fig 2). Satellite images used under a CC BY license, with permission from Esri, original copyright 2015. 8 of the 12 for Mojave. After transformation, the success rate improved to 8 of 12 (Derwent- water) and 12 of 12 (Mojave). doi:10.1371/journal.pone.0122077.g004 8 of the 12 for Mojave. After transformation, the success rate improved to 8 of 12 (Derwent- water) and 12 of 12 (Mojave). We ran several tests of the route algorithm on more complex paths; three of these are shown in Fig 6. We traced the “US” logo for the University of Sussex atop the Sussex campus and re- traced the route successfully with a 50 x 50, 100-gray sensor (Fig 6A). The tracker required Fig 5. Route recapitulation performance across different landscapes. We tested the ability of the auto-tracking algorithm to retrace simple, S-shape training routes (yellow dots) drawn across the Derwentwater, Mojave, and Amsterdam satellite regions (scale bar = 100 m). A Sample retraced paths are shown for two sensor resolutions (10 x 10 and 40 x 40) and 10 gray level pixel depth (insets show examples of sensor images). B Cumulative distance o departure of the retraced route from the training route is shown for the three regions at the four sensor resolutions and three pixel depths. Plotted are the summed geometric distances of each best-matched point (red) to the appropriate training path point (in sequence) based on a 100 x 100 point sampling imposed on each region (actual values given above bars). Satellite images used under a CC BY license, with permission from Esri, original copyright 20 Fig 5. Route recapitulation performance across different landscapes. We tested the Fig 5. Route recapitulation performance across different landscapes. g p p p y g training routes (yellow dots) drawn across the Derwentwater, Mojave, and Amsterdam satellite regions (scale bar = shown for two sensor resolutions (10 x 10 and 40 x 40) and 10 gray level pixel depth (insets show examples of sens departure of the retraced route from the training route is shown for the three regions at the four sensor resolutions an summed geometric distances of each best-matched point (red) to the appropriate training path point (in sequence) b imposed on each region (actual values given above bars). Satellite images used under a CC BY license, with permis doi:10.1371/journal.pone.0122077.g005 PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 10 / 14 Insect-Inspired Navigation Using Satellite Images Fig 6. Examples of tracking algorithm performance on complex routes. A Retracing of a route in the shape of the University of Sussex logo over the Sussex landscape. This route had several abrupt turns, including one of about 135 degrees. B Test of the tracking algorithm on a route that loops across itself. This route spells out a scripted “Jan” in honor of an uncle who lived in this Amsterdam neighborhood. C An expanding square spiral beginning at the Meeting House on the Sussex campus. The tracker accurately retraced the route until it passed through the region of low scene difference information over the field at the lower left of the image (near region e in Fig 1C; scale bar = 100 m). Satellite images used under a CC BY license, with permission from Esri, original copyright 2015. Fig 6. Examples of tracking algorithm performance on complex routes. A Retracing of a route in the shape of the University of Sussex logo over the Sussex landscape. This route had several abrupt turns, including one of about 135 degrees. B Test of the tracking algorithm on a route that loops across itself. This route spells out a scripted “Jan” in honor of an uncle who lived in this Amsterdam neighborhood. C An expanding square spiral beginning at the Meeting House on the Sussex campus. The tracker accurately retraced the route until it passed through the region of low scene difference information over the field at the lower left of the image (near region e in Fig 1C; scale bar = 100 m). Satellite images used under a CC BY license, with permission from Esri, original copyright 2015. doi:10.1371/journal.pone.0122077.g006 several samples to negotiate the corners of the pattern and even retraced the 135° turn on the upper right side of the “U.” We also tested the tracker on training routes that contained loops (Fig 6B). Although, such training routes are unlikely in actual homing animals, the tracker is able to retrace these routes because of the directional scene information as computed by the RIDF and the implicit forward momentum of the algorithm. In this situation, our algorithm out- performs that described in Baddeley et al. [13], in which each step is predicated on the compass bearing of the best RIDF match. Here we used a system that sets the next saccade arc based on the line determined by the best matched point on the current saccade, with the pivot determined by the position of the previous best matched point. As such, we take advantage of an aerial agent casting its gaze forward of its current path. This approach makes the agent more likely to follow its current path than that of a crossing path. We also tested the tracker on an expanding square spiral centered over the heterogeneous Sussex campus landscape. The tracker successfully nego- tiated the right-angle corners in the high information areas above the buildings, but failed in the low information area over the field at the lower left (Fig 6C). This behavior is consistent with the information difference topography shown in Fig 1. Subsequent tests with a more stringent matching threshold allowed the tracker to retrace the complete spiral. PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 The memory potential of neural tissue The NSFH model is elegantly simple: The animal or the agent simply flies or moves toward areas of highest familiarity. Of course, this depends on an agent being able to store all the scenes that are associated with a desirable route. Is this possible? Although a bee’s brain is less than a cubic millimeter in volume, the tissue between the eyes and motor centers is complex [10] and has approximately one million neurons [29] and a billion synapses. Therefore even a small brain has the potential to produce a large number of unique codes. Can this account for all the ‘glimpses’ that a bee might experience during her one- to two-month foraging lifetime? Active worker honeybees live about six to seven weeks. Rounding up to two months (60 days), multiplying by 24 hours in a day, 60 minutes in an hour, 60 seconds in a minute, and an eye frame capture rate of 60 per second, we get 180x24x60x60x60 = 311,040,000 scenes she could experience in a lifetime. How do these numbers compare to the architecture of a bee’s visual system? Each eye of a honeybee contains about 4500 ommatidia [10]. Visual information from ommatidial receptors relays through massively parallel neuropil structures with synaptic con- nections in the lamina, medulla, lobula, and among local interneurons before it reaches the mushroom bodies, central complex, and descending neurons. This visual network thus provides an enormous number of nodes and connections. Take for example, a simple array of two elements with two possible states: “on” or “off.” The number of unique configurations for this array is four ([0 0], [0 1], [1 0], [1 1]) and in general the number of permutations is 2n where n is the number of array elements. Obviously, with a visual system of thousands of ommatidia and tens of thousands of synaptic connections, we have a huge storage potential that exceeds the volume of potential lifetime visual memory required by a foraging in- sect. This is before one considers the efficiencies in visual coding (as opposed to storing raw pixel images [30]) and in deriving familiarity measures from, for instance, a neural network trained on the perceived images as opposed to storing all of them as a “perfect memory” (see [13]). These calculations assume that landscapes possess enough visual variability to address this neurological potential. Discussion The results of this study are both clear and surprising. We used the Navigation by Scene Famil- iarity Hypothesis to develop algorithms for the recapitulation of visually guided routes across Google Maps satellite images. We show proof of concept that even a simple, low-resolution sen- sor can navigate complex paths accurately and efficiently. Of course, it must be noted that our analyses depended on restricted visual imagery taken from a perspective higher than where most animals fly. Still, by analyzing the visual information available in satellite images from various re- gions with different sensor structures we show that the inherent richness of visual scenes suggests that this simple navigation algorithm may be applicable to natural environments as well. What’s more from a biological point of view, this work reinforces a way of thinking about visually guided behavior where tasks such as navigation can be accomplished with coarse visual information and an absence of perceptual processes such as object or place recognition. PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 11 / 14 Insect-Inspired Navigation Using Satellite Images Potential Applications We think the NSFH has many potential applications. It is a deliberately simple algorithm and thus is suited to applications where low power and weight are priorities such as autonomous robotics and wearable sensors. For example, this technology could be used to develop guidance aids for the visually impaired in the acquisition, processing, and retracing of real-world visually defined routes. We also see potential for unmanned vehicles (aerial, ground-based, or subma- rine) to use similar algorithms where map and GPS information is of little use or unreliable (such as space exploration or inside buildings) or in cases where previously captured images are of little or no value due to significant environmental disruptions. The memory potential of neural tissue Our scene analyses showed that the visual environment is indeed rich and that even a very simple sensor can access this information and suffer few missteps due to scene aliasing. Indeed even very low-resolution sensors with limited pixel depth produce steep difference information volcanoes when compared to surrounding scenes. Author Contributions Conceived and designed the experiments: DG PG AP. Performed the experiments: DG AP PG AD. Analyzed the data: DG AD. Contributed reagents/materials/analysis tools: DG AP PG AD. Wrote the paper: DG PG AP. Wrote MATLAB scripts for tracking algorithm: DG AD AP. Conceived and designed the experiments: DG PG AP. Performed the experiments: DG AP PG AD A l d h d DG AD C ib d / i l / l i l DG AP PG Conceived and designed the experiments: DG PG AP. Performed the experiments: DG AP PG AD. Analyzed the data: DG AD. Contributed reagents/materials/analysis tools: DG AP PG g p p AD. Analyzed the data: DG AD. Contributed reagents/materials/analysis tools: DG AP PG AD. Wrote the paper: DG PG AP. Wrote MATLAB scripts for tracking algorithm: DG AD AP. y g y AD. Wrote the paper: DG PG AP. Wrote MATLAB scripts for tracking algorithm: DG AD AP. Summary In this paper, we have given proof of concept that a deliberately simple visual sensor (a narrow low-resolution beam) can navigate using a familiarity principle in a visual environment (the world of satellite images taken from Google maps) that is quite different from that previously tested (a panoramic ground-level view of a simulated desert). In so doing, we have both shown the generality of the NSFH approach and also demonstrated the rich information available even in low resolution visual scenes that could be exploited by both small-brained animals and autonomous agents. Acknowledgments DDG thanks his brother Arthur Gaffin (Sightech Machine Vision Systems) for countless con- versations about neural processing and bee navigation. We thank Dr. Mariëlle Hoefnagels for valuable discussions during the development of the MATLAB algorithms and for critically re- viewing the manuscript. We also thank the staff of the Department of Informatics at the Uni- versity of Sussex for support in hosting the sabbatical visit of DDG and George Davis of the OU Department of Biology for arranging and setting up MATLAB for DDG, and Bradley Bray- field for helping us acquire appropriate satellite images. Increasing Algorithmic Efficiency The aim of this work was not to develop the most efficient navigational algorithm possible and therefore the current version could be improved in a number of ways. First, increased perfor- mance could be achieved by greater leveraging of a flying agent’s perspective, since being able to see the skyline at the horizon would provide strong directional information [27]. Second, an agent could use different resolution strategies depending on its situation. A high-resolution sensor works best when the agent is on the right path, as it decreases the chance of being fooled 12 / 14 PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 Insect-Inspired Navigation Using Satellite Images by a stray aliased scene. On the other hand, when the agent is off course, a low-resolution, wider-field sensor increases the catchment area and improves the chances of getting back on track [31]. Finally, biasing scene selections toward the direction of movement and by restricting the rotational field of view could improve efficiency by requiring fewer scene comparisons and reduce the chance of choosing an incorrect path. PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 References 1. Wystrach A, Graham P. What can we learn from studies of insect navigation? Animal Behaviour. 2012; 84: 13–20. 1. Wystrach A, Graham P. What can we learn from studies of insect navigation? Animal Behaviour. 2012; 84: 13–20. 2. Chittka L, Rossiter SJ, Skorupski P, Fernando C. What is comparable in comparative cognition? Philos Trans R Soc Lond B Biol Sci. 2012; 367: 2677–2685. doi: 10.1098/rstb.2012.0215 PMID: 22927566 2. Chittka L, Rossiter SJ, Skorupski P, Fernando C. What is comparable in comparative cognition? Philos Trans R Soc Lond B Biol Sci. 2012; 367: 2677–2685. doi: 10.1098/rstb.2012.0215 PMID: 22927566 3. Santschi F. Observations et remarques critiques sur le mécanisme de l'orientation chez les fourmis. Rev Suisse Zool. 1911; 19: 303–338. 4. von Frisch K. The dance language and orientation of bees. Cambridge: Harvard University Press; 1967. 4. von Frisch K. The dance language and orientation of bees. Cambridge: Harvard University Press; 1967. 5. Cartwright B, Collett TS. Landmark maps for honeybees. Biol Cybern. 1987; 57: 85–93. 5. Cartwright B, Collett TS. Landmark maps for honeybees. Biol Cybern. 1987; 57: 85–93. 6. Gould JL. The locale map of honeybees: Do insects have cognitive maps? Science. 1986; 232: 861– 863. PMID: 17755968 7. Menzel R, Greggers U, Smith A, Berger S, Brandt R, Brunke S, et al. Honey bees navigate according to a map-like spatial memory. Proc Natl Acad Sci USA. 2005; 102: 3040–3045. PMID: 15710880 8. Smith L, Philippides A, Graham P, Baddeley B, Husbands P. Linked local navigation for visual route guidance. Adapt Behav. 2007; 15: 257–271. 9. Wehner R. Desert ant navigation: How miniature brains solve complex tasks. J Comp Physiol A. 2003; 189: 579–588. PMID: 12879352 10. Strausfeld NJ. Arthropod brains: evolution, functional elegance, and historical significance. Cambridge: Harvard University Press; 2012. PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 13 / 14 Insect-Inspired Navigation Using Satellite Images 11. Graham P, Philippides A. Insect-inspired vision and visually guided behavior. In: Bhushan B, Winbigler HD, editors. Encyclopedia of nanotechnology. Dordrecht, Heidelberg, New York, London: Springer; 2012. pp. 1122–1127. 12. Baddeley B, Graham P, Husbands P, Philippides A. Holistic visual encoding of ant-like routes: Naviga- tion without waypoints. Adapt Behav. 2011; 19: 3–15. 13. Baddeley B, Graham P, Husbands P, Philippides A. A model of ant route navigation driven by scene fa- miliarity. PLoS Comput Biol 2012; 8(1): e1002336. doi: 10.1371/journal.pcbi.1002336 PMID: 22241975 14. Wystrach A, Mangan M, Philippides A, Graham P. References Snapshots in ants? New interpretations of paradig- matic experiments. J Exp Biol. 2013; 216: 1766–1770. doi: 10.1242/jeb.082941 PMID: 23348949 15. Muser B, Sommer S, Wolf H, Wehner R. Foraging ecology of the thermophilic Australian desert ant, Melophorus bagoti. Aust J Zool. 2005; 53: 301–311. 16. Sommer S, von Beeren C, Wehner R. Multiroute memories in desert ants. Proc Natl Acad Sci USA. 2008; 105: 317–322. PMID: 18160534 17. Müller M, Wehner R. Path integration provides a scaffold for landmark learning in desert ants. Curr Biol. 2010; 20: 1368–1371. doi: 10.1016/j.cub.2010.06.035 PMID: 20619653 18. Müller M, Wehner R. Path integration in desert ants, Cataglyphis fortis. Proc Natl Acad Sci USA. 1988; 85: 5287–5290. PMID: 16593958 19. Collett TS, Graham P, Durier V. Route learning by insects. Curr Opin Neurobiol. 2003; 13: 718–725. PMID: 14662374 20. Zeil J, Kelber A, Voss R. Structure and function of learning flights in bees and wasps. J Exp Biol. 1996; 199: 245–252. PMID: 9317729 21. Wehner R, Meier C, Zollikofer C. The ontogeny of foraging behaviour in desert ants, Cataglyphis bicol- or. Ecol Entomol. 2004; 29: 240–250. 22. Hentschel M, Wagner B. Autonomous robot navigation based on OpenStreetMap Geodata. In IEEE In- ternational Conference on Intelligent Transportation Systems (ITSC). 2010; 10.1109/ ITSC.2010.5625092 23. Senlet T, Elgammal A. Satellite image based precise robot localization on sidewalks. In IEEE Interna- tional Conference on Robotics and Automation. 2012; 10.1109/ICRA.2012.6225352. 24. Majdik A, Albers-Schoenberg Y, Scaramuzza D. MAV urban localization from Google street view data. In IEEE International Conference on Intelligent Robots and Systems. 2013; 10.1109/ IROS.2013.6696925 25. Land MF. Visual acuity in insects. Annu Rev Entomol. 1997; 42: 147–177. PMID: 15012311 26. Zeil J, Hofmann MI, Chahl JS. Catchment areas of panoramic snapshots in outdoor scenes. J Opt Soc Am A Opt Image Sci Vis. 2003; 20: 450–469. PMID: 12630831 27. Philippides A, Baddeley B, Cheng K, Graham P. How might ants use panoramic views for route naviga- tion? J Exp Biol. 2011; 214: 445–451. doi: 10.1242/jeb.046755 PMID: 21228203 28. Wystrach A, Philippides A, Aurejac A, Cheng K, Graham P. Visual scanning behaviours and their role in the navigation of the Australian desert ant Melophorus bagoti. J Comp Physiol A. 2014; 200: 615– 626. doi: 10.1007/s00359-014-0900-8 PMID: 24682419 29. Menzel R, Giurfa M. Cognitive architecture of a mini-brain: the honeybee. Trends Cogn Sci. 2001; 5: 62–71. PMID: 11166636 30. Wystrach A, Dewar ADM, Graham P. PLOS ONE \| DOI:10.1371/journal.pone.0122077 April 15, 2015 References Insect vision: Emergence of pattern recognition from coarse en- coding. Curr Biol. 2014; 24: R78–R80. doi: 10.1016/j.cub.2013.11.054 PMID: 24456981 31. Stürzl W, Mallot HA. Efficient visual homing based on Fourier transformed panoramic images. Rob Auton Syst. 2006; 54: 300–313. 14 / 14
https://openalex.org/W4251799780	https://bg.copernicus.org/articles/17/3991/2020/bg-17-3991-2020.pdf	English	null	Ecosystem physio-phenology revealed using circular statistics	null	2,019	cc-by	11,783	Daniel E. Pabon-Moreno1, Talie Musavi1, Mirco Migliavacca1, Markus Reichstein1,2, Christine Römermann2,3, and Miguel D. Mahecha1,2,4 1Department of Biogeochemical Integration, Max Planck Institute for Biogeochemistry, 07745 Jena, Germany 2German Centre for Integrative Biodiversity Research (iDiv), Deutscher Platz 5e, 04103 Leipzig, Germany 3Friedrich Schiller University, Institute of Ecology and Evolution, Philosophenweg 16, 07743 Jena, Germany 4Remote Sensing Center For Earth System Research, Leipzig University, 04103 Leipzig, Germany Correspondence: Daniel E. Pabon-Moreno (dpabon@bgc-jena.mpg.de) Received: 3 October 2019 – Discussion started: 15 October 2019 Revised: 12 June 2020 – Accepted: 26 June 2020 – Published: 6 August 2020 Received: 3 October 2019 – Discussion started: 15 October 2019 Revised: 12 June 2020 – Accepted: 26 June 2020 – Published: 6 August 2020 Abstract. Quantifying how vegetation phenology responds to climate variability is a key prerequisite to predicting how ecosystem dynamics will shift with climate change. So far, many studies have focused on responses of classical pheno- logical events (e.g., budburst or ﬂowering) to climatic vari- ability for individual species. Comparatively little is known on the dynamics of physio-phenological events such as the timing of maximum gross primary production (DOYGPPmax), i.e., quantities that are relevant for understanding terrestrial carbon cycle responses to climate variability and change. In this study, we aim to understand how DOYGPPmax depends on climate drivers across 52 eddy covariance (EC) sites in the FLUXNET network for different regions of the world. Most phenological studies rely on linear methods that cannot be generalized across both hemispheres and therefore do not allow for deriving general rules that can be applied for future predictions. One solution could be a new class of circular– linear (here called circular) regression approaches. Circular regression allows circular variables (in our case phenologi- cal events) to be related to linear predictor variables as cli- mate conditions. As a proof of concept, we compare the per- formance of linear and circular regression to recover origi- nal coefﬁcients of a predeﬁned circular model for artiﬁcial data. We then quantify the sensitivity of DOYGPPmax across FLUXNET sites to air temperature, shortwave incoming ra- diation, precipitation, and vapor pressure deﬁcit. Finally, we evaluate the predictive power of the circular regression model for different vegetation types. Our results show that the joint effects of radiation, temperature, and vapor pressure deﬁcit are the most relevant controlling factor of DOYGPPmax across sites. Woody savannas are an exception, where the most im- portant factor is precipitation. Daniel E. Pabon-Moreno1, Talie Musavi1, Mirco Migliavacca1, Markus Reichstein1,2, Christine Römermann2,3, and Miguel D. Mahecha1,2,4 Although the sensitivity of the DOYGPPmax to climate drivers is site-speciﬁc, it is possible to generalize the circular regression models across speciﬁc vegetation types. From a methodological point of view, our results reveal that circular regression is a robust alternative to conventional phenological analytic frameworks. The anal- ysis of phenological events at the global scale can beneﬁt from the use of circular statistics. Such an approach yields substantially more robust results for analyzing phenological dynamics in regions characterized by two growing seasons per year or when the phenological event under scrutiny oc- curs between 2 years (i.e., DOYGPPmax in the Southern Hemi- sphere). 1 Introduction Phenology is the study of the timing of biological events that can be observed at either the organismic level or the ecosystem scale (Lieth, 1974). For the latter, phenology is the study of some integral behavior across phenological states of the integrated canopy reﬂectance captured by re- mote sensing (Richardson et al., 2009; Zhang et al., 2003) or vegetation-driven ecosystem–atmosphere CO2 exchange ﬂuxes (Richardson et al., 2010). Ecosystem-scale physio- phenological processes of this kind are relevant quantities in global biogeochemical cycles and integrate both the seasonal dynamics of biophysical states (e.g., reﬂected in the canopy development) and the observed photosynthesis at the stand level (i.e., gross primary production). Here we are particu- larly interested in the timing when ecosystems reach their Biogeosciences, 17, 3991–4006, 2020 https://doi.org/10.5194/bg-17-3991-2020 © Author(s) 2020. This work is distributed under the Creative Commons Attribution 4.0 License. Biogeosciences, 17, 3991–4006, 2020 https://doi.org/10.5194/bg-17-3991-2020 © Author(s) 2020. This work is distributed under the Creative Commons Attribution 4.0 License. Published by Copernicus Publications on behalf of the European Geosciences Union. (2017) studied how the variability of annual GPP is in- ﬂuenced by GPPmax and the start and the end of the growing season. The authors found that GPPmax is a better explana- tory parameter for the interannual variability of annual GPP than the start and end days of the growing season. Bauerle et al. (2012) studied how photoperiod and temperature in- ﬂuence plants’ photosynthetic capacity for 23 tree species in temperate deciduous hardwoods, reporting that the photope- riod explains the variability of photosynthetic capacity bet- ter than temperature. So far, to the best of our knowledge, only one study has focused on understanding the temporal variability of GPPmax; Wang and Wu (2019) used a combi- nation of satellite remote-sensing and eddy covariance data to explore how DOYGPPmax is controlled by climatic con- ditions. The authors reported that higher temperatures ad- vance DOYGPPmax, while the inﬂuence of precipitation and radiation were biome-dependent. This study had a geograph- ical focus on China; a global approach considering several ecosystems across the whole latitudinal gradient is still lack- ing. forests) and when the summer is centered around the middle of the calendar year. The existing methods are, however, not sufﬁciently generic to describe (i) ecosystems in the Southern Hemisphere and (ii) ecosystems with multiple growing sea- sons per year as is often observed in, for example, semiarid regions. Figure 1 illustrates the problem of northern vs. southern hemispheric summers from a conceptual point of view, as- suming that some discrete event recurs annually, but the tim- ing varies according to some external drivers. We would then need to ﬁnd a predictive model explaining the interannual variability of phenology, i.e., the probability of this recurrent event in the course of the annual cycle. Figure 1 shows that linear regression models would be inappropriate to predict the day of the year (DOY) of some phenological event in the Southern Hemisphere as the actual target values to predict may alternate between ⪆3π 2 and ⪅π 2 . In recent years, circular statistics have gained some atten- tion as they offer a solution to problems of this kind (Morel- lato et al., 2010; Beyene et al., 2018). Unlike classical statis- tics, the predicted variables are expressed in terms of an- gular directions (degrees or radians) across a circumference (Fisher, 1995), allowing us to perform statistical analysis where the data space is not Euclidean. Published by Copernicus Publications on behalf of the European Geosciences Union. Published by Copernicus Publications on behalf of the European Geosciences Union. D. E. Pabon-Moreno et al.: Circular statistics in phenology Figure 1. Conceptual distribution of GPPmax timing (DOYGPPmax) for two hypothetical ecosystems: one in the Northern (blue) and one in the Southern Hemisphere (red). The dis- tance between the color line and the circle represents the frequency of the DOYGPPmax observations. The distance between the end and the beginning of the distribution represents the DOYGPPmax interannual variability. D. E. Pabon-Moreno et al.: Circular statistics in phenology 3992 maximum CO2 uptake within a growing season. Ecosystem physio-phenology is inﬂuenced by climate conditions but si- multaneously contributes to the regulation of different micro- and macrometeorological patterns. Physio-phenological cy- cles determine the temporal dynamics of land–atmosphere water and energy exchange ﬂuxes. Likewise, the terrestrial carbon cycle is affected by phenological controls on CO2 up- take and release (Peñuelas et al., 2009). The eddy covariance (EC) technique allows us to contin- uously measure the exchange of energy and matter between ecosystems and the atmosphere (Aubinet et al., 2012). The FLUXNET network collects EC data for most ecosystems of the world along with other meteorological variables, i.e., radiation, temperature, precipitation, atmospheric humidity, and often soil moisture (Baldocchi et al., 2001; Baldocchi, 2020). Particularly relevant to phenological studies is the sea- sonal trajectory of gross primary production (GPP), which allows us to derive phenological transition dates such as start and end of the growing season (e.g., Luo et al., 2018) as well as the timing of the maximum gross primary production, hereafter as referred to as DOYGPPmax (Zhou et al., 2016; Pe- ichl et al., 2018; Wang and Wu, 2019). Figure 1. Conceptual distribution of GPPmax timing (DOYGPPmax) for two hypothetical ecosystems: one in the Northern (blue) and one in the Southern Hemisphere (red). The dis- tance between the color line and the circle represents the frequency of the DOYGPPmax observations. The distance between the end and the beginning of the distribution represents the DOYGPPmax interannual variability. In this study we focus on understanding how climate vari- ability affects the time when ecosystems reach their maxi- mum potential for CO2 absorption. In order to reach this “op- timum state”, several preconditions must be met during the preceding part of the growing season. So far several studies have focused on studying the variability of maximum GPP during the growing season (GPPmax). For instance, Zhou et al. D. E. Pabon-Moreno et al.: Circular statistics in phenology 3993 decay function (Eq. 1), decay function (Eq. 1), decay function (Eq. 1), ⟨xt⟩= Pτ i=0xt−iwi Pτ i=0wi , (1) the following simply named circular regression) allow us to predict circular responses (e.g., the timing of phenological events) from other linear variables (Morellato et al., 2010). Given that any phenological event can be interpreted as an angular direction and should be modeled as such, we assume that these circular regressions are well suited in this context. Despite this evident suitability, circular statistics have not yet been extensively applied in the study of phenology and will therefore be presented here as an alternative to conventional linear techniques. ⟨xt⟩= Pτ i=0xt−iwi Pτ i=0wi , (1) (1) for estimating an exponentially weighted mean of the obser- vation vector xt = (xt,xt−1,...,xt−τ)T at time step t. The symbol ⟨...⟩denotes the weighted average; i indicates the number of days before t, going back to τ = 365 days. The weight decay is represented by wi = w0 exp −i ln(2) t1/2 . (2) In this paper, we aim to identify the factors controlling the timing of the maximal seasonal GPP (DOYGPPmax). The questions we want to answer are as follows: ﬁrst, can cir- cular statistics describe and predict DOYGPPmax per vege- tation type? This aspect requires testing the methodologi- cal advantages and caveats of circular statistics across hemi- spheres in comparison with linear methods. Second, can DOYGPPmax be explained using cumulative climate condi- tions? This question needs to consider different possibili- ties for generating temporally integrating features. And third, how is DOYGPPmax affected by the climatic conditions dur- ing the growing season? The last question requires a global cross-site analysis. Based on the ﬁndings of these three ques- tions, we then discuss the potential of circular regressions beyond this speciﬁc application case in related phenological problems and outline future applications. (2) The decay function gives the instantaneous value a weight of 1 (w0 = 1), and all preceding values receive an exponen- tially reduced weight as determined by the half-time param- eter t1/2. Finally, we make these variables comparable via centering standardization to unit variance. We perform a sen- sitivity analysis, evaluating the effect of the half-time param- eter, and identify the optimum as the value when the variance explained by the circular regression model is at a maximum. The results are presented in Supplement 1. 2.1 Data Since units of the circular response variable must be in radi- ans or degrees, we transform the days of the year to radians using Eq. (3). For leap years we remove the last day. We use 52 EC sites (with at least 7 years of data) located throughout the latitudinal gradient of the globe from the FLUXNET2015 database (Table A1; http://ﬂuxnet.ﬂuxdata. org/, last access: 11 July 2019 Pastorello et al., 2020). Each FLUXNET site is identiﬁed with an abbreviation of the coun- try and the name of the place, e.g., the EC tower AU-How, means that it is located in Howard Springs, Australia. From the dataset we use the GPP data that were derived using the nighttime partitioning method and considering the threshold of the variable u∗to discriminate values of insufﬁcient turbu- lence (Reichstein et al., 2005). In order to identify maximum daily GPP, we compute the quantile 0.9 for each day based on the half-hourly ﬂux observations. As potential explana- tory variables for DOYGPPmax we use the daily air tempera- ture (Tair), shortwave incoming radiation (SWin), precipita- tion (Precip), and vapor pressure deﬁcit (VPD). rad = DOY360 365 π 180, (3) (3) where DOY means day of the year. A basic circular regression model was proposed by Fisher and Lee (1992) as follows: y = µ + 2atan(βixi), (4) (4) y = µ + 2atan(βixi), where y is the target variable (i.e., DOYGPPmax) in radians, µ is the mean angular direction of the target variable x rep where y is the target variable (i.e., DOYGPPmax) in radians, µ is the mean angular direction of the target variable, xi rep- resents the values for the variable i, and βi is the regression coefﬁcient. The parameters µ and β are ﬁtted via the max- imum likelihood method using the reweighted least squares algorithm as proposed by Green (1984). Given that the past climate conditions affect the CO2 ex- change between the atmosphere and the ecosystems (eco- logical memory; Liu et al., 2019; Ryan et al., 2015), we assume that an aggregated form of these climatic variables needs to be considered in the prediction of the phenological responses. D. E. Pabon-Moreno et al.: Circular statistics in phenology Due to the high colinearity between the exponential weighted variables of Tair, SWin, and VPD, we perform a principal component analysis (PCA) on the matrix of vari- ables and FLUXNET sites and retain the leading principal component of these variables as well as precipitation as in- put for the circular statistics model (Hastie et al., 2009). The results of the PCA are presented in Supplement 2. Published by Copernicus Publications on behalf of the European Geosciences Union. In this framework, point events can be described as a von Mises distribution (Von Mises, 1918), the equivalent to the normal distribution in circular statistics. The von Mises distribution is described by two parameters: the mean angular direction (µ) and the concentration parameter (κ). Circular–linear regressions (in The challenge of understanding phenology is generally to characterize a discrete event that repeats periodically. Classi- cally, phenological analyses have been performed using lin- ear regression models (Morente-López et al., 2018; Zhou et al., 2016). Most of these studies analyze ecosystems char- acterized by one growing season (e.g., temperate or boreal Biogeosciences, 17, 3991–4006, 2020 https://doi.org/10.5194/bg-17-3991-2020 3993 2.3 Circular vs. linear regression Interpretation of the coefﬁcients in the circular regres- sion considering a reference point (black) generated with a circular– linear model with mean angular direction (µ = 0), two coefﬁcients (β1, β2) and two variables (x1, x2), where one of the coefﬁcients is negative (β1), and the other one is positive (β2). When the coefﬁ- cient is negative and the value of the parameter increases (blue), the result is an earlier observation compared with the reference point (the equivalent of −0.413 rad is 6.697 rad, as is shown below the equation). On the other hand, when the coefﬁcient is positive and the variable increases (yellow), the observation is later. To quantify the accuracy of each model per coefﬁcient we estimate the mean absolute error per model and coefﬁcient (Eq. 5). To compare the accuracy between models by co- efﬁcient we test the mean absolute errors between models (Eq. 6). To generate a single measure that allows us to com- pare both coefﬁcients and models we estimate the mean dif- ference accuracy (Eq. 7). The results can be understood as follows: if the difference is higher than 0, the circular model has a higher mean accuracy compared to the linear model and vice versa. To quantify which model has higher precision we estimate the difference between the SD of the mean absolute errors per model for each coefﬁcient (Eq. 8). Finally, we es- timate the mean differences of precision between the regres- sion coefﬁcients (Eq. 9), where again if the value is higher than 0, the circular model has a higher mean accuracy than the linear model; the inverse is true if the value is lower than 0. mean angular direction. The inverse would happen when the coefﬁcient is positive. Figure 2 conceptually illustrates how the coefﬁcients affect the predictions. Regarding the second aspect, we can state that if a coefﬁcient is not signiﬁcant, then its contribution would not be relevant to explaining the phe- nological observation. In our case we deﬁne the coefﬁcient to be signiﬁcant if the median of the distribution of p val- ues is less than 0.05. Finally, we can estimate the accuracy of the prediction using the Jammalamadaka–Sarma (JS) cor- relation coefﬁcient (Jammalamadaka and Sarma, 1988). As in any other regression framework, this approach helps us to quantify the effect of each climate variable on the interannual variability of DOYGPPmax. 2.3 Circular vs. linear regression To assess the performance of linear versus circular regres- sions we perform an experiment with simulated data in which we evaluate the accuracy and precision of both approaches to recover original regression coefﬁcients in a circular setting (Eq. 4). We add noise generated with a random von Mises distribution with the parameters n = 100 and κ = 30 to the model to ensure that the result follows a normal distribution. We predeﬁned a range of values for two regression coefﬁ- cients (β1 = ⟨0.01,...,3⟩, β2 = ⟨0.01,...,3⟩). We simulate the variables x1 and x2 as normal distributions with n = 100; a mean of 10 and 15, respectively; and standard deviations (SDs) of 1 and 2. We evaluate all possible combinations for the regression coefﬁcients 100 times, simulating different x1 and x2. In each iteration we generate y using the setup pre- viously described, and we recover the original regression co- efﬁcients using y as a response variable and x1 and x2 as predictors. Finally, we analyze two scenarios: (1) when the target timing occurs at the beginning of the year (µ = 0) and (2) when the target timing happens midyear (µ = π). The pa- rameters for the entire setup generate realistic data, where the standard deviation of y is not higher than 0.3 rad. An SD of 0.5 rad would be equivalent to having phenological observa- tions across half a year, which would not be realistic. Figure 2. Interpretation of the coefﬁcients in the circular regres- sion considering a reference point (black) generated with a circular– linear model with mean angular direction (µ = 0), two coefﬁcients (β1, β2) and two variables (x1, x2), where one of the coefﬁcients is negative (β1), and the other one is positive (β2). When the coefﬁ- cient is negative and the value of the parameter increases (blue), the result is an earlier observation compared with the reference point (the equivalent of −0.413 rad is 6.697 rad, as is shown below the equation). On the other hand, when the coefﬁcient is positive and the variable increases (yellow), the observation is later. Figure 2. 2.1 Data We aggregate the original time series of the Tair, SWin, Precip, and VPD for each DOYGPPmax using a half-life Relevant interpretations of ﬁtted circular regression mod- els are (1) the sign of the β coefﬁcients, (2) the statistical signiﬁcance of the coefﬁcients, and (3) the accuracy of the prediction. Regarding the ﬁrst point, a negative sign of the coefﬁcient would mean that an increasing value of the pre- dictor would lead to an earlier DOYGPPmax compared to the https://doi.org/10.5194/bg-17-3991-2020 Biogeosciences, 17, 3991–4006, 2020 Biogeosciences, 17, 3991–4006, 2020 3994 2.4 Analysis setup The target variable DOYGPPmax is the day of the year when GPP reaches its maximum during the growing season. Given that different ecosystems present more than one growing sea- son per year (e.g., semiarid ecosystems), it is necessary to identify the number of growing seasons per year. To iden- tify the number of growing seasons we apply a fast Fourier transformation (FFT; Cooley and Tukey, 1965) to the mean seasonal cycle of the GPP time series. The number of grow- ing seasons is equal to the maximum absolute value of the ﬁrst four FFT coefﬁcients (excluding the ﬁrst one). For each FLUXNET site, we reconstruct the GPP time series, taking the real numbers of the inverse FFT. We use these recon- structed time series to calculate the expected mean timing of DOYGPPmax and use this value as a template. To recover the real DOYGPPmax from the original time series, we deﬁne a window around the template of length inversely proportional to the number of cycles (180 d/number of growing seasons). To increase the robustness of the analysis we identify the days with the 10 highest GPP values. These days are used in the block bootstrapping mentioned above. Finally, since most of the sites are located in the Northern Hemisphere we expect that, in most cases, DOYGPPmax will be reached by the middle of the calendar year. To illustrate the method in practice, we compare the cir- cular and linear models using data from two sites: US-Ha1 (Northern Hemisphere, deciduous broadleaf forest) and AU- How (Southern Hemisphere, woody savanna). We relate the climate variables with DOYGPPmax (see Methods) and recon- struct the DOYGPPmax using the linear and circular regression models. We compare observed and predicted DOYGPPmax using JS correlation for the circular model and the Pear- son product moment for the linear model. For US-Ha1 both methods show similar performance in predicting DOYGPPmax (Fig. 4), while for AU-How, the circular model retrieves the original data better than the linear model, explaining 30 % more of the variance. In the event that the DOYGPPmax is reached at the beginning of the year, linear methods produce a strong bias that predicts the timing across the entire year (Fig. 4b). To quantify the contribution of each climate variable, we count the number of sites per vegetation type where the re- gression coefﬁcient is statistically signiﬁcant. 2.3 Circular vs. linear regression We estimate regression coefﬁcients for the bootstrap sam- ple i ∈{1,...,m} (m = 100) for the regression coefﬁcient βj, j ∈{1,2}, and the model M ∈{l,c} (denoted as ˆβM j,i). The model accuracy can then be estimated as the mean absolute error of the estimated regression parameter ˆβM j , j ∈{1,2} for the linear model M = l and the circular model M = c: To estimate the relative sensitivity of DOYGPPmax to the leading principal component representing Tair, SWin, and VPD as well as to Precip, we use the implementation of Eq. (4) in the R package “circular” (Agostinelli and Lund, 2017). To increase the robustness of the method we imple- mented a block bootstrapping per growing season, generat- ing a model parameter average based on 1000 iterations. In each analysis, we estimate the accuracy of the model using the JS correlation coefﬁcient. aM,j = 1 m m X i=1 \| ˆβM j,i −βj\|. (5) (5) The difference in accuracy for the coefﬁcient j between the circular and the linear model is shown in δa,j = al,j −ac,j. (6) (6) (6) δa,j = al,j −ac,j. Biogeosciences, 17, 3991–4006, 2020 https://doi.org/10.5194/bg-17-3991-2020 https://doi.org/10.5194/bg-17-3991-2020 3995 3 Results Here, we ﬁrst report results from simulated data to describe the performance of the circular regression approach com- pared to a linear model. Second, we compare the perfor- mance of circular and linear regression using empirical data. Third, we analyze the sensitivity of DOYGPPmax across vege- tation types and climate classes. Finally, we show the results of the predictive power of circular regression per vegetation type. δa = δa,1 + δa,2 2 . (7) (7) The difference in precision for the coefﬁcient j between the linear (l) and the circular model (c) is shown in The difference in precision for the coefﬁcient j between the linear (l) and the circular model (c) is shown in δp,j = sl,j −sc,j. (8) (8) 3.1 Circular vs. linear regression The mean difference in precision between the linear and the circular model is given by The mean difference in precision between the linear and the circular model is given by Figure 3a and c show that for µ = 0 (DOYGPPmax at the be- ginning of the year), circular regression has a higher accuracy and precision compared to the linear regression for the entire space of regression coefﬁcient values, with a maximum dif- ference of the order of 0.1 in terms of accuracy and of the order of 1 for precision. For µ = π (DOYGPPmax midyear) the linear model has a higher accuracy in most of the evalu- ated space, with a maximum difference of the order of 0.001 compared with the circular regression, while circular regres- sion has a higher precision for most of the regression coefﬁ- cients of the order of 0.001. These results show that circular regression has a higher precision in recovering the original regression coefﬁcients than linear regression no matter the moment of the year. On the other hand, circular regression has a higher accuracy than the linear model at the beginning of the year. While linear is better midyear, the differences are of the order of 0.001. δp = δp,1 + δp,2 2 , (9) (9) where sM,j is the sample SD of the vector ( ˆβM j,i)i, M ∈{l,c}. where sM,j is the sample SD of the vector ( ˆβM j,i)i, M ∈{l,c}. D. E. Pabon-Moreno et al.: Circular statistics in phenology Finally, the mean difference in accuracy between the linear and the circular model is given by 2.4 Analysis setup We perform a leave-one-out cross-validation per vegetation type to eval- uate the predictive power of the circular regression using climate conditions. We only consider vegetation types with more than ﬁve sites. In this case the standardization of the climate variables is not applied. Finally, we use the mean of the optimum half-time parameter per vegetation type to weigh the climate conditions. 3.2 Sensitivity of DOYGPPmax to climate variables From the 52 sites analyzed in this study, only one site (ES- LJu) shows bimodal growing seasons (see Supplement 1.2). As expected, in most cases DOYGPPmax occurs in the middle of the calendar year (Fig. S6 in the Supplement), reﬂecting the uneven site distribution in FLUXNET (Schimel et al., 2015). However, some ecosystems in the Northern Hemi- https://doi.org/10.5194/bg-17-3991-2020 https://doi.org/10.5194/bg-17-3991-2020 Biogeosciences, 17, 3991–4006, 2020 D. E. Pabon-Moreno et al.: Circular statistics in phenology D. E. Pabon-Moreno et al.: Circular statistics in phenology 3996 Figure 3. Accuracy and precision of linear and circular regression models by recovering the original regression coefﬁcients of a circular regression. (a, c) µ = 0 (maximum at the beginning of the year). (b, d) µ = pi (maximum midyear). Panels (a) and (b) correspond to the differences in accuracy between the models. Panels (c) and (d) correspond to the differences in the precision between the models. Blue means better performance of the circular model compared with the linear model, and red means higher performance of the linear model. Figure 3. Accuracy and precision of linear and circular regression models by recovering the original regression coefﬁcients of a circular regression. (a, c) µ = 0 (maximum at the beginning of the year). (b, d) µ = pi (maximum midyear). Panels (a) and (b) correspond to the differences in accuracy between the models. Panels (c) and (d) correspond to the differences in the precision between the models. Blue means better performance of the circular model compared with the linear model, and red means higher performance of the linear model. Figure 4. Correlation coefﬁcient between the observed and predicted DOYGPPmax using climatic variables. Two sites are presented: (a) US-Ha1 and (b) AU-How. The observed DOYGPPmax (green) is compared with the data retrieved using circular (orange) and linear (purple) regressions. Two correlation coefﬁcients are used: Jammalamadaka–Sarma (JS) and the Pearson product moment (Pearson). In the circular plot the months and the day of the year (DOY) are also plotted every 75 d. The green arrow indicates the mean angular direction of the original data distribution. Figure 4. Correlation coefﬁcient between the observed and predicted DOYGPPmax using climatic variables. Two sites are presented: (a) US-Ha1 and (b) AU-How. The observed DOYGPPmax (green) is compared with the data retrieved using circular (orange) and linear (purple) regressions. Two correlation coefﬁcients are used: Jammalamadaka–Sarma (JS) and the Pearson product moment (Pearson). In the circular plot the months and the day of the year (DOY) are also plotted every 75 d. The green arrow indicates the mean angular direction of the original data distribution. Biogeosciences, 17, 3991–4006, 2020 D. E. Pabon-Moreno et al.: Circular statistics in phenology Figure 5. Contribution of each climate variable to explain the inter- annual variation in DOYGPPmax per vegetation type. CSHs: closed shrublands (n = 1); DBF: deciduous broadleaf forest (n = 10); EBF: evergreen broadleaf forest (n = 5); ENF: evergreen needleleaf forest (n = 15); GRA: grassland (n = 8); MF: mixed forest (n = 5); OSHs: open shrublands (n = 1); SAV: savannas (n = 1); WET: per- manent wetlands (n = 2); WSAs: woody savannas (n = 3). Each bar shows the cumulative number of sites where each climate variable is statistically signiﬁcant. Table 1. Number of FLUXNET sites where each regression coefﬁ- cient is statistically signiﬁcant to explain the physio-phenology of GPPmax (DOYGPPmax). The table is divided by the sign of the co- efﬁcient. The ﬁrst column is the coefﬁcient for the dimensionality reduction between air temperature (Tair), shortwave incoming radi- ation (SWin), and vapor pressure deﬁcit (VPD); the second column is the coefﬁcient for precipitation (Precip). Climatic variable Sign Tair, SWin, VPD Precip (+) 8 2 (−) 38 14 sphere do reach DOYGPPmax at the beginning of the year; these are Mediterranean sites such as US-Var and ES-LJu. In general terms, most of the sites have an SD between 10 d and 40 d. The maximal SD is 46.9 d for the AU-Tum site. A detailed table with the mean angular direction and SD of DOYGPPmax of each site is presented in Sect. S1.2. For half of the sites, the JS correlation coefﬁcients are between 0.70 and 0.97 (Supplement 1, Fig. S5), showing that the interannual variability of DOYGPPmax is mainly ex- plained by the cumulative effect of the climate variables. Nineteen sites have a JS coefﬁcient of less than 0.7 (DK- Sor, FI-Hyy, US-MMS, DK-ZaH, FR-Pue, US-UMB, AU- Tum, US-Ton, FR-LBr, US-Me2, IT-Lav, AT-Neu, DE-Gri, IT-MBo, IT-Ro2, US-Wkg, BR-Sa1, FR-Fon, CZ-wet). For ES-LJu the JS coefﬁcient is 0.77 for the ﬁrst growing season and 0.78 for the second one (Table S2 in the Supplement). Figure 5. Contribution of each climate variable to explain the inter- annual variation in DOYGPPmax per vegetation type. D. E. Pabon-Moreno et al.: Circular statistics in phenology CSHs: closed shrublands (n = 1); DBF: deciduous broadleaf forest (n = 10); EBF: evergreen broadleaf forest (n = 5); ENF: evergreen needleleaf forest (n = 15); GRA: grassland (n = 8); MF: mixed forest (n = 5); OSHs: open shrublands (n = 1); SAV: savannas (n = 1); WET: per- manent wetlands (n = 2); WSAs: woody savannas (n = 3). Each bar shows the cumulative number of sites where each climate variable is statistically signiﬁcant. We ﬁnd that air temperature, shortwave incoming radi- ation, and vapor pressure deﬁcit appear as the dominant drivers worldwide at 43 of the total sites (84 %; Supple- ment 3). Precipitation is the main driver for ﬁve sites (AU- How US-Ton ZA-Kru US-SRM US-Wkg; Supplement 3). Interestingly, precipitation was the most important factor for all the woody savanna sites (Supplement 3). For three sites (DE-Gri, IT-Ro2, BRSa1), any climatic variable is signiﬁ- cant. In terms of the sign of the coefﬁcients, all the variables are predominantly negative (Table 1). This means that higher values of radiation, air temperature, VPD, and precipitation lead to an earlier DOYGPPmax. Individual sensitivities per site are shown in Supplement 3. Juanes” (ES-LJu), an open shrubland ecosystem in Spain. It is the only clearly bimodal ecosystem in our study (Fig. 6). In this case precipitation is not statistically signiﬁcant, while the combination of Tair, SWin, and VPD is signiﬁcant for both seasons. Furthermore, in both growing seasons Tair, SWin, and VPD have a negative coefﬁcient. The leave-one-out cross-validation for several vegetation types shows that the predictive power of the model for grassland (GRA) and evergreen broadleaf forest (EBF) is −0.3 and −0.31, respectively. For deciduous broadleaf forest (DBF) it is 0.46, and for evergreen needleleaf forest (ENF) it is 0.4, while for mixed forest (MF) the predictive power of the model is 0.88 (Fig. 7). The PCA between shortwave incoming radiation, air tem- perature, and vapor pressure deﬁcit has the highest frequency of signiﬁcant correlation coefﬁcients by number of sites for all the vegetation types with the exception of woody savan- nas (WSAs), where precipitation is shown to be more impor- tant for most sites than the dimensionality reduction between Tair, SWin, and VPD (Fig. 5). For closed shrublands (CSHs) and savannas (SAVs), both drivers have the same number of sites where the coefﬁcients are statistically signiﬁcant. Biogeosciences, 17, 3991–4006, 2020 https://doi.org/10.5194/bg-17-3991-2020 3997 4.2 Sensitivity of DOYGPPmax to climate variables of simulations with higher accuracy and precision than linear regressions. Hence, we would generally suggest that circular regressions may be advantageous when the aim is analyzing the effect of climatic variables on phenological events. We also found cases where the classical linear regression may be either more robust or equally suitable, e.g., when pheno- logical events are reached close to midyear. In the overall view, however, we consider that circular regressions are to be preferred over linear regression for their conceptual capacity to analyze the physio-phenology of ecosystems regardless of the day of the year when an event of interest occurs. This allows us to analyze phenological studies at the global scale regardless of geographic location or the distribution of the observations during the year. The geographical location of the FLUXNET2015 sites rep- resents an advantage when capturing the DOYGPPmax vari- ability at the global scale (Supplement 1, Fig. S6). Most of the analyzed sites (47) are located in the Northern Hemi- sphere. Two sites (GF-Guy and BR-Sa1) are located in the tropical region, and three sites (ZA-Kru, AU-How, AU-Tum) are in the Southern Hemisphere. However, because of the low number of sites reported in the tropical and southern re- gion with more than 7 years of data, our understanding of the DOYGPPmax variability in these regions is still limited. Increasing the number of tropical and Southern Hemisphere sites should be considered a high priority in the near future to complement our knowledge about the physio-phenological ecosystem state. Different phenological models have been developed, rang- ing from empirical approaches (Richardson et al., 2013) to process models (Asse et al., 2020) over the last decades. As we demonstrate here, circular statistics open new opportuni- ties to increase the robustness of phenological models, allow- ing us to analyze ecosystems across hemispheres within the same consistent framework. In fact, the results of the phe- nological sensitivity of DOYGPPmax indicate the complex- ity of ecosystem responses to climate variability. Our ap- proach provides motivation to integrate circular regressions into more complex statistical techniques like regression trees, Gaussian processes, or artiﬁcial neural networks, targeting a circular response variable. The high values of the JS correlation coefﬁcients for most of the sites demonstrate that the interannual variability of DOYGPPmax can be explained as the cumulative effect of the climate variables during the growing season. 4.1 Circular vs. linear regression We explored whether circular regression is a suitable tool for analyzing phenological events. Our results suggest that cir- cular regressions can recover predeﬁned coefﬁcients in a set A special case for understanding the sensitivity of DOYGPPmax to climate variables is the site “Llano de los https://doi.org/10.5194/bg-17-3991-2020 Biogeosciences, 17, 3991–4006, 2020 D. E. Pabon-Moreno et al.: Circular statistics in phenology 3998 Figure 6. DOYGPPmax sensitivity to different climate drivers in a Mediterranean ecosystem: Llano de los Juanes (ES-LJu), Spain, with two growing seasons (green and orange). (a) DOYGPPmax distribution across the year. The arrows indicate the mean angular direction of the growing season. (b) Regression coefﬁcients for each growing season and (c) the signiﬁcance values for each variable. The red line in panel (c) represents a p value of 0.05. Figure 6. DOYGPPmax sensitivity to different climate drivers in a Mediterranean ecosystem: Llano de los Juanes (ES-LJu), Spain, with two growing seasons (green and orange). (a) DOYGPPmax distribution across the year. The arrows indicate the mean angular direction of the growing season. (b) Regression coefﬁcients for each growing season and (c) the signiﬁcance values for each variable. The red line in panel (c) represents a p value of 0.05. 4.2 Sensitivity of DOYGPPmax to climate variables Sites where it was not possible to explain the variations in DOYGPPmax with enough conﬁdence (JS correlation < 0.7) might require the incorporation of biotic variables (e.g., species composition; Peichl et al., 2018) or soil property information that can im- prove the predictive power of the model. Our results suggest that there is no pattern between the DOYGPPmax sensitivity across vegetation types and climate classes (Sect. Fig. S1.7). In other words, the DOYGPPmax Biogeosciences, 17, 3991–4006, 2020 https://doi.org/10.5194/bg-17-3991-2020 3999 5 Conclusions sults are similar to those obtained by Wang and Wu (2019), who were the authors to conclude that an increase in the tem- perature produces an earlier DOYGPPmax. This phenomenon is likely explained by the leaf-out advancing during spring. Nevertheless, there is still no consensus on whether the in- crease in temperature will produce an earlier end of the grow- ing season. Several studies have demonstrated for different vegetation types that when temperature increases, spring on- set is earlier, and autumn senescence is later (Stocker et al., 2013; Linkosalo et al., 2009; Migliavacca et al., 2012; Morin et al., 2010; Post and Forchhammer, 2008), increasing the length of the growing season and the amount of CO2 that is taken up by ecosystems (Richardson et al., 2013). In this study we explored the potential of “circular regres- sions” to explain the physio-phenology of maximal CO2 up- take rates. We conclude that (1) shortwave incoming radi- ation, temperature, and vapor pressure deﬁcit are the main drivers of the timing of maximal CO2 uptake at the global scale (precipitation only plays a secondary role, with the ex- ception of woody savannas, where the most important vari- able is precipitation), and (2) although the sensitivity of the DOYGPPmax to the climate drivers is site-speciﬁc, it is possi- ble to extrapolate the circular regression model for different sites with the same vegetation type and similar latitudes. Fi- nally, we used simulated and empirical data to demonstrate that circular regression produces more accurate results than linear regression, in particular in cases when data need to be explored across hemispheres. Ecosystems with two growing seasons per year represent a very interesting case of the effect of climate drivers on DOYGPPmax across different growing seasons. In Llano de los Juanes, Spain (ES-LJu; Fig. 6), DOYGPPmax is reached in the ﬁrst growing season, when the rainy season is ﬁnishing, while in the second growing season DOYGPPmax is reached in the middle of the rainy season (data not shown). The effect of shortwave incoming radiation, temperature, and vapor pres- sure deﬁcit for both growing seasons is negative, suggesting that if we increase these variables during the period before, the DOYGPPmax will happen earlier. Phenology in Mediterranean ecosystems is mainly con- trolled by water availability (Kramer et al., 2000; Luo et al., 2018; Peñuelas et al., 2009). However, our results suggest that DOYGPPmax is mainly sensitive to SWin, Tair, and VPD. D. E. Pabon-Moreno et al.: Circular statistics in phenology Figure 7. Cross-validation of the circular regression model to predict DOYGPPmax for different vegetation types using air temperature, shortwave incoming radiation, precipitation and vapor pressure deﬁcit (see Methods). Deciduous broadleaf forest (DBF), evergreen broadleaf forest (EBF), grassland (GRA), mixed forest (MF), and evergreen needleleaf forest (ENF). For each vegetation type the Jammalamadaka– Sarma (JS) correlation coefﬁcient is shown in the title of each plot. The red line represents the perfect ﬁt. Figure 7. Cross-validation of the circular regression model to predict DOYGPPmax for different vegetation types using air temperature, shortwave incoming radiation, precipitation and vapor pressure deﬁcit (see Methods). Deciduous broadleaf forest (DBF), evergreen broadleaf forest (EBF), grassland (GRA), mixed forest (MF), and evergreen needleleaf forest (ENF). For each vegetation type the Jammalamadaka– Sarma (JS) correlation coefﬁcient is shown in the title of each plot. The red line represents the perfect ﬁt. pressure deﬁcit (VPD) at the global scale on the DOYGPPmax interannual variability, where for most of the sites these vari- ables have a negative regression coefﬁcient. This means that if the SWin, Tair, and VPD increase during the growing sea- son, the DOYGPPmax will be reached earlier. This effect can be a consequence of DOYGPPmax being reached when SWin and Tair are at a maximum. sensitivity is site-speciﬁc, probably produced by the unique combination of biotic (e.g., species composition, species phenology, species interaction, and phenotypic plasticity) factors that are not evaluated in our study. Several studies that focused on ecosystem phenology suggest that species composition plays a fundamental role in ecosystem physio- phenology of the CO2 uptake (Gonsamo et al., 2017; Peichl et al., 2018). On a global scale, our analysis shows that the combina- tion of air temperature, shortwave incoming radiation, and vapor pressure deﬁcit as well as precipitation has a negative sign. This means that if these variables increase during the growing season, the GPPmax will be reached earlier. Our re- While there is no clear relationship between the DOYGPPmax sensitivity and the vegetation type, we ﬁnd a predominant role of the combined effects of shortwave in- coming radiation (SWin), air temperature (Tair), and vapor https://doi.org/10.5194/bg-17-3991-2020 Biogeosciences, 17, 3991–4006, 2020 4000 https://doi.org/10.5194/bg-17-3991-2020 Biogeosciences, 17, 3991–4006, 2020 https://doi.org/10.5194/bg-17-3991-2020 D. E. Pabon-Moreno et al.: Circular statistics in phenology Appendix A: FLUXNET sites Table A1. FLUXNET sites used in our study. We report the name of the sites, the time period used for the analysis, and the climate class of each site following Köppen–Geiger classiﬁcation: tropical monsoon climate (Am), tropical savanna climate (Aw), cold semiarid climate (BSk), humid subtropical climate (Cfa), oceanic climate (Cfb), hot summer Mediterranean climate (Csa), warm summer Mediterranean climate (Csb), humid subtropical climate (Cwa), humid continental climate (Dfb), subarctic climate (Dfc, Dsc), and tundra climate (ET). We also report the vegetation type of the sites: closed shrubland (CSH), deciduous broadleaf forest (DBF), evergreen broadleaf forest (EBF), evergreen needleleaf forest (ENF), grassland (GRA), mixed forest (MF), open shrubland (OSH), savanna (SAV), permanent wetland (WET), and woody savanna (WSA). Site name Köppen– Vegeta- Period No. years Citation Data DOI Geiger tion analyzed class type AT-Neu Dfc GRA 2002:2012 11 Wohlfahrt et al. (2008) https://doi.org/10.18140/FLX/1440121 AU-How Aw WSA 2002:2014 13 Beringer et al. (2007) https://doi.org/10.18140/FLX/1440125 AU-Tum Cfb EBF 2001:2014 14 Leuning et al. (2005) https://doi.org/10.18140/FLX/1440126 BE-Bra Cfb MF 1999:2002, 2004:2014 15 Carrara et al. (2004) https://doi.org/10.18140/FLX/1440128 BE-Vie Cfb MF 1997:2014 18 Aubinet et al. (2001) https://doi.org/10.18140/FLX/1440130 BR-Sa1 Am EBF 2002:2005, 2009:2011 7 Saleska et al. (2003) https://doi.org/10.18140/FLX/1440032 CA-Man Dfc ENF 1994:1996, 1998:2003 12 Brooks et al. (1997) https://doi.org/10.18140/FLX/1440035 CH-Cha Cfb GRA 2005:2014 10 Merbold et al. (2014) https://doi.org/10.18140/FLX/1440131 CH-Dav ET ENF 1997:2014 18 Zielis et al. (2014) https://doi.org/10.18140/FLX/1440178 CH-Fru Cfb GRA 2005:2014 10 Imer et al. (2013) https://doi.org/10.18140/FLX/1440133 CH-Lae Cfb MF 2004:2014 11 Etzold et al. (2011) https://doi.org/10.18140/FLX/1440134 CZ-wet Cfb WET 2006:2014 9 Dušek et al. (2012) https://doi.org/10.18140/FLX/1440145 DE-Gri Cfb GRA 2004:2014 11 Prescher et al. (2010) https://doi.org/10.18140/FLX/1440147 DE-Hai Cfb DBF 2000:2012 13 Knohl et al. (2003) https://doi.org/10.18140/FLX/1440148 DE-Tha Cfb ENF 1996:2014 19 Grünwald and Bernhofer (2007) https://doi.org/10.18140/FLX/1440152 DK-Sor Cfb DBF 1996:2014 19 Pilegaard et al. (2011) https://doi.org/10.18140/FLX/1440155 DK-ZaH ET GRA 2000:2010, 2012:2014 14 Lund et al. (2012) https://doi.org/10.18140/FLX/1440224 ES-LJu Csa OSH 2005:2013 9 Serrano-Ortiz et al. (2009) https://doi.org/10.18140/FLX/1440226 FI-Hyy Dfc ENF 1996:2014 19 Suni et al. (2003) https://doi.org/10.18140/FLX/1440158 FI-Sod Dfc ENF 2001:2014 14 Thum et al. (2007) https://doi.org/10.18140/FLX/1440160 FR-Fon Cfb DBF 2005:2014 10 Delpierre et al. (2016) https://doi.org/10.18140/FLX/1440161 FR-LBr Cfb ENF 1996:2008 13 Berbigier et al. (2001) https://doi.org/10.18140/FLX/1440163 FR-Pue Csa EBF 2000:2015 15 Rambal et al. (2004) https://doi.org/10.18140/FLX/1440164 GF-Guy Am EBF 2004:2014 11 Bonal et al. (2008) https://doi.org/10.18140/FLX/1440165 IT-Col Csa DBF 1996:2014 19 Valentini et al. (1996) https://doi.org/10.18140/FLX/1440167 IT-Cpz Csa EBF 2000:2008 9 Garbulsky et al. 5 Conclusions These results agree with the analysis performed by Gordo and Sanz (2005), who were the authors to evaluate the phe- nological sensitivity of Mediterranean ecosystems to temper- ature and precipitation. They concluded that temperature was the most important driver. Although water is a limiting factor in Mediterranean ecosystems, its inﬂuence on plant physi- ology and plant phenology can be completely different. In terms of physiology, the GPPmax value can decrease, but in terms of phenology, DOYGPPmax can still be the same. Complex interactions between climate variables and phe- nological response and the interspeciﬁcity of the sensitivity at the site level explain in part the poor predictive power of the model for grasslands, evergreen broadleaf forest, ever- green needleleaf forest, and deciduous broadleaf forests in the cross-validation analysis (Fig. 7). However, the predic- tive power for mixed forest is high, even when the distri- bution of the latitudinal gradient is not the same for all the sites. These results reﬂect the fact that the circular regres- sion model can be extrapolated from different sites to pre- dict the DOYGPPmax interannual variability. This advantage could be a way to solve the common criticism that phenolog- ical models cannot be extrapolated by only generating ad hoc hypotheses (Richardson et al., 2013). Biogeosciences, 17, 3991–4006, 2020 https://doi.org/10.5194/bg-17-3991-2020 4001 https://doi.org/10.5194/bg-17-3991-2020 Competing interests. The authors declare that they have no conﬂict of interest. Competing interests. The authors declare that they have no conﬂict of interest. Aubinet, M., Vesala, T., and Papale, D. (Eds.): Eddy Covariance: A Practical Guide to Measurement and Data Analysis, Springer Atmospheric Sciences, Springer Netherlands, 2012. Baker, B., Guenther, A., Greenberg, J., Goldstein, A., and Fall, R.: Canopy ﬂuxes of 2-methyl-3-buten-2-ol over a ponderosa pine forest by relaxed eddy accumulation: Field data and model comparison, J. Geophys. Res.-Atmos., 104, 26107–26114, https://doi.org/10.1029/1999JD900749, 1999. Acknowledgements. We thank the reviewers for their helpful sug- gestions and Guido Kraemer for his help with the mathemati- cal notation. This project has received funding from the Euro- pean Union’s Horizon 2020 research and innovation program via the TRuStEE project under the Marie Skłodowska-Curie grant agreement no. 721995. This work used eddy covariance data acquired and shared by the FLUXNET community, including these networks: AmeriFlux, AfriFlux, AsiaFlux, CarboAfrica, Car- boEuropeIP, CarboItaly, CarboMont, ChinaFlux, Fluxnet-Canada, GreenGrass, ICOS, KoFlux, LBA, NECC, OzFlux-TERN, TCOS- Siberia, and USCCC. The ERA-Interim reanalysis data are provided by ECMWF and processed by LSCE. The FLUXNET eddy covari- ance data processing and harmonization were carried out by the Eu- ropean Fluxes Database Cluster, AmeriFlux Management Project, and the Fluxdata project of FLUXNET, with the support of CDIAC and ICOS Ecosystem Thematic Center, as well as the OzFlux, Chi- naFlux, and AsiaFlux ofﬁces. Baldocchi, D., Falge, E., Gu, L., Olson, R., Hollinger, D., Run- ning, S., Anthoni, P., Bernhofer, C., Davis, K., Evans, R., Fuentes, J., Goldstein, A., Katul, G., Law, B., Lee, X., Malhi, Y., Meyers, T., Munger, W., Oechel, W., Paw U, K. T., Pile- gaard, K., Schmid, H. P., Valentini, R., Verma, S., Vesala, T., Wilson, K., and Wofsy, S.: FLUXNET: A New Tool to Study the Temporal and Spatial Variability of Ecosystem-Scale Car- bon Dioxide, Water Vapor, and Energy Flux Densities, B. Am. Meteorol. Soc., 82, 2415–2434, https://doi.org/10.1175/1520- 0477(2001)082<2415:FANTTS>2.3.CO;2, 2001. Baldocchi, D. D.: How eddy covariance ﬂux measure- ments have contributed to our understanding of Global Change Biology, Glob. Change Biol., 26, 242–260, https://doi.org/10.1111/gcb.14807, 2020. Bauerle, W. L., Oren, R., Way, D. A., Qian, S. S., Stoy, P. C., Thornton, P. E., Bowden, J. D., Hoffman, F. M., and Reynolds, R. F.: Photoperiodic regulation of the seasonal pattern of photosynthetic capacity and the implications for carbon cycling, P. Natl. Acad. Sci. USA, 109, 8612–8617, https://doi.org/10.1073/pnas.1119131109, 2012. Financial support. This research has been supported by the H2020 Marie Skłodowska-Curie Actions (TRuStEE grant no. References Code availability. Code is available under GPL-3 license at https: //github.com/dpabon/ecosystem-physio-phenology-repo, last ac- cess: 29 July 2020 (https://doi.org/10.5281/zenodo.3921892 Pabon- Moreno et al., 2020). Agostinelli, C. and Lund, U.: R package circular: Circular Statistics (version 0.4-93), CA: Department of Environmental Sciences, Informatics and Statistics, Ca’ Foscari University, Venice, Italys UL: Department of Statistics, California Polytechnic State Uni- versity, San Luis Obispo, California, USA, available at: https:// r-forge.r-project.org/projects/circular/, last access: 29 June 2017. Data availability. FLUXNET database is available online at https://ﬂuxnet.ﬂuxdata.org/, last access: 11 July 2019 (https://doi.org/10.1038/s41597-020-0534-3 Pastorello et al., 2020). Archibald, S. A., Kirton, A., van der Merwe, M. R., Scholes, R. J., Williams, C. A., and Hanan, N.: Drivers of inter-annual variability in Net Ecosystem Exchange in a semi-arid sa- vanna ecosystem, South Africa, Biogeosciences, 6, 251–266, https://doi.org/10.5194/bg-6-251-2009, 2009. Supplement. The supplement related to this article is available on- line at: https://doi.org/10.5194/bg-17-3991-2020-supplement. Asse, D., Randin, C. F., Bonhomme, M., Delestrade, A., and Chuine, I.: Process-based models outcompete correlative mod- els in projecting spring phenology of trees in a future warmer climate, Agr. Forest Meteorol., 285–286, 107931, https://doi.org/10.1016/j.agrformet.2020.107931, 2020. Author contributions. DEPM, TM, MM, and MDM designed the study in collaboration with MR and CR. DEPM conducted the anal- ysis and wrote the manuscript, with substantial contributions from all coauthors. Aubinet, M., Chermanne, B., Vandenhaute, M., Longdoz, B., Yernaux, M., and Laitat, E.: Long term carbon dioxide ex- change above a mixed forest in the Belgian Ardennes, Agr. Forest Meteorol., 108, 293–315, https://doi.org/10.1016/S0168- 1923(01)00244-1, 2001. D. E. Pabon-Moreno et al.: Circular statistics in phenology (2008) https://doi.org/10.18140/FLX/1440168 IT-Lav Cfb ENF 2003:2014 12 Marcolla et al. (2003) https://doi.org/10.18140/FLX/1440169 IT-MBo Dfb GRA 2003:2013 11 Marcolla et al. (2011) https://doi.org/10.18140/FLX/1440170 IT-Noe Csa CSH 2004:2014 11 Marras et al. (2011) https://doi.org/10.18140/FLX/1440171 IT-Ren Dfc ENF 1999, 2002:2003, 2005:2013 12 Montagnani et al. (2009) https://doi.org/10.18140/FLX/1440173 IT-Ro1 Csa DBF 2001:2008 8 Rey et al. (2002) https://doi.org/10.18140/FLX/1440174 IT-Ro2 Csa DBF 2002:2008, 2010:2012 10 Tedeschi et al. (2006) https://doi.org/10.18140/FLX/1440175 IT-SRo Csa ENF 1999:2012 14 Chiesi et al. (2005) https://doi.org/10.18140/FLX/1440176 NL-Loo Cfb ENF 1996:2014 18 Moors (2012) https://doi.org/10.18140/FLX/1440178 RU-Cok Dsc OSH 2003:2013 11 van der Molen et al. (2007) https://doi.org/10.18140/FLX/1440182 RU-Fyo Dfb ENF 1998:2014 17 Kurbatova et al. (2008) https://doi.org/10.18140/FLX/1440183 US-Blo Csa ENF 1997:2007 11 Baker et al. (1999) https://doi.org/10.18140/FLX/1440068 US-GLE Dfc ENF 2005:2014 10 McDowell et al. (2000) https://doi.org/10.18140/FLX/1440069 US-Ha1 Dfb DBF 1992:2012 21 Urbanski et al. (2007) https://doi.org/10.18140/FLX/1440071 US-Los Dfb WET 2001:2008, 2010, 2014 10 Davis et al. (2003) https://doi.org/10.18140/FLX/1440076 US-Me2 Csb ENF 2002:2014 13 Treuhaft et al. (2004) https://doi.org/10.18140/FLX/1440079 US-MMS Cfa DBF 1999:2014 16 Schmid et al. (2000) https://doi.org/10.18140/FLX/1440083 US-NR1 Dfc ENF 1999:2014 16 Monson et al. (2002) https://doi.org/10.18140/FLX/1440087 US-PFa Dfb MF 1996:2014 19 Berger et al. (2001) https://doi.org/10.18140/FLX/1440089 US-SRM BSk WSA 2004:2014 11 Scott et al. (2008) https://doi.org/10.18140/FLX/1440090 US-Syv Dfb MF 2001:2007, 2012:2014 10 Desai et al. (2005) https://doi.org/10.18140/FLX/1440091 US-Ton Csa WSA 2001:2014 14 Xu and Baldocchi (2003) https://doi.org/10.18140/FLX/1440092 US-UMB Dfb DBF 2000:2014 15 Curtis et al. (2002) https://doi.org/10.18140/FLX/1440093 US-Var Csa GRA 2001:2014 14 Xu and Baldocchi (2004) https://doi.org/10.18140/FLX/1440094 US-WCr Dfb DBF 1999:2006, 2011:2014 12 Curtis et al. (2002) https://doi.org/10.18140/FLX/1440095 US-Wkg BSk GRA 2004:2014 11 Emmerich (2003) https://doi.org/10.18140/FLX/1440096 ZA-Kru Cwa SAV 2000:2005, 2007:2013 13 Archibald et al. (2009) https://doi.org/10.18140/FLX/1440188 https://doi.org/10.5194/bg-17-3991-2020 Biogeosciences, 17, 3991–4006, 2020 4002 D. E. Pabon-Moreno et al.: Circular statistics in phenology 4003 Switzerland: Similar Annual Trends, but Seasonal Differences, Ecosystems, 14, 1289–1309, https://doi.org/10.1007/s10021- 011-9481-3, 2011. Switzerland: Similar Annual Trends, but Seasonal Differences, Ecosystems, 14, 1289–1309, https://doi.org/10.1007/s10021- 011-9481-3, 2011. Beringer, J., Hutley, L. B., Tapper, N. J., and Cernusak, L. A.: Savanna ﬁres and their impact on net ecosystem productiv- ity in North Australia, Glob. Change Biol., 13, 990–1004, https://doi.org/10.1111/j.1365-2486.2007.01334.x, 2007. Fisher, N. I.: Statistical Analysis of Circular Data, Cambridge Uni- versity Press, Cambridge, 1995. Beyene, M. T., Jain, S., and Gupta, R. C.: Linear-Circular Statistical Modeling of Lake Ice-Out Dates, Water Resour. Res., 54, 7841– 7858, https://doi.org/10.1029/2017WR021731, 2018. Fisher, N. I. and Lee, A. J.: Regression Models for an Angular Response, Biometrics, 48, 665–677, https://doi.org/10.2307/2532334, 1992. Bonal, D., Bosc, A., Ponton, S., Goret, J.-Y., Burban, B., Gross, P., Bonnefond, J.-M., Elbers, J., Longdoz, B., Epron, D., Guehl, J.-M., and Granier, A.: Impact of severe dry sea- son on net ecosystem exchange in the Neotropical rainfor- est of French Guiana, Glob. Change Biol., 14, 1917–1933, https://doi.org/10.1111/j.1365-2486.2008.01610.x, 2008. Garbulsky, M. F., Peñuelas, J., Papale, D., and Filella, I.: Remote estimation of carbon dioxide uptake by a Mediterranean forest, Glob. Change Biol., 14, 2860–2867, https://doi.org/10.1111/j.1365-2486.2008.01684.x, 2008. Gonsamo, A., D’Odorico, P., Chen, J. M., Wu, C., and Buchmann, N.: Changes in vegetation phenology are not reﬂected in atmo- spheric CO2 and 13C/12C seasonality, Glob. Change Biol., 23, 4029–4044, https://doi.org/10.1111/gcb.13646, 2017. Brooks, J. R., Flanagan, L. B., Varney, G. T., and Ehleringer, J. R.: Vertical gradients in photosynthetic gas exchange characteris- tics and reﬁxation of respired CO2 within boreal forest canopies, Tree Physiol., 17, 1–12, https://doi.org/10.1093/treephys/17.1.1, 1997. Gordo, O. and Sanz, J. J.: Phenology and climate change: a long- term study in a Mediterranean locality, Oecologia, 146, 484–495, https://doi.org/10.1007/s00442-005-0240-z, 2005. Carrara, A., Janssens, I. A., Curiel Yuste, J., and Ceulemans, R.: Seasonal changes in photosynthesis, respiration and NEE of a mixed temperate forest, Agr. Forest Meteorol., 126, 15–31, https://doi.org/10.1016/j.agrformet.2004.05.002, 2004. Green, P. J.: Iteratively Reweighted Least Squares for Maximum Likelihood Estimation, and some Robust and Resistant Alterna- tives, J. Roy. Stat. Soc. B, 46, 149–192, 1984. Chiesi, M., Maselli, F., Bindi, M., Fibbi, L., Cherubini, P., Ar- lotta, E., Tirone, G., Matteucci, G., and Seufert, G.: Modelling carbon budget of Mediterranean forests using ground and re- mote sensing measurements, Agr. Forest Meteorol., 135, 22–34, https://doi.org/10.1016/j.agrformet.2005.09.011, 2005. Grünwald, T. D. E. Pabon-Moreno et al.: Circular statistics in phenology and Bernhofer, C.: A decade of carbon, wa- ter and energy ﬂux measurements of an old spruce for- est at the Anchor Station Tharandt, Tellus B, 59, 387–396, https://doi.org/10.1111/j.1600-0889.2007.00259.x, 2007. Hastie, T., Tibshirani, R., and Friedman, J. H.: The elements of sta- tistical learning: data mining, inference, and prediction, Springer Series in Statistics, 2nd Edn., Springer, New York, NY, 2009. Cooley, J. W. and Tukey, J. W.: An Algorithm for the Machine Cal- culation of Complex Fourier Series, Math. Comput., 19, 297– 301, https://doi.org/10.2307/2003354, 1965. Imer, D., Merbold, L., Eugster, W., and Buchmann, N.: Temporal and spatial variations of soil CO2, CH4 and N2O ﬂuxes at three differently managed grasslands, Biogeosciences, 10, 5931–5945, https://doi.org/10.5194/bg-10-5931-2013, 2013. Curtis, P. S., Hanson, P. J., Bolstad, P., Barford, C., Randolph, J. C., Schmid, H. P., and Wilson, K. B.: Biometric and eddy-covariance based estimates of annual carbon storage in ﬁve eastern North American deciduous forests, Agr. Forest Meteorol., 113, 3–19, https://doi.org/10.1016/S0168-1923(02)00099-0, 2002. Jammalamadaka, S. R. and Sarma, Y.: A correlation coefﬁcient for angular variables, Statistical theory and data analysis II, North- Holland, Amsterdam, 349–364, 1988. Davis, K. J., Bakwin, P. S., Yi, C., Berger, B. W., Zhao, C., Teclaw, R. M., and Isebrands, J. G.: The annual cycles of CO2 and H2O exchange over a northern mixed forest as observed from a very tall tower, Glob. Change Biol., 9, 1278–1293, https://doi.org/10.1046/j.1365-2486.2003.00672.x, 2003. Knohl, A., Schulze, E.-D., Kolle, O., and Buchmann, N.: Large carbon uptake by an unmanaged 250-year-old deciduous for- est in Central Germany, Agr. Forest Meteorol., 118, 151–167, https://doi.org/10.1016/S0168-1923(03)00115-1, 2003. Kramer, K., Leinonen, I., and Loustau, D.: The importance of phenology for the evaluation of impact of climate change on growth of boreal, temperate and Mediterranean forests ecosystems: an overview, Int. J. Biometeorol., 44, 67–75, https://doi.org/10.1007/s004840000066, 2000. Delpierre, N., Berveiller, D., Granda, E., and Dufrêne, E.: Wood phenology, not carbon input, controls the interannual variabil- ity of wood growth in a temperate oak forest, New Phytol., 210, 459–470, https://doi.org/10.1111/nph.13771, 2016. Desai, A. R., Bolstad, P. V., Cook, B. D., Davis, K. J., and Carey, E. V.: Comparing net ecosystem exchange of car- bon dioxide between an old-growth and mature forest in the upper Midwest, USA, Agr. Forest Meteorol., 128, 33–55, https://doi.org/10.1016/j.agrformet.2004.09.005, 2005. ˇ Kurbatova, J., Li, C., Varlagin, A., Xiao, X., and Vygodskaya, N.: Modeling carbon dynamics in two adjacent spruce forests with different soil conditions in Russia, Biogeosciences, 5, 969–980, https://doi.org/10.5194/bg-5-969-2008, 2008. Competing interests. The authors declare that they have no conﬂict of interest. 721995). Financial support. This research has been supported by the H2020 Marie Skłodowska-Curie Actions (TRuStEE grant no. 721995). The article processing charges for this open-access publication were covered by the Max Planck Society. Review statement. This paper was edited by David Bowling and re- viewed by two anonymous referees. Biogeosciences, 17, 3991–4006, 2020 The article processing charges for this open-access publication were covered by the Max Planck Society. Berbigier, P., Bonnefond, J.-M., and Mellmann, P.: CO2 and wa- ter vapour ﬂuxes for 2 years above Euroﬂux forest site, Agr. Forest Meteorol., 108, 183–197, https://doi.org/10.1016/S0168- 1923(01)00240-4, 2001. Review statement. This paper was edited by David Bowling and re- viewed by two anonymous referees. Berger, B. W., Davis, K. J., Yi, C., Bakwin, P. S., and Zhao, C. L.: Long-Term Carbon Dioxide Fluxes from a Very Tall Tower in a Northern Forest: Flux Measurement Methodology, J. Atmos. Ocean. Technol., 18, 529–542, https://doi.org/10.1175/1520- 0426(2001)018<0529:LTCDFF>2.0.CO;2, 2001. https://doi.org/10.5194/bg-17-3991-2020 Biogeosciences, 17, 3991–4006, 2020 Biogeosciences, 17, 3991–4006, 2020 D. E. Pabon-Moreno et al.: Circular statistics in phenology 4004 real trees (1846–2005) support the direct temperature observa- tions of climatic warming, Agr. Forest Meteorol., 149, 453–461, https://doi.org/10.1016/j.agrformet.2008.09.006, 2009. Atmos., 114, D07306, https://doi.org/10.1029/2008JD010650, 2009. Atmos., 114, D07306, https://doi.org/10.1029/2008JD010650, 2009. Moors, E.: Water use of forests in the Netherlands, Tech. rep., Vrije Universiteit, Amsterdam, available at: http://edepot.wur.nl/ 213926 (last access: 26 April 2019), 2012. Liu, Y., Schwalm, C. R., Samuels-Crow, K. E., and Ogle, K.: Ecological memory of daily carbon exchange across the globe and its importance in drylands, Ecol. Lett., 22, 1806–1816, https://doi.org/10.1111/ele.13363, 2019. Morellato, L. P. C., Alberti, L. F., and Hudson, I. L.: Applica- tions of Circular Statistics in Plant Phenology: a Case Stud- ies Approach, in: Phenological Research, Springer, Dordrecht, https://doi.org/10.1007/978-90-481-3335-2_16, 339–359, 2010. Lund, M., Falk, J. M., Friborg, T., Mbufong, H. N., Sigsgaard, C., Soegaard, H., and Tamstorf, M. P.: Trends in CO2 exchange in a high Arctic tundra heath, 2000–2010, J. Geophys. Res.-Biogeo., 117, G02001, https://doi.org/10.1029/2011JG001901, 2012. Morente-López, J., Lara-Romero, C., Ornosa, C., and Iriondo, J. M.: Phenology drives species interactions and modular- ity in a plant–ﬂower visitor network, Sci. Rep., 8, 9386, https://doi.org/10.1038/s41598-018-27725-2, 2018. p g Luo, Y., El-Madany, T. S., Filippa, G., Ma, X., Ahrens, B., Car- rara, A., Gonzalez-Cascon, R., Cremonese, E., Galvagno, M., Hammer, T. W., Pacheco-Labrador, J., Martín, M. P., Moreno, G., Perez-Priego, O., Reichstein, M., Richardson, A. D., Römer- mann, C., and Migliavacca, M.: Using Near-Infrared-Enabled Digital Repeat Photography to Track Structural and Physiolog- ical Phenology in Mediterranean Tree–Grass Ecosystems, Re- mote Sens., 10, 1293, https://doi.org/10.3390/rs10081293, 2018. Morin, X., Roy, J., Sonié, L., and Chuine, I.: Changes in leaf phenology of three European oak species in response to experimental climate change, New Phytol., 186, 900–910, https://doi.org/10.1111/j.1469-8137.2010.03252.x, 2010. Pabon-Moreno, D. E., Musavi, T., Migliavacca, M., Reich- stein, M., Römermann, C., and Mahecha, M. D.: Code- repository for the research paper: 11Ecosystem Physio- phenology revealed using circular statistics”, Zenodo, https://doi.org/10.5281/zenodo.3921892, 2020. Marcolla, B., Pitacco, A., and Cescatti, A.: Canopy Architecture and Turbulence Structure in a Coniferous Forest, Bound.-Layer Me- teorol., 108, 39–59, https://doi.org/10.1023/A:1023027709805, 2003. Pastorello, G., Trotta, C., Canfora, E., et al.: The FLUXNET2015 dataset and the ONEFlux processing pipeline for eddy covariance data, Scientiﬁc Data, 7, 225 pp., https://doi.org/10.1038/s41597- 020-0534-3, 2020. D. E. Pabon-Moreno et al.: Circular statistics in phenology Marcolla, B., Cescatti, A., Manca, G., Zorer, R., Cavagna, M., Fiora, A., Gianelle, D., Rodeghiero, M., Sottocornola, M., and Zampedri, R.: Climatic controls and ecosystem responses drive the inter-annual variability of the net ecosystem exchange of an alpine meadow, Agr. Forest Meteorol., 151, 1233–1243, https://doi.org/10.1016/j.agrformet.2011.04.015, 2011. Peichl, M., Gažoviˇc, M., Vermeij, I., Goede, E. d., Sonnen- tag, O., Limpens, J., and Nilsson, M. B.: Peatland vegeta- tion composition and phenology drive the seasonal trajectory of maximum gross primary production, Sci. Rep., 8, 8012, https://doi.org/10.1038/s41598-018-26147-4, 2018. Marras, S., Pyles, R. D., Sirca, C., Paw U, K. T., Snyder, R. L., Duce, P., and Spano, D.: Evaluation of the Advanced Canopy– Atmosphere–Soil Algorithm (ACASA) model performance over Mediterranean maquis ecosystem, Agr. Forest Meteorol., 151, 730–745, https://doi.org/10.1016/j.agrformet.2011.02.004, 2011. Peñuelas, J., Rutishauser, T., and Filella, I.: Phenology Feedbacks on Climate Change, Science, 324, 887–888, https://doi.org/10.1126/science.1173004, 2009. Pilegaard, K., Ibrom, A., Courtney, M. S., Hummelshøj, P., and Jensen, N. O.: Increasing net CO2 uptake by a Danish beech for- est during the period from 1996 to 2009, Agr. Forest Meteorol., 151, 934–946, https://doi.org/10.1016/j.agrformet.2011.02.013, 2011. McDowell, N. G., Marshall, J. D., Hooker, T. D., and Mus- selman, R.: Estimating CO2 ﬂux from snowpacks at three sites in the Rocky Mountains, Tree Physiol., 20, 745–753, https://doi.org/10.1093/treephys/20.11.745, 2000. Merbold, L., Eugster, W., Stieger, J., Zahniser, M., Nel- son, D., and Buchmann, N.: Greenhouse gas budget (CO2, CH4 and N2O) of intensively managed grassland fol- lowing restoration, Glob. Change Biol., 20, 1913–1928, https://doi.org/10.1111/gcb.12518, 2014. Post, E. and Forchhammer, M. C.: Climate change reduces re- productive success of an Arctic herbivore through trophic mismatch, Philos. T. Roy. Soc. B-Biol., 363, 2367–2373, https://doi.org/10.1098/rstb.2007.2207, 2008. Prescher, A.-K., Grünwald, T., and Bernhofer, C.: Land use regulates carbon budgets in eastern Germany: From NEE to NBP, Agr. Forest Meteorol., 150, 1016–1025, https://doi.org/10.1016/j.agrformet.2010.03.008, 2010. Migliavacca, M., Sonnentag, O., Keenan, T. F., Cescatti, A., O’Keefe, J., and Richardson, A. D.: On the uncertainty of phe- nological responses to climate change, and implications for a terrestrial biosphere model, Biogeosciences, 9, 2063–2083, https://doi.org/10.5194/bg-9-2063-2012, 2012. Rambal, S., Joffre, R., Ourcival, J. M., Cavender-Bares, J., and Rocheteau, A.: The growth respiration component in eddy CO2 ﬂux from a Quercus ilex mediterranean forest, Glob. Change Biol., 10, 1460–1469, https://doi.org/10.1111/j.1365- 2486.2004.00819.x, 2004. Monson, R. K., Turnipseed, A. A., Sparks, J. P., Harley, P. C., Scott- Denton, L. E., Sparks, K., and Huxman, T. D. E. Pabon-Moreno et al.: Circular statistics in phenology Leuning, R., Cleugh, H. A., Zegelin, S. J., and Hughes, D.: Carbon and water ﬂuxes over a temperate Eucalyptus forest and a tropi- cal wet/dry savanna in Australia: measurements and comparison with MODIS remote sensing estimates, Agr. Forest Meteorol., 129, 151–173, https://doi.org/10.1016/j.agrformet.2004.12.004, 2005. Dušek, J., ˇCížková, H., Stellner, S., Czerný, R., and Kvˇet, J.: Fluc- tuating water table affects gross ecosystem production and gross radiation use efﬁciency in a sedge-grass marsh, Hydrobiologia, 692, 57–66, https://doi.org/10.1007/s10750-012-0998-z, 2012. Emmerich, W. E.: Carbon dioxide ﬂuxes in a semiarid environment with high carbonate soils, Agr. Forest Meteorol., 116, 91–102, https://doi.org/10.1016/S0168-1923(02)00231-9, 2003. Lieth, H. (Ed.): Phenology and Seasonality Modeling, Ecological Studies, Springer-Verlag, Berlin Heidelberg, 1974. Etzold, S., Ruehr, N. K., Zweifel, R., Dobbertin, M., Zingg, A., Pluess, P., Häsler, R., Eugster, W., and Buchmann, N.: The Car- bon Balance of Two Contrasting Mountain Forest Ecosystems in Linkosalo, T., Häkkinen, R., Terhivuo, J., Tuomenvirta, H., and Hari, P.: The time series of ﬂowering and leaf bud burst of bo- https://doi.org/10.5194/bg-17-3991-2020 Biogeosciences, 17, 3991–4006, 2020 Biogeosciences, 17, 3991–4006, 2020 D. E. Pabon-Moreno et al.: Circular statistics in phenology F., Migliavacca, M., Ryu, Y., Sonnentag, O., and Toomey, M.: Climate change, phenol- ogy, and phenological control of vegetation feedbacks to the climate system, Agr. Forest Meteorol., 169, 156–173, https://doi.org/10.1016/j.agrformet.2012.09.012, 2013. Valentini, R., Angelis, P. D., Matteucci, G., Monaco, R., Dore, S., and Mucnozza, G. E. S.: Seasonal net carbon dioxide exchange of a beech forest with the atmosphere, Glob. Change Biol., 2, 199–207, https://doi.org/10.1111/j.1365- 2486.1996.tb00072.x, 1996. Ryan, E. M., Ogle, K., Zelikova, T. J., LeCain, D. R., Williams, D. G., Morgan, J. A., and Pendall, E.: Antecedent moisture and temperature conditions modulate the response of ecosystem res- piration to elevated CO2 and warming, Glob. Change Biol., 21, 2588–2602, https://doi.org/10.1111/gcb.12910, 2015. van der Molen, M. K., van Huissteden, J., Parmentier, F. J. W., Petrescu, A. M. R., Dolman, A. J., Maximov, T. C., Kononov, A. V., Karsanaev, S. V., and Suzdalov, D. A.: The growing sea- son greenhouse gas balance of a continental tundra site in the Indigirka lowlands, NE Siberia, Biogeosciences, 4, 985–1003, https://doi.org/10.5194/bg-4-985-2007, 2007. Saleska, S. R., Miller, S. D., Matross, D. M., Goulden, M. L., Wofsy, S. C., da Rocha, H. R., de Camargo, P. B., Crill, P., Daube, B. C., de Freitas, H. C., Hutyra, L., Keller, M., Kirch- hoff, V., Menton, M., Munger, J. W., Pyle, E. H., Rice, A. H., and Silva, H.: Carbon in Amazon Forests: Unexpected Seasonal Fluxes and Disturbance-Induced Losses, Science, 302, 1554– 1557, https://doi.org/10.1126/science.1091165, 2003. Von Mises, R.: Über die “Ganzzahligkeit” der Atomgewichte und verwandte Fragen, Phys. Z., 19, 490–500, 1918. Wang, X. and Wu, C.: Estimating the peak of growing season (POS) of China’s terrestrial ecosystems, Agr. Forest Meteorol., 278, 107639, https://doi.org/10.1016/j.agrformet.2019.107639, 2019. Schimel, D., Pavlick, R., Fisher, J. B., Asner, G. P., Saatchi, S., Townsend, P., Miller, C., Frankenberg, C., Hibbard, K., and Cox, P.: Observing terrestrial ecosystems and the car- bon cycle from space, Glob. Change Biol., 21, 1762–1776, https://doi.org/10.1111/gcb.12822, 2015. Wohlfahrt, G., Hammerle, A., Haslwanter, A., Bahn, M., Tappeiner, U., and Cernusca, A.: Seasonal and inter-annual variability of the net ecosystem CO2 exchange of a temperate mountain grassland: Effects of weather and management, J. Geophys. Res.-Atmos., 113, D08110, https://doi.org/10.1029/2007JD009286, 2008. Schmid, H. P., Grimmond, C. S. B., Cropley, F., Offerle, B., and Su, H.-B.: Measurements of CO2 and energy ﬂuxes over a mixed hardwood forest in the mid-western United States, Agr. Forest Meteorol., 103, 357–374, https://doi.org/10.1016/S0168- 1923(00)00140-4, 2000. Xu, L. and Baldocchi, D. D. E. Pabon-Moreno et al.: Circular statistics in phenology 4005 the ﬁfth assessment report of the intergovernmental panel on climate change, 1535, Cambridge university press Cambridge, United Kingdom and New York, NY, USA, 2013. the separation of net ecosystem exchange into assimilation and ecosystem respiration: review and improved algorithm, Glob. Change Biol., 11, 1424–1439, https://doi.org/10.1111/j.1365- 2486.2005.001002.x, 2005. Suni, T., Rinne, J., Reissell, A., Altimir, N., Keronen, P., Rannik, U., Dal Maso, M., Kulmala, M., and Vesala, T.: Long-term mea- surements of surface ﬂuxes above a Scots pine forest in Hyytiala, southern Finland, 1996–2001, Boreal Environ. Res., 8, 287–301, 2003. Rey, A., Pegoraro, E., Tedeschi, V., Parri, I. D., Jarvis, P. G., and Valentini, R.: Annual variation in soil respiration and its compo- nents in a coppice oak forest in Central Italy, Glob. Change Biol., 8, 851–866, https://doi.org/10.1046/j.1365-2486.2002.00521.x, 2002. Tedeschi, V., Rey, A., Manca, G., Valentini, R., Jarvis, P. G., and Borghetti, M.: Soil respiration in a Mediterranean oak forest at different developmental stages after coppicing, Glob. Change Biol., 12, 110–121, https://doi.org/10.1111/j.1365- 2486.2005.01081.x, 2006. Richardson, A. D., Braswell, B. H., Hollinger, D. Y., Jenkins, J. P., and Ollinger, S. V.: Near-surface remote sensing of spatial and temporal variation in canopy phenology, Ecol. Appl., 19, 1417– 1428, https://doi.org/10.1890/08-2022.1, 2009. g Richardson, A. D., Andy Black, T., Ciais, P., Delbart, N., Friedl, M. A., Gobron, N., Hollinger, D. Y., Kutsch, W. L., Longdoz, B., Luyssaert, S., Migliavacca, M., Montagnani, L., William Munger, J., Moors, E., Piao, S., Rebmann, C., Reich- stein, M., Saigusa, N., Tomelleri, E., Vargas, R., and Varlagin, A.: Inﬂuence of spring and autumn phenological transitions on forest ecosystem productivity, Philos. T. Roy. Soc. B-Biol., 365, 3227–3246, https://doi.org/10.1098/rstb.2010.0102, 2010. Thum, T., Aalto, T., Laurila, T., Aurela, M., Kolari, P., and Hari, P.: Parametrization of two photosynthesis models at the canopy scale in a northern boreal Scots pine forest, Tellus B, 59, 874– 890, https://doi.org/10.1111/j.1600-0889.2007.00305.x, 2007. Treuhaft, R. N., Law, B. E., and Asner, G. P.: Forest Attributes from Radar Interferometric Structure and Its Fusion with Optical Re- mote Sensing, BioScience, 54, 561–571, 2004. Urbanski, S., Barford, C., Wofsy, S., Kucharik, C., Pyle, E., Bud- ney, J., McKain, K., Fitzjarrald, D., Czikowsky, M., and Munger, J. W.: Factors controlling CO2 exchange on timescales from hourly to decadal at Harvard Forest, J. Geophys. Res.-Biogeosci., 112, G02020, https://doi.org/10.1029/2006JG000293, 2007. Richardson, A. D., Keenan, T. D. E. Pabon-Moreno et al.: Circular statistics in phenology E.: Carbon sequestra- tion in a high-elevation, subalpine forest, Glob. Change Biol., 8, 459–478, https://doi.org/10.1046/j.1365-2486.2002.00480.x, 2002. Reichstein, M., Falge, E., Baldocchi, D., Papale, D., Aubinet, M., Berbigier, P., Bernhofer, C., Buchmann, N., Gilmanov, T., Granier, A., Grünwald, T., Havránková, K., Ilvesniemi, H., Janous, D., Knohl, A., Laurila, T., Lohila, A., Loustau, D., Mat- teucci, G., Meyers, T., Miglietta, F., Ourcival, J.-M., Pumpanen, J., Rambal, S., Rotenberg, E., Sanz, M., Tenhunen, J., Seufert, G., Vaccari, F., Vesala, T., Yakir, D., and Valentini, R.: On Montagnani, L., Manca, G., Canepa, E., Georgieva, E., Acosta, M., Feigenwinter, C., Janous, D., Kerschbaumer, G., Lindroth, A., Minach, L., Minerbi, S., Mölder, M., Pavelka, M., Seufert, G., Zeri, M., and Ziegler, W.: A new mass conservation approach to the study of CO2 advection in an alpine forest, J. Geophys. Res.- G., Vaccari, F., Vesala, T., Yakir, D., and Valentini, R.: On https://doi.org/10.5194/bg-17-3991-2020 Biogeosciences, 17, 3991–4006, 2020 D. E. Pabon-Moreno et al.: Circular statistics in phenology D.: Seasonal trends in pho- tosynthetic parameters and stomatal conductance of blue oak (Quercus douglasii) under prolonged summer drought and high temperature, Tree Physiol., 23, 865–877, https://doi.org/10.1093/treephys/23.13.865, 2003. Scott, R. L., Cable, W. L., and Hultine, K. R.: The eco- hydrologic signiﬁcance of hydraulic redistribution in a semiarid savanna, Water Resour. Res., 44, W02440, https://doi.org/10.1029/2007WR006149, 2008. Xu, L. and Baldocchi, D. D.: Seasonal variation in car- bon dioxide exchange over a Mediterranean annual grass- land in California, Agr. Forest Meteorol., 123, 79–96, https://doi.org/10.1016/j.agrformet.2003.10.004, 2004. Serrano-Ortiz, P., Domingo, F., Cazorla, A., Were, A., Cuezva, S., Villagarcía, L., Alados-Arboledas, L., and Kowalski, A. S.: In- terannual CO2 exchange of a sparse Mediterranean shrubland on a carbonaceous substrate, J. Geophys. Res.-Biogeosci., 114, G04015, https://doi.org/10.1029/2009JG000983, 2009. Zhang, X., Friedl, M. A., Schaaf, C. B., Strahler, A. H., Hodges, J. C. F., Gao, F., Reed, B. C., and Huete, A.: Monitoring vegetation phenology using MODIS, Remote Sensing Environ., 84, 471– 475, https://doi.org/10.1016/S0034-4257(02)00135-9, 2003. Stocker, T. F., Qin, D., Plattner, G.-K. et al.: Climate change 2013: The physical science basis, Contribution of working group I to https://doi.org/10.5194/bg-17-3991-2020 Biogeosciences, 17, 3991–4006, 2020 D. E. Pabon-Moreno et al.: Circular statistics in phenology https://doi.org/10.5194/bg-17-3991-2020 Biogeosciences, 17, 3991–4006, 2020 D. E. Pabon-Moreno et al.: Circular statistics in phenology 4006 Zhou, S., Zhang, Y., Caylor, K. K., Luo, Y., Xiao, X., Ciais, P., Huang, Y., and Wang, G.: Explaining inter-annual vari- ability of gross primary productivity from plant phenology and physiology, Agr. Forest Meteorol., 226-227, 246–256, https://doi.org/10.1016/j.agrformet.2016.06.010, 2016. Zhou, S., Zhang, Y., Ciais, P., Xiao, X., Luo, Y., Caylor, K. K., Huang, Y., and Wang, G.: Dominant role of plant physiology in trend and variability of gross primary productivity in North America, Sci. Rep., 7, 41366, https://doi.org/10.1038/srep41366, 2017. Zielis, S., Etzold, S., Zweifel, R., Eugster, W., Haeni, M., and Buchmann, N.: NEP of a Swiss subalpine forest is signiﬁcantly driven not only by current but also by previous year’s weather, Biogeosciences, 11, 1627–1635, https://doi.org/10.5194/bg-11- 1627-2014, 2014. Biogeosciences, 17, 3991–4006, 2020 https://doi.org/10.5194/bg-17-3991-2020 Biogeosciences, 17, 3991–4006, 2020
https://openalex.org/W3014698044	https://www.emerald.com/insight/content/doi/10.1108/AAOUJ-09-2019-0043/full/pdf?title=analysis-of-antecedent-factors-in-academic-achievement-and-student-retention	English	null	Analysis of antecedent factors in academic achievement and student retention	AAOU Journal/AAOU journal	2,020	cc-by	6,237	© Rony Setiawan, Ariesya Aprillia and Nonie Magdalena. Published in Asian Association of Open Universities Journal. Published by Emerald Publishing Limited. This article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and create derivative works of this article (for both commercial and non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence may be seen at http:// creativecommons.org/licences/by/4.0/legalcode The current issue and full text archive of this journal is available on Emerald Insight at: https://www.emerald.com/insight/2414-6994.htm The current issue and full text archive of this journal is available on Emerald Insight at: https://www.emerald.com/insight/2414-6994.htm Abstract Purpose – Student achievement and retention are two important components that indicate the quality of a university. This is in line with one of the standards of university accreditations regarding the quality of students and graduates, which are set by BAN-PT, an institution that guarantees the quality of National Education Service delivery at every university in Indonesia. The purpose of this paper is analyzing the internal and external factors that contribute to academic achievement and student retention. Design/methodology/approach – This paper is a causal explanatory-based study, using multiple regression analysis among five independent variables and two dependent variables separately. Several sequential tests are conducted to maintain the proper research results, such as reliability, normality, multi- collinearity and heteroscedasticity. Findings – This paper provides empirical facts that student motivation has a significant positive impact on the level of student academic achievement, and the quality of lecturers has a significant positive impact both on student academic achievement and retention rates. Research limitations/implications – This study is conducted within the context of a university. The generalization in the study is low. Researchers are encouraged to explore further. Practical implications – This paper is expected to provide constructive feedback to university management in setting policies that are oriented toward strategic actions to optimize academic achievement and maintain their students. Originality/value – This paper provides insights of good university governance concerning in highe education management. Keywords Student, Family, University, Lecturer, Academic advisor Paper type Research paper 37 Received 30 September 2019 Revised 2 November 2019 12 November 2019 Accepted 19 November 2019 g Bandung, Indonesia Analysis of antecedent factors in academic achievement and student retention Student achievement and retention 37 Asian Association of Open Universities Journal Vol. 15 No. 1, 2020 pp. 37-47 Emerald Publishing Limited e-ISSN: 2414-6994 p-ISSN: 1858-3431 DOI 10.1108/AAOUJ-09-2019-0043 1. Introduction A fi Asonetypeofinstitution engagedinservices,universitiesprovideeducationalservicestotheir students, both at the undergraduate or diploma level, as well as postgraduate (master’s or doctoral) degrees. The sustainability of higher education is very much determined by the synergy of the performance of each faculty and the study programs it supports. The performance of a university will be guaranteed if it is nationally recognized by the relevant authorized institution, which in this context is the National Accreditation Board for Higher Education (BAN-PT). According to Wikipedia, BAN-PT is the only accreditation body that obtained authority from the Ministry of Research, Technology, and Higher Education of the Republic of Indonesia in improving the quality of higher education, introducing and disseminating “new paradigms in the management of higher education” and increasing relevance, academic atmosphere, institution management, efficiency and sustainability of Asian Association of Open Universities Journal Vol. 15 No. 1, 2020 pp. 37-47 Emerald Publishing Limited e-ISSN: 2414-6994 p-ISSN: 1858-3431 OI 10.1108/AAOUJ-09-2019-0043 AAOUJ 15,1 AAOUJ 15,1 university. Referring to the accreditation forms system established by BAN-PT, there are nine standardsthat becomeassessment factors, which include vision, mission, goals and strategies; governance, management, and cooperation; college student; human resources; finance, facilities and infrastructure; education; research and community service. Among the nine standards,thequality ofstudentsand graduatesarethecrucialaspectsthatmustbeconcerned by the university. The goodness of two factors contribute to the final result of BAN’s accreditation evaluation of a university, which is grade A reflects that the university shows its national eligibility (recommended) to every students because already fulfilled all those nine standards of a great university. The credibility of a university is determined by the quality of its students. Student achievements will be a benchmark for the success of a university in the eyes of end-users who will empower them in the corporate world. This is called “external to external stakeholders”. In terms of externalities, the reputation of a university from the perspective of graduate users is determined by the achievements of its graduate students. 38 Student performance can be measured from many aspects, one of which is through academic achievement. Student academic achievement can be appraised by the amount of their grade point average (GPA) achieved and the social activities they carry out through the intensity of their involvement in student activity units available at the university or study program where they study. 1. Introduction A fi In order to maintain the quality of graduates, study programs need to analyze and identify the factors that contribute to the academic achievement of their students. The study programs must be proactive because the graduate students will carry the “good name” of their alma mater when they begin to take part in the world of work in the company environment where they will work later. Continuity of the success of students’ academic achievement needs to be created and maintained through the effectiveness of the university’s management of various aspects that can influence the achievement. Students are university business sources and partners. The long-term sustainability of a university depends on the continuity of the satisfaction of its students on the performance of the institution in which they are studying. The positive perception of students regarding the quality of education delivery is reflected in their decision to remain committed in completing studies from the time they enter college until their graduation. A university can be said to be successful in realizing its quality management and customer relations if its students show high loyalty in their actions to keep their education contracts with the university. This is what is called the student retention rate. The success of a university in maintaining its students indicates that there is a consistency of students’ trust in the university they choose. The university’s interests are seen by various stakeholders, such as society, government, parents of students and students themselves. University really needs to pay attention to the ability of their students as a form of normative responsibility in which educational institutions should provide the best service, one of which is reflected in monitoring and handling student academic achievement. From the point of view of morality, the university’s concern for the academic achievements of its students is also a form of realization of social and moral responsibility, which needs to motivate their students so that they have the enthusiasm to achieve the best achievements in the education they are going through. High quality management of higher education will create an atmosphere of a comfortable, pleasant and positive university environment for students who are in it. This stimulus is expected to be an external source of energy which directly motivates students to continue their studies until they finish their university studies. 2. Literature review A d i f Academic performance of student can be seen from the GPA obtained. The index shows the achievement scores of students per semester in an aggregate manner from the time students go to college to the latest conditions. The index also reflects the accumulated consistency of student achievement for all courses that have been taken. Each course usually consists of three measurement components, namely, the Mid Semester Exam (UTS), Structured Academic Activity (KAT) and the Final Semester Examination (UAS). The composition of the three components varies according to the policies of each lecturer, with weights ranging from 20 to 50%, but there are also a number of lecturers who emphasize the KAT value because the subject is considered to emphasize more on aspects of practice and active participation. The final grade criteria for each course are grades A (equivalent to 4) with a range of values equal or greater than 80, grades Bþ (equivalent to 3.5) with a range of values greater than 73 to less than 80, grades B (equivalent to 3) with a range of values greater than 67 to less than 73, grades Cþ (equivalent to 2.5) with a range of values greater than 61 to with less than 67, grades C (equivalent to 2) with a range of values greater than 55 to less than 61, grades D (equivalent to 1) with a range of values greater than 41 to less than 55 and grades E (equivalent to zero) with a range of values less than 41. Achievement Index (GPA) is obtained from the accumulation of the final conversion value multiplication in the form of numbers with weighted credits (semester credit units, usually around two or three credits where one credit is equivalent to 50 min of weekly meeting in the class) for each course, while the GPA is the average student GPA for each semester. Aside from the GPA, academic achievement can be evaluated from the participation of students in various activity units at the university. Student participation in various organizations on campus, both at the level of study programs or faculty members and universities, such as student affairs, nature lovers, choirs, spirituality or religion and the like; be an indication that the student has a positive desire to develop and actualize his personality, both for himself or herself, the institution and others. Student achievement and retention 1. Introduction A fi University reliability is proven by how well they can generate graduates with very satisfying academic performance and maintain the commitment of their students in their education. The success of university relied on how well its management responds to change. Today, disruption era crucially affects many aspects of life, education is one of them. The advancement of technology drives university in creating and conducting teaching and learning system within digital space, where every lecturer and student interact indirectly through media, such as electronic devices, software and the Internet. The fact is education disruption already happened, that is, through open universities. Both open and non-open universities continuously need to focus seriously in running their education system, focusing on the performance of their students. The goodness of a sustainable university is seen from the greatness of its students from different generations, in terms of numbers and capabilities. 39 AAOUJ 15,1 involvement, faculty and staff members approachability, business procedures, learning experiences and student support services (Chuah and Lim, 2018). There are many factors that contribute to the advancement of academic achievement and student retention, both from internal and external factors. External factors derive from the personal and family of the students themselves, while external factors derive from universities, lecturers in the class and academic advisor. Good motivation actually comes from us. Motivation speaks of will, effort and persistence. The performance of a student is sustained by himself or herself. Although this is true, students are also human beings who cannot be separated from the emotional elements that can influence him or her in showing attitudes and behavior. The atmosphere of emotions can be influenced by environmental conditions around the students themselves, both direct and indirect environments. In this context, the immediate environment means the families of students who can play a role in creating an atmosphere that influences the spirit of students in carrying out their education, especially if between students and family students have a close relationship both physically and/or psychologically. The indirect environment of students is related to the condition of the university as a “second home” for him or her as a learner. The quality of the university, the competence of lecturers and the role of academic advisors as mentors can be a catalyst for student success in achieving academic achievement as well as consideration in his or her decision to continue his education at the university. 40 According to Suhaily and Soelasih (2015), physical and psychological stability (manifest or behavior) of students determines their academic success, which includes how well the level of student health through adequate sleep and student perceptions in addressing current lectures and students’ perspectives in connecting between his or her success in the world of education with the world of work or business later through the level of attendance in the classroom, readiness to complete each lecture assignment and focus in providing maximum effort in following the teaching and learning process at the university (Bakar et al., 2010; Garkaz et al., 2011; Thawabieh, 2016). 2. Literature review A d i f Students can be smart in hard skills through teaching and learning in the classroom and capable in soft skills through empowering themselves in the community on campus, so they have a whole academic achievement. Students are the main customers of the university. The existence of a university is not merely determined by the academic achievements of its students but also by its ability to retain students to continue their studies at the university. The student retention rate shows the student’s confidence in his or her university. A high student retention rate indicates that the university has a good performance regarding management’s commitment to service quality in maintaining students’ trust in the quality of education provision from the university which they have decided to enter from the beginning. A low student retention rate indicates that there are aspects of service that need to be addressed immediately by the university which undermine the commitment of students to continue their education at the university. Student retention really needs attention from a university because this can be contagious, both to fellow students at the university and also prospective students who have the potential to approach or stay away. Hennig-Thurau et al., (2001) prove that the quality of teaching and student emotional commitment to the institution is very important to the student’s loyalty to his or her university. Another study finds that there are several things affecting student retention, such as academic advising, social connectedness, student AAOUJ 15,1 AAOUJ 15,1 According to Suhaily and Soelasih (2015), family conditions (both in terms of quantity and quality) play a role in determining the academic success of students, which includes how many children in a student’s family support or disrupt student concentration in pursuing lectures, economic level and background education of student parents in providing financial and educational support, closeness and involvement of student parents who act as personal counselors in listening to problems, giving advice and encouraging and harmonious relationships between parents as a positive atmosphere in the student family environment (Singh et al., 2016). According to Suhaily and Soelasih (2015), the atmosphere of the university (both physically and psychologically) influences the academic success of its students, which includes the availability and condition of university facilities and infrastructure (buildings, elevators, classrooms, air conditioners, etc.) in providing comfort in the campus environment, the university’s good name in providing pride for students, as well as the number of students and interactions with classmates in providing comfort while attending class (Mushtaq and Khan, 2012). According to Suhaily and Soelasih (2015), the competency of hard skills and soft skills held by lecturers as educators determines student academic success, which includes the readiness of lecturers in preparing and delivering lecture material so that students can understand what is explained by lecturers accompanied by the use of helping devices that support and comfort students in concentrating while in the classroom, the positive attitude of the lecturer toward students shown by the ability of lecturers to maintain good relations with students and create interactive classes, as well as qualified classroom management that is seen from the mastery of lecturers in creating student discipline while in class. Academic advisors are regular lecturers assigned by the study program in providing academic services to the advisee students they support. They are likened to “parents on campus” for their advisee students. Academic advisors play a role in providing academic direction and advice, including discussions in dealing with personal problems faced by their advisee students, especially those related to their college life while on campus and the surrounding environment. The study program has a policy that there must be at least four times face-to-face meetings between academic advisors and their advisee students, each of which includes future study plans, online advising, DKBS signature (Study Load Contract Documents) and ongoing study evaluations and discussion. AAOUJ 15,1 The academic advisor plays a role in providing academic advising to the students he or she teaches, he or she acts like a mentor. Mentoring is a guidance process, in which a person, who has more qualified skills and experience, acts as a role-model, lecturer or teacher, sponsor, motivator, counselor and friend with the aim of enhancing and developing the personal development and professionalism of others who do not have experience (Demir et al., 2014). Mentoring is a means of providing visionary support, motivation and feedback to students (Demir et al., 2014). Mentoring has a positive impact on students’ ability to deal with stress during study and their success in completing their education at the university (Demir et al., 2014). Students highly value academic advisors who are easy to meet or contact, are friendly, and help in providing direction in connecting their academic experience with future life plans (Muola et al., 2011; Ismail and Jui, 2014). direction and advice, including discussions in dealing with personal problems faced by their advisee students, especially those related to their college life while on campus and the surrounding environment. The study program has a policy that there must be at least four times face-to-face meetings between academic advisors and their advisee students, each of which includes future study plans, online advising, DKBS signature (Study Load Contract Documents) and ongoing study evaluations and discussion. The academic advisor plays a role in providing academic advising to the students he or she teaches, he or she acts like a mentor. Mentoring is a guidance process, in which a person, who has more qualified skills and experience, acts as a role-model, lecturer or teacher, sponsor, motivator, counselor and friend with the aim of enhancing and developing the personal development and professionalism of others who do not have experience (Demir et al., 2014). Mentoring is a means of providing visionary support, motivation and feedback to students (Demir et al., 2014). Mentoring has a positive impact on students’ ability to deal with stress during study and their success in completing their education at the university (Demir et al., 2014). Students highly value academic advisors who are easy to meet or contact, are friendly, and help in providing direction in connecting their academic experience with future life plans (Muola et al., 2011; Ismail and Jui, 2014). 41 J ) The research model is presented in Figure 1 below. Student achievement and retention 3. Research methodology Th l i i hi d The population in this study was all active students of a university in Indonesia at the time of the study, which was estimated at around 7,000 people. By using an alpha level of 10% and referring to the Harry King Nomogram (Hair et al., 2009), the number of samples taken should be 261 people. To be more representative, by considering that the university which was the object of this study consisted of nine faculties and 27 study programs, the researchers doubled the number by 2.5 times. The researcher will use 700 (rounding up of 652.5) university students as respondents in this study. The sampling procedure that will be used in this study is simple random sampling. Simple random sampling is a sampling technique that is carried out randomly and simply, which method is used so that each respondent has an equal chance of being selected (Cooper and Schindler, 2011). This research is a causal explanatory research. This study examines the effect of each of the five independent variables (student, family, university, lecturer and academic advisor) on academic achievement and student retention separately. The method to be used to carry out this test is multiple linear regressions, which is an analysis of the effect (cause–effect) of more than one independent variable on a dependent variable (Neuman, 2014). In order to get decent and precise results, researchers conducted a series of analytical methods on research instruments and data, which included testing validity and reliability, normality, heteroscedasticity and outliers (if needed). y ( ) Physical and psychological conditions of students were measured using nine items of question translated from a questionnaire that was used by Suhaily and Soelasih (2015) with a Cronbach’s alpha reliability value of 0.684 and a confirmatory factor analysis (CFA) loading factor value ranging from 0.430 to 0.617. The quantity and quality of student families is measured using four items of question that have been used by Suhaily and Soelasih (2015) with a Cronbach’s alpha reliability value of 0.601 and a CFA loading factor value ranging from 0.473 to 0.562. The quality of university facilities and infrastructure was measured using 5 items of question that had been used by Suhaily and Soelasih (2015) with a Cronbach’s alpha reliability value of 0.679 and a CFA loading factor value ranging from 0.604 to 0.724. AAOUJ 15,1 42 3. Research methodology Th l i i hi d Lecturer hard skills and soft skills were measured using 7 items of question that had been used by Suhaily and Soelasih (2015) with a Cronbach’s alpha reliability value of 0.718 and a CFA loading factor value ranging from 0.429 to 0.686. The quality of academic advisors was measured using 12 items of question of student assessment of academic advising instruments (advisor assessment instruments) used by Miller and Irons (2014). The scale used for the five independent variables and 37 question indicator items is a Likert scale consisting of 1 (strongly disagree) to 4 (strongly agree). The level of student academic achievement is measured by the magnitude of the GPA achieved by students as respondents at the time of the study. The GPA is in the range of 0.00–4.00. The retention rate of students from each study program is measured by comparing the number of students who are still active in a lecture (actual) in a period with the number of students who should be (ideally) still active. AAOUJ 15,1 The figure shows that there are five factors (either internal or external) contribute to academic achievement and retention of student. Based on literature reviews conducted by researchers, here are 10 hypotheses derived from theoretical paradigm and empirical studies: H1. Student has positive impact on academic achievement. H1. Student has positive impact on academic achievement. H2. Family has positive impact on academic achievement. H3. University has positive impact on academic achievement. H4. Lecturer has positive impact on academic achievement. H5. Academic advisor has positive impact on academic achievement. H6. Student has positive impact on student retention. H7. Family has positive impact on student retention. H8. University has positive impact on student retention. H9. Lecturer has positive impact on student retention. H10. Academic advisor has positive impact on student retention. Academic Achievement Student Family University Lecturer Academic Advisor Student Retention Source(s): Processed from many sources + + + + + + + + + + Figure 1. Research model Student Retention Academic Achievement Academic Advisor Source(s): Processed from many sources AAOUJ 15,1 The first five hypotheses describe positive influence of several factors toward student academic achievement while another last five hypotheses describe positive influence of several factors toward student retention rate. The directions of those hypotheses are positive; meaning the higher quality or intensity of those factors, the higher students’ academic performance and their retention at the university. Student achievement and retention 4. Data analysis and results y Researchers have conducted a survey through the distribution of question questionnaires to 713 students studying at University X (100% response rate). They came from 17 study programs under the auspices of the university which were the object of this research. Based on the results of the descriptive analysis, respondents in this study were dominated by women (55.3%), aged between 18 to 20 years (57.9%), only students (87.4%) from the city where the university was (63.7%), reside at home with family (56.2%), have a complete family (92.1%) and are not active in organizational activities at the university (58.9%) and GPA ranges from 2.76 to 3.50 (46.6%). The results of reliability testing using Cronbach’s alpha coefficient are presented in Table 1 below. 43 The table shows that almost all items (as many as 34 items) have good reliability, with values ranging from 0.770 to 0.963 (more than 0.6). g g ( ) The results of normality testing using the Kolomogrov–Smirnov’s test are presented in Table 2 below. The table shows that the research data that have proven their normality because they have significance value above 10%, which is equal to 0.664. Because the research data are proven normal, testing of outliers is not necessary. The results of multicollinearity testing, using the variance in factor (VIF) and toleranc values are presented in Table 3 below. The table shows that the research data collected has met the criteria for multi-collinearity because it has a tolerance value exceeding 0.10 (ranging from 0.815 to 0.496) and a VIF value of less than 10 (ranging from 1.227 to 2.016). Heteroscedasticity test ensures that dependent variables exhibit equal levels of variance across the range of those variables (Hair et al., 2009). The results of heteroscedasticity testing, using Glejser method, are presented in Table 4 below. The table shows that the research data collected has met the criteria for heteroscedasticity because more than a half of independent variables have significant value more than 10%, ranging from 0.169 to 0.417, pertaining of linear regression analysis of those independent variables and absolute residual value (abs). For the first test, the researcher conducted a multiple linear regression analysis between five independent variables with GPA as the dependent variable, with the results presented in Table 5 below. 4. Data analysis and results The table above shows that there are two proven research hypotheses, which have a significant positive effect (beta value of 0.248 with significance value of 0.000) on academic achievement of student and lecturer significantly positive effect on student academic achievement (beta value of 0.015 with significance value of 0.080) at an alpha level of 10%. For the second test, the researcher conducted a multiple linear regression analysis between five independent variables with the student retention rate as the dependent variable, with the results presented in Table 6 below. The table above shows that there is a proven research hypothesis, which has a positive effect on student retention rate (beta value of 1.536 with significance value of 0.070) at an alpha level of 10%. Variable Cronbach’s alpha Number of items Student 0.770 7 Family 0.796 4 University 0.774 5 Lecturer 0.895 7 Academic advisor 0.963 11 Source(s): Processed research data (2019) Table 1. Reliability test results 5. Research implications Unstandardized residual N 713 Normal parametersa,b Mean OE-7 Standard deviation 0.32109290 Most extreme differences Absolute 0.027 Positive 0.027 Negative –0.016 Kolmogorov–Smirnov Z test 0.728 Asymp. sig. (two-tailed) 0.664 Note(s): (a) test distribution is normal; (b) calculated from data Source(s): Processed research data (2019) Coefficientsa Model Collinearity statistics Tolerance VIF 1 Student 0.734 1.363 Family 0.815 1.227 University 0.544 1.837 Lecturer 0.496 2.016 Academic advisor 0.812 1.231 Note(s): aDependent variable: GPA Source(s): Processed research data (2019) Coefficientsa Model Unstandardized coefficients Standardized coefficients B Std. Error Beta t Sig. 1 (Constant) 0.340 0.132 2.579 0.010 Student 0.004 0.005 0.035 0.812 0.417 Family 0.002 0.007 0.009 0.221 0.825 University 0.010 0.007 0.070 1.375 0.169 Lecturer 0.009 0.006 -0.090 1.703 0.089 Academic advisor 0.006 0.002 0.108 2.600 0.010 Note(s): aDependent variable: ABS Source(s): Processed research data (2019) Table 2. Normality test results Table 3. Multi-collinearity test results Table 4. Heteroscedasticity test results AAOUJ 15,1 44 5. Research implications h l f hi d i p The results of this study are in line with the previous studies conducted by Thawabieh (2016), Singh et al., (2016), Suhaily and Soelasih (2015), Ayuni and Mulyana (2015), Garkaz et al., (2011) and Bakar et al. (2010). Student academic achievement is determined by the personal intention of the student and the quality of the lecturers. Every study program at university should provide formal training programs or informal appeals to its lecturers (including academic advisors) so that they pay more attention to the students who are taught and mentored by them. Every lecturer (in every class meeting) and academic advisor (in every interaction, both through face to face and various forms of electronic communication media available) need to instill awareness so that students can better manage themselves in totality. Coefficientsa Model Unstandardized coefficients Standardized coefficients B Std. Error Beta t Sig. 1 (Constant) 2.715 0.209 12.999 0.000 Student 0.047 0.008 0.248 5.831 0.000 Family 0.004 0.011 0.016 0.398 0.691 University 0.034 0.011 0.148 2.990 0.003 Lecturer 0.015 0.009 0.091 1.755 0.080 Academic Advisor 0.010 0.004 0.103 2.558 0.011 Note(s): aDependent variable: GPA Source(s): Processed research data (2019) Coefficientsa Model Unstandardized coefficients Standardized coefficients B Std. Error Beta t Sig. 1 (Constant) 1.260 0.344 3.662 0.004 Student 0.051 0.090 0.206 0.566 0.582 Family 0.033 0.075 0.210 0.437 0.671 University 0.105 0.092 0.537 1.146 0.276 Lecturer 0.232 0.116 1.536 2.010 0.070 Academic advisor 0.131 0.080 0.827 1.642 0.129 Note(s): aDependent variable: student retention rate Source(s): Processed research data (2019) Table 5. Results of regression analysis I Table 6. Results of regression analysis II Student achievement and retention 45 Each teaching staff members and academic advisor should continuously educate students so that they are able to have regular physical activities, such as sleeping, eating and exercising routinely. To balance this out, each teaching staff members and academic advisor should also demonstrate exemplary behavior to students so that they can show positive attitudes and behaviors, such as discipline, initiative, independence and involvement in student activities in the campus environment. Lecturers at the university have the responsibility as “second parents” in directing and shaping the personalities of their students who are willing and able to apply moral and spiritual values through management and self-leadership in the life of study that are aligned with their academic achievements. y g The level of student retention is determined by the quality of the lecturers. 7. Conclusion A d i hi Academic achievement and student retention are two important indicators that show the excellence of a university. University academic achievement is determined more by the intrinsic motivation of its students. The enthusiasm in students at the university is a trigger for their determination and persistence in achieving themselves in the field of education as a form of self-actualization on the campus where they work. The retention of students is more determined by the extrinsic motivation that sources from the campus environment they are in. The professionalism of lecturers at the university is a reason that affirms the desire of students to complete their education at the university. Researchers hope that the university (synergy with all study programs) regularly conducts an active review of the quality of its lecturers and students. It is not only evaluating but it is also hoped that they will take proactive actions in improving and maintaining the motivation of each teaching staff and their students so that they are willing and able to try to provide the best for the progress of their university. Furthermore, the wind of change is coming faster than its prediction; globalization and digitalization are more real. Open universities are not the specific type of universities any longer, but they have become the future trend of universities all around the world. Hopefully, the tertiary education government can proactively provide any necessary supports and inspire every university to prepare and develop every resource needed to adopt the system that an open university has in the life of its college community. Thus, every university is living in the digital space of higher education ecosystem. The success of each element of the university citizens determines the continuation of the university. 6. Limitations Research that has been done is inseparable from several limitations. This study only involved five independent variables that were predicted empirically had an influence on the level of academic achievement and student retention. The researcher suggests that future research should involve more endogenous variables in order to be able to get a more comprehensive picture of the data. In testing its hypothesis, this study uses linear multiple regression analysis to examine the causal relationship between five independent variables and two dependent variables. The researcher suggests that future research should use a more dynamic research model, such as including moderation or mediation variables in its processing using more diverse methods, such as ANOVA or path analysis. Because of access constraints, this research has not yet reached all students of each study program. Researchers suggest that future research should involve all students who are studying in each study program chosen at a university. Moreover, this quantitative study conducted at a private university, therefore this research results cannot be generalized on every university. Researchers suggest that future research should involve more universities, including public and open universities, to expand and strengthen the applicability range of research’s results. 46 Cooper, D.R. and Schindler, P.S. (2011), Business Research Methods, McGraw-Hill Irwin, New York, NY. 5. Research implications h l f hi d i Each study program at the university should provide a self-development program for its lecturers so that they become human resources that provide qualified teaching, ranging from planning student-centered lecture systems, organizing media diversity and the methods used during the teaching and learning process in the classroom, leadership role models and co-learners in which students are able to apply what they get from the presentation of the material and the behavior of their lecturers, as well as controlling the harmonious relationship among lecturers and students. The lecturer is a figure who becomes an encouragement for students in their diligence in carrying out to accomplish their study. The quality of lecturers is one of the considerations of students in committing to their universities. Each lecturer is a frontline employee at the university. They conduct intensive direct interaction with their students. Such conditions reinforce the fact that the decision of students to continue their studies at the university is largely determined by their perceptual assessment of their lecturers, ranging from aspects of knowledge and experience of interpersonal skills. Chuah, P. and Lim, P. (2018), “Applying quality tools to improve student retention supporting process: a case study from WOU”, Asian Association of Open Universities Journal, Vol. 13 No. 1, pp. 60-72. References Bakar, K.A., Tarmizi, R.A., Mahyuddin, R., Elias, H., Luan, W.S. and Ayub, A.F.M. (2010), “Relationships between university students’ achievement motivation, attitude, and academic performance in Malaysia”, Procedia Social and Behavioral Sciences, Vol. 2, pp. 4906-4910. Chuah, P. and Lim, P. (2018), “Applying quality tools to improve student retention supporting process: a case study from WOU”, Asian Association of Open Universities Journal, Vol. 13 No. 1, pp. 60-72. Cooper, D.R. and Schindler, P.S. (2011), Business Research Methods, McGraw-Hill Irwin, New York, NY. Demir, S., Demir, S.G., Bulut, H. and Hilsar, F. (2014), “Effect of mentoring program on ways of coping with stress and locus of control for nursing students”, Asian Nursing Research, Vol. 8, pp. 254-260. Student achievement and retention Student achievement and retention Garkaz, M., Banimahd, B. and Esmaeili, H. (2011), “Factors affecting accounting students’ performance: the case of students at the islamic azad university”, Procedia Social and Behavioral Sciences, Vol. 29, pp. 122-128. Hair, J.F., Black, W.C., Babin, B.J., Anderson, R.E. and Tathan, R.L. (2009), Multivariate Data Analysis, Pearson Education International, New Jersey, NJ. 47 Hennig-Thurau, T., Langer, M.F. and Hansen, U. (2001), “Modeling and managing student loyalty: an approach based on the concept of relationship quality”, Journal of Service Research, Vol. 3, No. 4, pp. 331-334. Ismail, A. and Jui, M.K.K. (2014), “The role of mentoring program in enhancing mentees’ academic performance”, Journal of Education and Learning, Vol. 8 No. 1, pp. 13-22. Miller, R.L. and Irons, J.G. (2014), “Academic advising: a handbook for advisors and students volume 1: models, students, topics, and issues”, available at: http://teachpsych.org/ebooks/academic- advising-2014-vol1 (accessed 17 April 2019). Muola, J.M., Maithya, R. and Mwinzi, A.M. (2011), “The effect of academic advising on academic performance of university students in Kenyan universities”, International Multidisciplinary Journal Ethiopia, Vol. 5 No. 5, pp. 332-345. Mushtaq, I. and Khan, S.N. (2012), “Factors affecting students’ academic performance”, Global Journa of Management and Business Research, Vol. 12 No. 9, available at: https://globaljournals.org GJMBR V l 12/3 F Aff i S d A d i df ( d 1 A il 2019) Mushtaq, I. and Khan, S.N. (2012), “Factors affecting students’ academic performance”, Global Journal of Management and Business Research, Vol. 12 No. 9, available at: https://globaljournals.org/ GJMBR_Volume12/3-Factors-Affecting-Students-Academic.pdf (accessed 17 April 2019). GJMBR_Volume12/3-Factors-Affecting-Students-Academic.pdf (accessed 17 April 2019). Neuman, W.L. (2014), Social Research Methods: Qualitative and Quantitative Approaches, Alyn and Bacon, New York, NY. Singh, S.P., Malik, S. and Singh, P. References (2016), “Factors affecting academic performance of students”, Paripex - Indian Journal of Research, Vol. 5 No. 4, pp. 176-178. Suhaily, L. and Soelasih, Y. (2015), “Factors affecting student achievement in faculty of economics ‘X’ university”, Journal The Winners, Vol. 16 No. 1, pp. 25-35. Thawabieh, A.M. (2016), “Factors affecting university students’ achievement”, British Journal of Education, Society and Behavioural Science, Vol. 14 No. 4, pp. 1-11. Student achievement and retention Corresponding author R S i b Rony Setiawan can be contacted at: Ronysetiawan.xie@gmail.com For instructions on how to order reprints of this article, please visit our w www.emeraldgrouppublishing.com/licensing/reprints.htm Or contact us for further details: permissions@emeraldinsight.com
https://openalex.org/W4246016777	https://www.cambridge.org/core/services/aop-cambridge-core/content/view/C7B4A0BB81E86236F0E41C6764EBDCFA/S0140078900020927a.pdf/div-class-title-miscellany-div.pdf	English	null	Miscellany	The bulletin of the Royal College of Psychiatrists/Bulletin of the Royal College of Psychiatrists	1,988	cc-by	670	302 302 Bulletin of the Royal College of Psychiatrists Bulletin of the Royal College of Psychiatrists Diploma in Alcohol Counselling and Consultation that it is also of considerable interest to qualified professionals such as social workers, nurses, health visitors, GPs and medical students. It addresses a mixed audience with an emphasis on reading about and understanding the points of view of people in different situations. The group meetings are designed to underline this process of mutual understanding. Further information: Jenny Rook, Department of Health and Social Welfare, North Spur Building, The Open University, Milton Keynes, MK7 6AA (telephone 0908 653743). A diploma qualification course, part-time from October 1988 to September 1989, will be held at the University of Kent. It is designed for those in the helping professions who wish to develop their skills in work with problem drinkers. The course begins in October 1988 with a three week module at the University, followed by four two-week modules over the following 11 months, finishing in September 1989. Students will be expected to have at least one year's relevant experience and for the duration of the course to be working with clients who have alcohol- related problems. Fee: Â£840(subject to review). Enquiries and application forms: Mrs Gail Jones, Alcohol Interventions Training Unit, Secretary, School of Continuing Education, Rutherford Col lege, University of Kent, Canterbury, Kent CT2 7NX (telephone 0227 764000, extension 3691). National Institute of Crisis Intervention Therapy & Research Huddcrsfield Health District has set up the above Institute from 1 April 1988. It will be based at 63 Nabcroft Lane, Crosland Moor, Huddersfield. Dr N. Rao Punukollu, Consultant Psychiatrist and Clinical Director of the Huddersfield Crisis Inter vention Team has been appointed as Director. The institute's aims and objectives include: organising educational programmes on crisis intervention theory and practice for mental health professionals; organising introductory courses on crisis inter vention approach; developing a specialist library on crisis intervention theory and practice in mental health; launching a journal on crisis intervention and recruiting membership. Dr Punukollu would welcome approaches direct to him at the Institute headquarters from those interested in becoming involved in this organisation. Preventing Mental Illness is the first stage of a project to encourage the development of preventive mental health services. Dr Jennifer Newton (MIND's Prevention Research Officer) reviews the research literature relating to prevention and focuses on depression and schizophrenia as examples of what might be possible. The book is published by Routlcdgc & Kegan Paul and isavailable from book shops and MIND's Mail Order Department, 4th Floor, 24-32 Stephenson Way, London NWI 2HD. It costs Â£25(post free from MIND). Miscellany Diploma in Alcohol Counselling and Consultation Recent publications Women's Problems with Alcohol and Other Drugs: Improving our Response. Edited by Jan Waterson, Betsy Ettorrc, Ronno Griffiths and Rosemary Kent. These Proceedings of a Conference organised by the Alcohol Interventions Training Unit, University of Kent at Canterbury, with the Addiction Research Unit, Institute of Psychiatry, London and held on 5 July 1986 at Eliot College, University of Kent are available from Mrs Gail Jones, School of Continuing Education, Rutherford Extension, University of Kent, Canterbury CT2 7NX. Price: single copies Â£2.50;Â£1.75per copy for orders of 10 or more; Â£1.50per copy for orders of 20 or more (all prices include postage and packing). Cheques should be made payable to UNIKENT (Those who attended the conference are entitled to a complimentary copy). Mental Handicap: Patterns for Living This individual study pack is produced by the Open University's Department of Health and Social Welfare. It is built around case study material and supplemented by a programme of optional group meetings. Although initially the course was aimed at unqualified care staff, patients and volunteers, feed back since its launch in November 1986has indicated (p ) The new edition of A-Z of Welfare Benefitsfor People with Mental Illness which includes changes in force from April 1988 can also be purchased from MIND's Mail Order Department at the above address; price Â£1per copy, postage and packing included. https://doi.org/10.1192/S0140078900020927 Published online by Cambridge University Press
https://openalex.org/W2900684137	https://www.scienceopen.com/document_file/7fe0e654-8e03-4d37-b4a4-2ea289b62239/ScienceOpen/BHCI-2018_Garcia.pdf	English	null	Everybody is talking about Virtual Assistants, but how are people really using them?	Electronic workshops in computing	2,018	cc-by	3,836	http://dx.doi.org/10.14236/ewic/HCI2018.96 http://dx.doi.org/10.14236/ewic/HCI2018.96 Everybody is talking about Virtual Assistants, but how are people really using them? Dr Marta Perez Garcia Sarita Saffon Lopez Hecto Telefonica R&D Telefonica R&D Telefoni Madrid, Spain Madrid, Spain Madri Marta.perezgarcia@telefonica.com Sarita.saffonlopez@telefonica.com Hector.Donis Hector Donis Telefonica Digital Madrid, Spain Hector.Donis@telefonica.com Sarita Saffon Lopez Telefonica R&D Madrid, Spain ita saffonlopez@telefo Voice activated virtual assistants are growing rapidly in number, variety and visibility, driven by media coverage, corporate communications, and inclusion in a growing variety of devices. This trend can also be observed by how difficult it is becoming to find, among internet users, people who have not used or even heard of this new technology. Having said this, there is a visible shortage of academic research on this topic. Therefore, in the interest of creating a knowledge base around voice activated virtual assistants based on artificial intelligence, this multi-country exploratory research study was carried out. It concludes by providing information about the usage of four main voice activated virtual assistants (Siri, Google Assistant, Cortana and Alexa), comparisons between virtual assistant most frequently used and countries, and finally it establishes the base for further research activities. Voice Activated Virtual Assistant, Artificial Intelligence, Technology adoption. 1. INTRODUCTION and most frequent interactions with it is the virtual personal assistants (Arafa and Mamdani, 2000). Humans’ interest in building a machine that is able to perform some kind of reasoning is not something new. In fact, we can go back to 300 B.C. to find Talos’ story, a giant automaton made of bronze that protected the island of Crete. Dismissing tales, Ramon Lull explored the first traces of Artificial Intelligence (AI) with the mechanical calculator (calculus ratiocinator) around 1300 CE, which anticipated aspects of the Turing Machine and intended to make operations on concepts, not just numbers (Russell & Norvig 2009, p. 16). Regardless of the wave of technology adoption with virtual assistants, little has been written in academia about it so that theory can be built upon. Most information regarding virtual assistants comes from brands websites or online magazines, but it is rather difficult to find theoretical foundations and solid hypothesis of users’ behaviours around these technologies, beyond information such as “the most popular features” or “new functionalities”. A necessity then is stressed to do formal research on their actual usage, differences among assistants, and how its use vary by users’ age, sex or culture. Several centuries later, AI research took form in the 1950s within a group of theorists and researchers, who predicted machines would become capable, in just a few decades, of carrying out any task humans can do (Simon, 1965; Minsky, 1967). However, it was until 1980s when AI research obtained the recognition it sought by launching the expert systems (Jackson, 1998), which simulated the decision-making ability of humans. © Garcia et al. Published by BCS Learning and Development Ltd. Proceedings of British HCI 2018. Belfast, UK. 2.2 Google Assistant The days when Google was only a search engine have long passed. One of their most remarkable launches was Google Now in 2012, its intelligent personal assistant that uses natural language to provide a voice search. However, it did not engage in two-way conversations, which represented a drawback, so in May 2016 Google Assistant made its debut, as part of Google Allo, its messaging app. In 2017 Google announced a series of sneak peeks into the future by stating they are working on its assistant to support visual responses (LeFebvre, 2017), gather search sight information through the camera (Welch, 2017) or send money (Miller, 2017). The sample was distributed similarly in terms of gender (59% female, 41% male) and included ages between 18 and 55 (M= 36.26, SD=10.45). All the respondents were active users of apps and internet through their mobile phones, and had diverse affinity with technology. For the descriptive and statistical analysis of VAVA’s use, using SPSS platform, only the respondents that were active users were taken in account, resulting in a sample of 1047 users. Furthermore, in order to explore differences between the four main VAVAs, the study assured a sufficiently large sample per each of them. 2. VOICE ACTIVATED VIRTUAL ASSISTANTS Finally, in June 2017, Amazon Echo Show is launched in the US, a screen-device that multiplied the functionalities that Alexa could provide with a new interface. A key differentiator of Amazon strategy is the Alexa Skill Set, which enables anybody to design, build and help launch a new functionality in Alexa such as calling a cab or changing the colour of smart lights. statistic corporation Statista (201), Alexa and Google alone gather 87% of the global market share. Additionally, they use the same type of interaction with users, voice, and they are all talking AI databases. Also, they are accessible in affordable and widespread devices such as laptops, tablets, smartphones or own-brand devices, such as Amazon´s Echo (which is different from the others, but we found interesting that Google and Apple are also moving towards this type of device, which could help give us a hint on future trends). These VAVAs are the AI play of four of the most powerful companies in the world, so it is vital to understand how its usage is being articulated. 4.1 Knowledge and usage of VAVAs The first relevant result is that a great majority of online users in all countries of the study either uses or knows about VAVAs (Figure 1). Globally, just over half of online users say they know about VAVAs (55.9%) and over a quarter of them say they use 2.1 Siri Telefonica, the telecommunications multinational with its headquarters in Spain, is deeply involved in understanding and pursuing the vast opportunities that AI and VAVAs can bring to the company to deliver additional value and help create a new relationship with its customers. Therefore, the R&D research team carried out this study as part of the exploration and establishment of a benchmark of VAVAs use across important markets in Telefonica’s footprint. Siri is Apple’s VAVA, which was launched in 2011, the first of its kind, representing a competitive threat for competitors such as Google (Barnett, 2011), receiving both positive and negative reviews. For instance, some people found Siri was not able to understand and execute the tasks users were asking her to do, but they were also able to foresee its potential (Siegler, 2011). Currently, Siri can help users to set a reminder, an alarm, a calendar appointment, tell about the weather, take users to Google to find an answer, tell a joke, and so on. A total of 3749 self-administered online questionnaires were sent during the first week of December 2017 to a panel of online users across seven countries: UK, Germany, Spain, Brazil, Argentina and Chile. Additionally, the US was included, not only as an assumed leader in VAVA adoption, but also to gather data on Alexa, which is not yet fully launched and established in several of the other markets studied. The survey was constructed by the research team and was composed of 11 multiple choice questions (excluding demographics) regarding subjects as knowledge and use of VAVAs, frequency of use, amount of tasks and place of use, among others. 2.3 Cortana In April 2013, at the BUILD developer conference, Microsoft introduced its VAVA Cortana, launched in January 2015. Microsoft defines Cortana as a personal digital assistant that is designed to help users carry out basic tasks as well as provide answers to users’ questions. Cortana’s focus is to work on devices with Microsoft OS, above all from the computer, but it is also available in Android smartphones. Cortana can schedule alarms or alerts, to send reminders based on user’s location, or answer questions about traffic and sport scores. 2. VOICE ACTIVATED VIRTUAL ASSISTANTS Another problem encountered in the literature, is an absence of a common terminology: AI agents (Castelfranchi, 1998), virtual assistants (Martin and Allende, 2015), intelligent assistants (Kiseleva et al, 2016), and so on. What is important is that while the different terms are used, they are all referring to the same thing. For the purpose of this study, we will refer to them as Voice Activated Virtual Assistants (VAVAs). The 2010s arrived focusing on algorithms of machine learning by enabling computers to have access to large amounts of data, which comes back to what was expected in the 1950s (Samuel, 1959; Koza, 1996). This kind of application through a simplified interaction in games and hobbies is what enabled the adoption of AI at a user level. What is happening with AI implementation today on our daily basis then? One of many examples of our closest For this study, we have gathered information of four key VAVAs: Siri (Apple), Google Assistant (Google), Cortana (Microsoft) and Alexa (Amazon). The reason behind this choice is that they have all been at least four years in the market, so they are more mature than other competitors, and, according to the © Garcia et al. Published by BCS Learning and Development Ltd. Proceedings of British HCI 2018. Belfast, UK. 1 Everybody is talking about Virtual Assistants, but how are people really using them? Perez ● Saffon ● Donis Alexa is the first voice activated virtual assistant linked to a stand-alone home device rather than integrated into existing electronic devices. UK and Germany receive Amazon Echo in September of that year, expanding the market in which they play in search of a major device adoption. Finally, in June 2017, Amazon Echo Show is launched in the US, a screen-device that multiplied the functionalities that Alexa could provide with a new interface. A key differentiator of Amazon strategy is the Alexa Skill Set, which enables anybody to design, build and help launch a new functionality in Alexa such as calling a cab or changing the colour of smart lights. Alexa is the first voice activated virtual assistant linked to a stand-alone home device rather than integrated into existing electronic devices. UK and Germany receive Amazon Echo in September of that year, expanding the market in which they play in search of a major device adoption. 2.4 Alexa In November 2014, Amazon introduced Alexa and Amazon Echo in the US, but only available for Prime members. It officially launched in the US in 2015. 2 Everybody is talking about Virtual Assistants, but how are people really using them? Perez ● Saffon ● Donis of use (p<0,001); even the country that has the lowest percentage, Spain, still shows over half its users using it almost daily. them (27.9%). Only one in ten say that they have never heard of them (11.5%) and a minority has used them in the past but stopped (4,6%). Figure 2: Frequency of use by countries and rate of user population that uses their VAVA for one to three tasks. Figure 1: Distribution of knowledge and use of VAVAs in the seven countries of the study and total. Figure 2: Frequency of use by countries and rate of user population that uses their VAVA for one to three tasks. Figure 1: Distribution of knowledge and use of VAVAs in the seven countries of the study and total. Nonetheless, it is also important to remark that the majority of VAVA users in each country only tend to carry out between one to three different tasks (or use cases such as setting up an alarm, asking what the weather is like, playing music, etc.). A possible explanation for the small number of tasks used could be that users just started using the VAVAs and have not had enough time to discover all of its use cases. However, this could be challenged by the fact that 55% reported using it for a year or longer. The research study has highlighted the fact that even if the first VAVA was launched several years ago, there is still more than two times the population that knows about VAVAs but does not use them than those who know and use them. As the data shows, VAVA adoption is not a question of knowledge about the service, as most non-users are aware of them, so it may be more related to a missing a trigger from knowledge to actual use. This is an area for continued research, to confirm not only the reasons for not using VAVAs from a user perspective, but also to understand what must happen in order to stimulate VAVA adoption to move from early adopters to early majority, and what can the industry do to make this easier. 2.4 Alexa Interestingly, Alexa is the VAVA that is being used for more tasks by its users (Figure 3). The number of users that execute six or seven tasks with Alexa, 17%, almost triples that of other VAVAs users (6% Google, 7% Cortana and 8% Siri). These results could be explored in a qualitative study to dive into the reason why more tasks are being used with Alexa, such as the fact that it is because it has more skills, or because the stand-alone device does not have to compete with the smartphone or computer other functionalities, or for its relationship with domestic tasks. Some important country differences exist, portrayed in Figure 1. As expected, the US is the country where most online users report using VAVAs (44.2%) while Chile is at the other end of the spectrum with only 22% of VAVAs users and the highest percentage of people who has never heard of them (17.1%). Spain is the country with the highest rate of people that are aware of the existence of VAVAs, but do not use them with 66.5% while the US has the lowest with 38.1%. Figure 3: Tasks carried out by users with their VAVAs These results confirm the hypothesis about the US being the most advanced market in the sample for VAVA adoption. In fact, the US is the only market where the number of VAVAs users is higher than those who know about but do not use them. Furthermore, considering this country is where Alexa is mostly present (24% of its user base), it is possible to hypothesize that this VAVA, in the form of a standalone device, has enabled a further expansion of VAVAs into the early and late majority. Figure 3: Tasks carried out by users with their VAVAs Figure 4: Frequency of use per VAVA. 4.3 Place and device of use Another relevant finding is that the use of VAVAs takes place mostly at home (Figure 5). In the case of Cortana, because of its link to Windows, people use it more at home than Siri or Google Assistant. However, even the latter two, with their significant mobile integration on smartphone devices, seem also to concentrate their use at home. Due to its Amazon Echo standalone device proposition, Alexa stands out even more for its use at home. Nonetheless, there is a small niche that declares to use it mostly or exclusively out of the home (14% of Siri users and 8% of Google Assistant users). Figure 6: Frequency of use for the four age ranges, and rate of sample that uses VAVA for more than six tasks. Figure 5: Place of use for the four major VAVAs Figure 6: Frequency of use for the four age ranges, and rate of sample that uses VAVA for more than six tasks. 5. CONCLUSIONS This study aimed to pursue a thorough understanding of current VAVA usage in seven different countries to establish a knowledge base from which more specific research studies will originate. The first finding of this study has been to establish that regardless young application of VAVAs, there is a high awareness across countries, even if its usage is unequal across countries. Figure 5: Place of use for the four major VAVAs This study has also explored the most common device in which the VAVAs are used. The sample data shows that Siri and Google Assistant are mainly used in mobile devices (average globally of 75% and 73% of the time of use, respectively), while Cortana concentrates its main use on the PC with an average of 37% of the time of use being in this device. Alexa has the lion share of use on its own device with an average of 50%, far superior and proven statistically significant by Chi-square test (p<0.001), compared to the 3% to 6% average for the other VAVAs. None of these four VAVAs seems to dominate the tablet, which is the device where VAVAs are used the least, possibly due easier tactile screen interaction. This study has also identified that most people who start using this type of AI tend to make frequent use of it, which suggests the functionalities that they are able to carry out are valuable for users. In terms of age and use, this study has highlighted that, against what one might have thought, younger people are less attracted and are less heavy users than the oldest segment of the sample, both in frequency of use and number of tasks. Finally, the study has also identified the key devices people typically use for interacting with their VAVAs. What draws attention to this finding is that although the smartphone is a rather common device to access the VAVA, most people use it more frequently at home, rather than when they are on the go, which it was expected and reported, mostly from Alexa, as it is linked to a home device. 4.2 Frequency of use and tasks requested The study has identified that current VAVA users are making use of it regularly (Figure 2); over 60% of the 1047 global users report using their VAVAs daily or several times a week (Almost daily). And, even though Chi-square test proves statistical significant the differences between counties for the frequency Figure 4: Frequency of use per VAVA. 3 Everybody is talking about Virtual Assistants, but how are people really using them? Perez ● Saffon ● Donis VAVAs for more than six tasks, on average, only 12.8% of the younger groups did. These differences were also statically significant on a Chi-square test at a level of confidence of 99% (p-value <0.001), and the contrast establishes the base for a further qualitative phase exploring the barriers of younger segments to increase the quantity of tasks as well as to determine the motivations of older segments for a more frequent and varied VAVAs usage. Another possible reason, hypothesised under the analysis performed for this study, could be a higher frequency of use of this VAVA. Alexa’s users report using their VAVA daily more than twice, or even three times, as much as any other VAVA (Figure 4). 4.4 Age comparisons In terms of generational differences in the use of VAVAs, four age-range groups were created (18 to 24, 25 to 34, 35 to 44 and 45 to 55). A key discovery was that the latter group is not only one that uses most frequently their VAVA, but the fact they use it for more tasks than the youngest group (Figure 6). While 34.9% of the 45 to 55yrs reported using their VAVA daily, only 26.1% of the 18-24 group did, and 6% reported using it less than once a month, being the highest of all four age groups. The differences between these two age groups was proven statistically significant (p-value=0.04). Additionally, while 22.4% of the 45 to 55yrs reported using their 8. REFERENCES https://techcrunch.com/2011/10/11/iphone-4s- review/ (2018/02/15). Russell, Stuart J. and Norvig, P. (2009) Artificial Intelligence: A Modern Approach (3rd ed.). Prentice Hall, Upper Saddle River, New Jersey. Barnett, E. (2011) Google's Eric Schmidt: Apple's Siri could pose 'threat'. http://www.telegraph.co.uk/technology/eric- schmidt/8873664/Googles-Eric-Schmidt-Apples- Siri-could-pose-threat.html (2018/02/15). Russell, Stuart J. and Norvig, P. (2003) Artificial Intelligence: A Modern Approach (2nd ed.), Prentice Hall, Upper Saddle River, New Jersey. O'Kane, S. (2016) Siri's big upgrades won't matter if it can't understand its users. https://www.theverge.com/2016/6/14/11925076/s iri-apple-wwdc-ios-updates-voice-recognition (2018/02/15). Simon, H.A. (1965) The Shape of Automation for Men and Management. Harper & Row, New York, New York. Minsky, M. (1967) Computation: Finite and Infinite Machines. Prentice-Hall, Englewood Cliffs, New Jersey. Kahn, J. (2017) Apple engineers share behind-the- scenes evolution of Siri & more on Apple Machine Learning Journal. https://9to5mac.com/2017/08/23/evolution-siri- machine-learning-journal/ (2018/01/22). Jackson, P. (1998) Introduction to Expert Systems (3rd ed.). Addison-Wesley Longman Publishing Co., Boston, Massachusetts. McCorduck, P. (2004) Machines Who Think (2nd ed.). A. K. Peters, Ltd., Natick, Massachusetts. LeFebvre, R. (2017) Google Assistant now accepts typed and verbal cues. https://www.engadget.com/2017/05/17/google- assistant-now-accepts-typed-and-verbal-cues/ (2018/02/18). Samuel, A. (1959) Some Studies in Machine Learning Using the Game of Checkers. IBM Journal of Research and Development, 3(3), 211-229. Welch, C. (2017) Google Assistant will soon search by sight with your smartphone camera. https://www.theverge.com/2017/5/17/15648128/g oogle-assistant-camera-sight-search-feature-io- 2017 (2018/02/18). Koza, J.R., Bennett, F.H.. Andre, D. and Keane, M.A. (1996) Automated Design of Both the Topology and Sizing of Analog Electrical Circuits Using Genetic Programming. Artificial Intelligence in Design ’96, 151–170. Miller, P. (2017) You'll soon be able to send money with Google Assistant. https://www.theverge.com/2017/5/18/15660728/g oogle-assistant-payment-api-peer-to-peer- money-money-money-io-2017 (2018/02/18). Lally, A., Prager, J.M., McCord, M.C., Boguraev, B. K., Patwardhan, S., Fan, J. and Chu-Carroll, J. (2012) Question analysis: How Watson reads a clue. IBM Journal of Research and Development, 56(3.4), 2:1-2:14. Pega Systems Inc. (2017) What Consumers Really Think About AI: A Global Study. Retreived May 18, 2018 from https://www.pega.com/ai-survey Zhang, Z. (2012) Microsoft kinect sensor and its effect. IEEE MultiMedia, 19(2), 4-10. Statista (2017) Worldwide intelligent/digital assistant market share in 2017 and 2020, by product. Retreived May 18, 20118 from https://www.statista.com/statistics/789633/worldw ide-digital-assistant-market-share/ Statista (2017) Worldwide intelligent/digital assistant market share in 2017 and 2020, by product. Retreived May 18, 20118 from https://www.statista.com/statistics/789633/worldw ide-digital-assistant-market-share/ Chen, J. X. (2016) The evolution of computing: AlphaGo. Computing in Science & Engineering, 18(4), 4-7. Arafa, Y. and Mamdani, A. (2000) Virtual personal service assistants: towards real-time characters with artificial hearts. 6. FUTURE RESEARCH OPPORTUNITIES This research study has triggered some remarkable aspects that are worth exploring separately. We are currently pursuing a multi-country study around personality attributes users perceive in these four VAVAs as well as identifying what would be the desired ones for the future. We are also exploring the current barriers for younger segments for not using VAVAs as frequently and for as many tasks as older segments. 4 Everybody is talking about Virtual Assistants, but how are people really using them? Perez ● Saffon ● Donis Everybody is talking about Virtual Assistants, but how are people really using them? Perez ● Saffon ● Donis 8. REFERENCES 5th International Conference on Intelligent User Interfaces, New Orleans, Louisiana, January 9-12, 2000, 9-12, ACM, New York, New York. Castelfranchi, C. (1998) Modelling social action for AI agents. Artificial Intelligence, 103(1-2), 157-182. Martín, P. J. and Allende, J. S. (2015) De Eliza a Siri: La evolución. Tecnología y desarrollo, 8, 3-30. Kiseleva, J., Williams, K., Hassan Awadallah, A., Crook, A. C., Zitouni, I., and Anastasakos, T. (2016) Predicting user satisfaction with intelligent assistants. 39th International ACM SIGIR conference on Research and Development in Information Retrieval, Pisa, Italy, July 17-21, 2016, 45-54, ACM, New York, New York. Siegler, M.G. (2011). The iPhone 4S: Faster, More Capable, And You Can Talk To It. TechCrunch. 5
https://openalex.org/W2900229004	https://www.intechopen.com/citation-pdf-url/63776	English	null	Anaplastic Large Cell Lymphoma	IntechOpen eBooks	2,019	cc-by	10,604	Abstract Anaplastic large cell lymphoma (ALCL) describes a distinct group of T cell lymphomas characterised by cell surface expression of CD30. At least three entities of ALCL exist, with similar cellular morphology but varying clinical courses and pathology: systemic ALCL, anaplastic lymphoma kinase (ALK)-positive, systemic ALCL ALK− and primary cutaneous ALCL. A fourth provisional entity associated with breast implants has been proposed, named breast implant-associated (BIA)- ALCL. ALCL have varying clinical outcomes, affect both children and adults, and range from being well-characterised at the genetic level to relatively unknown, predominantly due to the relative rarity of this group of malignancies. Current therapeutic approaches include standard chemotherapeutic agents as well as novel drugs including monoclonal antibodies and kinase inhibitors. Keywords: anaplastic large cell lymphoma, anaplastic lymphoma kinase, tyrosine kinase inhibitors, peripheral T cell lymphoma, BIA-ALCL Anaplastic Large Cell Lymphoma Suzanne D. Turner i ell i i 2.1 Clinical course The large majority of ALCL, ALK+ are diagnosed in a younger patient population with a median age of 10.2–11 and have a relatively good prognosis (>80% overall survival; OS) [10–15]. In contrast ALCL, ALK− more often affects an older demo- graphic (40–65 years of age) and has a poor prognosis (<50% OS) [16–19]. Whether these different clinical outcomes are age-related or due to inherent properties of the malignancies remains to be determined although in support of the latter, ALK− ALCL carrying DUSP22 rearrangements have been reported to have a superior 5-year OS of 90% (compared to 17% for TP63 rearranged cases and 42% for ALK−/DUSP22−/ TP63− cases) although if patients are stratified according to age rather than ALK status, the outcome in response to treatment is the same [17, 19, 20]. The relatively high survival rates of patients diagnosed with ALCL, ALK+ may also be attributable to the host immune response whereby cytotoxic T lymphocytes, helper T cells and B cells responding to ALK have been detected in patients [21, 22]. Patients with ALCL, ALK+ mount an immune response to the ALK protein in the form of a humoral antibody response [23]. In fact, the titre of ALK autoantibodies in a patient’s serum can be predictive of outcome with an inverse correlation between ALK antibodies and relapse [24]. This prognostic factor can be extended further when combined with the presence or absence of minimal disseminated disease (MDD), with children having low ALK autoantibody titres combined with presence of MDD being of high risk, with the converse indicative of low risk [25]. 2.2 Histopathological presentation and immunophenotype ALCL spans a broad morphological spectrum with sub-types including common (65%), small cell and lymphohistiocytic variants (32% combined) with the latter constituting a poor prognostic variable [26–28]. The unifying feature of ALCL is the presence of CD30 expression on the surface of the tumour cells, particularly the larger ones. CD30 is a marker of activated immune cells but does not distinguish between a T or B cell origin when applied in isolation. Hence, for a diagnosis of a T cell lymphoma, a cell surface protein, or combination of proteins unique to T cells must be detected. In this regard, many ALCL express CD4, CD2 and/or CD5 but often lack CD3. The positive expression of CD4 in the absence of CD8 combined with the pres- ence of cytotoxic proteins such as TIA-1, Granzyme B and/or perforin is at odds with the presumed cytotoxic T cell origin of ALCL [7, 29]. However, in some cases, no T cell specific proteins are detectable and these are categorised as being ‘null cell’, although the majority demonstrate molecular rearrangements of the T cell receptor (TCR) [30]. 1. Introduction Anaplastic large cell lymphoma (ALCL) was first described in 1985 as a CD30- positive (or ki-1+) histiocytic lymphoma, later re-classified as a distinct clinical entity, ALCL [1]. The presence of a chromosomal translocation in this malignancy was described independently by several authors in 1989–1990 [2–5]. This was further refined in 1994 on cloning of the t(2;5)(p23;q35) translocation breakpoint product, identified as a fusion protein of Nucleophosmin 1 (NPM) and anaplastic lymphoma kinase (ALK), the latter a previously uncharacterized protein named after the disease from which it was cloned [6]. Sometime later in 2008, systemic (s) ALCL was divided into two provisional entities: ALCL, ALK+ and ALCL, ALK− which were confirmed as distinct entities in the revised 4th edition of the WHO classification of tumours of haemopoietic and lymphoid tissues [7]. The revised 4th edition also includes a new provisional entity of ALCL associated with breast implants, breast implant-associated (BIA)-ALCL which may consist of at least two clinically distinguishable forms, if not a spectrum of disease, ranging from sub-capsular seroma fluid to aggressive, infiltrating masses with good and poor prognoses respectively [8, 9]. As well as systemic forms of the disease, there exists a cutaneous type belonging to the class of primary cutaneous CD30-positive T cell lymphoproliferative disorders—primary cutaneous (pc) ALCL [7]. In this chapter, the clinical and pathological presentations of each of these disease entities will be presented and discussed as will the biology underlying these malignancies. 1 Peripheral T-cell Lymphomas 2.3.1 ALCL, ALK+ ALCL is, in general, a genetically stable cancer with few common defining genetic alterations besides translocations involving ALK [31, 32]. In this regard, the t(2;5) (p23;q35) generating NPM-ALK at the breakpoint is the most common event with many variants having been published over the years (Table 1) [33]. The common expression of NPM-ALK, and its nuclear and cytoplasmic location as opposed to cytoplasmic-alone position as seen with many of the other variants, may account for its predominance in ALCL, ALK+; nuclear location may provide a competitive advan- tage over cytoplasmic alone. Alternatively, the NPM1 gene on chromosome 5 may be 2 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 p g p Chromosomal alteration Fusion protein Cellular location References t(2;5)(p23;q35) NPM-ALK Nucleus and cytoplasm [2–6] t(2;3)(p23;q12.2) TFG-ALK (short, long and extra-long isoforms) Cytoplasm [134, 135] t(1;2)(q25;p23) TPM3-ALK Cytoplasm [136] Inv(2)(p23;q35) ATIC-ALK Cytoplasm [137, 138] t(X;2)(q11–12;p23) MSN-ALK Membrane [139] t(2;17)(p23;q23) CLTC-ALK Cytoplasm (granular) [140, 141] t(2;22)(p23;q11.2) MYH9-ALK Cytoplasm [142] t(2;19)(p23;q13.1) TPM4-ALK Cytoplasm [143] t(2;17)(p23;q25) RNF213/ALO17-ALK Cytoplasm [144] Table 1. Overview of ALK fusion partners identified in ALCL, ALK+. Table 1. Overview of ALK fusion partners identified in ALCL, ALK+. more prone to breakage and fusion with new partners due to its active transcription at the same time as ALK, although there is no evidence to suggest this is the case. What is clear, is that all reported ALK fusion proteins generate a hyperactive tyrosine kinase that is ligand-independent, driving cellular proliferation and survival [33]. Taking the example of NPM-ALK, this fusion protein retains the oligomerisation domains of NPM1 and the entire intracellular portion of ALK encoding the kinase domain, resulting in dimerization, auto-phosphorylation and subsequent hyperactivity initiat- ing a whole plethora of signal transduction pathways (Figure 1) [6, 34]. more prone to breakage and fusion with new partners due to its active transcription at the same time as ALK, although there is no evidence to suggest this is the case. What is clear, is that all reported ALK fusion proteins generate a hyperactive tyrosine kinase that is ligand-independent, driving cellular proliferation and survival [33]. Taking the example of NPM-ALK, this fusion protein retains the oligomerisation domains of NPM1 and the entire intracellular portion of ALK encoding the kinase domain, resulting in dimerization, auto-phosphorylation and subsequent hyperactivity initiat- ing a whole plethora of signal transduction pathways (Figure 1) [6, 34]. Figure 1. 2.3.1 ALCL, ALK+ NPM-ALK activates a plethora of signalling pathways conferring many of the cancer hallmarks on tumour cells. NPM-ALK autophosphorylates tyrosine residues providing docking sites for SH2 domain-containing proteins and the development of a signalosome consisting of at least 46 proteins [35]. Key pathways involved in cell survival and proliferation include the PI 3-Kinase/Akt, Ras/MAP Kinase and JAK/STAT pathways as well as PLCγ [36–41]. While activation of JNK and PI 3-Kinase by NPM-ALK can drive cell proliferation, they also inactivate p53 by ubiquitin-mediated degradation [42]. NPM-ALK also activates immunomodulatory pathways including up-regulation of PDL1 mediated by STAT3, as well as silencing some proteins by epigenetic means (green arrow, Figure 1), including those associated with signalling downstream of a functional TCR [43–47]. In addition, NPM-ALK directs metabolic activity of the cells shifting to aerobic glycolysis with increased lactate and biomass production promoting cell survival [48]. Figure 1. g NPM-ALK activates a plethora of signalling pathways conferring many of the cancer hallmarks on tumour cells. NPM-ALK autophosphorylates tyrosine residues providing docking sites for SH2 domain-containing proteins and the development of a signalosome consisting of at least 46 proteins [35]. Key pathways involved in cell survival and proliferation include the PI 3-Kinase/Akt, Ras/MAP Kinase and JAK/STAT pathways as well as PLCγ [36–41]. While activation of JNK and PI 3-Kinase by NPM-ALK can drive cell proliferation, they also inactivate p53 by ubiquitin-mediated degradation [42]. NPM-ALK also activates immunomodulatory pathways including up-regulation of PDL1 mediated by STAT3, as well as silencing some proteins by epigenetic means (green arrow, Figure 1), including those associated with signalling downstream of a functional TCR [43–47]. In addition, NPM-ALK directs metabolic activity of the cells shifting to aerobic glycolysis with increased lactate and biomass production promoting cell survival [48]. 3 3 Peripheral T-cell Lymphomas While ALK translocations are diagnostic of ALCL, ALK+ and are central to disease pathogenesis, the role of other contributing mutations is largely unknown as few consistent genetic abnormalities besides those generating ALK translocations have been reported. This may, in part, be due to the plethora a cancer hallmarks that can be driven by NPM-ALK alone (Figure 1). However, array comparative genomic hybridization (aCGH) studies have highlighted some commonalities [31, 32]. For example, gains of chromosomes 7, 6q, 17p, 17q24-qter and losses of chromosomes 4q13-q21, 11q14 and 13q although the significance of these is unknown [32]. 2.3.1 ALCL, ALK+ However, a higher number of genomic imbalances as detected by aCGH at a resolution of 1 MB, has been associated with a worse prognosis [31]. p g The recognition of NPM-ALK as a driving oncogenic event and the paucity of other reported consistent genomic/genetic abnormalities in ALCL, ALK+ has led to studies of the epigenetics of ALCL [31, 49, 50]. Profiling of CpG methylation in ALCL defined a number of genes silenced in these malignancies including the TCR signalling-related proteins Zap70, LAT, CD3ε, SLP76 and the IL2Rγ chain [43, 45–47, 49, 51]. Given that NPM-ALK can substitute for signalling normally induced via an engaged TCR, activation of these proximal TCR signalling proteins may be detrimental to cell survival resulting in their evolutionary down-regulation [38, 52]. Furthermore, a number of miRNA have been implicated in tumorigenesis including miR17-92, miR135b, miR29a and miR16 [53–56]. 2.3.2 ALCL, ALK− By their very definition, ALCL, ALK− lack expression of ALK fusion proteins, but until recently, few studies had found major contributory and consistent muta- tions. DUSP22 rearrangements leading to loss of expression of DUSP22 have been reported in as many as 30% of cases and activating JAK1/STAT3 mutations in 20% [19, 57, 58]. In addition, rearrangements leading to TP63 mutation (8% of cases) and ERBB4 truncation have been demonstrated as have novel, rare rearrangements leading to the generation of NcoR2-ROS1, NFkB2-ROS1 and NFkB2-TYK2 fusion proteins [19, 57, 59, 60]. In addition, similar to ALK+ ALCL, miRNA have been implicated in disease pathogenesis including miR155 as well as others that enable a molecular distinction between ALCL, ALK+ and ALK− as well as peripheral T cell lymphoma, not otherwise specified (PTCL-NOS) [61–64]. Likewise, genomic classifiers of ALCL, ALK− amongst other peripheral T cell lymphomas have been demonstrated using a variety of genomic analysis techniques and includes the dif- ferentiating 3-gene signature of TNFRSF8, BATF3 and TMOD1 [65–68]. SNP arrays have also led to the identification of recurrent losses at 17p13 and/or 6q21 where the TP53 and PRDM1 genes are located respectively, in as many as 52% of cases sugges- tive of a role for the loss of the p53 and BLIMP1 proteins in disease pathogenesis [69] Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 The Netherlands, UK, Australia and the USA being prevalent [74–80]. Most recently, seven cases have been reported in Latin America [81]. There are approximately 5–10 million women with breast implants worldwide with rates of BIA-ALCL being pro- portionately rare although difficult to put an exact figure to. Dependent on the study conducted, incidence rates range from 1 to 89 cases per million women with breast implants [82, 83]. This reaches a much higher incidence if one considers women with textured implants alone. Almost all cases reported to date have been associated with a breast implant of a textured surface at some point during the history of the patient; whilst rare cases have been reported in women with smooth implants, the patient had been in receipt of a textured implant at some stage [78, 84]. In addition, both saline and silicone filled implants have been implicated in patients with BIA-ALCL. The tumour cells generally present as a monoclonal expansion of CD30-positive cells, as an effusion within the fibrous capsule surrounding the implant [78]. 3.2 Histopathological presentation and immunophenotype Like sALCL, BIA-ALCL is characterised by CD30 expression on lymphoid cells, in the latter situation contained within the peri-prosthetic effusion [28, 87]. These cells can be detected by immunohistochemistry, cytology and flow cytometry of seroma fluid or any solid mass [85]. A Th17/Th1 origin has been proposed whereby tumour cells secrete IFNγ, IL6, IL8, IL17 and TGFβ although a Th2 derivation has also been put forward [88–90]. 4. Primary cutaneous ALCL While skin involvement can occur as an extranodal manifestation of sALCL, isolated cutaneous disease can also occur, although this is largely ALK-negative [17]. Primary cutaneous ALCL belongs to the spectrum of CD30 positive lymphopro- liferative disorders (LPDs) and like BIA-ALCL is largely indolent in nature. While largely affecting adults who present with isolated, ulcerating nodules, children can also develop pcALCL. 3.1 Clinical course BIA-ALCL appears to represent at least two clinical entities if not a spectrum of malignancies; patients present on most occasions with an indolent seroma with rarer incidences of invasive solid masses [77]. Indeed, cases have been reported of tumour growth into the ribs with metastases to distant lymph nodes [85, 86]. 3.3 Underlying genetic alterations Like ALCL, ALK−, BIA-ALCL has not to date been associated with genomic events leading to activation of ALK. However, in concert with ALCL, ALK−, activating mutations of JAK/STAT proteins have been reported in a very few cases [91, 92]. Given the relative rarity of this disease, larger scale studies are required to elucidate the underlying genetics. 3. BIA-ALCL BIA-ALCL is a relatively new addition to the spectrum of ALCL, although the first case was reported in 1997, but did not receive much attention until further cases were identified and published, and the FDA acknowledged an association in 2011 [70, 71]. In March 2015, the French health minister issued a warning following reports of 18 cases in France [72]. A further follow-up report released by the FDA in 2017 described 414 medical device reports and 9 deaths associated with BIA-ALCL [73]. Many case series have been reviewed and reported since, with data from France, Italy, 4 5. Treatment of ALCL As for most peripheral T cell lymphomas, standard combination chemotherapy has been the mainstay of treatment for many years, specifically in the case of systemic disease [14]. In contrast, the relatively indolent cutaneous and breast implant-associated forms are primarily treated by surgical removal [86]. However, as this spectrum of diseases crosses age boundaries, there are age-specific differ- ences in therapeutic approaches. 4.1 Clinical course Like systemic ALCL, ALK−, cutaneous ALCL is also a disease of an older demographic with the majority of patients being over 50 years of age, yet is closer 5 Peripheral T-cell Lymphomas to ALCL, ALK+ in its prognosis, reaching a 5-year OS of over 90% [17]. However, relapse is relatively common in this patient group occurring in as many as 30–40% of patients and some rare cases (12–16%) can progress to systemic disease [93–95]. Spontaneous regression has been reported, although in rare cases with partial regressions being more common [96]. 4.2 Histopathological presentation and immunophenotype Diagnosis can be difficult with other cutaneous T cell lymphomas such as lymphomatoid papulosis (LyP) and transformed mycosis fungoides (MF) pro- viding differential diagnoses [9]. However, like systemic ALCL, CD30 expression is a defining feature of this malignancy as it is for the other CD30-associated LPDs. 4.3 Underlying genetic alterations Due to its relative rarity, sometimes-difficult diagnosis and indolent course, studies of the underlying genetics are few. However, limited studies have elucidated some of the genetic events that may be driving this disease process some of which are also common to sALCL. For example, as in sALCL, DUSP22-IRF4 rearrange- ments have been detected in 20–57% of pcALCL and ALK expression is seen in rare cases [19, 97–101]. In addition, aCGH has identified gains of 7q31 and losses of 6q16-21 as well as 13q34, collectively in 45% of examined patient specimens [102]. As well as similarities to ALCL, ALK-negative with regards to DUSP22 transloca- tions, upregulation of miR155 has also been observed in both pcALCL and sALCL, ALK− [61, 103]. The functional and clinical significance of these genetic events is still subject to investigation. 5.3 New and novel treatment options for ALCL In the post-genomic era, targeted agents have become the mainstay of chemo- therapy, largely in addition to standard cytotoxic drugs. In the case of ALCL, ALK+, inhibitors of ALK are the obvious choice and many have been developed since the discovery of ALK expression in Non-Small Cell Lung Cancer [14]. The first ALK inhibitor to be developed was Pfizer’s PF-2341066, now known as crizotinib, a dual ALK/cMet inhibitor with efficacy in experimental models of ALCL, ALK+ [110]. However, these drugs have been slow to make their way into the clinic for the treatment of ALCL, largely due to its relative rarity and paediatric presentation. A phase I study of crizotinib for children with relapsed/refractory ALK+ malignancies including ALCL, reported seven out of nine patients to achieve a complete response (CR) [111]. A phase II expansion cohort showed overall response rates of 83 and 90% respectively for those children receiving crizotinib at dosages of 165 and 280 mg/m2 respectively [112]. However, discontinuation of therapy has led to rapid relapse of both children and adults with ALCL, ALK+ questioning the required window of therapy [113]. py Naturally, ALK inhibitors only apply to the therapy of ALCL, ALK+. In con- trast, the common expression of CD30 on all ALCL sub-types means that targeted agents to this cell surface protein should be broadly applicable [114]. In this vein, BV, an anti-CD30 antibody tethered to the microtubule inhibitor monomethyl auristatin E, has shown promising results in clinical trials, although relapse is again an issue with down-regulation of CD30 expression seen [115–117]. However, results of the Phase 3 ALCANZA trial for pcALCL and MF showed impressive results with an objective response rate of 67% in the BV arm (versus 20% in the standard treatment arm: methotrexate or bexarotene) [118]. However, BV is not without its side-effects with peripheral neuropathy being prominent (affect- ing 67% of patients in the afore-mentioned trial) [118]. Likewise, results of a Phase 2 trial of relapsed/refractory sALCL showed peripheral neuropathy to be a considerable side-effect in 91% of patients although a 5-year OS of 79% was achieved (69% CR, 80% ORR for ALK+ patients and 52% CR, 81% ORR for ALK−) [119]. 5.1 Treatment of children with ALCL As mentioned before, the large majority of patients diagnosed with ALCL, ALK+ are children and young adults. As such, the therapeutic approach is tuned to this patient population with children receiving a combination of chemothera- peutic agents with survival rates in excess of 90% [10, 13, 14]. The ALCL99 trial, the largest trial ever to be conducted for children diagnosed with ALCL (n = 352) applied a therapeutic regimen consisting of a B cell protocol (based on NHL- BFM-B) with randomisation of vinblastine [13]. The success of this trial has led to most centres adopting the ALCL99 treatment protocol. Additionally, the success of the ALCL99 trial and the plethora of biological data produced suggest that patients might be stratified according to ALK autoantibody titre and the presence of MDD as discussed above. Indeed, vinblastine monotherapy might be more appropriate fo low risk patients reducing both acute and chronic side-effects of the combination chemotherapy protocol [104, 105]. As mentioned before, the large majority of patients diagnosed with ALCL, ALK+ are children and young adults. As such, the therapeutic approach is tuned to this patient population with children receiving a combination of chemothera- peutic agents with survival rates in excess of 90% [10, 13, 14]. The ALCL99 trial, the largest trial ever to be conducted for children diagnosed with ALCL (n = 352) applied a therapeutic regimen consisting of a B cell protocol (based on NHL- pp p g g p BFM-B) with randomisation of vinblastine [13]. The success of this trial has led to most centres adopting the ALCL99 treatment protocol. Additionally, the success of the ALCL99 trial and the plethora of biological data produced suggest that patients might be stratified according to ALK autoantibody titre and the presence of MDD as discussed above. Indeed, vinblastine monotherapy might be more appropriate for low risk patients reducing both acute and chronic side-effects of the combination chemotherapy protocol [104, 105]. 6 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 5.2 Treatment of adults with ALCL Adults with ALCL tend to be ALK− and are treated with the standard T cell lymphoma regimen CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone) although CHOEP (CHOP + etoposide) has been demonstrated to be superior in the treatment of adult ALCL, ALK+ patients [106]. In the case of BIA-ALCL, surgical excision with complete capsulectomy is recommended and is often sufficient to induce remission particularly for patients that present with a contained seroma [85, 86]. However, patients with aggressive BIA-ALCL that has metastasised require radiotherapy if not chemotherapy, and anecdotal evidence suggests that upfront use of brentuximab vedotin (BV) may benefit these patients [107]. In the case of pcALCL, localised excision and/or radiotherapy is largely prescribed due to the obvious skin presentation, although cases with multi-focal lesions may require more aggressive treatment involving chemo- therapy [108, 109]. 6.1 Cell of origin Systemic ALCL presents in the periphery suggestive of a peripheral T cell origin, although as many as 50% of children show mediastinal involvement [29]. In this latter vein, a thymic origin has been proposed whereby gene expression signatures associated with early thymic progenitors (ETP) are detected in ALCL cancer stem cells, in fitting with the detection of transcripts for the t(2;5)(p23;q35) translocation breakpoint product in 2% of cord blood specimens from healthy babies [124, 125]. In addition, studies of epigenetic signatures are in keeping with an ETP origin [49]. As such, it is not inconceivable that ALCL, ALK+ has a thymic, perhaps in utero origin in-line with the pathogenesis of paediatric leukaemias [29]. Additionally, this is in keeping with a paediatric presentation and the early-life involution of the thymus. Furthermore, studies of murine models show that events in the periphery once incipient tumour cells emerge from the thymus contribute to disease pathogenesis as discussed below [30]. p g While ALCL, ALK+ is proposed to emerge from the thymus, a similar origin likely does not apply to ALK-negative disease, including pcALCL, BIA-ALCL and ALCL, ALK−. In these latter cases, circulating peripheral T cells are most prob- ably the cells of origin given the older age of diagnosis and peripheral location, particularly with regards to BIA-ALCL and pcALCL. If this is the case, if the type of T cell that becomes transformed can be identified, this may give clues as to disease pathogenesis. While histopathology indicating an activated CD30-expressing T cell producing cytotoxic proteins, yet also often retaining CD4 expression, has given rise to a presumed cytotoxic T cell origin, recent data challenges this perception [29, 126]. Specifically, analysis of gene expression data suggests a Th17 origin, a T cell that usually responds to large extracellular infectious agents such as bacteria and is often implicated in autoimmune disease [89, 127]. However, given that ALCL often lack expression of TCR-related signalling proteins as well as a functional cell surface TCR, analogies to innate lymphoid cells (ILC), specifically ILC type 3 cells are also apparent [127]. Naturally, the eventual cell phenotype is not necessarily reflec- tive of the cell of origin with environmental events likely contributing to the final observed identity. 5.3 New and novel treatment options for ALCL A randomised Phase 3 trial to establish the efficacy of BV in combination with cyclophosphamide, doxorubicin and prednisolone, in comparison to these chemotherapeutic agents given with vincristine in place of BV, is ongoing for the frontline treatment of CD30-positive lymphomas including ALCL (ECHELON 2; NCT01777152). Other potential therapeutic targets for the treatment of ALCL include PDGFR, JAK/STAT, PD-1/PDL1 and reactivation of p53 [42, 44, 120, 121]. 7 Peripheral T-cell Lymphomas Indeed, biological studies have identified a number of potential therapeutic targets, which in some cases, and with time, have been matched to available drugs. However, with relatively few patients, coupled with a good prognosis, at least for children with ALK+ systemic disease, it is difficult to formulate trials to test these agents. A further approach given the immune response to ALK in patients with ALK- positive disease, is a vaccination strategy [122]. This is especially relevant as ALK expression seems to be limited to tissues of neonatal origin suggesting that side- effects will be limited [123]. 6.1 Cell of origin In this regard, whether in ALCL, ALK+ this is shaped by ALK- mediated activities (or is the consequence of other induced (epi)genetic events) remains to be fully elucidated as it does for other ALCL sub-types. In evidence, it has been shown that NPM-ALK induces expression of cytotoxic proteins suggesting that their presence reflects the activities of this inherent transforming event rather than a property of the cell of origin, at least for ALCL, ALK+ [128]. This would 8 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 partly explain the ‘confused’ T cell phenotype with both helper and cytotoxic T cell properties apparent. Indeed, plasticity amongst helper T cell subsets is immense and is dependent on the relative expression levels of key transcription factors such as T-bet, RORγ, GATA-3 and Foxp3 as well as cytokines in the microenvironment [129]. Hence, for a T cell aberrantly expressing a variety of genetic changes, embed- ded in specific inflammatory microenvironments, the resultant cell surface pheno- type may no longer reflect the cell of origin. Another factor to consider is genetic predisposition or health status of the patients whereby some, with for example, autoimmune disease or allergies and a preponderance of Th17 or Th1/Th2 cells respectively may be more at risk, with the resultant tumour phenotype dependent on this. In evidence, at least for BIA-ALCL Th1, Th2 and Th17 origins have been proposed based on the profile of secreted cytokines and expression of specific transcription factors, although of course none of these factors in isolation are necessarily truly indicative of the cell of origin, and as mentioned before, the contribution of the microenvironment cannot be discounted [88–90]. 6.2 An infectious aetiology? The common expression of CD30 on all entities of ALCL is suggestive of an infectious aetiology whereby activation of the underlying T cells triggers expres- sion of this cell surface protein. However, individual cell surface proteins in isola- tion are not necessarily indicative of the cell of origin of any given cancer, which combined with the propensity of cancer cells to aberrantly up- or down-regulate expression of proteins according to evolutionary fitness necessitates further evidence to draw conclusive decisions. Yet, in evidence of an infectious aetiology, Figure 2. Proposed mechanisms of tumorigenesis for ALCL. Data suggest that the NPM-ALK generating chromosomal translocation occurs in primitive haemopoietic cells, such as early thymic progenitors, whereby aberrant TCR rearrangements are tolerated [30]. Incipient tumour cells then exit into the periphery where secondary events lead to transformation. Conversely, systemic ALCL, ALK−, pcALCL and BIA-ALCL more likely initiate in circulating peripheral T cells whereby chronic antigenic stimulation mediated by infectious agents, an inflammatory milieu and/or toxic insult leads to the acquisition of malignancy-promoting mutations and cellular transformation. 7. Conclusions ALCL is a diverse disease entity affecting a range of patients ranging from children to women with breast implants. What is clear, is that all ALCL share some common immunohistopathological features, most prominently CD30 expression, but the clinical courses of these diseases vary considerably from the indolent LPD, pcALCL through to aggressive, poor prognostic malignan- cies such as sALCL, ALK−. Our understanding of the underlying biology is improving year on year and has had a significant impact on clinical decision making including therapeutic approaches. While for many forms of ALCL, therapy has not altered considerably over the past decade, novel targeted approaches to treatment are entering the clinical arena ranging from mono- clonal antibodies to kinase inhibitors. Indeed, we are now in the fortunate position whereby there are a plethora of therapeutic agents, but too few patients to trial them. Figure 2. g Proposed mechanisms of tumorigenesis for ALCL. Data suggest that the NPM-ALK generating chromosomal translocation occurs in primitive haemopoietic cells, such as early thymic progenitors, whereby aberrant TCR rearrangements are tolerated [30]. Incipient tumour cells then exit into the periphery where secondary events lead to transformation. Conversely, systemic ALCL, ALK−, pcALCL and BIA-ALCL more likely initiate in circulating peripheral T cells whereby chronic antigenic stimulation mediated by infectious agents, an inflammatory milieu and/or toxic insult leads to the acquisition of malignancy-promoting mutations and cellular transformation. 9 Peripheral T-cell Lymphomas sALCL have been reported in the context of insect and tick bites, as well as bacte- rial infections on the surface of breast implants in BIA-ALCL and in association with cutaneous T cell lymphomas whereby TLRs 2, 4 and 7 are expressed by tumour cells [130–132]. Such infectious aetiologies would also produce an inflam- matory microenvironment dictated by the infectious agent whereby cytokines, growth factors and many cell types involved in inflammation would be present and may contribute to disease pathogenesis. In this regard, the lymphohistiocytic subtype of sALCL is, as its name suggests, infiltrated with macrophages and many cytokines have been detected at elevated levels in patients diagnosed with ALCL, ALK+ [26, 133] (Figure 2). Acknowledgements I would like to acknowledge all authors that have contributed to our understand- ing of ALCL who could not be cited due to page length restrictions. Conflict of interest The author declares no conflicts of interest. Conflict of interest The author declares no conflicts of interest. 10 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 Author details © 2018 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/ by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Author details Suzanne D. Turner Department of Pathology, University of Cambridge, Cambridge, UK *Address all correspondence to: sdt36@cam.ac.uk 11 Peripheral T-cell Lymphomas [15] d'Amore ES et al. Anaplastic large cell lymphomas: A study of 75 pediatric patients. Pediatric and Developmental Pathology. 2007;10(3):181-191 [4] Bitter MA et al. Morphology in Ki-1(CD30)-positive non-Hodgkin's lymphoma is correlated with clinical features and the presence of a unique chromosomal abnormality, t(2;5) (p23;q35). The American Journal of Surgical Pathology. 1990;14(4): 305-316 References [1] Stein H et al. The expression of the Hodgkin's disease associated antigen Ki-1 in reactive and neoplastic lymphoid tissue: Evidence that Reed-Sternberg cells and histiocytic malignancies are derived from activated lymphoid cells. Blood. 1985;66(4):848-858 [9] Swerdlow SH. World Health Organization, and International Agency for Research on Cancer. In: WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. Revised 4th ed. World Health Organization Classification of Tumours; 2017. p. 585 [2] Kaneko Y et al. A novel translocation, t(2,5)(p23;q35), in childhood phagocytic large T-cell lymphoma mimicking malignant histiocytosis. Blood. 1989;73(3):806-813 [2] Kaneko Y et al. A novel translocation, t(2,5)(p23;q35), in childhood phagocytic large T-cell lymphoma mimicking malignant histiocytosis. Blood. 1989;73(3):806-813 [10] Minard-Colin V et al. Non- Hodgkin lymphoma in children and adolescents: Progress through effective collaboration, current knowledge, and challenges ahead. Journal of Clinical Oncology. 2015;33(27):2963-2974 [3] Le Beau MM et al. The t(2;5) (p23;q35): A recurring chromosomal abnormality in Ki-1-positive anaplastic large cell lymphoma. Leukemia. 1989;3(12):866-870 [3] Le Beau MM et al. The t(2;5) (p23;q35): A recurring chromosomal abnormality in Ki-1-positive anaplastic large cell lymphoma. Leukemia. 1989;3(12):866-870 [11] Turner SD et al. Anaplastic large cell lymphoma in paediatric and young adult patients. British Journal of Haematology. 2016;173(4):560-572 [11] Turner SD et al. Anaplastic large cell lymphoma in paediatric and young adult patients. British Journal of Haematology. 2016;173(4):560-572 [4] Bitter MA et al. Morphology in Ki-1(CD30)-positive non-Hodgkin's lymphoma is correlated with clinical features and the presence of a unique chromosomal abnormality, t(2;5) (p23;q35). The American Journal of Surgical Pathology. 1990;14(4): 305-316 [12] Le Deley MC et al. Prognostic factors in childhood anaplastic large cell lymphoma: Results of a large European intergroup study. Blood. 2008;111(3):1560-1566 [13] Le Deley MC et al. Vinblastine in children and adolescents with high-risk anaplastic large-cell lymphoma: Results of the randomized ALCL99-vinblastine trial. Journal of Clinical Oncology. 2010;28(25):3987-3993 [5] Rimokh R et al. A translocation involving a specific breakpoint (q35) on chromosome 5 is characteristic of anaplastic large cell lymphoma ('Ki-1 lymphoma'). British Journal of Haematology. 1989;71(1):31-36 [14] Prokoph N et al. Treatment options for paediatric anaplastic large cell lymphoma (ALCL): Current standard and beyond. Cancers (Basel). 2018 Mar 30;10(4). pii: E99. DOI: 10.3390/ cancers10040099 [6] Morris SW et al. Fusion of a kinase gene, ALK, to a nucleolar protein gene, NPM, in non-Hodgkin's lymphoma. Science. 1994;263(5151):1281-1284 [7] Swerdlow SH. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. Lyon: International Agency for Research on Cancer; 2008. [14] Prokoph N et al. Treatment options for paediatric anaplastic large cell lymphoma (ALCL): Current standard and beyond. Cancers (Basel). 2018 Mar 30;10(4). pii: E99. DOI: 10.3390/ cancers10040099 [22] Ait-Tahar K et al. B and CTL responses to the ALK protein in patients with ALK-positive ALCL. International Journal of Cancer. 2006;118(3): 688-695 References 439 p [15] d'Amore ES et al. Anaplastic large cell lymphomas: A study of 75 pediatric patients. Pediatric and Developmental Pathology. 2007;10(3):181-191 [8] Swerdlow SH et al. The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood. 2016;127(20):2375-2390 [8] Swerdlow SH et al. The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood. 2016;127(20):2375-2390 [16] ten Berge RL et al. ALK-negative anaplastic large-cell lymphoma demonstrates similar poor prognosis to peripheral T-cell lymphoma, 12 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 unspecified. Histopathology. 2003;43(5):462-469 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 unspecified. Histopathology. 2003;43(5):462-469 oncoantigen in pediatric anaplastic lymphoma kinase-positive anaplastic large cell lymphoma with tumor dissemination and relapse risk. Blood. 2010;115(16):3314-3319 [17] Savage KJ et al. ALK-anaplastic large-cell lymphoma is clinically and immunophenotypically different from both ALK+ ALCL and peripheral T-cell lymphoma, not otherwise specified: Report from the international peripheral T-cell lymphoma project. Blood. 2008;111(12):5496-5504 [25] Mussolin L et al. Kinetics of humoral response to ALK and its relationship with minimal residual disease in pediatric ALCL. Leukemia. 2009;23(2):400-402 [25] Mussolin L et al. Kinetics of humoral response to ALK and its relationship with minimal residual disease in pediatric ALCL. Leukemia. 2009;23(2):400-402 [26] Benharroch D et al. ALK-positive lymphoma: A single disease with a broad spectrum of morphology. Blood. 1998;91(6):2076-2084 [26] Benharroch D et al. ALK-positive lymphoma: A single disease with a broad spectrum of morphology. Blood. 1998;91(6):2076-2084 [26] Benharroch D et al. ALK-positive lymphoma: A single disease with a broad spectrum of morphology. Blood. 1998;91(6):2076-2084 [18] Gascoyne RD et al. Prognostic significance of anaplastic lymphoma kinase (ALK) protein expression in adults with anaplastic large cell lymphoma. Blood. 1999;93(11):3913-3921 [27] Lamant L et al. Prognostic impact of morphologic and phenotypic features of childhood ALK-positive anaplastic large-cell lymphoma: Results of the ALCL99 study. Journal of Clinical Oncology. 2011;29(35):4669-4676 [19] Parrilla Castellar ER et al. ALK- negative anaplastic large cell lymphoma is a genetically heterogeneous disease with widely disparate clinical outcomes. Blood. 2014;124(9):1473-1480 [28] Montes-Mojarro IA et al. The pathological Spectrum of systemic anaplastic large cell lymphoma (ALCL). Cancers (Basel). 2018 Apr 4;10(4). pii: E107. DOI: 10.3390/cancers10040107 [20] Sibon D et al. Long-term outcome of adults with systemic anaplastic large-cell lymphoma treated within the Groupe d'Etude des Lymphomes de l'Adulte trials. Journal of Clinical Oncology. 2012;30(32):3939-3946 [29] Malcolm TI et al. Challenging perspectives on the cellular origins of lymphoma. Open Biology. 2016 Sep;6(9). [23] Pulford K et al. Immune response to the ALK oncogenic tyrosine kinase in patients with anaplastic large-cell lymphoma. Blood. 2000;96(4):1605-1607 Haematology. 2008;140(5): 516-526 [33] Duyster J, Bai RY, Morris SW. Translocations involving anaplastic lymphoma kinase (ALK). Oncogene. 2001;20(40):5623-5637 [42] Cui YX et al. NPM-ALK inhibits the p53 tumor suppressor pathway in an MDM2 and JNK-dependent manner. Blood. 2009;113(21):5217-5227 [42] Cui YX et al. NPM-ALK inhibits the p53 tumor suppressor pathway in an MDM2 and JNK-dependent manner. Blood. 2009;113(21):5217-5227 [34] Shiota M, Mori S. Anaplastic large cell lymphomas expressing the novel chimeric protein p80NPM/ALK: A distinct clinicopathologic entity. Leukemia. 1997;11(Suppl 3):538-540 [34] Shiota M, Mori S. Anaplastic large cell lymphomas expressing the novel chimeric protein p80NPM/ALK: A distinct clinicopathologic entity. Leukemia. 1997;11(Suppl 3):538-540 [43] Ambrogio C et al. NPM-ALK oncogenic tyrosine kinase controls T-cell identity by transcriptional regulation and epigenetic silencing in lymphoma cells. Cancer Research. 2009;69(22):8611-8619 [35] Crockett DK et al. Identification of NPM-ALK interacting proteins by tandem mass spectrometry. Oncogene. 2004;23(15):2617-2629 [35] Crockett DK et al. Identification of NPM-ALK interacting proteins by tandem mass spectrometry. Oncogene. 2004;23(15):2617-2629 [44] Marzec M et al. Oncogenic kinase NPM/ALK induces through STAT3 expression of immunosuppressive protein CD274 (PD-L1, B7-H1). Proceedings of the National Academy of Sciences of the United States of America. 2008;105(52):20852-20857 [36] Marzec M et al. Oncogenic tyrosine kinase NPM/ALK induces activation of the rapamycin-sensitive mTOR signaling pathway. Oncogene. 2007;26(38):5606-5614 [36] Marzec M et al. Oncogenic tyrosine kinase NPM/ALK induces activation of the rapamycin-sensitive mTOR signaling pathway. Oncogene. 2007;26(38):5606-5614 [37] Marzec M et al. Oncogenic tyrosine kinase NPM/ALK induces activation of the MEK/ERK signaling pathway independently of c-Raf. Oncogene. 2007;26(6):813-821 [45] Zhang Q et al. IL-2R common gamma-chain is epigenetically silenced by nucleophosphin-anaplastic lymphoma kinase (NPM-ALK) and acts as a tumor suppressor by targeting NPM-ALK. Proceedings of the National Academy of Sciences of the United States of America. 2011;108(29):11977-11982 [38] Turner SD et al. The NPM-ALK tyrosine kinase mimics TCR signalling pathways, inducing NFAT and AP-1 by RAS-dependent mechanisms. Cellular Signalling. 2007;19(4):740-747 [46] Zhang Q et al. STAT5A is epigenetically silenced by the tyrosine kinase NPM1-ALK and acts as a tumor suppressor by reciprocally inhibiting NPM1-ALK expression. Nature Medicine. 2007;13(11):1341-1348 [39] Turner SD et al. Vav-promoter regulated oncogenic fusion protein NPM-ALK in transgenic mice causes B-cell lymphomas with hyperactive Jun kinase. Oncogene. 2003;22(49):7750-7761 [47] Zhang Q et al. STAT3 induces transcription of the DNA methyltransferase 1 gene (DNMT1) in malignant T lymphocytes. Blood. 2006;108(3):1058-1064 [40] Zamo A et al. Anaplastic lymphoma kinase (ALK) activates Stat3 and protects hematopoietic cells from cell death. Oncogene. References pii: 160232 [21] Ait-Tahar K, Barnardo MC, Pulford K. CD4 T-helper responses to the anaplastic lymphoma kinase (ALK) protein in patients with ALK-positive anaplastic large-cell lymphoma. Cancer Research. 2007;67(5):1898-1901 [30] Malcolm TI et al. Anaplastic large cell lymphoma arises in thymocytes and requires transient TCR expression for thymic egress. Nature Communications. 2016;7:10087 [22] Ait-Tahar K et al. B and CTL responses to the ALK protein in patients with ALK-positive ALCL. International Journal of Cancer. 2006;118(3): 688-695 [22] Ait-Tahar K et al. B and CTL responses to the ALK protein in patients with ALK-positive ALCL. International Journal of Cancer. 2006;118(3): 688-695 [31] Youssif C et al. Genomic profiling of pediatric ALK-positive anaplastic large cell lymphoma: A Children's Cancer and Leukaemia Group Study. Genes, Chromosomes & Cancer. 2009;48(11):1018-1026 [23] Pulford K et al. Immune response to the ALK oncogenic tyrosine kinase in patients with anaplastic large-cell lymphoma. Blood. 2000;96(4):1605-1607 [32] Salaverria I et al. Genomic profiling reveals different genetic aberrations in systemic ALK-positive and ALK-negative anaplastic large cell lymphomas. British Journal of [24] Ait-Tahar K et al. Correlation of the autoantibody response to the ALK 13 Peripheral T-cell Lymphomas Haematology. 2008;140(5): 516-526 kinase that utilizes phospholipase C-gamma to mediate its mitogenicity. Molecular and Cellular Biology. 1998;18(12):6951-6961 [38] Turner SD et al. The NPM-ALK tyrosine kinase mimics TCR signalling pathways, inducing NFAT and AP-1 by RAS-dependent mechanisms. Cellular Signalling. 2007;19(4):740-747 [40] Zamo A et al. Anaplastic lymphoma kinase (ALK) activates Stat3 and protects hematopoietic cells from cell death. Oncogene. 2002;21(7):1038-1047 Haematology. 2008;140(5): 516-526 2002;21(7):1038-1047 [48] McDonnell SR et al. Integrated phosphoproteomic and metabolomic profiling reveals NPM-ALK- mediated phosphorylation of PKM2 and metabolic reprogramming in [41] Bai RY et al. Nucleophosmin- anaplastic lymphoma kinase of large-cell anaplastic lymphoma is a constitutively active tyrosine 14 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 anaplastic large cell lymphoma. Blood. 2013;122(6):958-968 anaplastic large cell lymphoma. Blood. 2013;122(6):958-968 [56] Spaccarotella E et al. STAT3- mediated activation of microRNA cluster 17~92 promotes proliferation and survival of ALK-positive anaplastic large cell lymphoma. Haematologica. 2014;99(1):116-124 [56] Spaccarotella E et al. STAT3- mediated activation of microRNA cluster 17~92 promotes proliferation and survival of ALK-positive anaplastic large cell lymphoma. Haematologica. 2014;99(1):116-124 [49] Hassler MR et al. Insights into the pathogenesis of anaplastic large-cell lymphoma through genome-wide DNA methylation profiling. Cell Reports. 2016;17(2):596-608 [49] Hassler MR et al. Insights into the pathogenesis of anaplastic large-cell lymphoma through genome-wide DNA methylation profiling. Cell Reports. 2016;17(2):596-608 [57] Crescenzo R et al. Convergent mutations and kinase fusions lead to oncogenic STAT3 activation in anaplastic large cell lymphoma. Cancer Cell. 2015;27(4):516-532 [57] Crescenzo R et al. Convergent mutations and kinase fusions lead to oncogenic STAT3 activation in anaplastic large cell lymphoma. Cancer Cell. 2015;27(4):516-532 [50] Merkel O et al. Identification of differential and functionally active miRNAs in both anaplastic lymphoma kinase (ALK)+ and ALK− anaplastic large-cell lymphoma. Proceedings of the National Academy of Sciences of the United States of America. 2010;107(37):16228-16233 [58] Feldman AL et al. Discovery of recurrent t(6;7)(p25.3;q32.3) translocations in ALK-negative anaplastic large cell lymphomas by massively parallel genomic sequencing. Blood. 2011;117(3):915-919 [58] Feldman AL et al. Discovery of recurrent t(6;7)(p25.3;q32.3) translocations in ALK-negative anaplastic large cell lymphomas by massively parallel genomic sequencing. Blood. 2011;117(3):915-919 [51] Bonzheim I et al. Anaplastic large cell lymphomas lack the expression of T-cell receptor molecules or molecules of proximal T-cell receptor signaling. Blood. 2004;104(10):3358-3360 [59] Vasmatzis G et al. Genome- wide analysis reveals recurrent structural abnormalities of TP63 and other p53-related genes in peripheral T-cell lymphomas. Blood. 2012;120(11):2280-2289 [52] Marzec M et al. Malignant transformation of CD4+ T lymphocytes mediated by oncogenic kinase NPM/ ALK recapitulates IL-2-induced cell signaling and gene expression reprogramming. Journal of Immunology. 2013;191(12):6200-6207 [60] Scarfo I et al. Identification of a new subclass of ALK-negative ALCL expressing aberrant levels of ERBB4 transcripts. Blood. 2016;127(2): 221-232 [60] Scarfo I et al. Identification of a new subclass of ALK-negative ALCL expressing aberrant levels of ERBB4 transcripts. Blood. [61] Merkel O et al. Oncogenic role of miR-155 in anaplastic large cell lymphoma lacking the t(2;5) translocation. The Journal of Pathology. 2015 Aug;236(4):445-456. DOI: 10.1002/ path.4539. Epub 2015 Apr 27 [60] Scarfo I et al. Identification of a new subclass of ALK-negative ALCL expressing aberrant levels of ERBB4 transcripts. Blood. 2016;127(2): 221-232 [63] Laginestra MA et al. Pathogenetic and diagnostic significance of microRNA deregulation in peripheral T-cell lymphoma not otherwise specified. Blood Cancer Journal. 2014;4:259 [62] Steinhilber J et al. Next- generation sequencing identifies deregulation of microRNAs involved in both innate and adaptive immune response in ALK+ ALCL. PLoS One. 2015;10(2):e0117780 Haematology. 2008;140(5): 516-526 2016;127(2): 221-232 [53] Matsuyama H et al. miR- 135b mediates NPM-ALK-driven oncogenicity and renders IL-17- producing immunophenotype to anaplastic large cell lymphoma. Blood. 2011;118(26):6881-6892 [61] Merkel O et al. Oncogenic role of miR-155 in anaplastic large cell lymphoma lacking the t(2;5) translocation. The Journal of Pathology. 2015 Aug;236(4):445-456. DOI: 10.1002/ path.4539. Epub 2015 Apr 27 [54] Desjobert C et al. MiR-29a down-regulation in ALK-positive anaplastic large cell lymphomas contributes to apoptosis blockade through MCL-1 overexpression. Blood. 2011;117(24):6627-6637 [62] Steinhilber J et al. Next- generation sequencing identifies deregulation of microRNAs involved in both innate and adaptive immune response in ALK+ ALCL. PLoS One. 2015;10(2):e0117780 [62] Steinhilber J et al. Next- generation sequencing identifies deregulation of microRNAs involved in both innate and adaptive immune response in ALK+ ALCL. PLoS One. 2015;10(2):e0117780 [55] Dejean E et al. Hypoxia- microRNA-16 downregulation induces VEGF expression in anaplastic lymphoma kinase (ALK)-positive anaplastic large-cell lymphomas. Leukemia. 2011 Dec;25(12):1882-1890. DOI: 10.1038/ leu.2011.168. Epub 2011 Jul 22 [63] Laginestra MA et al. Pathogenetic and diagnostic significance of microRNA deregulation in peripheral T-cell lymphoma not otherwise specified. Blood Cancer Journal. 2014;4:259 15 Peripheral T-cell Lymphomas www.fda.gov/MedicalDevices/ ProductsandMedicalProcedures/ ImplantsandProsthetics/ BreastImplants/ucm239996.htm [64] Liu C et al. MicroRNA expression profiling identifies molecular signatures associated with anaplastic large cell lymphoma. Blood. 2013;122(12):2083-2092 [72] Ministere des affaires sociales, d.l.s.e.d.d.d.f., La France reste vigilante pour detecter rapidement les cas de lymphomes parmi les femmes porteuses d'implants mammaires. 2015 [65] Piva R et al. Gene expression profiling uncovers molecular classifiers for the recognition of anaplastic large-cell lymphoma within peripheral T-cell neoplasms. Journal of Clinical Oncology. 2010;28(9):1583-1590 [65] Piva R et al. Gene expression profiling uncovers molecular classifiers for the recognition of anaplastic large-cell lymphoma within peripheral T-cell neoplasms. Journal of Clinical Oncology. 2010;28(9):1583-1590 [73] Administration, U.F.a.D., Breast Implant-Associated Anaplastic Large Cell Lymphoma (BIA-ALCL). 2017 [73] Administration, U.F.a.D., Breast Implant-Associated Anaplastic Large Cell Lymphoma (BIA-ALCL). 2017 [66] Piccaluga PP et al. Molecular profiling improves classification and prognostication of nodal peripheral T-cell lymphomas: Results of a phase III diagnostic accuracy study. Journal of Clinical Oncology. 2013;31(24):3019-3025 [66] Piccaluga PP et al. Molecular profiling improves classification and prognostication of nodal peripheral T-cell lymphomas: Results of a phase III diagnostic accuracy study. Journal of Clinical Oncology. 2013;31(24):3019-3025 [74] Antonella C, Rosaria B, Marcella M. 22 cases of BIA-ALCL: Awareness and outcome tracking from the Italian Ministry of Health. Plastic and Reconstructive Surgery. 2017 [75] Johnson L et al. Haematology. 2008;140(5): 516-526 Breast implant associated anaplastic large cell lymphoma: The UK experience. Recommendations on its management and implications for informed consent. European Journal of Surgical Oncology. 2017;43(8):1393-1401 [75] Johnson L et al. Breast implant associated anaplastic large cell lymphoma: The UK experience. Recommendations on its management and implications for informed consent. European Journal of Surgical Oncology. 2017;43(8):1393-1401 [67] Agnelli L et al. Identification of a 3-gene model as a powerful diagnostic tool for the recognition of ALK-negative anaplastic large-cell lymphoma. Blood. 2012;120(6):1274-1281 [68] Iqbal J et al. Molecular signatures to improve diagnosis in peripheral T-cell lymphoma and prognostication in angioimmunoblastic T-cell lymphoma. Blood. 2010;115(5):1026-1036 [68] Iqbal J et al. Molecular signatures to improve diagnosis in peripheral T-cell lymphoma and prognostication in angioimmunoblastic T-cell lymphoma. Blood. 2010;115(5):1026-1036 [76] de Jong D et al. Anaplastic large-cell lymphoma in women with breast implants. JAMA. 2008;300(17):2030-2035 [77] Laurent C et al. Breast implant-associated anaplastic large cell lymphoma: Two distinct clinicopathological variants with different outcomes. Annals of Oncology. 2016;27(2):306-314 [69] Boi M et al. PRDM1/BLIMP1 is commonly inactivated in anaplastic large T-cell lymphoma. Blood. 2013;122(15):2683-2693 [70] Keech JA Jr, Creech BJ. Anaplastic T-cell lymphoma in proximity to a saline-filled breast implant. Plastic and Reconstructive Surgery. 1997;100(2):554-555 [70] Keech JA Jr, Creech BJ. Anaplastic T-cell lymphoma in proximity to a saline-filled breast implant. Plastic and Reconstructive Surgery. 1997;100(2):554-555 [78] Clemens MW, Miranda RN. Coming of age: Breast implant-associated anaplastic large cell lymphoma after 18 years of investigation. Clinics in Plastic Surgery. 2015;42(4):605-613 [78] Clemens MW, Miranda RN. Coming of age: Breast implant-associated anaplastic large cell lymphoma after 18 years of investigation. Clinics in Plastic Surgery. 2015;42(4):605-613 [71] Administration, U.F.a.D. Anaplastic large cell lymphoma (ALCL) in women with breast implants: Preliminary findings and analyses. 2011. Available from: http://wayback.archive-it. org/7993/20171115053750/https:/ [79] Doren EL et al. U.S. epidemiology of breast implant-associated anaplastic large cell lymphoma. Plastic and Reconstructive Surgery. 2017;139(5):1042-1050 16 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 Breaking down the evidence. Mutation Research, Reviews in Mutation Research. 2014;762:123-132 Breaking down the evidence. Mutation Research, Reviews in Mutation Research. 2014;762:123-132 [80] Srinivasa DR et al. Global adverse event reports of breast implant- associated ALCL: An international review of 40 government authority databases. Plastic and Reconstructive Surgery. 2017;139(5):1029-1039 [80] Srinivasa DR et al. Global adverse event reports of breast implant- associated ALCL: An international review of 40 government authority databases. Plastic and Reconstructive Surgery. [86] Clemens MW et al. Complete surgical excision is essential for the management of patients with breast implant-associated anaplastic large- cell lymphoma. Journal of Clinical Oncology. 2016;34(2):160-168 Haematology. 2008;140(5): 516-526 2017;139(5):1029-1039 [88] Lechner MG et al. Survival signals and targets for therapy in breast implant-associated ALK—Anaplastic large cell lymphoma. Clinical Cancer Research. 2012;18(17):4549-4559 [81] Ramos-Gallardo G et al. Is Latin America Ready to Identify Anaplastic Large Cell Lymphoma in Breast Implants Patients? Regional Encounter during the National Plastic Surgery Meeting in Cancun. Mexico: Aesthetic Plast Surg; 2018 [89] Kadin ME et al. Biomarkers provide clues to early events in the pathogenesis of breast implant-associated anaplastic large cell lymphoma. Aesthetic Surgery Journal. 2016;36(7):773-781 [82] McGuire P, Reisman NR, Murphy DK. Risk factor analysis for capsular contracture, malposition, and late Seroma in subjects receiving Natrelle 410 form-stable silicone breast implants. Plastic and Reconstructive Surgery. 2017;139(1):1-9 [90] Kadin ME et al. IL-13 is produced by tumor cells in breast implant associated anaplastic large cell lymphoma: Implications for pathogenesis. Human Pathology. 2018 Aug;78:54-62. DOI: 10.1016/j.humpath.2018.04.007. Epub 2018 Apr 22 [90] Kadin ME et al. IL-13 is produced by tumor cells in breast implant associated anaplastic large cell lymphoma: Implications for pathogenesis. Human Pathology. 2018 Aug;78:54-62. DOI: 10.1016/j.humpath.2018.04.007. Epub 2018 Apr 22 [83] de Boer M et al. Breast implants and the risk of anaplastic large-cell lymphoma in the breast. JAMA Oncology. 2018;4:335-341 [83] de Boer M et al. Breast implants and the risk of anaplastic large-cell lymphoma in the breast. JAMA Oncology. 2018;4:335-341 [84] Loch-Wilkinson A et al. Breast implant-associated anaplastic large cell lymphoma in Australia and New Zealand: High-surface-area textured implants are associated with increased risk. Plastic and Reconstructive Surgery. 2017;140(4):645-654 [84] Loch-Wilkinson A et al. Breast implant-associated anaplastic large cell lymphoma in Australia and New Zealand: High-surface-area textured implants are associated with increased risk. Plastic and Reconstructive Surgery. 2017;140(4):645-654 [92] Di Napoli A et al. Targeted next generation sequencing of breast implant-associated anaplastic large cell lymphoma reveals mutations in JAK/STAT signalling pathway genes, TP53 and DNMT3A. British Journal of Haematology. 2018;180(5):741-744 [85] Clemens MW, Horwitz SM. NCCN consensus guidelines for the diagnosis and management of breast implant- associated anaplastic large cell lymphoma. Aesthetic Surgery Journal. 2017;37(3):285-289 [85] Clemens MW, Horwitz SM. NCCN consensus guidelines for the diagnosis and management of breast implant- associated anaplastic large cell lymphoma. Aesthetic Surgery Journal. 2017;37(3):285-289 [93] Bekkenk MW et al. Primary and secondary cutaneous CD30(+) lymphoproliferative disorders: A report from the Dutch cutaneous lymphoma group on the long-term follow-up data of 219 patients and guidelines for diagnosis and treatment. Blood. 2000;95(12):3653-3661 [86] Clemens MW et al. [91] Blombery P et al. Whole exome sequencing reveals activating JAK1 and STAT3 mutations in breast implant-associated anaplastic large cell lymphoma anaplastic large cell lymphoma. Haematologica. 2016;101(9):e387-e390 Haematology. 2008;140(5): 516-526 Complete surgical excision is essential for the management of patients with breast implant-associated anaplastic large- cell lymphoma. Journal of Clinical Oncology. 2016;34(2):160-168 [94] Liu HL et al. CD30+ cutaneous lymphoproliferative disorders: The Stanford experience in lymphomatoid [87] Ye X et al. Anaplastic large cell lymphoma (ALCL) and breast implants: 17 Peripheral T-cell Lymphomas papulosis and primary cutaneous anaplastic large cell lymphoma. Journal of the American Academy of Dermatology. 2003;49(6): 1049-1058 [102] Szuhai K et al. Array-CGH analysis of cutaneous anaplastic large cell lymphoma. Methods in Molecular Biology. 2013;973:197-212 [102] Szuhai K et al. Array-CGH analysis of cutaneous anaplastic large cell lymphoma. Methods in Molecular Biology. 2013;973:197-212 papulosis and primary cutaneous anaplastic large cell lymphoma. Journal of the American Academy of Dermatology. 2003;49(6): 1049-1058 [103] Benner MF et al. Primary cutaneous anaplastic large cell lymphoma shows a distinct miRNA expression profile and reveals differences from tumor-stage mycosis fungoides. Experimental Dermatology. 2012;21(8):632-634 [103] Benner MF et al. Primary cutaneous anaplastic large cell lymphoma shows a distinct miRNA expression profile and reveals differences from tumor-stage mycosis fungoides. Experimental Dermatology. 2012;21(8):632-634 [95] Kempf W et al. EORTC, ISCL, and USCLC consensus recommendations for the treatment of primary cutaneous CD30-positive lymphoproliferative disorders: Lymphomatoid papulosis and primary cutaneous anaplastic large-cell lymphoma. Blood. 2011;118(15):4024-4035 [95] Kempf W et al. EORTC, ISCL, and USCLC consensus recommendations for the treatment of primary cutaneous CD30-positive lymphoproliferative disorders: Lymphomatoid papulosis and primary cutaneous anaplastic large-cell lymphoma. Blood. 2011;118(15):4024-4035 [104] Garner R et al. Long-term disease control of refractory anaplastic large cell lymphoma with vinblastine. Journal of Pediatric Hematology/Oncology. 2009;31(2):145-147 [96] Krishnan J, Tomaszewski MM, Kao GF. Primary cutaneous CD30-positive anaplastic large cell lymphoma. Report of 27 cases. Journal of Cutaneous Pathology. 1993;20(3):193-202 [105] Brugieres L et al. Single-drug vinblastine as salvage treatment for refractory or relapsed anaplastic large-cell lymphoma: A report from the French Society of Pediatric Oncology. Journal of Clinical Oncology. 2009;27(30):5056-5061 [97] Kiran T et al. The significance of MUM1/IRF4 protein expression and IRF4 translocation of CD30(+) cutaneous T-cell lymphoproliferative disorders: A study of 53 cases. Leukemia Research. 2013;37(4):396-400 [106] Cederleuf H et al. The addition of etoposide to CHOP is associated with improved outcome in ALK+ adult anaplastic large cell lymphoma: A Nordic Lymphoma Group study. British Journal of Haematology. 2017;178(5):739-746 [98] Geller S et al. ALK-positive primary cutaneous anaplastic large cell lymphoma: A case report and review of the literature. International Journal of Dermatology. [100] Pham-Ledard A et al. IRF4 gene rearrangements define a subgroup of CD30-positive cutaneous T-cell lymphoma: A study of 54 cases. The Journal of Investigative Dermatology. 2010;130(3):816-825 [98] Geller S et al. ALK-positive primary cutaneous anaplastic large cell lymphoma: A case report and review of the literature. International Journal of Dermatology. 2018;57(5):515-520 [97] Kiran T et al. The significance of MUM1/IRF4 protein expression and IRF4 translocation of CD30(+) cutaneous T-cell lymphoproliferative disorders: A study of 53 cases. Leukemia Research. 2013;37(4):396-400 [100] Pham-Ledard A et al. IRF4 gene rearrangements define a subgroup of CD30-positive cutaneous T-cell lymphoma: A study of 54 cases. The Journal of Investigative Dermatology. 2010;130(3):816-825 [115] Nielson C et al. Loss of CD30 expression in anaplastic large cell lymphoma following brentuximab therapy. Journal of Drugs in Dermatology. 2016;15(7): 894-895 [114] Senter PD, Sievers EL. The discovery and development of brentuximab vedotin for use in relapsed Hodgkin lymphoma and systemic anaplastic large cell lymphoma. Nature Biotechnology. 2012;30(7):631-637 [110] Christensen JG et al. Cytoreductive antitumor activity of PF-2341066, a novel inhibitor of anaplastic lymphoma kinase and c-Met, in experimental models of anaplastic large-cell lymphoma. Molecular Cancer Therapeutics. 2007;6(12 Pt 1): 3314-3322 Haematology. 2008;140(5): 516-526 2018;57(5):515-520 [99] Wada DA et al. Specificity of IRF4 translocations for primary cutaneous anaplastic large cell lymphoma: A multicenter study of 204 skin biopsies. Modern Pathology. 2011;24(4):596-605 [107] Alderuccio JP et al. Frontline brentuximab vedotin in breast implant-associated anaplastic large- cell lymphoma. Clinical Case Reports. 2018;6(4):634-637 [108] Melchers RC et al. Evaluation of treatment results in multifocal primary cutaneous anaplastic large cell lymphoma: Report of the Dutch Cutaneous Lymphoma Group. The British Journal of Dermatology. 2018 Sep;179(3):724-731. DOI: 10.1111/ bjd.16501. Epub 2018 Jun 21 [108] Melchers RC et al. Evaluation of treatment results in multifocal primary cutaneous anaplastic large cell lymphoma: Report of the Dutch Cutaneous Lymphoma Group. The British Journal of Dermatology. 2018 Sep;179(3):724-731. DOI: 10.1111/ bjd.16501. Epub 2018 Jun 21 [100] Pham-Ledard A et al. IRF4 gene rearrangements define a subgroup of CD30-positive cutaneous T-cell lymphoma: A study of 54 cases. The Journal of Investigative Dermatology. 2010;130(3):816-825 [101] Feldman AL et al. Recurrent translocations involving the IRF4 oncogene locus in peripheral T-cell lymphomas. Leukemia. 2009;23(3):574-580 [109] Melchers RC et al. Recommendations for the optimal radiation dose in patients with [109] Melchers RC et al. Recommendations for the optimal radiation dose in patients with 18 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 primary cutaneous anaplastic large cell lymphoma: A report of the Dutch cutaneous lymphoma group. International Journal of Radiation Oncology, Biology, Physics. 2017;99(5):1279-1285 primary cutaneous anaplastic large cell lymphoma: A report of the Dutch cutaneous lymphoma group. International Journal of Radiation Oncology, Biology, Physics. 2017;99(5):1279-1285 [116] Arai H et al. ALK-negative anaplastic large cell lymphoma with loss of CD30 expression during treatment with brentuximab vedotin. Rinshō Ketsueki. 2016;57(5):634-637 primary cutaneous anaplastic large cell lymphoma: A report of the Dutch cutaneous lymphoma group. International Journal of Radiation Oncology, Biology, Physics. 2017;99(5):1279-1285 [116] Arai H et al. ALK-negative anaplastic large cell lymphoma with loss of CD30 expression during treatment with brentuximab vedotin. Rinshō Ketsueki. 2016;57(5):634-637 [117] Pro B et al. Brentuximab vedotin (SGN-35) in patients with relapsed or refractory systemic anaplastic large- cell lymphoma: Results of a phase II study. Journal of Clinical Oncology. 2012;30(18):2190-2196 [110] Christensen JG et al. Cytoreductive antitumor activity of PF-2341066, a novel inhibitor of anaplastic lymphoma kinase and c-Met, in experimental models of anaplastic large-cell lymphoma. Molecular Cancer Therapeutics. 2007;6(12 Pt 1): 3314-3322 [110] Christensen JG et al. Haematology. 2008;140(5): 516-526 Anaplastic lymphoma kinase expression as a marker of malignancy. Application to a case of anaplastic large cell lymphoma with huge granulomatous reaction. Haematologica. 2000;85(9):978-981 [131] Piccaluga PP et al. Anaplastic lymphoma kinase expression as a marker of malignancy. Application to a case of anaplastic large cell lymphoma with huge granulomatous reaction. Haematologica. 2000;85(9):978-981 [124] Moti N et al. Anaplastic large cell lymphoma-propagating cells are detectable by side population analysis and possess an expression profile reflective of a primitive origin. Oncogene. 2015 Apr 2;34(14):1843- 1852. DOI: 10.1038/onc.2014.112. Epub 2014 May 12 [132] Hu H et al. Bacterial biofilm infection detected in breast implant- associated anaplastic large-cell lymphoma. Plastic and Reconstructive Surgery. 2016;137(6):1659-1669 [132] Hu H et al. Bacterial biofilm infection detected in breast implant- associated anaplastic large-cell lymphoma. Plastic and Reconstructive Surgery. 2016;137(6):1659-1669 [125] Laurent C et al. Circulating t(2;5)- positive cells can be detected in cord blood of healthy newborns. Leukemia. 2012;26(1):188-190 [133] Knorr F et al. Blood cytokine concentrations in pediatric patients with anaplastic lymphoma kinase- positive anaplastic large cell lymphoma. Haematologica. 2018;103(3):477-485 [126] Kasprzycka M et al. Nucleophosmin/anaplastic lymphoma kinase (NPM/ALK) oncoprotein induces the T regulatory cell phenotype by activating STAT3. Proceedings of the National Academy of Sciences of the United States of America. 2006;103(26):9964-9969 [134] Hernandez L et al. Diversity of genomic breakpoints in TFG-ALK translocations in anaplastic large cell lymphomas: Identification of a new TFG-ALK(XL) chimeric gene with transforming activity. The American Journal of Pathology. 2002;160(4):1487-1494 [127] Schleussner N et al. The AP-1- BATF and -BATF3 module is essential for growth, survival and TH17/ ILC3 skewing of anaplastic large cell lymphoma. Leukemia. 2018 Sep;32(9):1994-2007. DOI: 10.1038/ s41375-018-0045-9. Epub 2018 Mar 28 [135] Hernandez L et al. TRK-fused gene (TFG) is a new partner of ALK in anaplastic large cell lymphoma producing two structurally different TFG-ALK translocations. Blood. 1999;94(9):3265-3268 [128] Pearson JD et al. NPM-ALK and the JunB transcription factor regulate the expression of cytotoxic molecules in ALK-positive, anaplastic large cell lymphoma. International Journal of Clinical and Experimental Pathology. 2011;4(2):124-133 [136] Lamant L et al. A new fusion gene TPM3-ALK in anaplastic large cell lymphoma created by a (1;2) (q25;p23) translocation. Blood. 1999;93(9):3088-3095 [129] Tripathi SK, Lahesmaa R. Transcriptional and epigenetic regulation of T-helper lineage specification. Immunological Reviews. 2014;261(1):62-83 [137] Wlodarska I et al. The cryptic inv(2)(p23q35) defines a new molecular genetic subtype of ALK-positive anaplastic large-cell lymphoma. Blood. 1998;92(8):2688-2695 [130] Lamant L et al. Haematology. 2008;140(5): 516-526 Cytoreductive antitumor activity of PF-2341066, a novel inhibitor of anaplastic lymphoma kinase and c-Met, in experimental models of anaplastic large-cell lymphoma. Molecular Cancer Therapeutics. 2007;6(12 Pt 1): 3314-3322 [118] Prince HM et al. Brentuximab vedotin or physician's choice in CD30-positive cutaneous T-cell lymphoma (ALCANZA): An international, open-label, randomised, phase 3, multicentre trial. Lancet. 2017;390(10094):555-566 [111] Mosse YP et al. Safety and activity of crizotinib for paediatric patients with refractory solid tumours or anaplastic large-cell lymphoma: A Children's Oncology Group phase 1 consortium study. The Lancet Oncology. 2013;14(6):472-480 [111] Mosse YP et al. Safety and activity of crizotinib for paediatric patients with refractory solid tumours or anaplastic large-cell lymphoma: A Children's Oncology Group phase 1 consortium study. The Lancet Oncology. 2013;14(6):472-480 [119] Pro B et al. Five-year results of brentuximab vedotin in patients with relapsed or refractory systemic anaplastic large cell lymphoma. Blood. 2017;130(25):2709-2717 [112] Mosse YP et al. Targeting ALK with Crizotinib in pediatric anaplastic large cell lymphoma and inflammatory myofibroblastic tumor: A Children's oncology group study. Journal of Clinical Oncology. 2017;35(28):3215-3221 [120] Laimer D et al. PDGFR blockade is a rational and effective therapy for NPM-ALK-driven lymphomas. Nature Medicine. 2012;18(11):1699-1704 [113] Gambacorti-Passerini C, Mussolin L, Brugieres L. Abrupt relapse of ALK-positive lymphoma after discontinuation of Crizotinib. The New England Journal of Medicine. 2016;374(1):95-96 [113] Gambacorti-Passerini C, Mussolin L, Brugieres L. Abrupt relapse of ALK-positive lymphoma after discontinuation of Crizotinib. The New England Journal of Medicine. 2016;374(1):95-96 [121] Chen J et al. Cytokine receptor signaling is required for the survival of ALK− anaplastic large cell lymphoma, even in the presence of JAK1/STAT3 mutations. Proceedings of the National Academy of Sciences of the United States of America. 2017;114(15):3975-3980 [114] Senter PD, Sievers EL. The discovery and development of brentuximab vedotin for use in relapsed Hodgkin lymphoma and systemic anaplastic large cell lymphoma. Nature Biotechnology. 2012;30(7):631-637 [114] Senter PD, Sievers EL. The discovery and development of brentuximab vedotin for use in relapsed Hodgkin lymphoma and systemic anaplastic large cell lymphoma. Nature Biotechnology. 2012;30(7):631-637 [122] Chiarle R et al. The anaplastic lymphoma kinase is an effective oncoantigen for lymphoma vaccination. Nature Medicine. 2008;14(6): 676-680 [123] Pulford K et al. Detection of anaplastic lymphoma kinase (ALK) and nucleolar protein nucleophosmin (NPM)-ALK proteins in normal and neoplastic cells with the 19 Peripheral T-cell Lymphomas in five cases. Haematologica. 2010;95(3):449-455 monoclonal antibody ALK1. Blood. 1997;89(4):1394-1404 monoclonal antibody ALK1. Blood. 1997;89(4):1394-1404 [131] Piccaluga PP et al. [144] van, der Krogt JA et al. Anaplastic lymphoma kinase-positive anaplastic large cell lymphoma with the variant RNF213-, ATIC- and TPM3-ALK fusions is characterized by copy number gain of the rearranged ALK gene. Haematologica. 2017;102(9):1605-1616 Haematology. 2008;140(5): 516-526 Cutaneous presentation of ALK-positive anaplastic large cell lymphoma following insect bites: Evidence for an association [138] Ma Z et al. Inv(2)(p23q35) in anaplastic large-cell lymphoma induces 20 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 Anaplastic Large Cell Lymphoma DOI: http://dx.doi.org/10.5772/intechopen.81382 constitutive anaplastic lymphoma kinase (ALK) tyrosine kinase activation by fusion to ATIC, an enzyme involved in purine nucleotide biosynthesis. Blood. 2000;95(6):2144-2149 constitutive anaplastic lymphoma kinase (ALK) tyrosine kinase activation by fusion to ATIC, an enzyme involved in purine nucleotide biosynthesis. Blood. 2000;95(6):2144-2149 [139] Tort F et al. Molecular characterization of a new ALK translocation involving moesin (MSN-ALK) in anaplastic large cell lymphoma. Laboratory Investigation. 2001;81(3):419-426 [140] Cools J et al. Identification of novel fusion partners of ALK, the anaplastic lymphoma kinase, in anaplastic large-cell lymphoma and inflammatory myofibroblastic tumor. Genes, Chromosomes & Cancer. 2002;34(4):354-362 [140] Cools J et al. Identification of novel fusion partners of ALK, the anaplastic lymphoma kinase, in anaplastic large-cell lymphoma and inflammatory myofibroblastic tumor. Genes, Chromosomes & Cancer. 2002;34(4):354-362 [141] Touriol C et al. Further demonstration of the diversity of chromosomal changes involving 2p23 in ALK-positive lymphoma: 2 cases expressing ALK kinase fused to CLTCL (clathrin chain polypeptide-like). Blood. 2000;95(10):3204-3207 [142] Lamant L et al. Non-muscle myosin heavy chain (MYH9): A new partner fused to ALK in anaplastic large cell lymphoma. Genes, Chromosomes & Cancer. 2003;37(4):427-432 [143] Lawrence B et al. TPM3-ALK and TPM4-ALK oncogenes in inflammatory myofibroblastic tumors. The American Journal of Pathology. 2000;157(2):377-384 [144] van, der Krogt JA et al. Anaplastic lymphoma kinase-positive anaplastic large cell lymphoma with the variant RNF213-, ATIC- and TPM3-ALK fusions is characterized by copy number gain of the rearranged ALK gene. Haematologica. 2017;102(9):1605-1616 21
https://openalex.org/W4286662478	http://repository.ubaya.ac.id/38781/7/Siti%20Rahayu_Faktor-Faktor%20Yang%20Mempengaruhi%20Perilaku%20Word%20Of%20Mouth%20Intention%20Pelanggan.pdf	Indonesian	null	Faktor-Faktor Yang Mempengaruhi Perilaku Word Of Mouth Intention Pelanggan PadaTempat Makan Masakan Khas Jawa Di Surabaya	Jurnal Manajerial	2,021	cc-by	8,048	ABSTRACT B k d Background – The restaurant industry is a very competitive industry to attract and retain consumers. Restaurant owners need to understand how consumers want, need and perceptions. Word of mouth (WOM) has an important role in any effective marketing strategy for the restaurant industry. Diterima : 03 September 2020 Direview : 13 Oktober 2020 Direvisi : 20 Oktober 2020 Disetujui : 24 Januari 2021 Diterima : 03 September 2020 Direview : 13 Oktober 2020 Direvisi : 20 Oktober 2020 Disetujui : 24 Januari 2021 y Purpose – This study aims to determine the influence of food quality, the quality of personal interaction, the quality of the physical environment, the perceived value, and the quality of the relationship (satisfaction, trust, and commitment) on the behavior of mouth-to-mouth intention of customers where to eat Javanese specialties in Surabaya. Disetujui : 24 Januari 2021 Purpose – This study aims to determine the influence of food quality, the quality of personal interaction, the quality of the physical environment, the perceived value, and the quality of the relationship (satisfaction, trust, and commitment) on the behavior of mouth-to-mouth intention of customers where to eat Javanese specialties in Surabaya. Disetujui : 24 Januari 2021 Design / Methodology / Approach – This study uses a sample of 200 respondents. The data processing method used in this research is Structural Equation Modeling (SEM) with LISREL 8.8 for Window software. The target population in this study is all people who have eaten and drank in one of the 10 places to eat Javanese specialties in Surabaya. The type of sampling in this study is accidental sampling or convenience sampling. Result and Discussion – Food quality, physical environment, personal interactional quality, perceived value have a significant positive effect on satisfaction. Satisfaction has a significant positive effect on Word of Mouth and trust. Satisfaction does not have an impact on the customer commitment to eating Javanese specialties in Surabaya. Trust has a significant positive effect on the customers' commitment to eating Javanese specialties in Surabaya. Trust and commitment have no influence on the word of mouth intention of customers to eat Javanese specialties in Surabaya. Conclusion – The results of the study found that customers of Javanese cuisine in Surabaya rated satisfaction as not the main assessment in forming a commitment to return to enjoy the food served. P-ISSN : 2354-8592 E-ISSN : 2621-5055 P-ISSN : 2354-8592 E-ISSN : 2621-5055 P-ISSN : 2354-8592 E-ISSN : 2621-5055 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 Siti Rahayu2 Departemen Manajemen Layanan dan Pariwisata, Universitas Surabaya, Surabaya, Indonesia, ernajani@staff.ubaya.ac.id Siti Rahayu2 Departemen Manajemen Layanan dan Pariwisata, Universitas Surabaya, Surabaya, Indonesia, ernajani@staff.ubaya.ac.id Erna Andajani3 Departemen Manajemen Layanan dan Pariwisata, Universitas Surabaya, Surabaya, Indonesia, ernajani@staff.ubaya.ac.id y Departemen Manajemen Layanan dan Pariwisata, Universitas Surabaya, Surabaya, Indonesia, ernajani@staff.ubaya.ac.id y Departemen Manajemen Layanan dan Pariwisata, Universitas Surabaya, S ernajani@staff.ubaya.ac.id Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia pyright© Creative Commons Attribution 4.0 International License gram Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Copyright© Creative Commons Attribution 4.0 International License FAKTOR-FAKTOR YANG MEMPENGARUHI PERILAKU WORD OF MOUTH INTENTION PELANGGAN PADA TEMPAT MAKAN MASAKAN KHAS JAWA DI SURABAYA Nurfathan Hasbiy Purwanto1 Nurfathan Hasbiy Purwanto Departemen Manajemen Layanan dan Pariwisata, Universitas Surabaya, Surabaya, Indonesia, ernajani@staff.ubaya.ac.id ABSTRAK Latar Belakang - Industri restoran merupakan industri yang sangat kompetitif, untuk menarik dan mempertahankan konsumen maka pemilik restoran perlu memahami bagaimana keinginan, kebutuhan dan persepsi konsumen. Word of Mouth (WOM) memiliki peran penting dalam setiap strategi pemasaran yang efektif untuk industri restoran. Tujuan - Penelitian ini bertujuan untuk mengetahui pengaruh variabel kualitas makanan, kualitas interaksi personal, kualitas lingkungan fisik, nilai yang dirasakan, dan kualitas hubungan (kepuasan, kepercayaan, dan komitmen) terhadap perilaku mulut. Niat mulut-ke-mulut pelanggan tempat makan makanan khas Jawa di Surabaya.. Desain / Metodologi / Pendekatan - Penelitian ini menggunakan sampel 200 responden. Metode pengolahan data yang digunakan dalam penelitian ini adalah Structural Equation Modeling (SEM) dengan software LISREL 8.8 for Window. Populasi sasaran dalam penelitian ini adalah seluruh masyarakat yang pernah makan dan minum di salah satu dari 10 tempat makan makanan khas Jawa yang ada di Surabaya. Jenis pengambilan sampel dalam penelitian ini adalah accidental sampling atau convenience sampling. Hasil dan Pembahasan - Kualitas makanan, lingkungan fisik, kualitas interaksi pribadi, nilai yang dirasakan berpengaruh positif signifikan terhadap kepuasan. Kepuasan berpengaruh positif signifikan terhadap Word of Mouth dan kepercayaan. Kepuasan tidak berdampak pada komitmen pelanggan untuk menyantap makanan khas Jawa di Surabaya. Kepercayaan berpengaruh positif signifikan terhadap komitmen konsumen terhadap makanan khas Jawa di Surabaya. Kepercayaan dan komitmen tidak berpengaruh terhadap niat mulut ke mulut pelanggan untuk menyantap makanan khas Jawa di Surabaya. Hasil dan Pembahasan - Kualitas makanan, lingkungan fisik, kualitas interaksi pribadi, nilai yang dirasakan berpengaruh positif signifikan terhadap kepuasan. Kepuasan berpengaruh positif signifikan terhadap Word of Mouth dan kepercayaan. Kepuasan tidak berdampak pada komitmen pelanggan untuk menyantap makanan khas Jawa di Surabaya. Kepercayaan berpengaruh positif signifikan terhadap komitmen konsumen terhadap makanan khas Jawa di Surabaya. Kepercayaan dan komitmen tidak berpengaruh terhadap niat mulut ke mulut pelanggan untuk menyantap makanan khas Jawa di Surabaya. Kesimpulan - Hasil penelitian menemukan bahwa pelanggan kuliner Jawa di Surabaya menilai kepuasan bukan penilaian utama dalam membentuk komitmen untuk kembali menikmati makanan yang disajikan. Pelaku usaha perlu mencari dan mengembangkan hal-hal baru dari aspek lain seperti mempercantik interior dan eksterior bangunan atau memperbanyak variasi makanan serta meningkatkan pelayanan dan hal-hal kecil seperti kebersihan. Dalam bisnis restoran perlu dikembangkan nilai lebih yang tidak dimiliki pesaing. Hal ini menyulitkan pelanggan untuk mencari alternatif lain. ABSTRAK Batasan Penelitian - Pada penelitian ini ditemukan bahwa variabel Commitment saja tidak cukup dalam mengukur hubungan antara Commitment dan Word of mouth intention, peneliti perlu mengukur variabel Commitment menjadi tiga bagian yaitu Affective commitment, Calculative commitment dan Normative commitment. Dengan mengukur dimensi commitment ini penelitian selanjutnya diharapkan bisa melihat bagaimana hasil dari pengaruh dimensi commitment terhadap Word of mouth intention di industri layanan. Kata Kunci : Kepuasan, Kepercayaan, Komitmen, Intensitas, Word of Mouth. ABSTRACT B k d Business actors need to find and develop new things from other aspects such as beautifying interior and exterior buildings or increasing the variety of food and improving services and small things such as cleanliness. In the restaurant business, it is necessary to develop more value which is not owned by competitors. This makes it difficult for customers to find other alternatives. Researh Limitations – In this study, it was found that the Commitment variable alone was not sufficient in measuring the relationship between commitment and word of mouth intention, researchers needed to measure the commitment variable into three parts, namely affective commitment, calculative commitment and normative commitment. By measuring the commitment dimension, further research is expected to see how the results of the impact of the commitment dimension on word of mouth intention in the service industry. Keyword : Satisfaction, Trust, Commitment, Word of Mouth, Intentions 37 P-ISSN : 2354-8592 E-ISSN : 2621-5055 P-ISSN : 2354-8592 E-ISSN : 2621-5055 P-ISSN : 2354-8592 E-ISSN : 2621-5055 g t© C e t e Co o s tt but o 0 te t o ce se am Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Copy g t© C e t e Co o s tt but o 0 te t o ce se Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timu pyright© Creative Commons Attribution 4.0 International License Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia PENDAHULUAN pemasaran yang efektik untuk industri restoran. Restoran dengan anggaran promosi terbatas sangat bergantung pada media promosi word of mouth, oleh karena itu, WOM dapat dianggap sebagai sarana penting untuk menarik minat serta mempertahaknan konsumen (Wirtz dan Chew, 2002). Baloglu dan Mccleary (1999) Industri restoran merupakan industri yang sangat kompetitif, untuk menarik minat dan mempertahankan konsumen, pemilik restoran perlu memahami bagaimana keinginan, kebutuhan serta persepsi konsumen. word of mouth (WOM) memiliki peranan penting dari setiap strategi 38 P-ISSN : 2354-8592 E-ISSN : 2621-5055 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 dalam Jalilvand et al., (2017) berpendapat word of mouth memiliki dampak paling positif terhadap citra yang dirasakan konsumen dari pada tiga lainnya yaitu, saran dari professional, iklan, buku, film, dan berita. Selain itu terdapat penelitian yang menunujukkan bahwa 76% dari semua keputusan pembelian dipengaruhi oleh word of mouth dan diperkirakan terdapat 3,4 Miliar percakapan word of mouth setiap harinya dan 2,3 Miliar diantaranya adalah percakapan yang membahas sebuah merek. merupakan konsep yang terpenting dalam menciptakan suatu produk. Lemke et al., 2011 dalam Choi dan Kim (2012) menyatakan bahwa Personal interaction quality merupakan hubungan intreraksi interaktif antara pelanggan dengan pemberi layanan, dimana karyawan dan staff sebagai perwakilan perusahaan tersebut, yang akan berkomunikasi secara langsung dengan pelanggan mulai dari proses pemberian informasi sampai proses layanan diberikan. Penelitian yang dilakukan oleh Nielsen melalui studinya menjelaskan bahwa 59% pelanggan di Indonesia lebih suka memilih hindangan makanan khas lokal. Serta pada wilayah Asia tenggara ditemukan bahwa tindakan dapat melebihi kepercayaan pada 19 format iklan yang telah disurvei, dimana responden yang mempercayai rekomendasi dari orang yang dikenal mengaku bahwa keputusan atau tindakan pembelian didasari oleh pendapat atau rekomendasi dari orang- orang yang dikenal dengan nilai 91% tindakan (take action) dan 88% kepercayaan (trust), sementara itu iklan-iklan di TV dan situs bermerek masing-masing memiliki persentase sebesar 81%. Menurut Mowen dan Minor (2002) berpendapat Aspek fisik dan tempat dari lingkungan fisik dapat mempengaruhi perilaku dari pelanggan. seperti warna, suara, penerangan, dan susunan tatanan ruang, orang atau benda. Lingkungan fisik dapat mempengaruhi persepsi konsumen dengan cara mekanisme melalui inrdra penglihatan, pendengaran, penciuman, dan indra sentuhan atau peraba. Menurut Zeithaml (1998) dalam Ryu et al., (2012), nilai yang dirasakan oleh pelanggan dapat diartikan sebagai hasil perbandingan antara manfaat yang dirasakan secara dengan pengorbanan yang dirasakan atau biaya yang dibayar oleh pelanggan. PENDAHULUAN Parasuraman et al., (1995) dalam Ryu et al., (2012) Berpendapat pelanggan dapat mengevaluasi sebuah produk atau layanan yang telah dibeli dapat. menyediakan nilai dan harapan yang sesuai, sehingga hasil dari evaluasi nilai yang ght© Creative Commons Attribution 4.0 International License am Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Hipotesis Law et al., (2004) dalam Namin (2016) berpendapat bahwa “… food quality and the various food offered are an important elemet og customer satisfaction”. Dimana pada hal ini menunjukkan food quality serta berbagai macam makanan yang ditawarkan merupakan unsur penting dari kepuasan pelanggan. Selain itu Yuksel (2002) dalam Jalivand et al., (2017) menegaskan jika pelanggan cenderung menghabiskan uang dan waktu di lingkungan layanan yang dapat memicu perasaan senang. Atribut restoran seperti kualitas makanan dan tema merupakan hal yang penting dalam mendapatkan kepuasan pelanggan. Pada penelitian ini dapat ditarik pernyataan sebagai berikut. Morgan dan Hunt (1994) dalam Jalivand et al., (2017) berargumen kepercayaan didefinisikan sebagai satu orang percaya bahwa orang lain akan memenuhi kebutuhannya. Dalam hal layanan, kepercayaan merupakan pendapat yang dipegang oleh pelanggan bahwa penyedia layanan akan menyediakan layanan yang dapat memenuhi kebutuhan pelanggan. TINJAUAN PUSTAKA Menurut Namkung dan Jang (2007) bahwa kualitas makanan merupakan syarat yang diperlukan untuk memenuhi kebutuhan dan harapan pelanggan. Kualitas makanan 39 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 P-ISSN : 2354-8592 E-ISSN : 2621-5055 diberikan oleh pelanggan bersifat subjektif dan pribadi. pihak untuk membangun hubungan kerjasama yang lebih baik dan meyakinkan. Huntley (2006) dalam jalivand et al., (2017), menyatakan bahwa Relationship Quality sebagai tingkat kepuasan pembeli dari waktu ke waktu dengan kemitraan keseluruhan, diwujudkan dalam kualitas produk, kualitas layanan dan nilai uang. Menurut Kotler and Keller (2012) word of mouth marketing adalah kegiatan pemasaran melalui perantara orang ke orang baik secara lisan, tulisan, maupun alat komunikasi elektronik berhubungan dengan pengalaman pembelian jasa atau dalam penggunaannya. Deng et al., (2009) dalam Jalivand et al., (2017) berpendapat Karena kepuasan adalah keadaan emosional, reaksi pasca-pembelian dapat berupa kemarahan, ketidakpuasan, kejengkelan, netralitas, kegembiraan, atau kesenangan. Selain itu kupuasan pelanggan merupakan bentuk evaluasi pelanggan terhadap produk dan layanan dalam hal apakah sudah sesuai dengan persyaratan pelanggan tersebut. ght© Creative Commons Attribution 4.0 International License am Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia H1: Food quality berpengaruh positif terhadap Satisfaction pelanggan restoran makanan khas jawa di Surabaya Menurut Moorman et al., (1992) mendefinisikan komitmen sebagai hasrat abadi untuk mempertahankan hubungan yang berharga, dari definisi ini dapat diketahui jika komitmen tidak dapat melibatkan hanya satu pihak saja. Komitmen akan memotivasi dari kedua belah makanan khas jawa di Surabaya 40 Gabarino dan Johnson (1999) dalam Jalivand et al., (2017) berpendapat bahwa kepuasan pelanggan dalam industri restoran dapat dipengaruhi oleh kualitas dari lingkungan fisik restoran. Selain itu Chang P-ISSN : 2354-8592 E-ISSN : 2621-5055 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 P-ISSN : 2354-8592 E-ISSN : 2621-5055 (2000) dalam Ryu dan Han (2009) mengemukakan bahwa kualitas lingkungan fisik yang dipersepsikan merupakan indikator langsung dari kepuasan pelanggan. Wakefield dan Blodgett (1996) dalam Ryu dan Han (2009) mengungkapkan bahwa persepsi akan kualitas lingkungan fisik secara signifikan mempengaruhi kepuasan pelanggan. Pada penelitian ini dapat ditarik pernyataan sebagai berikut. dengan perusahaan maka semakin tinggi kepuasan konsumen terhadap produk atau layanan yang dimiliki perusahaan. Selain itu Jamel dan Naser (2002) dalam Choi dan Kim (2012) menyatakan kualitas relasional, mengacu pada kualitas interaksi antara karyawan dengan pelanggan, memiliki pengaruh langsung pada kepuasan pelanggan. Choi dan Kim (2012) memberi saran untuk meningkatkan relationship quality perlu upaya untuk menjamin personal interaction quality yang baik dengan pelanggan. Pada penelitian ini dapat ditarik pernyataan sebagai berikut. g am Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia ght© Creative Commons Attribution 4.0 International License Surabaya Patterson dan Spreng (1997) dalam Ryu et al., (2012) menyatakan nilai yang dirasakan pelanggan memiliki hubungan positif dan langsung dengan kepuasan pelanggan. McDougall dan Levesque (2000) dalam Ryu et al., (2012) berpendapat bahwa persepsi terhadap kualitas layanan dan nilai yang dirasakan pelanggan merupakan dua indikator kepuasan pelanggan yang paling menonjol di empat industri jasa yaitu, restoran, layanan otomatis, penata rambut, dan layanan gigi. Pada penelitian ini dapat ditarik pernyataan sebagai berikut. Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Surabaya Menurut Getty dan Thompson (1994) dalam Jalivand et al., (2017) menunjukkan bahwa tingkat kepuasan yang tinggi dapat meningkatkan niat pelanggan untuk membeli kembali serta niat untuk merekomendasikan produk dan layana. Pelanggan yang merasa puas serta dapat menyebarkan WOM secara positif maka secara tidak langsung pelanggan tersebut membantu perusahaan dalam menyediakan iklan secara gratis yang di sebut dengan word of mouth (Jalivand dan Samiei, 2012 dalam Jalivand et al., 2017). Pada penelitian ini dapat ditarik pernyataan sebagai berikut. H3: Satisfaction berpengaruh positif terhadap word of mouth intention pelanggan restoran makanan khas jawa di Surabaya Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 P-ISSN : 2354-8592 E-ISSN : 2621-5055 P-ISSN : 2354-8592 E-ISSN : 2621-5055 Walsh et al., (2010) dalam Jalivand et al., (2017), dalam pasar ritel secara online dan offline kepuasan memiliki dampak positif terhadap kepercayaan dan komitmen konsumen. Chien-Lung Hsu et al., (2010) dalam Purnasari dan Yuliando (2015) berpendapat “Customers satisfaction with enterprises has a positive influence on customers trust”, artinya kepuasan pelanggan dengan perusahaan memiliki pengaruh positif pada kepercayaan pelanggan. Pada penelitian ini dapat ditarik pernyataan sebagai berikut. H6: Satisfaction berpengaruh positif terhadap Trust pelanggan restoran makanan khas jawa di Surabaya H7: Satisfaction berpengaruh positif terhadap Commitment pelanggan restoran makanan khas jawa di Surabaya Walsh et al., (2010) dalam Jalivand et al., (2017), dalam pasar ritel secara online dan offline kepuasan memiliki dampak positif terhadap kepercayaan dan komitmen konsumen. Chien-Lung Hsu et al., (2010) dalam Purnasari dan Yuliando (2015) berpendapat “Customers satisfaction with enterprises has a positive influence on customers trust”, artinya kepuasan pelanggan dengan perusahaan memiliki pengaruh positif pada kepercayaan pelanggan. Pada penelitian ini dapat ditarik pernyataan sebagai berikut. H8: Trust berpengaruh positif terhadap commitment pelanggan restoran makanan khas jawa di Surabaya Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia H5: Perceived value berpengaruh positif terhadap satisfaction pelanggan restoran makanan khas jawa di Surabaya H5: Perceived value berpengaruh positif terhadap satisfaction pelanggan restoran makanan khas jawa di Surabaya Fang et al., (2011) menyatakan bahwa semakin baik kualitas interaksi personal 41 41 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 H8: Trust berpengaruh positif terhadap commitment pelanggan restoran makanan khas jawa di Surabaya Menurut Baloglu (2002) dalam Jalivand et al., (2017) menyatakan bahwa pelanggan setia memiliki ikatan yang kuat serta komitmen yang kuat. Pelanggan dapat menciptakan WOM positif serta persepsi terhadap harga dapat berkurang. Geyskens et al., (1996), singh dan Sirdeshmukh (2000) dalam Ribbink et al., (2004) menjelaskan bahwa kepercayaan sangat memiliki pengaruh terhadap hubungan antara restoran dengan pelanggannya dan juga dapat mempengaruhi perilaku dari pelanggannya di waktu yang akan datang. Moorman et al., (1992) menyatakan bahwa sebuah komitmen membuat keberhasilan dalam berhubungan, serta hubungan antara mitra bisa saling menguntungkan satu dengan yang lain dan puas. Komitmen yang muncul dari pelanggan membuat mereka konsisten dalam mengunjungi restoran serta loyalitas pelanggan dapat memberikan rekomendasi kepada orang lain. Pada penelitian ini dapat ditarik pernyataan sebagai berikut. H6: Satisfaction berpengaruh positif terhadap Trust pelanggan restoran makanan khas jawa di Surabaya H7: Satisfaction berpengaruh positif terhadap Commitment pelanggan restoran makanan khas jawa di Surabaya Kepercayaan lebih didasarkan pada proses sosial yang terjadi dalam hubungan bisnis, sementara komitmen lebih berorientasi pada rencana berhubungan satu sama lain di masa depan. Trust dan commitment merupakan hasil dari pengembagnan hubungan kolaboratif antara dua perusahaan. Mengembangkan sebuah hubungan bisnis yang dapat dipercaya membutuhkan proses jangka panjang, yang berlangsung tahap demi tahap, risiko, dan ketidak menentuan berkurang, sedangkan commitment dan trust meningkat (Zineldin et al., 1997). Pada penelitian ini dapat ditarik pernyataan sebagai berikut. H9: Trust berpengaruh positif terhadap word of mouth intention pelanggan restoran makanan khas jawa di Surabaya H10: Commitment berpengaruh positif terhadap word of mouth intention pelanggan restoran makanan khas jawa di Surabaya. 42 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 P-ISSN : 2354-8592 E-ISSN : 2621-5055 Sumber: Jalivand et al., (2017). Gambar 1. Model Penelitian Sumber: Jalivand et al., (2017). Sumber: Jalivand et al., (2017). Gambar 1. Model Penelitian Gambar 1. Model Penelitian Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia ght© Creative Commons Attribution 4.0 International License am Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur In METODOLOGI PENELITIAN kuesioner tersebut. Besaran jumlah sampel pada penelitian ditujukan berdasarkan kecukupan teknik analisis yang digunakan merupakan metode maximum likehood. Menurut Wijanto (2008), ukuran sampel yang diperlukan untuk metode estimasi maximum likehood adalah 5 dan maksimal 10 responden untuk jumlah variabel teramati (indikator) yang ada pada model. Aras dan skala yang digunakan pada penelitian ini merupakan aras interval, yaitu aras pengukuran yang memiliki jarak yang sama dan selisih yag sama pada skala pengukuran (Trimudi dan Harini, 2008). aras interval disusun berdasarkan numerical scale yaitu 5 point likert scale yang digunakan untuk memberikan penilaian kepada masing-masing pernyataan. Semakin besar nilai yang mendekati angka 5 maka menunjukkan tanggapan positif atau setuju terhadap pernyataan dalam kuesioner, namun apabila semakin kecil nilai yang mendekati angka 1 maka menunjukkan tanggapan yang negatif atau tidak setuju dengan pernyataan yang ada dalam Dalam penelitian ini jumlah variabel teramati (indikator) adalah 31. Maka jumlah sampel yang digunakan minimal 5 dikalikan 31 indikator yaitu sebanyak 155 responden dan maksimal adalah 10 dikalikan 31 yaitu sebanyak 310 responden. Namun menurut Merdsker et al., (1994) dalam Jalivand et al., (2017) merekomendasikan agar besaran 43 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 P-ISSN : 2354-8592 E-ISSN : 2621-5055 P-ISSN : 2354-8592 E-ISSN : 2621-5055 (8) Bebek Palupi (Jl. Raya Rungkut Asri Tengah), (9) Rawon Setan (Jl. Embong Malang), (10) Rujak Cingur Ahmad Jais, dengan karakteristik sebagai berikut: sampel tidak lebih dari 200 responden, maka dari itu penelitian ini menggunakan sampel sebesar 200 responden. (8) Bebek Palupi (Jl. Raya Rungkut Asri Tengah), (9) Rawon Setan (Jl. Embong Malang), (10) Rujak Cingur Ahmad Jais, dengan karakteristik sebagai berikut: (8) Bebek Palupi (Jl. Raya Rungkut Asri Tengah), (9) Rawon Setan (Jl. Embong Malang), (10) Rujak Cingur Ahmad Jais, dengan karakteristik sebagai berikut: Metode pengolahan data yang digunakan dalam penelitian ini adalah Structural Equation Modeling (SEM) dengan software LISREL 8.8 for Windows. Syarat jumlah good fit index menurut Hair et al., (2010) yang baik kurang lebih menggunakan paling tidak 3-4 indeks dari indeks absolute dan incremental agar dapat dilakukan pengujian lanjutan model penelitian. Terdapat beberapa indeks kesesuaian (absolute dan incremental) untuk menguju model diterima atau ditolak, Uji Validitas merupakan sejauh mana sebuah pengukuran/ sekumpulan pengukuran dapat mengukur secara akurat konsep yang diteliti (Hair et al., 2010). Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 P-ISSN : 2354-8592 E-ISSN : 2621-5055 P-ISSN : 2354-8592 E-ISSN : 2621-5055 P-ISSN : 2354-8592 E-ISSN : 2621-5055 Berdasarkan hasil pada tabel 1 dapat diketahui hasil uji kecocokan struktural model sudah menunjukkan kriteria yang baik dimana ukuran nilai kecocokan telah memenuhi kriteria yang ditetapkan. Dimana pada hasil tersebut nilai CMIN/DF (1.485), RMSEA (0.049), TLI/NNFI (0.96), CFI (0.96) dimana hasil tersebut termasuk kategori good Fit, dan nilai GFI (0.84) termasuk dalam kategori majinal fit (mendekati), dengan demikian untuk malakukan uji hipotesis dapat menggunakan hasil uji kecocokan model struktural yang sudah ada. Setelah proses pengujian yang telah dilaksanakan maka uji hipotesis dapat dilakukan, dimana pada uji hipotesis ini menggunakan software Lisrel 8.8, sesuai dengan teori acuan dimana besaran nilai T-value memiliki nilai mutlak 1,96, sehinggan batasan nilai hipotesis dapat diterima atau ditolak apabila nilai T-value lebih besar dari 1,96. populasi yang akan diambil tidak bisa teridentifikasi dengan jelas berapa jumlah keseluruahnnya. Jenis pengambilan sampel dalan penelitian ini merupakan accidental sampling atau convencience sampling yakni teknik penentuan sampel berdasarkan kebetulan, yaitu siapa saja yang secara kebetulan bertemu dengan peneliti dapat digunakan sebagai sampel, jika orang yang ditemui tersebut cocok sebagai sumber data (Sugiyono, 2013). Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indone Copyright© Creative Commons Attribution 4 0 International License METODOLOGI PENELITIAN Cara yang digunakan untuk mengukur validitas dan realibilitas adalah dengan menggunkan bantuan program computer SPSS 25 for windows untuk mengevaluasi apakah instrument penelitian kuesioner sudah tepat digunakan untuk mengukur variabel penelitian. 1. Berdomisili di Surabaya. 2. Berusia minimal 17 tahun. 3. Pernah berkunjung, makan dan minum di salah satu dari 10 tempat makan masakan khas Jawa di Surabaya, sebanyak 2 kali dalam 6 bulan terakhir. 4. Apabila responden pernah berkunjung ke semua objek yang ditentukan maka responden wajib memilih salah satu objek yang terakhir kali dikunjungi. 5. Memiliki pendidikan minimal SMA/SMK/Sederajat dengan tujuan responden mampu memahami dan menjawab pernyataan yang ada pada kuesioner secara objektif mengenai Food quality, physical environment quality, personal interaction quality dan perceived value, Satisfaction, Trust, Commitment dan Word of Mouth Intention. 6. Pernah memberikan rekomendasi kepada orang lain mengenai salah satu dari 10 tempat makan masakan khas Jawa di Surabaya. Target populasi dalam penelitian ini adalah semua orang yang pernah makan dan minum di salah satu 10 tempat makan masakan khas Jawa di Surabaya yaitu: (1) Lontong Balap Garuda pak Gendut, (2) Sate Klopo Ondomohen bu Asih, (3) Soto Ayam Lamongan Cak Har, (4) Soto Ambengan Pak Sadi, (5) Tahu Telur Pak Jayen, (6) Pecel Ponorogo bu Yatin, (7) Pecel Madiun Bu Kus, Teknik pengambilan sampling yang digunakan pada penelitian ini adalah non- probability sampling, merupakan teknik pengambilan sampel dimana setiap anggota dari populasi tidak memiliki peluang yang sama untuk menjadi sampel atau subjek penelitian (sekaran and bougie, 2010). Serta 44 g am Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia HASIL DAN PEMBAHASAN Model struktural yang akan diteliti yaitu mengenai pengaruh dari variabel food quality, physical environment quality, personal interaction quality, perceived value, satisfaction, trust, commitment, terhadap word of mouth intention dari pelanggan restoran makanan khas Jawa yang ada di Surabaya. Pada penelitian ini model struktural dianalisis melalui aplikasi program LISREL 8.8 for Windows. Tabel 1 Uji Goodness of Fit Struktural Model Tabel 1 Uji Goodness of Fit Struktural Model Uji Kecocokan Kriteria Hasil Ket. CMIN/DF RMSEA GFI TLI/ NNFI CFI CMIN/DF≤2 RMSEA≤ 0.08 ≥ 0.90 ≥ 0.90 ≥ 0.90 1.485 0.049 0.84 0.96 0.96 Good Fit Good Fit Marjinal Fit Good Fit Good Fit Sumber : Data Diolah, 2020 45 Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 P-ISSN : 2354-8592 E-ISSN : 2621-5055 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 P-ISSN : 2354-859 http://dx.doi.org/10.30587/manajerial.v8i01.1858 E-ISSN : 2621-505 Tabel 2 Hasil Uji Hipotesis Hipotesis T-Value Std.Est Ket H1 Food Quality → Satisfaction 3.38 0.29 Terdukung H2 Phsical Environment Quality → Satisfaction 3.85 0.33 Terdukung H3 Satisfaction → Word of Mouth 3.25 0.53 Terdukung H4 Personal Interction Quality → Satisfaction 3.97 0.32 Terdukung H5 Percieved Value → Satisfaction 3.06 0.22 Terdukung H6 Satisfaction → Trust 6.48 0.78 Terdukung H7 Satisfaction → Commitment 0.70 0.12 Tidak Terdukung H8 Tust → Cmmitment 3.52 0.68 Terdukung H9 Trust → Word of Mouth 0.93 0.93 Tidak Terdukung H10 Commitment → Word of Mouth 0.80 0.12 Tidak Terdukung Sumber : Data Diolah, 2020 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 P-ISSN : 2354-8592 http://dx.doi.org/10.30587/manajerial.v8i01.1858 E-ISSN : 2621-5055 Tabel 2 Hasil Uji Hipotesis Hipotesis T-Value Std.Est Ket H1 Food Quality → Satisfaction 3.38 0.29 Terdukung H2 Phsical Environment Quality → Satisfaction 3.85 0.33 Terdukung H3 Satisfaction → Word of Mouth 3.25 0.53 Terdukung H4 Personal Interction Quality → Satisfaction 3.97 0.32 Terdukung H5 Percieved Value → Satisfaction 3.06 0.22 Terdukung H6 Satisfaction → Trust 6.48 0.78 Terdukung H7 Satisfaction → Commitment 0.70 0.12 Tidak Terdukung H8 Tust → Cmmitment 3.52 0.68 Terdukung H9 Trust → Word of Mouth 0.93 0.93 Tidak Terdukung H10 Commitment → Word of Mouth 0.80 0.12 Tidak Terdukung Sumber : Data Diolah, 2020 Gambar 2. Value pada tiap hipotesa penelitian Berdasarkan hasil dari penelitian ini ditemukan beberapa kesamaan hasil dari value memiliki pengaruh terhadap variabel satisfaction. Copyright© Creative Commons Attribution 4.0 International License ive Commons Attribution 4.0 International License HASIL DAN PEMBAHASAN Tabel 2 Hasil Uji Hipotesis Hipotesis T-Value Std.Est Ket H1 Food Quality → Satisfaction 3.38 0.29 Terdukung H2 Phsical Environment Quality → Satisfaction 3.85 0.33 Terdukung H3 Satisfaction → Word of Mouth 3.25 0.53 Terdukung H4 Personal Interction Quality → Satisfaction 3.97 0.32 Terdukung H5 Percieved Value → Satisfaction 3.06 0.22 Terdukung H6 Satisfaction → Trust 6.48 0.78 Terdukung H7 Satisfaction → Commitment 0.70 0.12 Tidak Terdukung H8 Tust → Cmmitment 3.52 0.68 Terdukung H9 Trust → Word of Mouth 0.93 0.93 Tidak Terdukung H10 Commitment → Word of Mouth 0.80 0.12 Tidak Terdukung Sumber : Data Diolah, 2020 Gambar 2. Value pada tiap hipotesa penelitian Gambar 2. Value pada tiap hipotesa penelitian Gambar 2. Value pada tiap hipotesa penelitian Gambar 2. Value pada tiap hipotesa penelitian value memiliki pengaruh terhadap variabel satisfaction. Berdasarkan hasil dari penelitian ini ditemukan beberapa kesamaan hasil dari penelitian sebelumnya yang dilakukan oleh Jalivand et al., (2017), yaitu variabel Food quality, Physical environment quality, Personal interaction quality dan Perceived Dimana pada hal ini jika semua indikator-indikator yang terdapat pada setiap variabel seperti salah satunya restoran makanan khas jawa menyajikan makanan yang hangat dan memiliki rasa 46 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 P-ISSN : 2354-8592 E-ISSN : 2621-5055 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 P-ISSN : 2354-8592 E-ISSN : 2621-5055 berkunjung. Pernyataan ini juga sesuai dengan hasil penelitian yang dilakukan Ryu dan Han (2009) dimana ketika pelanggan merasakan bahwa lingkungan fisik dapat mencerminkan sebuah kualitas yang baik seperti dekorasi yang menarik dan pencahayan yang baik maka tingkat kepuasan pelanggan akan meningkat. Sehingga pelanggan yang pernah berkunjung dan menikmati makanan dan minuman di salah satu dari sepuluh (10) tempat makan masakan khas Jawa di Surabaya setuju bahwa restoran makanan khas Jawa menampilkan kualitas lingkungan fisik yang manarik dapat meningkatkan kepuasan pelanggan yang berukunjung dan menikmati makanan di tempat makan masakan khas Jawa di Surabaya. yang enak yang terdapat pada variabel Food quality, restoran makanan khas jawa memiliki peralatan makanan yang bersih serta menyediakan area makan yang nyaman yang terdapat variabel Physical environment quality, pelayan dapat memperlakukan pelanggan dengan ramah yang terdapat pada variabel Personal interaction quality, dan harga makanan yang ditawarkan sesuai dengan kualitas yang terdapat pada variabel Perceived value. Maka memungkingkan kepuasan (satisfaction) pelanggan dapat terpenuhi. ght© Creative Commons Attribution 4.0 International License am Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia ght© Creative Commons Attribution 4.0 International License m Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Copyright© Creative Commons Attribution 4.0 International License g am Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia HASIL DAN PEMBAHASAN Bitner dan Hubbert (1994) dalam Jalivand et al., (2017) pada konseptual dan pengembangan hipotesis menyatakan bahwa kualitas makanan (food quality) memiliki hubungan langsung dengan kepuasan pelanggan (customer satisfaction), dimana pada hal ini merupakan faktor inti untuk memenuhi kebutuhan pelanggan. Selain itu Teng dan Chang (2013) dalam Jalivand et al., (2017) menyebutkan bahwa rasa, pilihan makanan yang sehat, penyajian makanan, kesegaran bahan makanan, variasi menu makanan, dan suhu makanan saat disajikan merupakan faktor penting dalam kualitas makanan (food quality). Bowen dan shoemaker (1998) dalam Jalivand et al., (2017) menyatakan rasa senang pelanggan untuk merekomendasikan berasal dari nilai pengalaman konsumsi dimana pelanggan cenderung mengekspresikan pendapat dengan merekomendasikan pengalaman pelanggan kepada pelanggan lain, Dalam hal ini pelanggan yang pernah berkunjung dan menikmati makanan dan minuman di salah satu dari sepuluh (10) tempat makan masakan khas Jawa di Surabaya setuju bahwa kepuasan yang didapatkan pelanggan dapat mengembangkan keinginan untuk mengatakan hal-hal positif tentang restoran Temuan dalam penelitian yang dilakukan Jalivand et al., (2017) menyatakan bahwa kualitas lingkungan fisik serperti dekorasi yang menarik, musik, warna serta pencahayaan lingkungan restoran dapat meningkatkan kepuasan pelanggan yang 47 47 P-ISSN : 2354-8592 E-ISSN : 2621-5055 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 ght© Creative Commons Attribution 4.0 International License Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timu Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 P-ISSN : 2354-8592 E-ISSN : 2621-5055 P-ISSN : 2354-8592 E-ISSN : 2621-5055 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 P-ISSN : 2354-8592 E-ISSN : 2621-5055 Trust tidak memiliki pengaruh terhadap variabel Word of Mouth. Han dan Ryu (2012) menyatakan bahwa tingkat nonmonetary switching cost tidak signifikan berpengaruh pada niat nonmonetary switching cost, karena tingkat kepercayaan dan komitmen pelanggan dalam konteks restoran full service tidak bergantung pada adanya kedua jenis switching cost sehingga operator restoran harus dapat mengembangkan kepercayaan dan komitmen pelanggan. Trust tidak memiliki pengaruh terhadap variabel Word of Mouth. Han dan Ryu (2012) menyatakan bahwa tingkat nonmonetary switching cost tidak signifikan berpengaruh pada niat nonmonetary switching cost, karena tingkat kepercayaan dan komitmen pelanggan dalam konteks restoran full service tidak bergantung pada adanya kedua jenis switching cost sehingga operator restoran harus dapat mengembangkan kepercayaan dan komitmen pelanggan. menambahkan komitmen yang merupakan komponen kualitas hubungan lain memiliki peran sebagai mediator terhadap nilai kepercayaan pada niat perilaku pelanggan. Bowen dan Shoemaker (1998) dalam Ercis et al., (2012) menyatakan bahwa kepuasan saja bukan merupakan faktor yang cukup untuk mengukur komitmen dan kepuasan pelanggan dalam menciptakan pelanggan yang loyal. Bellseter dan Zelman (2001) dalam Jalivand et al., (2017) menemukan bahwa kepercayaan terhadap merek memiliki pengaruh secara langsung terhadap komitmen pelanggan serta secara tidak langsung dapat mempengaruhi tingkat kestabilan harga. Casalo et al., (2007) dalam Sahagun et al., (2014) menyatakan kepuasan dan kepercayaan memiliki hubungan positif dengan komitmen. Garbirano dan Johnson (1999) dalam Sahagun et al., (2014) menambahkan ketika salah satu pihak rentan maka hasil yang didapat tidak pasti, maka dari itu komitmen hanya akan berposisi ketika kontrol atas resiko, sehingga kontrol semacam ini ditawarkan melalui kepercayaan. Moorman et al., (1992) dalam Sahagun et al., (2014) mengatakan tingkatan kepercayaan yang lebih tinggi dapat meningkatkan komitmen pelanggan. Hasil penelitian ini sejalan dengan penelitian yang dilakukan oleh Sumedi et al., (2015) dimana pada penelitiannya menyebutkan terdapat tiga komponen dalam variabel komitmen pelaggan yaitu, affective, calculative dan normative yang dapat mencerminkan motivasi pelanggan untuk melanjutkan hubungan (Cater dan Zabkar, 2009). Komitmen afektif merupakan komitmen hubungan pelanggan dengan penyedia layanan yang disebabkan oleh ikatan emosional. Ikatan emosional ini muncul karena perasaan positif, gembira dan kesetaraan dari hubungan yang dimiliki antara pelangggan dan penyedia layanan. Semakin tinggi komitmen afektif pelanggan pada penyedia layanan semakin tinggi ikatan emosional pelanggan dengan penyedia layanan. g am Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia ght© Creative Commons Attribution 4.0 International License Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 kepuasan pelanggan. Semakin banyak pelanggan merasakan kualitas layanan melebihi biaya yang didapatkan, maka semakin tinggi persepsi terhadap nilai layanan yang dapat menghasilkan kepuasan yang lebih besar, selain itu nilai yang dirasakan juga menunjukkan dampak yang kuat dan signifikan terhadap kepuasan pelanggan yang dapat mempengaruhi pelanggan untuk melakukan niat pembelian berulang. makanan khas Jawa serta adanya keinginan pelanggan untuk merekomendasikan salah satu dari (10) tempat makan masakan khas Jawa di Surabaya kepada pelanggan lain. Jalivand Et al.,(2017) menyimpulkan bahwa interaksi pribadi dengan pelanggan merupakan salah satu elemen terpenting selain produk dan layanan di industri restoran, interaksi ini akan menciptakan nilai kepercayaan, kepuasan, kalkulatif, emosional dan komitmen serta memungkinkan pelanggan melakukan pemasaran WOM. Sehingga pelanggan yang pernah berkunjung dan menikmati makanan dan minuman di salah satu dari sepuluh (10) tempat makan masakan khas Jawa di Surabaya setuju bahwa kualitas interaksi yang baik, melayani pelanggan dengan ramah dan memiliki respon yang cepat dalam memenuhi keinginan pelanggan dapat terpenuhi, dimana karyawan atau pelayan sebagai front office yang berinteraksi secara langsung dengan pelanggan dapat meningkatkan kepuasan pelanggan. Hubungan pembeli dan penjual dalam konteks offline menunjukkan bahwa kepuasan memiliki efek positif pada kepercayaan, alasannya adalah pengalaman memuaskan pelanggan menjadikan sumber kepercayaan bagi pelanggan tentang perusahaan, Garbaniro dan Johnson (1999) dalam Dabholkar dan Sheng (2012). Temuan lain juga menyatakan semakin tinggi kepuasan pembeli dengan pemasok, semakin besar kepercayaan pembeli mempercayai pemasok. Hasil dari uji hipotesis ini didukung oleh hasil dari penelitian yang dilakukan oleh Dev dan Han (2011) dimana hasil dari penelitian tersebut menghasilkan variabel satisfaction tidak memiliki pengaruh terhadap variabel Commitment. Dev dan Han (2011) pada temuannya menyatakan bahwa kepercayaan dan komponen kualitas hubungan merupakan mediator yang sempurna untuk mempengaruhi nilai kepuasan pelanggan. Ok et al., (2005) dalam Dev dan Han (2011) Rust dan Oliver (1994) dalam Jalivand et al., (2017) nilai yang dirasakan pelanggan akan muncul dari hasil evaluasi antara manfaat (input) yang didapat dengan biaya (output) yang dikorbankan untuk mendapatkan produk atau layanan yang ditawarkan. Petterson dan Spreng (1997); Enggert dan Ulega (2002) dalam Jackie (2004) menyatakan nilai yang dirasakan dianggap sebagai nilai penentu dalam 48 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 Hasil dari uji hipotesis ini didukung oleh hasil dari penelitian yang dilakukan oleh Han dan Ryu (2012) dimana hasil dari penelitian tersebut menghasilkan variabel Komitmen kalkulatif merupakan komitmen pelanggan yang didasari oleh pertimbangan ekonomi, yaitu perhitungan untung dan rugi ketika mempertahaknan 49 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 P-ISSN : 2354-8592 E-ISSN : 2621-5055 hubungan atau memutuskan hubungan dengan penyedia layanan. Komitmen kalkulatif muncul karena pelanggan menilai hubungan dengan penyedia layanan menggunakan perspektif penilaian rasional daripada emosional (Cater dan Zakbar,2009 dalam Sumedi et al., 2015). ini dapat dibuktikan dengan hasil kuseioner yaitu pelanggan telah mengunjungi restoran sebanyak minimal dua kali dalam enam bulan terakhir. Komitmen pelanggan secara normatif tidak ada larangan secara norma sosial atau norma agama yang melarang pelanggan untuk makan dan minum di salah satu dari kesepuluh tempat makan masakan khas Jawa di Surabaya. namun hasil yang dapat mempengaruhi komitmen pelanggan dengan Word of Mouth adalah komitmen secara kalkulatif dimana pelanggan bisa saja mengganti pilihan restoran lain, mengingat jika pelanggan bisa melihat masih terdapat alternatif lain dengan biaya yang lebih rendah untuk mekan dengan menu yang sama di tempat lain. Sehinggan pelanggan tidak merasa dipaksa untuk memiliki hubungan dengan restoran yang berbasis ekonomi. Komitmen normatif merupakan komitmen yang berkaitan dengan norma sosial, moral, dan kewajiban pada pihak lain. Pelanggan yang memiliki komitmen normatif bisa memiliki perasaan bersalah jika pelanggan memilih untuk memutuskan hubungan dengan penyedia layanan. Oleh karena itu pelanggan yang memiliki komitmen normatif menunjukkan tingkat komitmen pelanggan untuk tetap berhubunga dengan penyedia layanan berdasarkan perasaan kewajiban untuk melakukannya (Cater dan Zakbar,2009 dalam Sumedi et al., 2015) Hasil dari penelitian Sumedi et al., (2015) menunjukkan jika hipotesis penelitian pada variabel komitmen afektif terhadap Word of Mouth diterima, sedangkan hipotisis pada variabel komitmen kalkulatif dan normatif terhadap Word of Mouth ditolak pada konteks industri layanan bank Syariah. Dalam hal ini memunkinkan jika pelanggan restoran makanan khas Jawa sudah memenuhi komitmen secara afektif dimana pelanggan sudah memiliki ikatan emosional dengan salah satu dari kesepuluh tempat makan masakan khas Jawa di Surabaya hal g am Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia ght© Creative Commons Attribution 4.0 International License KESIMPULAN Hasil dari peneltian ini menunjukkan bahwa food quality memiliki pegaruh positif terhadap satisfaction. Dalam hal ini dapat dinyatakan bahwa tempat makan masakan khas Jawa di Surabaya telah memberikan kualitas makanan sesuai dengan keinginan dan kebutuhan pelanggan yang meliputi rasa makanan yang enak, suhu makanan yang hangat, memiliki tampilan yang menarik dan menggunakan bahan baku segar dapat meningkatkan kepuasan 50 P-ISSN : 2354-8592 E-ISSN : 2621-5055 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 pelanggan yang menikmati makanan ditempat. Hasil dari physical environment quality juga menunjukkan bahwa memiliki hubungan positif terhadap satisfaction. Dapat dinyatakan bahwa tempat makan masakan khas Jawa di Surabaya dapat memiliki area makan yang nyaman, pencahayaan yang sesuai dan peralatan makan yang bersih telah terpenuhi, sehingga tingkat kepeuasan pelanggan yang berkunjung dapat terus meningkat. Sehingga dalam hal ini semakin tinggi nilai yang dirasakan pelanggan maka semakin tinggi juga tingkat kepuasan pelanggan. Hasil dari penelitian ini menunjukkan bahwa satisfaction memiliki pengaruh positif terhadap word of mouth intentions. Dalam hal ini dapat dinyatakan bahwa tingkat kepuasan pelanggan yang tinggi dapat mempengaruhi perilaku keinginan pelanggan untuk memeberikan rekomendasi kepada pelanggan lain untuk mengunjungi tempat makan masakan khas Jawa di Surabaya. Selain itu penelitian ini juga menunjukkan satisfaction memiliki pengaruh positif terhadap trust. Dapat dinyatakan bahwa tingkat kepuasan pelanggan ketika berkunjung ke tempat makan masakan khas Jawa di Surabaya dapat meningkatkan kepercayaan pelanggan terhadap restoran yang dikunjungi. Hasil dari penelitian menunjukkan bahwa personal interaction quality memiliki hubungan positif terhadap satisfaction. Dapat dinyatakan bahwa tempat makan masakan khas Jawa di Surabaya antara pelayan restoran dan pelanggan menjalin kualitas hubungan komunikasi atau interaksi yang baik. Dalam hal ini pelanggan telah menilai bahwa pelayan restoran telah memberikan pelayanan yang ramah serta pelayan dapat merespon dengan cepat keingianan pelanggan dengan baik. Sehingga tingkat kepuasan yang dirasakan melalui interaksi pelayan restoran dapat terus meningkat. Selain itu hasil penelitian juga menemukan perceived value memiliki hubungan yang positif terhadap satisfaction. Dapat dinyatakan bahwa pelanggan yang berkunjung ke tempat makan masakan khas Jawa di Surabaya menilai bahwa harga makanan yang ditawarkan, pelayanan yang diberikan dan pengalaman yang ditawarkan telah sesuai dengan keinginan pelanggan. Namun hasil yang berbeda dapat ditunjukkan pada penelitian ini bahwa satisfaction tidak memiliki pengaruh secara langsung terhadap commitment. Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia DAFTAR PUSTAKA Choi, Beom Joon and Kim, Hyun Sik., 2012. “The Impact of outcome Quality, Interaction Quality, Peer to Peer Quality on Customer Satisfaction With a Hospital Service”. Journal of Health Care Quality Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 mouth, selain itu pelanggan tempat makan masakan khas Jawa di Surabaya melihat bahwa terdapat alternatif layanan lain dengan biaya switching yang rendah. ke salah satu dari sepuluh restoran khas jawa di Surabaya. Konsumen tidak secara rutin mengkonsumsi makanan khas jawa, sehingga diperlukan variasi makanan yang disantap setiap hari pyright© Creative Commons Attribution 4.0 International License y g gram Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia KESIMPULAN Hai ini dimungkinkan karena terdapat faktor lain yang perlu diperhatikan yaitu terdapat alternatif atau pilihan lain yang menyajikan produk atau layanan yang sama namun dengan harga yang ditawarkan lebih murah, namun tetap bisa memenuhi kepuasan pelanggan, sehingga hal ini lah yang bisa menyebabkan kepuasan pelanggan terhadap restoran makanan khas jawa tidak akan selalu bisa membentuk komitmen pelanggan 51 P-ISSN : 2354-8592 E-ISSN : 2621-5055 P-ISSN : 2354-8592 E-ISSN : 2621-5055 Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia IMPLIKASI PENELITIAN Hasil lain juga menunjukkan bahwa trust dan commitment tidak memiliki pengaruh secara langsung terhadap word of mouth intention. Dalam variabel trust, hal ini dimungkinkan karena kurangnya peningkatan kinerja interaksi yang dapat berkonstribusi untuk membangun ikatan sosial antara karyawan layanan dengan pelanggan, dan meningkatkan kepercayaan pelanggan. Selain itu perlunya menjaga hubungan antar pelanggan dan pelayan dapat meningkatkan niat pelanggan untuk merekomendasikan dan menyebarkan word of mouth positif. Terkait variabel commitment, dalam hal ini dimungkinkan pengaruh komitmen yang didasarkan pada motivasi ekonomi tidak berkorelasi dengan word of Hasil penelitian menemukan bahwa pelanggan kuliner Jawa di Surabaya menilai kepuasan bukan penilaian utama dalam membentuk komitmen untuk kembali menikmati makanan yang disajikan. Pelaku usaha perlu mencari dan mengembangkan hal-hal baru dari aspek lain seperti mempercantik interior dan eksterior bangunan atau memperbanyak variasi makanan serta meningkatkan pelayanan dan hal-hal kecil seperti kebersihan. Dalam bisnis restoran perlu dikembangkan nilai lebih yang tidak dimiliki pesaing. Hal ini menyulitkan pelanggan untuk mencari alternatif lain. DAFTAR PUSTAKA Choi, Beom Joon and Kim, Hyun Sik., 2012. “The Impact of outcome Quality, Interaction Quality, Peer to Peer Quality on Customer Satisfaction With a Hospital Service”. Journal of Health Care Quality Fang, Yu-Hui, Chao-Min Chiu dan Eric T.G. Wang. 2011. Understanding Customers Satisfaction and Repurchase Intentions. Journal of Internet Research, 21 (4), pp. 479-503. Hair, J.F., et al. (2010). Multivariate Data Analysis. (7th edition). New Jersey : Pearson Education Inc. Jalilvand, M. R., Salimipour, S., Elyasi, M., & Mohammadi, M. (2017). Factors influencing word of mouth behaviour in the restaurant industry. Marketing Intelligence & Planning, 35(1), 81– 110. Kotler, P. and K. Keller. (2012). Marketing Management, Fourteenth Edition, Pearson Education Ltd. 52 Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indone Copyright© Creative Commons Attribution 4.0 International License Program Studi Manajemen Universitas Muhammadiyah Gresik Jawa Timur Indonesia Jurnal Manajerial, Volume 08 Nomor 01 tahun 2021 http://dx.doi.org/10.30587/manajerial.v8i01.1858 P-ISSN : 2354-8592 E-ISSN : 2621-5055 P-ISSN : 2354-8592 E-ISSN : 2621-5055 P-ISSN : 2354-8592 E-ISSN : 2621-5055 Moorman, C., Deshpande´, R. and Zaltman, G. (1992), Relationship Between Providers and Users of Market Research: The Role of Personal Trust, Marketing Science Institute, Cambridge, MA. Mowen, Jhon C dan Minor, Michael. 2002. Perilaku Konsumen. Penerbit Erla Namin, A. (2017). Revisiting customers’ perception of service quality in fast food restaurants. Journal of Retailing and Consumer Services, 34, 70–81. Namkung, Y. and Jang, S. (2007), “Does food quality really matter in restaurant? Its impact on customer satisfaction and behavioral intentions”, Journal of Hospitality and Tourism Research, Vol. 31 No. 3. Purnasari, H., dan Yuliando, H. (2015). ‘How Relationship Quality on Customer Commitment Influences Positive e-WOM”. Agriculture and Agricultural Science Procedia, 3, 149–153. Ribbink, D., Van Riel, A.C.R., Liljander, V. & Streukens, S. (2004). “Comfort your online customer: quality, trust and loyalty on the internet”. Managing Service Quality, 14. Ryu, K. and Han, H. (2012), “Influence of the quality of food, service, and physical environment on customer satisfaction in quick-casual restaurants: moderating role of perceived price”, Journal of Hospitality & Tourism Research, Vol. 34 No. 3. Wirtz Jochen and Chew Patricia Yp, 2002, “The Effect Of Oncentives, Deal Pronesss, Satisfaction And Tie Strenght On Word Of Mouth Behavior”, international journal of service industry management, Vol. 13 No. 2, pp.141-162 Sekaran. U. dan R. Bougie, 2010, Research Methods For Business, Fifth Edition, A Jhon Wiley and Sons Ltd. Sugiyono, 2013, Statistika untuk Penelitian, Bandung, Alfabeta. Sugiyono, 2013, Statistika untuk Penelitian, Bandung, Alfabeta. Turmudi dan Sri Harini. 2008. Metode Statistika Pendekatan Teoritis dan Aplikatif. Malang: UIN-Malang Press. Zineldin, M, Johannisson, B. and Dandridge, T. 1997. Strategic Relationship Management: a Multi Dimensional Perspective. Almqvist & Wiksell International. 53 Home I Archives I Vo18No01 (2021):j uma1Manajerial M --- ........ ._ __ --- ... --.. - Articles 001: httpj /dx.doi.org/ 10.30587[ manajerial.v8i01 Published: 2021-01-19 Foreign Owne.rship Effe<t to Stock Market liquidity in Indonesia Annisa Yasmin Ill POF 01-21 Pengaruh Oisipl.in Kerja Te.rl'\adap Kinerja Karyawan PT. Perke.bu nan Nusantara XDI Pabrik Minyak Sawit Paser Belengkong Amir Ha.:mzah. Wahyudi Wahyum, Eli ana Eliana 22-36 Faktor ·Faktor Yang Mempengaruhi Perilaku Word Of Mouth Inten tion Pelanggan PadaTem pat Makan Masakan Khas jawa Di Surabaya Nurfathan Haroiy Purwanto. Siti Rahayu, Ema Anclajani Oete.rminan l ndividu Me.ngadopsi Layanan E·Payment (Studi Pada Millenia Is Oi Kabupaten Gresik) Marisya Ma.Mia Khoirina. alfina a!fina, Hans Febrianto Setyo 37-53 54-67 Pengaruh Karakte.ristik Jndividu Dan l eade.rsh ip, Terhadap Kine.rj a Organisasi Oe.ngan Teamwork Sebagai Variabel Intervening Pacta Puskesm as Terakreditasi Paripurna Di Kabu pate.n Pasu ruan Mira Kurni.nvati, Lidia Andiani. Ha.nif Mau1udin Ill POF Anal isis Str ategi Positioning Produk Berdasarkan Persepsi Konsu men Pada l ndustri Batik Oijawa Timur Kristiningsih Kristiningsih, Lestari l estar\ 'Ni'ltik Herawati Ill POF Formulasi Strategi Ukm jilbab Azky Colle«ion Untuk Me.ningkatkann Oaya Saing Oi Masa Pande.mi Covid-19 8;.101 Sua:batl.d Aslamiya.h 1 02-117 Home I Archives I Vo18No01 (2021):j uma1Manajerial M --- ........ ._ __ --- ... --.. - Articles 001: httpj /dx.doi.org/ 10.30587[ manajerial.v8i01 Published: 2021-01-19 Foreign Owne.rship Effe<t to Stock Market liquidity in Indonesia Annisa Yasmin Ill POF 01-21 Pengaruh Oisipl.in Kerja Te.rl'\adap Kinerja Karyawan PT. Perke.bu nan Nusantara XDI Pabrik Minyak Sawit Paser Belengkong Amir Ha.:mzah. Wahyudi Wahyum, Eli ana Eliana 22-36 Faktor ·Faktor Yang Mempengaruhi Perilaku Word Of Mouth Inten tion Pelanggan PadaTem pat Makan Masakan Khas jawa Di Surabaya Nurfathan Haroiy Purwanto. Siti Rahayu, Ema Anclajani Oete.rminan l ndividu Me.ngadopsi Layanan E·Payment (Studi Pada Millenia Is Oi Kabupaten Gresik) Marisya Ma.Mia Khoirina. alfina a!fina, Hans Febrianto Setyo 37-53 54-67 Pengaruh Karakte.ristik Jndividu Dan l eade.rsh ip, Terhadap Kine.rj a Organisasi Oe.ngan Teamwork Sebagai Variabel Intervening Pacta Puskesm as Terakreditasi Paripurna Di Kabu pate.n Pasu ruan Mira Kurni.nvati, Lidia Andiani. Ha.nif Mau1udin Ill POF Anal isis Str ategi Positioning Produk Berdasarkan Persepsi Konsu men Pada l ndustri Batik Oijawa Timur Kristiningsih Kristiningsih, Lestari l estar\ 'Ni'ltik Herawati Ill POF Formulasi Strategi Ukm jilbab Azky Colle«ion Untuk Me.ningkatkann Oaya Saing Oi Masa Pande.mi Covid-19 8;.101 Sua:batl.d Aslamiya.h 1 02-117 Home I Archives I Vo18No01 (2021):j uma1Manajerial M --- ........ ._ __ --- ... --.. - Articles Foreign Owne.rship Effe<t to Stock Market liquidity in Indonesia Ill POF Pengaruh Oisipl.in Kerja Te.rl'\adap Kinerja Karyawan PT. Perke.bu nan Nusantara XDI Pabrik Minyak Sawit Paser Belengkong Amir Ha.:mzah. Wahyudi Wahyum, Eli ana Eliana 22-36 Faktor ·Faktor Yang Mempengaruhi Perilaku Word Of Mouth Inten tion Pelanggan PadaTem pat Makan Masakan Khas jawa Di Surabaya Nurfathan Haroiy Purwanto. Siti Rahayu, Ema Anclajani Oete.rminan l ndividu Me.ngadopsi Layanan E·Payment (Studi Pada Millenia Is Oi Kabupaten Gresik) Marisya Ma.Mia Khoirina. alfina a!fina, Hans Febrianto Setyo 37-53 54-67 Pengaruh Karakte.ristik Jndividu Dan l eade.rsh ip, Terhadap Kine.rj a Organisasi Oe.ngan Teamwork Sebagai Variabel Intervening Pacta Puskesm as Terakreditasi Paripurna Di Kabu pate.n Pasu ruan Mira Kurni.nvati, Lidia Andiani. Ha.nif Mau1udin Ill POF Anal isis Str ategi Positioning Produk Berdasarkan Persepsi Konsu men Pada l ndustri Batik Oijawa Timur Kristiningsih Kristiningsih, Lestari l estar\ 'Ni'ltik Herawati Ill POF Formulasi Strategi Ukm jilbab Azky Colle«ion Untuk Me.ningkatkann Oaya Saing Oi Masa Pande.mi Covid-19 8;.101 Sua:batl.d Aslamiya.h 1 02-117 Pengaruh Oisipl.in Kerja Te.rl'\adap Kinerja Karyawan PT. Perke.bu nan Nusantara XDI Pabrik Minyak Sawit Paser Belengkong Amir Ha.:mzah. Wahyudi Wahyum, Eli ana Eliana 22-36 Faktor ·Faktor Yang Mempengaruhi Perilaku Word Of Mouth Inten tion Pelanggan PadaTem pat Makan Masakan Khas jawa Di Surabaya Nurfathan Haroiy Purwanto. Siti Rahayu, Ema Anclajani Oete.rminan l ndividu Me.ngadopsi Layanan E·Payment (Studi Pada Millenia Is Oi Kabupaten Gresik) Marisya Ma.Mia Khoirina. alfina a!fina, Hans Febrianto Setyo 37-53 54-67 Pengaruh Karakte.ristik Jndividu Dan l eade.rsh ip, Terhadap Kine.rj a Organisasi Oe.ngan Teamwork Sebagai Variabel Intervening Pacta Puskesm as Terakreditasi Paripurna Di Kabu pate.n Pasu ruan MiraKurni nvati LidiaAndiani Ha nif Mau1udin Amir Ha.:mzah. Wahyudi Wahyum, Eli ana Eliana Oete.rminan l ndividu Me.ngadopsi Layanan E·Payment (Studi Pada Millenia Is Oi Kabupaten Gresik) Marisya Ma.Mia Khoirina. alfina a!fina, Hans Febrianto Setyo Ill POF
https://openalex.org/W2023287569	https://europepmc.org/articles/pmc2229834?pdf=render	English	null	An atypical orthologue of 6‐pyruvoyltetrahydropterin synthase can provide the missing link in the folate biosynthesis pathway of malaria parasites	Molecular microbiology	2,007	cc-by	9,765	Introduction Folate cofactors are essential molecules for all living organisms, required for the transfer of one-carbon units in a number of metabolic steps, including the key methylation of dUMP to give dTMP, an essential nucleotide for DNA synthesis. Most microorganisms can synthesize the required folates from the simple precursors GTP, p-aminobenzoate (pABA) and glutamate. Higher eukary- otes, with the exception of plants, are incapable of this and depend on a dietary intake of folate. Drugs targeting folate metabolism have long been used as highly successful antimicrobial and anticancer agents (Nzila et al., 2005). The lethal human malaria parasite, Plasmodium falci- parum, is capable of both de novo folate biosynthesis and salvage of pre-formed folate from the plasma of its human host, as demonstrated by radiolabelling studies with folate precursors and intact folates (Krungkrai et al., 1989; Wang et al., 2004a,b). Current evidence suggests that both routes are essential for sustained healthy parasite growth, although the reasons for this are unclear (Hyde, 2005). In other microorganisms (Bermingham and Derrick, 2002) and in plants (Cossins and Chen, 1997), the enzymic steps in the folate biosynthetic pathway are well established from decades of biochemical and genetic analysis. The purine ring system of GTP is rearranged by GTP cyclohydrolase I (GTPCH-I; EC 3.5.4.16) to that of a pterin (7,8- dihydroneopterin triphosphate; DHNTP), the three-carbon side-chain of which is subsequently cleaved to leave one carbon atom, after which pABAand glutamate are linked to the resulting pterin (6-hydroxymethyl-7,8 dihydropterin; 6HMDP) in successive steps involving 6-hydroxymethyl- 7,8-dihydropterin pyrophosphokinase (PPPK or HPPK; EC 2.7.6.3), dihydropteroate synthase (DHPS; EC 2.5.1.15) and dihydrofolate synthase (DHFS; EC 6.3.2.12) [Fig. 1, scheme (a)]. However, a long-standing mystery concern- ing this pathway in P. falciparum is the apparent lack of a gene encoding the enzyme required for the third step in this pathway, dihydroneopterin aldolase (DHNA; EC 4.1.2.25), OnlineOpen: This article is available free online at www.blackwell-synergy.com OnlineOpen: This article is available free online at www.blackwell-synergy.com PTPS to a Cys to Glu change at its active site relative to all previously characterized PTPS molecules, includ- ing that of the human host. Sabine Dittrich, Sarah L. Mitchell, Andrew M. Blagborough,† Qi Wang, Ping Wang, Paul F. G. Sims and John E. Hyde* Manchester Interdisciplinary Biocentre, Faculty of Life Sciences, University of Manchester, 131 Princess Street, Manchester M1 7DN, UK. Accepted 25 November, 2007. For correspondence. E-mail john. hyde@manchester.ac.uk; Tel. (+44) 161 306 4185; Fax (+44) 161 306 5201. †Present address: Department of Biological Sciences, Sir Alexander Fleming Building, Exhibition Road, South Kensington, London SW7 2AS, UK. Re-use of this article is permitted in accordance with the Creative Commons Deed, Attribution 2.5, which does not permit commercial exploitation. © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd Molecular Microbiology (2008) 67(3), 609–618 Molecular Microbiology (2008) 67(3), 609–618 Molecular Microbiology (2008) 67(3), 609–618 doi:10.1111/j.1365-2958.2007.06073.x First published online 18 December 2007 An atypical orthologue of 6-pyruvoyltetrahydropterin synthase can provide the missing link in the folate biosynthesis pathway of malaria parasites Accepted 25 November, 2007. For correspondence. E-mail john. hyde@manchester.ac.uk; Tel. (+44) 161 306 4185; Fax (+44) 161 306 5201. †Present address: Department of Biological Sciences, Sir Alexander Fleming Building, Exhibition Road, South Kensington, London SW7 2AS, UK. Summary Folate metabolism in malaria parasites is a long- standing, clinical target for chemotherapy and prophy- laxis. However, despite determination of the complete genome sequence of the lethal species Plasmodium falciparum, the pathway of de novo folate biosynthesis remains incomplete, as no candidate gene for dihydro- neopterin aldolase (DHNA) could be identiﬁed. This enzyme catalyses the third step in the well-charact- erized pathway of plants, bacteria, and those eukary- otic microorganisms capable of synthesizing their own folate. Utilizing bioinformatics searches based on both primary and higher protein structures, together with biochemical assays, we demonstrate that P. falci- parum cell extracts lack detectable DHNA activity, but that the parasite possesses an unusual orthologue of 6-pyruvoyltetrahydropterin synthase (PTPS), which simultaneously gives rise to two products in compa- rable amounts, the predominant of which is 6-hydroxymethyl-7,8-dihydropterin, the substrate for the fourth step in folate biosynthesis (catalysed by 6-hydroxymethyl-7,8-dihydropterin pyrophosphoki- nase; PPPK). This can provide a bypass for the missing DHNAactivity and thus a means of completing the biosynthetic pathway from GTP to dihydrofolate. Supported by site-directed mutagenesis experiments, we ascribe the novel catalytic activity of the malarial Fig. 1. The conventional folate (a) and biopterin (b) biosynthetic pathways as found in (a) plants, bacteria and lower eukaryotes that are capable of de novo folate synthesis, and (b) in mammals and other organisms that utilize 5,6,7,8-tetrahydrobiopterin (BH4) as a cofactor. Certain organisms, such as some fungi, cyanobacteria and pseudomonads, possess both pathways. Pathway (c) involving the P. falciparum PTPS orthologue is demonstrated in this work and (d) shows the substrates (i), (iii) and products (ii), (iv) of conventional DHNA and PTPS enzymes respectively. Underlined product (ii) in pathways (a) and (c) is 6-hydroxymethyl-7,8-dihydropterin (6HMDP), the required substrate for PPPK. Asterisked product (iv) in pathway (c) was identiﬁed from its oxidation product (see text). Abbreviations: GTPC, GTP cyclohydrolase I; P, poorly deﬁned phosphatase activity (thought in some systems to ﬁrst involve loss of pyrophosphate then subsequent removal of the ﬁnal phosphate); DHNA, dihydroneopterin aldolase; PPPK, hydroxymethyldihydropterin pyrophosphokinase; DHPS, dihydropteroate synthase; DHFS, dihydrofolate synthase; PTPS, pyruvoyltetrahydropterin synthase; SR, sepiapterin reductase. 610 S. Dittrich et al. 610 S. Dittrich et al. Fig. 1. The conventional folate (a) and biopterin (b) biosynthetic pathways as found in (a) plants, bacteria and lower eukaryotes that are capable of de novo folate synthesis, and (b) in mammals and other organisms that utilize 5,6,7,8-tetrahydrobiopterin (BH4) as a cofactor. Summary Certain organisms, such as some fungi, cyanobacteria and pseudomonads, possess both pathways. Pathway (c) involving the P. falciparum PTPS orthologue is demonstrated in this work and (d) shows the substrates (i), (iii) and products (ii), (iv) of conventional DHNA and PTPS enzymes respectively. Underlined product (ii) in pathways (a) and (c) is 6-hydroxymethyl-7,8-dihydropterin (6HMDP), the required substrate for PPPK. Asterisked product (iv) in pathway (c) was identiﬁed from its oxidation product (see text). Abbreviations: GTPC, GTP cyclohydrolase I; P, poorly deﬁned phosphatase activity (thought in some systems to ﬁrst involve loss of pyrophosphate then subsequent removal of the ﬁnal phosphate); DHNA, dihydroneopterin aldolase; PPPK, hydroxymethyldihydropterin pyrophosphokinase; DHPS, dihydropteroate synthase; DHFS, dihydrofolate synthase; PTPS, pyruvoyltetrahydropterin synthase; SR, sepiapterin reductase. parum possesses a thus-far unique variant of an enzyme associated in other organisms with the biosynthesis of tetrahydrobiopterin, which, via a novel mechanism involv- ing a key amino acid alteration in the active site, can provide a bypass route to the substrate of the subsequent enzyme (PPPK) in the folate pathway. the protein that removes two carbon atoms from the pterin side-chain as described above [Fig. 1, reaction (d) (i)–(ii)]. In contrast to the situation with all other folate biosynthetic pathway genes in P. falciparum (Brooks et al., 1994; Triglia and Cowman, 1994; Lee et al., 2001; Salcedo et al., 2001), our attempts to clone a malarial dhna gene by degenerate oligonucleotide PCR based on conserved amino acid motifs in orthologues from other species were unsuccess- ful, and originally ascribed to low levels of conservation among such orthologues (e.g. Escherichia coli, Bacillus subtilis, Pneumocystis carinii and Arabidopsis thaliana DHNAs share identities of only c. 20–30%, with no contigu- ous sequences of conserved residues). However, upon subsequent completion of the genome sequence of P. fal- ciparum, no dhna gene was apparent by BLAST searches (Gardner et al., 2002a), nor has one been identiﬁed by similar analyses in other species of Plasmodium or in the related apicomplexan parasite Toxoplasma gondii that have been sequenced more recently. We therefore adopted several strategies, using both bioinformatics and biochemical assays, to further explore this observation and to identify how malaria parasites might cope with this apparent gap in their folate biosynthetic machinery. The results from such experiments demonstrate that P. falci- © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Bioinformatic searches for DHNA Consistent with the initial analysis of the complete genome sequence of P. falciparum (Gardner et al., 2002a), we found that PSI-BLAST (Altschul et al., 1997), PRATT (Jonassen et al., 1995) and similar search algo- rithms applied to the predicted complete proteome of P. falciparum failed to identify any statistically signiﬁcant hits for a candidate DHNA enzyme. We therefore employed more powerful bioinformatics approaches based on secondary and tertiary structural queries using the programs 3D-PSSM (Kelley et al., 2000) and GEN- THREADER (GT) (McGuffin and Jones, 2003), which utilize different algorithms to predict structural analogues of a query sequence. This strategy exploited the fact that all © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Plasmodium falciparum PTPS orthologue and folate biosynthesis 611 Table 1. Top six hits using the 3D-PSSM program (Kelley et al., 2000) and searching the entire predicted P. falciparum proteome for putative T-fold protein types. Table 1. Top six hits using the 3D-PSSM program (Kelley et al., 2000) and searching the entire predicted P. falciparu protein types. g the 3D-PSSM program (Kelley et al., 2000) and searching the entire predicted P. falciparum proteome for putative T-fold Table 1. Top six hits using the 3D-PSSM program (Kelley et al., 2000) and searching the entire predicted P. falciparum proteome for putative T-fold protein types. Gene locus in PlasmoDBa T-fold protein type with highest match 3D-PSSM e-valueb 3D-PSSM rankc Annotation in PlasmoDBa PFF1360w PTPS 2.19 ¥ 10-12 1 6-Pyruvoyl tetrahydropterin synthase, putative (PTPS) PFL1155w GTPCH-I 1.33 ¥ 10-4 1 GTP cyclohydrolase I (GTPCH-I) PFF0955c DHNA 1.18 3 Hypothetical protein (with ribonuclease Rh-like InterPro domain) PF14_0627 DHNA 1.78 6 Ribosomal protein S3, putative PFB0420w UO 2.32 4 2C-methyl-D-erythritol 2,4-cyclodiphosphate synthase PF10_0105 DHNA 2.50 7 Hypothetical protein (with ribosomal L22 protein InterPro domain) a. From the Plasmodium sequence database at http://www.plasmodb.org/ b. Matches with e-values greater than 1 are considered not to be statistically signiﬁcant. c. Position in structural match table; this indicates whether the P. falciparum protein in question shows better matches to other structural families, as is the case for all but the top two entries above. PPPK and DHPS, in the form of the bifunctional PPPK- DHPS molecule cloned from T. gondii (Pashley et al., 1997), and monitored the production of radiolabelled 7,8- dihydropteroate from [14C] pABA [see Fig. S2, reaction scheme (a)]. Bioinformatic searches for DHNA DHNAs with determined three-dimensional crystal struc- tures are members of a small family known as ‘tunnelling- fold’ (T-fold) proteins (Colloc’h et al., 2000), established members of which are DHNA, GTPCH-I, urate oxidase (UO; EC 1.7.3.3), 6-pyruvoyltetrahydropterin synthase (PTPS; EC 4.2.3.12) and dihydroneopterin triphosphate epimerase (no EC number assigned) (Colloc’h et al., 2002). All protein sequences longer than 50 amino acids (5893 sequences) predicted from the whole genome sequence by GlimmerM (Salzberg et al., 1999; Gardner et al., 2002b), developed for identifying coding sequences in P. falciparum, were processed through 3D-PSSM and GT. Both identiﬁed unambiguously only two T-fold proteins in the parasite, GTPCH-I (3D-PSSM ‘expect’ value 1.33 ¥ 10-4; GT ‘probability’ value 4 ¥ 10-8) and a putative PTPS (values of 2.19 ¥ 10-12 and 4 ¥ 10-5 respectively) (Table 1). All other predicted proteins scored 1.18 in 3D-PSSM and > 0.01 in GT. The 35 proteins falling between 1.18 and 20 in 3D-PSSM and the 27 between 0.01 and 0.1 in GT (classed by the latter as ‘low conﬁdence’) were examined individually, but none was a credible candidate for a DHNA orthologue. In addition to the statistical scores returned by the two programs, this conclusion was based on the absence in any candidate sequence of a Gxxxx- ExxxxQ or closely related motif, found near the N- terminus in all known DHNAs (Fig. S1). Table 2. Plasmodium falciparum cell extracts have SHMT and PPPK-DHPS but not DHNA activities. SHMTa PPPK-DHPSb DHNAb P. falciparum 4110 100 930 60 ndc (2.97 0.07) (0.66 0.04) (0) E. coli 5110 80 3800 60 9830 140 (3.69 0.06) (2.70 0.04) (6.97 0.10) a. Counts accumulated per 10 ml reaction aliquot after 1 h incubation and 1 h screen exposure. b. Counts accumulated per 10 ml reaction aliquot after 10 min incu- bation and 1 h screen exposure. c. No detectable counts above background. All values corrected for counts recorded at time zero, normalized to equal numbers of cells. Means of three determinations SD; pmol equivalents in parentheses. Table 2. Plasmodium falciparum cell extracts have SHMT and PPPK-DHPS but not DHNA activities. SHMTa PPPK-DHPSb DHNAb P. falciparum 4110 100 930 60 ndc (2.97 0.07) (0.66 0.04) (0) E. coli 5110 80 3800 60 9830 140 (3.69 0.06) (2.70 0.04) (6.97 0.10) Bioinformatic searches for DHNA This coupling strategy avoided the need for custom synthesis of labelled DHN for a direct assay. The protocol was veriﬁed using endogenous DHNA in extracts of a standard E. coli strain [BL21(DE3)] as a positive control. To establish that extracts from P. falciparum cul- tures were generally active for known enzymes, they were assayed in parallel for both endogenous PPPK-DHPS and serine hydroxymethyltransferase (SHMT; EC 2.1.2.1) activities, the latter by monitoring incorporation of 14C from labelled serine into 5,10-methylenetetrahydrofolate. As these activities are all involved in folate metabolism, we reasoned that if a DHNA were also present, it would be likely to exhibit a comparable level of activity. However, despite the expected behaviour of all positive controls, where we could readily measure the reaction products in both E. coli and malarial extracts from similar cell numbers (~2.0 ¥ 109 to 1.0 ¥ 1010), no DHNA activity could be detected in the latter preparations under the standard assay conditions used (Table 2). DHNAs with determined three-dimensional crystal struc- tures are members of a small family known as ‘tunnelling- fold’ (T-fold) proteins (Colloc’h et al., 2000), established members of which are DHNA, GTPCH-I, urate oxidase (UO; EC 1.7.3.3), 6-pyruvoyltetrahydropterin synthase (PTPS; EC 4.2.3.12) and dihydroneopterin triphosphate epimerase (no EC number assigned) (Colloc’h et al., 2002). All protein sequences longer than 50 amino acids (5893 sequences) predicted from the whole genome sequence by GlimmerM (Salzberg et al., 1999; Gardner et al., 2002b), developed for identifying coding sequences in P. falciparum, were processed through 3D-PSSM and GT. Both identiﬁed unambiguously only two T-fold proteins in the parasite, GTPCH-I (3D-PSSM ‘expect’ value 1.33 ¥ 10-4; GT ‘probability’ value 4 ¥ 10-8) and a putative PTPS (values of 2.19 ¥ 10-12 and 4 ¥ 10-5 respectively) (Table 1). All other predicted proteins scored 1.18 in 3D-PSSM and > 0.01 in GT. The 35 proteins falling between 1.18 and 20 in 3D-PSSM and the 27 between 0.01 and 0.1 in GT (classed by the latter as ‘low conﬁdence’) were examined individually, but none was a credible candidate for a DHNA orthologue. In addition to the statistical scores returned by the two programs, this conclusion was based on the absence in any candidate sequence of a Gxxxx- ExxxxQ or closely related motif, found near the N- terminus in all known DHNAs (Fig. S1). © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Investigation of a putative malarial PTPS orthologue With both the bioinformatics and biochemical data provid- ing strong and complementary evidence that a conven- tional DHNA may be lacking in P. falciparum, we next explored the possibility that the DHNA step could be bypassed in malaria parasites, focusing on a protein puta- tively identiﬁed as a PTPS orthologue, which is encoded at the PFF1360w locus of the genome (http://www. plasmodb.org). The logical basis for this was twofold. First, this PTPS orthologue had been identiﬁed in the bioinformatics searches above as the only T-fold family protein in P. falciparum other than GTPCH-I (Table 1). Second, PTPS enzymes in other organisms use DHNTP as substrate, which is closely related to DHN, the sub- strate of DHNA [Fig. 1, reactions (d) (i)–(ii) and (iii)–(iv)], although at the primary sequence level, the DHNA and PTPS families are completely different, with no common motifs (Figs S1 and 3). In other organisms, including animals, fungi and certain bacteria, PTPS is the second enzyme in the pathway of tetrahydrobiopterin (BH4) syn- thesis, an essential cofactor for aromatic amino acid hydroxylases, glyceryl-ether monooxygenases and nitric oxide synthases (Thony et al., 2000). Conventionally, PTPS converts DHNTP to 6-pyruvoyltetrahydropterin (PTP), which is then reduced by sepiapterin reductase (SR; EC 1.1.1.153) to BH4 [Fig. 1, scheme (b)]. Upon alignment of the putative P. falciparum PTPS with all other known PTPSs, a high level of conservation was apparent, including residues known from crystallographic studies of rat PTPS (Burgisser et al., 1995; Ploom et al., 1999) and Caenorhabditis elegans PTPS [Protein Data Bank (PDB) structures 1B66 and 2G64 respectively; http://www.rcsb. org/pdb/home/home.do] to be determinants for DHNTP binding (Fig. S3). However, a striking divergence in resi- dues at the active site was evident (Fig. 2). In particular, the Cys residue (Cys-38/42/43 in C. elegans/rat/human PTPS; Fig. S3), which is completely conserved in all other known PTPSs, is absent from that of P. falciparum, of other Plasmodium species and of the related apicom- plexan parasite T. gondii. The side-chain of this key residue ionizes to form the active nucleophile required for base-catalysed proton abstraction from the substrate (Burgisser et al., 1995; Ploom et al., 1999). Structural modelling using crystallographic data for the P. Investigation of a putative malarial PTPS orthologue falciparum enzyme (PDB structure 1Y13) conﬁrmed spatial differ- ences between the malarial enzyme and those of other eukaryotes, and in particular, the presence of a Glu residue (Glu-38) in the former, closely similar in position and orientation to the active site Cys in the latter (Fig. 2 and Fig. S4). We thus hypothesized that, because of this change, the malarial PTPS enzyme may have different catalytic properties compared with conventional ortho- logues in other organisms. Fig. 2. Aligned sequences of PTPS orthologues between the conserved His residues that co-ordinate the active site Zn2+ ion (arrows). The active site Cys residue conserved in all non-apicomplexan enzymes to date is marked by an asterisk above the sequences; the Glu residue proposed to act as nucleophile instead of Cys in Plasmodium is marked by an asterisk below (see also Fig. S3). Accession numbers for non-apicomplexan enzymes: human, Q03393; rat, P27213; guppy, Q90W95; Drosophila melanogaster, P48611; Caenorhabditis elegans, O02058; Escherichia coli, P65870; Synechocystis, Q55798. Gene loci for malarial sequences (http://www.plasmodb.org): Plasmodium gallinaceum (BLAST search); Plasmodium berghei, PB000950.03.0; Plasmodium vivax, Pv114505; Plasmodium falciparum, PFF1360w. The Toxoplasma gondii sequence was also determined from BLAST searching (http://www.toxodb.org). © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Assay for DHNA activity in P. falciparum cell extracts While strongly indicative of the absence of a DHNA- encoding gene in P. falciparum, the above bioinformatic methods all depend on a degree of similarity at the primary or secondary structural level. We therefore assayed for DHNA activity biochemically in P. falciparum cell extracts without any assumptions as to the possible nature of the protein. We initially coupled this reaction, using 7,8-dihydroneopterin (DHN) as the (normal) sub- strate, to the subsequent two enzymes (i.e. catalysing the fourth and ﬁfth steps) in the folate biosynthetic pathway, b. Counts accumulated per 10 ml reaction aliquot after 10 min incu- bation and 1 h screen exposure. All values corrected for counts recorded at time zero, normalized to equal numbers of cells. Means of three determinations SD; pmol equivalents in parentheses. © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 612 S. Dittrich et al. Investigation of a putative malarial PTPS orthologue Activity of the P. falciparum PTPS orthologue falciparum produces this com- pound, as the HPLC retention time indicated, then the product of the ﬁnal enzyme in the coupled assay, radio- labelled dihydropteroate, should be detectable. A time- dependent accumulation of this product was indeed observed when the PfPTPS reaction was linked to either T. gondii PPPK-DHPS or the P. falciparum equivalent [Fig. 4, sample (a)]. In contrast, when PTPS from either human or E. coli was linked to the PPPK-DHPS assay, no radioactive product was detected [Fig. 4, samples (b) and (c)], consistent with their producing only PTP in the con- ventional reaction, which is not a substrate for PPPK. Fig. 3. HPLC separation of pterin products arising from the reactions of the PTPS orthologues from P. falciparum (Pf1 and Pf2; solid line), E. coli (Ec; dotted line) and human (Hs; short-dashed line) using the normal substrate dihydroneopterin triphosphate (DHNTP), followed by oxidation to maximize ﬂuorescence from the pterin ring system (Nichol et al., 1985). This process converts the products of the reaction 6-pyruvoyl-5,6,7,8-tetrahydropterin to pterin, and 6HMDP to 6-hydroxymethylpterin. No substrate (long-dashed line) and no enzyme (dot-dashed line) controls showed no ﬂuorescence in these positions. Vertical arrows indicate the retention times of the pterin and 6-hydroxymethylpterin standards. Fig. 3. HPLC separation of pterin products arising from the Fig. 3. HPLC separation of pterin products arising from the reactions of the PTPS orthologues from P. falciparum (Pf1 and Pf2; solid line), E. coli (Ec; dotted line) and human (Hs; short-dashed line) using the normal substrate dihydroneopterin triphosphate (DHNTP), followed by oxidation to maximize ﬂuorescence from the pterin ring system (Nichol et al., 1985). This process converts the products of the reaction 6-pyruvoyl-5,6,7,8-tetrahydropterin to pterin, and 6HMDP to 6-hydroxymethylpterin. No substrate (long-dashed line) and no enzyme (dot-dashed line) controls showed no ﬂuorescence in these positions. Vertical arrows indicate the retention times of the pterin and 6-hydroxymethylpterin standards. To rule out the possibility of any contaminating E. coli DHNA activity producing 6HMDP from DHN that might be present as a minor component of our DHNTP prepara- tions, further control experiments were necessary. First, non-recombinant E. coli lysate was incubated separately with equal concentrations of DHN and DHNTP for the standard period. Activity of the P. falciparum PTPS orthologue To test this hypothesis, we cloned the P. falciparum ptps gene into an E. coli expression system and puriﬁed the His-tagged product to near homogeneity. We then analy- sed this enzyme, together with His-tagged PTPS enzymes from E. coli and human as controls, for its ability to act on the natural substrate DHNTP. Pterin products were separated on high-performance liquid chromatogra- phy (HPLC) and monitored by the ﬂuorescence of the pterin ring, which was oxidized prior to chromatography by acidic iodine to the fully conjugated form, to enhance ﬂuorescence (Nichol et al., 1985). Both the bacterial and human enzymes converted DHNTP to PTP as expected [an unstable compound that is detected after oxidation and consequent loss of the side-chain at position 6 as unsubstituted pterin (Woo et al., 2002)]. However, the predominant peak of ﬂuorescence (Pf1) arising from the malarial enzyme eluted at the same position as authentic 6-hydroxymethylpterin (the oxidized product of 6HMDP), together with a less intense, more rapidly eluted peak (Pf2) that coincided with the pterin derived from the bac- terial and human enzyme reactions (Fig. 3). Control experiments using commercial 6HMDP conﬁrmed that the hydroxymethyl side-chain, unlike the 1′, 2′-dioxopropyl side-chain of PTP, is stable to the ring oxidation conditions. Thus, in contrast to human and E. coli PTPS, which yield only the single expected product, the malarial © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Plasmodium falciparum PTPS orthologue and folate biosynthesis 613 Fig. 3. HPLC separation of pterin products arising from the reactions of the PTPS orthologues from P. falciparum (Pf1 and Pf2; solid line), E. coli (Ec; dotted line) and human (Hs; short-dashed line) using the normal substrate dihydroneopterin triphosphate (DHNTP), followed by oxidation to maximize ﬂuorescence from the pterin ring system (Nichol et al., 1985). This process converts the products of the reaction 6-pyruvoyl-5,6,7,8-tetrahydropterin to pterin, and 6HMDP to 6-hydroxymethylpterin. No substrate (long-dashed line) and no enzyme (dot-dashed line) controls showed no ﬂuorescence in these positions. Vertical arrows indicate the retention times of the pterin and 6-hydroxymethylpterin standards. has yielded consistently negative results (S.L. Mitchell and J.E. Hyde, unpublished). In other organisms, 6HMDP is produced by DHNA in the third step of the conventional folate biosynthesis pathway [Fig. 1, scheme (a)]. If the PTPS orthologue from P. Activity of the P. falciparum PTPS orthologue A high level of DHNA activity was observed when DHN was provided as substrate, but the almost complete lack of reaction with DHNTP showed that < 2.5% of the latter had degraded by complete dephos- phorylation to DHN [Fig. 5, sample (a)]. As expected, when recombinant E. coli lysate expressing PfPTPS was similarly tested, signiﬁcant product could be made from both substrates [Fig. 5, sample (b)]. Critically, however, when affinity chromatography/ion-exchange- puriﬁed PfPTPS from the same lysate was assayed in this way, product was now only observed with DHNTP as the substrate [Fig. 5, sample (c)]. Together, these data (i) demonstrate that essentially all of the activity seen with DHNTP as substrate can be ascribed to the malarial gene product and (ii) exclude any contribution of E. coli DHNA carried over from the bacterial lysate acting on DHN derived from DHNTP breakdown to the production of 6HMDP in our assays. Thus, of the human, bacterial and parasite PTPS molecules tested, only the malarial ortho- logue displays the ability to produce 6HMDP, which it synthesizes speciﬁcally from DHNTP, and can thereby provide the bypass to PPPK necessitated by the apparent lack of a DHNA activity in this organism. enzyme gives rise to two different products. Although the ratios of these products were quite variable from run to run, by measuring the relative ﬂuorescence intensities of equal concentrations of the 6-hydroxymethylpterin and pterin standards (1.72:1), we calculated from a series of assays (n = 13) that the two products are formed in broadly similar molar quantities, with the former exhibiting a slight predominance (1.48 0.68 SD mol of 6- hydroxymethylpterin to 1 mol of pterin). © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Veriﬁcation of PfPTPS production of 6HMDP by linkage to PPPK-DHPS reactions The retention time for 6-hydroxymethylpterin upon HPLC was well separated from those of other common pterins under our elution conditions and so was straightforward to identify. However, to conﬁrm the assignment made for this slower eluting ﬂuorescent product of the malarial enzyme from its retention time, we coupled the PTPS reaction to the PPPK and DHPS activities of both T. gondii and P. falciparum, as described above for DHNA [Fig. S2, scheme (b)]. The only known substrate for PPPK enzymes is 6HMDP (Brown, 1971), and both P. falciparum (Kasekarn et al., 2004) and T. gondii (Smith, 2006) PPPK have been shown to process this compound in the normal way. Moreover, testing of a wide range of other pterins as potential substrates for malarial PPPK Site-directed mutagenesis to explore the role of Glu-38 Site-directed mutagenesis to explore the role of Glu-38 Site-directed mutagenesis to explore the role of Glu-38 From the primary sequence (Fig. 2) and structural (Fig. S4) alignments of the P. falciparum PTPS orthologue with those of other organisms, we hypothesized that the substitution of a glutamate residue (Glu-38) in P. falciparum for the active-site cysteine residue of all other known PTPS molecules was likely to be a major © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 factor in the novel ability of the malarial enzyme to produce 6HMDP from DHNTP, particularly as no other residue in this region has an ionizable side-chain capable of acting as a nucleophile for base catalysis. To investi- gate this experimentally, site-directed mutagenesis of the wild-type Pfptps clone was carried out to give a series of m Fig. 4. Conﬁrmation of 6HMDP production by the P. falciparum PTPS (Pf) orthologue, but not by those from human (Hs) and E. coli (Ec). Products of the PTPS reaction were coupled to the next two enzymes in the pathway, PPPK and DHPS, in the form of recombinant bifunctional enzyme from T. gondii, and production of 14C-radiolabelled 7,8-dihydropteroate assayed in triplicate on a Typhoon imager [see Fig. S2, scheme (b), for the reaction set-up and more experimental detail]. Top, Typhoon image; bottom, plot of counts recorded after reaction times up to 45 min and expressed per 1 ml of reaction mix per hour of Typhoon screen exposure (mean SD, relative to zero time); (a) PfPTPS with DHNTP substrate, (b) HsPTPS with DHNTP substrate, (c) EcPTPS with DHNTP substrate, (d) 6HMDP, (e) no enzyme, but DHNTP substrate present, (f) PfPTPS, but no substrate present. In the bottom panel, negative controls (e) and (f) are omitted for clarity. In the positive control (d), commercial 6HMDP was provided as substrate for the PPPK step. One hundred counts on the ordinate are equivalent to 0.071 pmol product per ml of reaction mix. Qualitatively similar results were obtained when P. falciparum PPPK-DHPS was used in the coupled reaction instead of T. gondii PPPK-DHPS. 614 S. Dittrich et al. m 614 S. Dittrich et al. residue (Gln) that retains the polar carbonyl group in the side-chain, but lacks the acidic proton whose dissociation renders the side-chain of Glu negatively charged at physi- ological pH, and (iii) substitution with a completely apolar side-chain (Leu) of comparable volume to that of Glu. Site-directed mutagenesis to explore the role of Glu-38 The three mutant clones expressed soluble recombinant pro- teins at levels comparable to the wild type in E. coli, and these products all gave circular dichroic (CD) spectra closely similar to the wild type and characteristic of prop- erly folded proteins (data not shown). Puriﬁed prepara- tions of the four PTPS variants were assayed in parallel at equal concentrations after the standard incubation with DHNTP as substrate and the pterin products monitored by their ﬂuorescence after HPLC separation (Fig. 6). As expected, the Glu38Leu mutant, lacking any nucleophilic centre in the side-chain, gave no detectable product, whereas the Glu38Cys mutant gave a clear peak of con- ventional pterin product, but importantly, no 6HMDP. This was also observed for the Glu38Gln mutant, albeit at a signiﬁcantly reduced level. Only the wild-type enzyme yielded 6HMDP as well as the pterin peak. These data demonstrate the critical role of the Glu-38 side-chain in the production of the PPPK substrate. m m Discussion In the positive control (d), commercial 6HMDP was provided as substrate for the PPPK step. One hundred counts on the ordinate are equivalent to 0.071 pmol product per ml of reaction mix. Qualitatively similar results were obtained when P. falciparum PPPK-DHPS was used in the coupled reaction instead of T. gondii PPPK-DHPS. m Fig. 5. Formation of 14C-radiolabelled 7,8-dihydropteroate starting from either DHNTP (black) or DHN (grey) substrate (20 mM, 1 h incubation, 37°C) as recorded on the Typhoon imager (per 1 h screen exposure) from comparable amounts of (a) lysate from untransformed E. coli cells (and thus containing only EcDHNA) and (b) lysate from transformed E. coli cells overexpressing PfPTPS. The same experiment is repeated in (c) using affinity chromatography/ion-exchange-puriﬁed PfPTPS obtained from an equal volume of lysate as used in (b). One hundred counts on the ordinate are equivalent to 0.071 pmol product per ml of reaction mix. © 2007 Th A th m factor in the novel ability of the malarial enzyme to produce 6HMDP from DHNTP, particularly as no other residue in this region has an ionizable side-chain capable of acting as a nucleophile for base catalysis. To investi- gate this experimentally, site-directed mutagenesis of the wild-type Pfptps clone was carried out to give a series of mutants, Glu38Cys, Glu38Gln and Glu38Leu. The ratio- nale for these modiﬁcations was (i) substitution with the conventional Cys residue into the active site of the malarial molecule, (ii) substitution with the closely related factor in the novel ability of the malarial enzyme to produce 6HMDP from DHNTP, particularly as no other residue in this region has an ionizable side-chain capable of acting as a nucleophile for base catalysis. To investi- gate this experimentally, site-directed mutagenesis of the wild-type Pfptps clone was carried out to give a series of mutants, Glu38Cys, Glu38Gln and Glu38Leu. The ratio- nale for these modiﬁcations was (i) substitution with the conventional Cys residue into the active site of the malarial molecule, (ii) substitution with the closely related Fig. 5. Formation of 14C-radiolabelled 7,8-dihydropteroate starting from either DHNTP (black) or DHN (grey) substrate (20 mM, 1 h incubation, 37°C) as recorded on the Typhoon imager (per 1 h screen exposure) from comparable amounts of (a) lysate from untransformed E. coli cells (and thus containing only EcDHNA) and (b) lysate from transformed E. coli cells overexpressing PfPTPS. © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Discussion Most microorganisms (Bermingham and Derrick, 2002) and plants (Cossins and Chen, 1997) synthesize folate de novo via a well-deﬁned pathway involving GTPCH-I, Fig. 4. Conﬁrmation of 6HMDP production by the P. falciparum PTPS (Pf) orthologue, but not by those from human (Hs) and E. coli (Ec). Products of the PTPS reaction were coupled to the next two enzymes in the pathway, PPPK and DHPS, in the form of recombinant bifunctional enzyme from T. gondii, and production of 14C-radiolabelled 7,8-dihydropteroate assayed in triplicate on a Typhoon imager [see Fig. S2, scheme (b), for the reaction set-up and more experimental detail]. Top, Typhoon image; bottom, plot of counts recorded after reaction times up to 45 min and expressed per 1 ml of reaction mix per hour of Typhoon screen exposure (mean SD, relative to zero time); (a) PfPTPS with DHNTP substrate, (b) HsPTPS with DHNTP substrate, (c) EcPTPS with DHNTP substrate, (d) 6HMDP, (e) no enzyme, but DHNTP substrate present, (f) PfPTPS, but no substrate present. In the bottom panel, negative controls (e) and (f) are omitted for clarity. In the positive control (d), commercial 6HMDP was provided as substrate for the PPPK step. One hundred counts on the ordinate are equivalent to 0.071 pmol product per ml of reaction mix. Qualitatively similar results were obtained when P. falciparum PPPK-DHPS was used in the coupled reaction instead of T. gondii PPPK-DHPS. Fig. 4. Conﬁrmation of 6HMDP production by the P. falciparum PTPS (Pf) orthologue, but not by those from human (Hs) and E. coli (Ec). Products of the PTPS reaction were coupled to the next two enzymes in the pathway, PPPK and DHPS, in the form of recombinant bifunctional enzyme from T. gondii, and production of 14C-radiolabelled 7,8-dihydropteroate assayed in triplicate on a Typhoon imager [see Fig. S2, scheme (b), for the reaction set-up and more experimental detail]. Top, Typhoon image; bottom, plot of counts recorded after reaction times up to 45 min and expressed per 1 ml of reaction mix per hour of Typhoon screen exposure (mean SD, relative to zero time); (a) PfPTPS with DHNTP substrate, (b) HsPTPS with DHNTP substrate, (c) EcPTPS with DHNTP substrate, (d) 6HMDP, (e) no enzyme, but DHNTP substrate present, (f) PfPTPS, but no substrate present. In the bottom panel, negative controls (e) and (f) are omitted for clarity. © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Discussion HPLC separated pterin products arising from the reactions of the wild type (solid line), E38C variant (dashed line), E38Q variant (dotted line) and E38L variant (dot-dashed line) using dihydroneopterin triphosphate (DHNTP) as substrate, followed by processing and calibration as for Fig. 3. DHNA, PPPK, DHPS and DHFS [Fig. 1, scheme (a)]. The predicted functions of all but one of these enzymes, DHNA, that are also expected to be present in P. falciparum, have been conﬁrmed by biochemical assays on parasite-derived cloned gene products [GTPCH-I (P. Wang and J.E. Hyde, unpublished); PPPK (Kasekarn et al., 2004); DHPS (Triglia et al., 1997); DHFS (Salcedo et al., 2001)]. Here, we were unable to detect DHNA activity in P. falciparum extracts that were demon- strably active in other enzymes of folate metabolism, although we cannot exclude the formal possibility of a parasite variant that is inactive under our standard assay conditions, in which DHNA is readily detected from E. coli lysates. However, these biochemical data are consistent with the uniformly negative bioinformatic searches for a DHNA-encoding gene using a range of algorithms that depend on similarities at either the primary sequence, or secondary/tertiary structural, levels. With respect to its component enzymes, P. falciparum is thus unusual in apparently lacking DHNA in an otherwise complete pathway that is essentially identical to that found in other folate-producing organisms. In addition, we demonstrate that, in the apparent absence of such a gene, the parasite encodes an unusual member of the PTPS family, which is a viable alternative to DHNA capable of synthesizing the 6HMDP required as substrate for PPPK from DHNTP. Moreover, the direct experimental coupling of this reaction to P. falciparum PPPK-DHPS to produce dihydropteroate Despite signiﬁcant primary sequence variation, all bac- terial and eukaryotic PTPSs characterized to date pre- serve the key Cys residue in their active sites (Fig. 2), and we show here that, unlike the malarial enzyme, neither E. coli nor human PTPS is capable of producing 6HMDP from DHNTP (Figs 3 and 4). We ascribe this reaction to the replacement in P. falciparum of the Cys residue with Glu-38, whose side-chains in our models occupy a very similar spatial position and orientation relative to the active site (Fig. S4). © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Discussion The same experiment is repeated in (c) using affinity chromatography/ion-exchange-puriﬁed PfPTPS obtained from an equal volume of lysate as used in (b). One hundred counts on the ordinate are equivalent to 0.071 pmol product per ml of reaction mix. Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Plasmodium falciparum PTPS orthologue and folate biosynthesis 615 Fig. 6. Effect of mutation of the Glu-38 residue in the P. falciparum PTPS orthologue on the reaction products. HPLC separated pterin products arising from the reactions of the wild type (solid line), E38C variant (dashed line), E38Q variant (dotted line) and E38L variant (dot-dashed line) using dihydroneopterin triphosphate (DHNTP) as substrate, followed by processing and calibration as for Fig. 3. now provides the basis for a complete deﬁnition of a pathway for folate biosynthesis in this organism [Fig. 1, scheme (c)]. We also note that deployment of the PfPTPS orthologue in the synthesis of 6HMDP would eliminate the need for a separate phosphatase activity to act on DHNTP, as depicted in the second step of the conven- tional pathway in Fig. 1, scheme (a). Microarray analyses (Llinas et al., 2006) show that the gene encoding PfPTPS (PFF1360w) is expressed across the blood stages of all parasite lines investigated with a shallow peak at 18–22 h post invasion, corresponding to the early to mid- trophozoite stages. This resembles the patterns seen for the other folate biosynthesis genes, all of which peak during the 20–30 h phase of the 48 h asexual cycle (http:// www.plasmodb.org and Nirmalan et al., 2002; Llinas et al., 2006). Interestingly, not only is PfPTPS capable of providing the PPPK substrate, it simultaneously gener- ates a comparable amount of the normal PTP product of this class of enzyme, at least under our in vitro conditions. This suggests a possible dual role for PTPS in parasite pterin metabolism, and may relate to the discovery of a P. falciparum pterin 4a-carbinolamine dehydratase (PCD; EC 4.2.1.96) with conventional pterin recycling activity (Wang et al., 2006). However, such a second role for PfPTPS may not lie in the standard BH4 synthesis pathway [Fig. 1, scheme (b)] because we detect little or no BH4 in parasite extracts, whereas 6HMDP is readily observed (P. Wang and J.E. Hyde, unpublished). Fig. 6. Effect of mutation of the Glu-38 residue in the P. falciparum PTPS orthologue on the reaction products. © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Cell extracts Plasmodium falciparum-infected erythrocytes were saponin- lysed and the harvested parasite pellets (~1–5 ¥ 1010 para- sites) re-suspended in 500 ml of Reporter Gene Lysis Buffer in the presence of Complete C Protease Inhibitor Cocktail, 1 U DNase and 1 U RNase (all reagents from Roche, Swit- zerland) and incubated at room temperature with gentle agi- tation for 15 min. The resultant lysates were centrifuged (14 000 g, 30 min) at 4°C and the soluble fraction decanted and concentrated to 200 ml, from which 40 ml was used per assay. E. coli pellets from cultures of known density were lysed using BugBuster detergent (Novagen, UK) according to the manufacturer’s guidelines in the presence of Complete C Protease Inhibitor Cocktail, 1 U DNase and 1 U RNase. The soluble fraction was decanted into a fresh tube after centrifu- gation and reduced to 1/50 of the volume of initial culture, from which 40 ml was used per assay. Cloning and expression of the P. falciparum ptps gene The intronless ptps gene from P. falciparum genomic DNA (K1 strain) was PCR ampliﬁed using primers tailored to the Ek/LIC (Novagen) expression system: GACGACGACAA- GATGATGAAAGAGGAAACCCTAAATTCAG (forward) and GAGGAGAAGCCCGGTTTATATATATTTGTGGACTATGGCC TTTTGTG (reverse). The PCR product was extracted after gel separation using Zymoclean (Zymo Research, USA) and cloned into the pET-46 Ek/LIC vector (Novagen, UK) to produce the plasmid Pfptps-pET46 by ligation-independent cloning using T4 polymerase. 616 S. Dittrich et al. and Kotb (1989). The assay mixture (100 ml) contained 50 mM Tris-HCl (pH 8.0), 3 mM DDT, 0.25 mM pyridoxal 5′-phosphate (Sigma, UK), 2.5 mM EDTA, 2 mM tetrahydro- folate (Schircks, Switzerland), 0.024 mM [3-14C] serine (57 mCi mmol-1, Moravek, CA), 0.2 mM unlabelled serine and enzyme, and radiolabelled product counts determined as above, except that the chromatography paper was washed in running dH2O for 2 h to remove unconverted 14C-serine. Counts detected in this way were converted to picomoles of product by comparison with standard curves generated by directly imaging known quantities of the radiolabelled sub- strates and assuming equimolar conversion to product during the enzyme reactions. PfPTPS currently available, where the wild-type enzyme is bound to biopterin in PDB structure 1Y13. A previous study has indicated that disruption of the folate biosynthetic pathway by gene deletion in this haploid organism is apparently lethal, even in the pres- ence of exogenous (salvageable) folate, and thus unlikely to be a feasible route to conﬁrm the in vivo role of PfPTPS in this pathway (Wang et al., 2004b). Moreover, no spe- ciﬁc inhibitors of PTPS molecules have yet been reported that could be utilized to this end. However, the bioinfor- matic and biochemical evidence we report here is consis- tent with the absence of a conventional DHNA activity, while revealing the presence of an unusual apicomplexan PTPS orthologue whose product provides the necessary substrate for PPPK. This suggests that P. falciparum and related parasite species have evolved a novel and unex- pected route of folate biosynthesis and that the structural and biochemical differences between the malarial and mammalian PTPS orthologues identiﬁed here might be exploitable in the search for new parasite targets within this critical pathway. However, given the major importance of the BH4 synthesis pathway in humans, any candidate inhibitors would need to show a marked discrimination between the host and parasite molecules. DHNA, PPPK, DHPS and SHMT activities DHPS activity was measured by the incorporation of [ring-14C] pABA into DHP, as previously described (Aspinall et al., 2002); DHNA and PPPK were monitored by coupling their activities to that of DHPS, using the appropriate substrates: 6HMDP for PPPK activity and 7,8-dihydroneopterin for DHNA activity (Schircks, Switzerland) [Fig. S2, scheme (a)]. Stan- dard reactions (100 ml) contained 20 mM pterin substrate, 0.1 M Tris-HCl (pH 8.0), 5 mM DTT, 10 mM MgCl2, 20 mM ATP (for DHNA and PPPK activities), 20 mM [14C] pABA (spe- ciﬁc activity 58 mCi mmol-1; Moravek, California) and enzyme. To couple the DHNA reaction, mixes were supple- mented with 2 mg of recombinant PPPK-DHPS from T. gondii (Aspinall et al., 2002) or from P. falciparum (Triglia et al., 1997), as appropriate. The reaction was incubated at 37°C for 1 h, together with negative controls lacking either enzyme or substrate. Aliquots (25 ml) were removed and reactions stopped by EDTA added to 2.5 mM. Aliquots were then spotted onto DE-81 paper (Whatman, UK) pre-soaked in 2.5 mM EDTA (pH 8.0) and air-dried. Unincorporated [14C] pABA was removed by washing for 1 h in 80 mM NaCl, 0.1 M Tris-HCl (pH 7.9) at room temperature (3 ¥ 500 ml). The radioactivity incorporated into [14C] DHP on the dried paper was detected on a Typhoon scanner (Amersham, UK) after ~24 h exposure. Counts were quantiﬁed using the ImageQuant program and corrected for background counts seen in negative controls and thus are expressed as absolute product counts. SHMT activity was measured by incorpora- tion of 14C from serine to 5,10-methylenetetrahydrofolate using a modiﬁed protocol (Smith, 2006) based on Geller Discussion That substitution of Glu-38 for Cys is the critical evolutionary modiﬁcation of the malarial enzyme is strongly supported by our site-directed mutagenesis experiments, where reverse engineering of the wild-type PfPTPS to place Cys in the active site abol- ishes the production of 6HMDP and causes the enzyme to behave as a conventional PTPS molecule (Fig. 6). The shift to a Glu residue in the natural malarial enzyme replaces the single nucleophilic centre of Cys with nega- tive charge distributed over the two side-chain oxygens of Glu, which could explain our observation of two products generated in comparable quantities by the parasite orthologue (see Fig. S5 and legend for a proposed mechanism). This hypothesis could be explored further by crystallography of the wild type and three mutant versions of PfPTPS with bound substrate/product or analogues, which would also extend the only structural data of e Authors mpilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Experimental procedures DHNA, PPPK, DHPS and SHMT activities © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 References Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J.H., Zhang, Z., Miller, W., and Lipman, D.J. (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res 25: 3389–3402. Aspinall, T.V., Joynson, D.H.M., Guy, E., Hyde, J.E., and Sims, P.F.G. (2002) The molecular basis of sulfonamide resistance in Toxoplasma gondii and implications for the clinical management of toxoplasmosis. J Inf Dis 185: 1637–1643. Bermingham, A., and Derrick, J.P. (2002) The folic acid bio- synthesis pathway in bacteria: evaluation of potential for antibacterial drug discovery. Bioessays 24: 637–648. DHNTP substrate was produced from GTP in a separate reaction using Synechocystis GTPCH-I as described (Lee et al., 1999). DHNTP concentration was estimated from its absorbance at 330 nm using an extinction coefficient of 6300 M-1 cm-1 (Bracher et al., 1998). Aliquots of the DHNTP reaction mixture were frozen at -80°C and used once for PTPS enzyme assays. The coupled PTPS-PPPK/DHPS assay was performed as described above for DHNA assays with minor modiﬁcations [Fig. S2, scheme (b)]. The reaction (200 ml) was performed in 100 mM Tris-HCl (pH 8.0), 5 mM DTT, 20 mM DHNTP, 10 mM MgCl2, 20 mM ATP, 20 mM [14C] pABA, 1 mg ml-1 puriﬁed PTPS and 7.5 mg ml-1 TgPPPK/ DHPS or 3 mg ml-1 of concentrated lysate from cells express- ing PfPPPK/DHPS. Reactions were incubated at 37°C for 1 h. The positive control contained 20 mM 6HMDP (Schircks, Switzerland) instead of PTPS enzyme and substrate. Bracher, A., Eisenreich, W., Schramek, N., Ritz, H., Gotze, E., Herrmann, A, et al. (1998) Biosynthesis of pteridines – NMR studies on the reaction mechanisms of GTP cyclohy- drolase I, pyruvoyltetrahydropterin synthase, and sepiapterin reductase. J Biol Chem 273: 28132–28141. Brooks, D.R., Wang, P., Read, M., Watkins, W.M., Sims, P.F.G., and Hyde, J.E. (1994) Sequence variation of the hydroxymethyldihydropterin pyrophosphokinase – dihy- dropteroate synthase gene in lines of the human malaria parasite, Plasmodium falciparum, with differing resistance to sulfadoxine. Eur J Biochem 224: 397–405. Brown, G.M. (1971) The biosynthesis of pteridines. In Advances in Enzymology, Vol. 35. Meister, A. (ed.) New York: John Wiley, pp. 35–77. Burgisser, D.M., Thony, B., Redweik, U., Hess, D., Heiz- mann, C.W., Huber, R., and Nar, H. Puriﬁcation of recombinant PTPS Escherichia coli strain Rosetta2(DE3)pLysS (Novagen, UK) was transformed with Pfptps-pET46, induced by 1 mM IPTG at an OD600 of 0.6–1.0, and the cells were grown for a further 5 h at 30°C, harvested by centrifugation and lysed using BugBuster (Novagen, UK). The PfPTPS N-terminal His- tagged fusion protein in the soluble crude extract was puriﬁed by absorption onto a column of Ni-NTA resin (Qiagen, UK), washing with 20 mM imidazole in 300 mM NaCl, 50 mM NaH2PO4, pH 8.0, elution with 250 mM imidazole in the same buffer, dialysis into 100 mM Tris-HCl (pH 8.0), 2 mM DTT and storage in 10% glycerol at -80°C until further use. His-tagged PTPSs from E. coli and human were produced in the same © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Plasmodium falciparum PTPS orthologue and folate biosynthesis 617 way from pET-28a-based clones pET-ePTPS and pET- hPTPS respectively (Woo et al., 2002). Further puriﬁcation of the enzyme was carried out on a Resource Q ion exchange column (GE Healthcare, UK), binding in 50 mM Tris (pH 8.0), 5 mM DTT, 25 mM NaCl, and eluting in the same buffer except with 1 M NaCl. mutagenesis system (Promega, UK) on the expression plasmid PfPTPS-pET46 described above. All inserts were completely sequenced to verify the desired alterations and ensure that no others had been introduced. Acknowledgements We thank Young Shik Park, Inje University, South Korea, for clones of human and E. coli ptps, and of Synechocystis gtpc; Tony Triglia (WEHI, Melbourne) for the Pfpppk-dhps clone; Claire Wilson and Simon Hubbard (UMIST/University of Manchester) for assistance with bioinformatics searches; Fiona Flett (University of Manchester) for producing T. gondii PPPK-DHPS, and Jeremy Derrick (University of Manchester) and Andrew Hanson (University of Florida, Gainesville) for helpful discussions. We acknowledge ﬁnancial support from the WellcomeTrust, UK (Grant No. 073896 and studentship for A.M.B.) and BBSRC, UK (studentships for S.L.M. and S.D.). Assay of PTPS activity PTPS activity was assayed according to Lee et al. (1999) with minor modiﬁcations. Assays were performed in 100 mM Tris-HCl (pH 8.0), 10 mM MgCl2, 9 mM DHNTP, 2 mM DTT with 1–2 mg ml-1 puriﬁed PTPS in a total volume of 50 ml for 1 h at 37°C. The reaction mixture was treated with 100 ml of acidic iodine solution (2% KI/1% I2 in 1 M HCl) for 1 h in the dark to oxidize the pterin ring and thereby maximize ﬂuores- cent intensity (Nichol et al., 1985). After centrifugation, excess iodine in the supernatant was reduced to iodide by 50 ml of 5% ascorbic acid, the sample passed through a 0.2 mm Micro-Spin ﬁlter (Alltech Associates, UK) and 20 ml analysed for pterins by HPLC on a Dionex Summit system. A reverse-phase column, Adsorbosil C18, 10 mm, 250 ¥ 4.6 mm internal diameter (Alltech Associates, UK), was used with a mobile phase of 10 mM sodium phosphate, pH 6.0, at a ﬂow rate of 1 ml min-1 for 40 min. Fluorescence excitation and emission wavelengths were set to 350 nm and 450 nm, respectively, and areas under peaks used to calcu- late product ratios. Retention times were compared with pterin standards obtained from Schircks, Switzerland. Abso- lute retention times were critically dependent on the condition of the column from previous usage, so all samples and stan- dards for a given experiment were run in immediate succession. References (1995) 6-Pyruvoyl tetrahydropterin synthase, an enzyme with a novel type of active site involving both zinc binding and an intersubunit catalytic triad motif; site-directed mutagenesis of the pro- posed active center, characterization of the metal binding site and modeling of substrate binding. J Mol Biol 253: 358–369. 618 S. Dittrich et al. acterization of the bifunctional hydroxymethyldihydropterin pyrophosphokinase-dihydropteroate synthase gene from Toxoplasma gondii. Mol Biochem Parasitol 86: 37–47. oxidase sequence matches many structural features from urate oxidase. FEBS Lett 526: 5–10. Cossins, E.A., and Chen, L.F. (1997) Folates and one-carbon metabolism in plants and fungi. Phytochemistry 45: 437– 452. Ploom, T., Thony, B., Yim, J., Lee, S., Nar, H., Leimbacher, W, et al. (1999) Crystallographic and kinetic investigations on the mechanism of 6-pyruvoyl tetrahydropterin synthase. J Mol Biol 286: 851–860. Gardner, M.J., Hall, N., Fung, E., White, O., Berriman, M., Hyman, R.W, et al. (2002a) Genome sequence of the human malaria parasite Plasmodium falciparum. Nature 419: 498–511. Salcedo, E., Cortese, J.F., Plowe, C.V., Sims, P.F.G., and Hyde, J.E. (2001) A bifunctional dihydrofolate synthetase- folylpolyglutamate synthetase in Plasmodium falciparum identiﬁed by functional complementation in yeast and bacteria. Mol Biochem Parasitol 112: 241–254. Gardner, M.J., Shallom, S.J., Carlton, J.M., Salzberg, S.L., Nene, V., Shoaibi, A., et al. (2002b) Sequence of Plasmo- dium falciparum chromosomes 2, 10, 11 and 14. Nature 419: 531–534. Salzberg, S.L., Pertea, M., Delcher, A.L., Gardner, M.J., and Tettelin, H. (1999) Interpolated Markov models for eukary- otic gene ﬁnding. Genomics 59: 24–31. Geller, A.M., and Kotb, M.Y. (1989) A binding assay for serine hydroxymethyltransferase. Anal Biochem 180: 120–125. Hyde, J.E. (2005) Exploring the folate pathway in Plasmo- dium falciparum. Acta Trop 94: 191–206. Smith, J. (2006) Identiﬁcation and characterisation of folate pathway genes in Toxoplasma gondii. PhD Thesis, Univer- sity of Manchester. Jonassen, I., Collins, J.F., and Higgins, D.G. (1995) Finding ﬂexible patterns in unaligned protein sequences. Protein Sci 4: 1587–1595. Thony, B., Auerbach, G., and Blau, N. (2000) Tetrahydrobiop- terin biosynthesis, regeneration and functions. Biochem J 347: 1–16. Kasekarn, W., Sirawaraporn, R., Chahomchuen, T., Cowman, A.F., and Sirawaraporn, W. (2004) Molecular charac- terization of bifunctional hydroxymethyldihydropterin pyrophosphokinase-dihydropteroate synthase from Plas- modium falciparum. Mol Biochem Parasitol 137: 43–53. Triglia, T., and Cowman, A.F. (1994) Primary structure and expression of the dihydropteroate synthetase gene of Plas- modium falciparum. Proc Natl Acad Sci USA 91: 7149– 7153. Kelley, L.A., MacCallum, R.M., and Sternberg, M.J.E. (2000) Enhanced genome annotation using structural proﬁles in the program 3D-PSSM. J Mol Biol 299: 499–520. Triglia, T., Menting, J.G.T., Wilson, C., and Cowman, A.F. (1997) Mutations in dihydropteroate synthase are respon- sible for sulfone and sulfonamide resistance in Plasmodium falciparum. Proc Natl Acad Sci USA 94: 13944–13949. Krungkrai, J., Webster, H.K., and Yuthavong, Y. 618 S. Dittrich et al. (1989) De novo and salvage biosynthesis of pteroylpentaglutamates in the human malaria parasite, Plasmodium falciparum. Mol Biochem Parasitol 32: 25–37. Wang, P., Nirmalan, N., Wang, Q., Sims, P.F.G., and Hyde, J.E. (2004a) Genetic and metabolic analysis of folate salvage in the human malaria parasite Plasmodium falciparum. Mol Biochem Parasitol 135: 77–87. Lee, C.S., Salcedo, E., Wang, Q., Wang, P., Sims, P.F.G., and Hyde, J.E. (2001) Characterization of three genes encoding enzymes of the folate biosynthetic pathway in Plasmodium falciparum. Parasitology 122: 1–13. Wang, P., Wang, Q., Aspinall, T.V., Sims, P.F.G., and Hyde, J.E. (2004b) Transfection studies to explore essential folate metabolism and antifolate drug synergy in the human malaria parasite Plasmodium falciparum. Mol Microbiol 51: 1425–1438. Lee, S.W., Lee, H.W., Chung, H.J., Kim, Y.A., Kim, Y.J., Hahn, Y., et al. (1999) Identiﬁcation of the genes encoding enzymes involved in the early biosynthetic pathway of pte- ridines in Synechocystis sp PCC 6803. FEMS Microbiol Lett 176: 169–176. Wang, Q., Hauser, V., Read, M., Wang, P., Hanson, A.D., Sims, P.F.G., and Hyde, J.E. (2006) Functional identiﬁca- tion of orthologous genes encoding pterin recycling activity in Plasmodium falciparum and Toxoplasma gondii. Mol Biochem Parasitol 146: 109–112. Llinas, M., Bozdech, Z., Wong, E.D., Adai, A.T., and DeRisi, J.L. (2006) Comparative whole genome transcriptome analysis of three Plasmodium falciparum strains. Nucleic Acids Res 34: 1166–1173. Woo, H.J., Hwang, Y.K., Kim, Y.J., Kang, J.Y., Choi, Y.K., Kim, C.G., and Park, Y.S. (2002) Escherichia coli 6-pyruvoyltetrahydropterin synthase ortholog encoded by ygcM has a new catalytic activity for conversion of sepiap- terin to 7,8-dihydropterin. FEBS Lett 523: 234–238. Woo, H.J., Hwang, Y.K., Kim, Y.J., Kang, J.Y., Choi, Y.K., Kim, C.G., and Park, Y.S. (2002) Escherichia coli 6-pyruvoyltetrahydropterin synthase ortholog encoded by ygcM has a new catalytic activity for conversion of sepiap- terin to 7,8-dihydropterin. FEBS Lett 523: 234–238. McGuffin, L.J., and Jones, D.T. (2003) Improvement of the GENTHREADER method for genomic fold recognition. Bioin- formatics 19: 874–881. Nichol, C.A., Smith, G.K., and Duch, D.S. (1985) Biosyn- thesis and metabolism of tetrahydrobiopterin and molybdopterin. Ann Rev Biochem 54: 729–764. © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 Site-directed mutagenesis of the Pfptps gene Three oligonucleotides were designed to mutate the codon for Glu-38 in the ‘FRETLH’ motif of the putative active site of the P. falciparum ptps gene (Fig. 2). GAA (Glu) was mutated to CTT (Leu) with 5′-TACAATGGTTTTCGActtACCTTAC ATGGTCATA, to CAG (Gln) with 5′-TACAATGGTTTTCGA cagACCTTACATGGTCATA and to TGC (Cys) with 5′-TA CAATGGTTTTCGAtgcACCTTACATGGTCATA. Mutagenesis was carried out with the GeneEditor in vitro site-directed Colloc’h, N., Poupon, A., and Mornon, J.P. (2000) Sequence and structural features of the T-fold, an original tunnelling building unit. Proteins 39: 142–154. Colloc’h, N., Mornon, J.P., and Camadro, J.M. (2002) Towards a new T-fold protein? The coproporphyrinogen III © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 © 2007 The Authors Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 rnal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618 618 S. Dittrich et al. Supplementary material Nirmalan, N., Wang, P., Sims, P.F.G., and Hyde, J.E. (2002) Transcriptional analysis of genes encoding enzymes of the folate pathway in the human malaria parasite Plasmodium falciparum. Mol Microbiol 46: 179–190. This material is available as part of the online article from: http://www.blackwell-synergy.com/doi/abs/10.1111/ j 1365 2958 2007 06073 Nzila, A., Ward, S.A., Marsh, K., Sims, P.F.G., and Hyde, J.E. (2005) Comparative folate metabolism in humans and malaria parasites (part I): pointers for malaria treatment from cancer chemotherapy. Trends Parasitol 21: 292–298. Pashley, T.V., Volpe, F., Pudney, M., Hyde, J.E., Sims, P.F. G., and Delves, C.J. (1997) Isolation and molecular char- Nzila, A., Ward, S.A., Marsh, K., Sims, P.F.G., and Hyde, J.E. (2005) Comparative folate metabolism in humans and malaria parasites (part I): pointers for malaria treatment from cancer chemotherapy. Trends Parasitol 21: 292–298. (This link will take you to the article abstract). Please note: Blackwell Publishing is not responsible for the content or functionality of any supplementary materials sup- plied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article. Pashley, T.V., Volpe, F., Pudney, M., Hyde, J.E., Sims, P.F. G., and Delves, C.J. (1997) Isolation and molecular char- © 00 e u o s Journal compilation © 2007 Blackwell Publishing Ltd, Molecular Microbiology, 67, 609–618
https://openalex.org/W3175304561	https://repository.publisso.de/resource/frl:6429625/data	English	null	The effects of transcranial direct current stimulation on gait in patients with Parkinson’s disease: a systematic review	Translational neurodegeneration	2,021	cc-by	13,033	© The Author(s). 2021 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Abstract Background: Gait problems are an important symptom in Parkinson’s disease (PD), a progressive neurodegenerative disease. Transcranial direct current stimulation (tDCS) is a neuromodulatory intervention that can modulate cortical excitability of the gait-related regions. Despite an increasing number of gait-related tDCS studies in PD, the efficacy of this technique for improving gait has not been systematically investigated yet. Here, we aimed to systematically explore the effects of tDCS on gait in PD, based on available experimental studies. Methods: Using the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) approach, PubMed, Web of Science, Scopus, and PEDro databases were searched for randomized clinical trials assessing the effect of tDCS on gait in patients with PD. Results: Eighteen studies were included in this systematic review. Overall, tDCS targeting the motor cortex and supplementary motor area bilaterally seems to be promising for gait rehabilitation in PD. Studies of tDCS targeting the dorosolateral prefrontal cortex or cerebellum showed more heterogeneous results. More studies are needed to systematically compare the efficacy of different tDCS protocols, including protocols applying tDCS alone and/or in combination with conventional gait rehabilitation treatment in PD. Conclusions: tDCS is a promising intervention approach to improving gait in PD. Anodal tDCS over the motor areas has shown a positive effect on gait, but stimulation of other areas is less promising. However, the heterogeneities of methods and results have made it difficult to draw firm conclusions. Therefore, systematic explorations of tDCS protocols are required to optimize the efficacy. Keywords: Transcranial direct current stimulation, Gait, Parkinson’s disease excitatory signaling from the thalamus to manifold cor- tical areas is decreased, leading to widespread cortical dysfunctions [3, 4]. Bradykinesia, dystonia, tremor, and postural balance disorders are prominent motor symp- toms of PD [5]. Gait disturbances are debilitating im- pairments that increase the risk of falling in patients and negatively impact the quality of life [6]. Dopaminergic medication and deep brain stimulation are current standard interventions for PD [7, 8]. However, some motor symptoms do not respond well to medication and deep brain stimulation, and these treatments can also Pol et al. Translational Neurodegeneration (2021) 10:22 https://doi.org/10.1186/s40035-021-00245-2 Pol et al. Translational Neurodegeneration (2021) 10:22 https://doi.org/10.1186/s40035-021-00245-2 Pol et al. Translational Neurodegeneration https://doi.org/10.1186/s40035-021-00245-2 Background Parkinson’s disease (PD) is a progressive neurodegenera- tive disorder [1] caused by degeneration of the substan- tia nigra and dysfunction of the striatal pathway [2]. This leads to the increased GABAergic signaling from the output nuclei of the basal ganglia to the subcortical structures, including the thalamus. Consequently, the * Correspondence: hamzehbaharlouei@gmail.com 1Musculoskeletal Research Center, Isfahan University of Medical Sciences, Isfahan, Iran Full list of author information is available at the end of the article sfahan, Iran Full list of author information is available at the end of the article * Correspondence: hamzehbaharlouei@gmail.com 1Musculoskeletal Research Center, Isfahan University of Medical Sciences, Isfahan, Iran Full list of author information is available at the end of the article The effects of transcranial direct current stimulation on gait in patients with Parkinson’s disease: a systematic review Fateme Pol1, Mohammad Ali Salehinejad2, Hamzeh Baharlouei1* and Michael A. Nitsche2,3 Pol et al. Translational Neurodegeneration (2021) 10:22 Page 2 of 19 Page 2 of 19 tDCS on larger motor networks, given the dense con- nectivity of the motor cortex and the basal ganglia, and an impact of tDCS over the cortical regions to target the basal ganglia-thalamocortical motor circuits [26, 29–32]. In accordance, a functional MRI study has shown that the anodal tDCS over the primary motor cortex (M1) in- creases the functional connectivity between the left caudate nucleus and parietal association cortices and modulates the functional connectivity of cortico-striatal and thalamo-cortical circuits [33]. Furthermore, tDCS affects the functional connectivity between cortico- striatal and thalamo-cortical circuits [33], which is im- paired in PD [34]. result in motor and sensory symptoms [7, 8]. Therefore, non-pharmacological, noninvasive therapies are cur- rently increasingly being probed for their therapeutic value, including the non-invasive brain stimulation approaches. Transcranial direct current stimulation (tDCS) is a non-invasive brain stimulation technique in which a weak electrical current is applied through the scalp. It al- ters cortical excitability by modulating the neuronal rest- ing membrane potentials toward hyperpolarization or depolarization [9] and can produce acute and neuroplas- tic alterations of cortical excitability at the macroscale level of brain regions [10]. While anodal stimulation with standard protocols increases the cortical excitabil- ity, cathodal stimulation decreases it [11]. Stimulation for a few minutes produces neuroplastic after-effects, which share some characteristics with long-term po- tentiation and depression, including the involvement of glutamatergic synapses and calcium-dependency [12, 13]. Beyond these regional effects, tDCS modulates local intracortical circuits [13] and induces modifications of large-scale functional networks, which might also be useful for improving PD symptoms [14]. The efficacy of tDCS for gait improvement has not been reviewed specifically with respect to clinical effects and suit- ability of specific intervention protocols. In this systematic review, we set out to evaluate the effect of tDCS alone and in combination with other rehabilitation techniques on gait in PD patients, with consideration of specific parameters of tDCS that are assumed to affect the outcomes, such as the electrode position, stimulation intensity/duration, timing of medication, and performance. The main questions of this systemtic review are: 1) does tDCS improve gait parameters in PD patients? and 2) which protocols are best suited to improve gait in PD patients? p g y p PD involves degeneration of dopaminergic neurons of the substantia nigra and impairment of dopaminergic circuits, especially motor circuits [15, 16]. Brain imaging studies with positron emission tomography and func- tional magnetic resonance imaging (MRI) have shown subcortical striato-nigral deficits in PD, which affect the activity of the cortical motor network [17]. In addition, the movement-related activity of the supplementary motor area (SMA) is significantly reduced in PD [17, 18]. It has been hypothesized that structural and func- tional connectivity between the SMA and the mesen- cephalic locomotor region, a region that contributes to the control of locomotion, is abnormal in PD [19, 20], and reduced activity of the SMA contributes to the pathogenesis of freezing of gait (FoG) [6]. Given the re- duced activity of premotor and primary motor cortical regions in PD [21], there has been a growing interest in clinical application of tDCS to counterbalance respective alterations, and improve gait in PD. Data sources and search strategy This systematic review was performed following the PRISMA (Preferred Reporting Items for Systematic Re- views and Meta-Analyses) guidelines [35] and was regis- tered on 23 October, 2020, in the PROSPERO database (CRD42020177459). We conducted an electronic search in the following databases: PubMed, Web of Science, Scopus, and PEDro, with the last search updated in February, 2021. The search terms were “Parkinson’s disease”, “Parkinsonian”, “transcranial direct current stimulation”, “gait”, “walking” and their respective synonyms (i.e., timed up and go, step, cadence, stride), and acronyms (i.e., tDCS). Data extraction After identifying relevant articles for inclusion in this study, data extraction was carried out independently by the two evaluators (HB, FP). The data included au- thors, year of publication, demographics of the partic- ipants, study design, tDCS protocol (electrode placement, stimulation intensity, duration, electrode size, and the number of sessions), combined treat- ments, outcome measures, main findings, and occur- rence of adverse effects of tDCS. Disagreements between the evaluators were resolved through a third researcher (AS). Data overview The initial search resulted in 156 articles. After elim- inating 37 duplicates, 99 articles were excluded after screening by titles and abstracts. Two studies did not report sham stimulation results and were removed after reading the full text [40, 41]. Although our search space included not only PD but also Parkinsonian syndromes, all studies identified with a randomized design were conducted in PD. Eighteen articles, published between 2010 and 2021, were included in the final analysis of this study (Fig 1). Study selection h Gait is a useful indicator for the therapeutic effects of motor rehabilitation in PD [22]. Anodal tDCS has the potential to enhance excitability and activity of motor regions in the brain and thus improve gait initiation. In animal models, tDCS even increased extracellular striatal dopamine levels [23], which might further ameliorate the motor symptoms of PD. Recent systematic reviews have confirmed that tDCS improves motor functions of PD patients [24, 25]. Moreover, some studies have sug- gested that tDCS combined with conventional gait re- habilitation therapy can have superior effects [26–28]. These effects might be partially due to the effects of Our research question was based on the PICOS (Popula- tion, Intervention, Comparison, Outcome measures, and study design) principle. Studies were included if the fol- lowing inclusion criteria were met: (a) included sham- controlled tDCS; (b) employed patients with the diagno- sis of PD or Parkinson syndromes; and (c) had a ran- domized controlled trial design (parallel groups or cross- over). Studies were excluded if they involved non-human subjects, written in a non-English language, or involved other techniques of transcranial stimulation (e.g., trans- cranial magnetic stimulation). The titles and abstracts of the retrieved papers were initially screened by two Page 3 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 Page 3 of 19 independent reviewers (HB and FP), and duplicates were eliminated by Endnote. After that, a full-text analysis was performed to determine whether these studies met the inclusion and exclusion criteria. scale includes 11 items that rate the internal and ex- ternal validity of a study. The first item, which refers to external validity, is not considered as part of the final PEDro score [38, 39]. Two researchers (HB and FP) rated the articles independently, and any disagree- ments were resolved by discussion or, if required, with the consensus of a third reviewer (AS). The PE- Dro cut-points for determination of study quality were 9–10 (excellent), 6–8 (good), 4–5 (fair) and, below 4 (poor) [37]. The effect of motor area stimulation on gait g Ten studies assessed the effect of motor area stimulation on gait, and seven of them showed that tDCS improved gait. In the studies with positive findings, the anodal electrode was positioned 1–2 cm anterior to the vertex [26, 27, 50, 54], over the vertex [38], at the M1 corre- sponding to the leg with which the patient used to start walking after a freezing episode [53], or over the hotspot of the first dorsal interosseus muscle of the more af- fected side [48]. The stimulation intensity was 2 mA in six studies [26, 27, 38, 50, 53, 54]. Only one study, which tackled the motor cortex hand area, applied tDCS with 1-mA intensity [48]. The stimulation duration was 30 min in one study [38], 20 min in two studies [50, 53], 15 min in two studies [27, 54], 13 min in one study [26] and 10 min in one study [48]. Most of these studies in- cluded multiple-session interventions, comprising 10 sessions [26], 8 sessions [50], 6 sessions [38], and 5 ses- sions [53]. A single session approach was used in two studies [27, 54]. Six studies with positive findings mea- sured gait immediately after the intervention, while in one study the first post-intervention assessment was per- formed one day after intervention [50]. The positive ef- fects of tDCS lasted at least for three months in one g Different from those studies with positive findings, Dagan et al. [44] reported that tDCS over the motor cor- tex alone did not improve gait, but a single session of combined stimulation over the motor and the dorsolat- eral prefrontal cortex resulted in positive effects. How- ever, this study differed from the above-mentioned studies with regard to the electrode size and type, and the stimulation intensity, which utilized relatively small electrodes and comparatively lower stimulation intensity (max, 1.5 mA). In the study by Schabrun et al. [32], tDCS at the left M1 applied for the first 20 min of each of the 9 dual task gait training sessions did not improve gait compared to the sham tDCS group. In this study, gait was not measured immediately after intervention and the first post-intervention assessment was per- formed one week after treatment. Gait was assessed under the dual task condition [32] (Table 3). Risk of bias To assess the methodological quality of trials, the PEDro scale was used (http://www.pedro.org.au), which has been shown to represent high reliability and validity for this purpose [36, 37]. The PEDro Fig. 1 The PRISMA flow chart of included studies investigating the effects of transcranial direct current stimulation on gait symptoms in Parkinson's Disease Page 4 of 19 Page 4 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 study [50], for two months in one study [38], for one month in four studies [26, 38, 50, 53], and for two weeks in one study [53]. In two of these studies, tDCS was used as a stand-alone treatment [50, 53]. Other studies ap- plied tDCS before gait training with visual cues [26], or before 1 Hz transcranial magnetic stimulation [48]. Other studies showed that tDCS had an impact on gait in PD patients who participated in a group-based exer- cise program [27], gait was improved in groups that re- ceived only real tDCS or a combination of real tDCS and physical therapy [38], or that only simultaneous tDCS and physical therapy improved gait in PD [54]. The state of medication during intervention was on in three studies [26, 27, 50]. The other four studies did not report the state of medication during intervention [38, 48, 53, 54]. Gait improvement was assessed via various methods, including the 10 m walk test [26, 50], timed up and go (TUG) test [26, 54], cadence [27, 38, 48], double support time [48], gait velocity [38, 54], stride length [48, 54], step length [38, 48], step width [38], 6-min walk test [54], stand walk sit test [53], and number of steps [48]. These outcomes were obtained during the medication-on state in two studies [27, 48]. In one study the gait was measured in both best medication-on state (considered by the patients and blinded rater to be the best response to their usual dopaminergic medication) and medication-off state (more than 12 h of withdrawal of dopaminergic medication) [50]. In another study out- comes were assessed in the off state and one hour after drug intake [26]. Two studies did not report the state of medication during assessment [38, 53] (Table 3). Participant characteristics A total of 322 individuals participated in the included studies. The mean age in each study ranged 62–72.4 years, and 173 of the participants were male (one study did not report the number of males and females [54]). The mean disease duration ranged 4.3–11 years among the studies (Table 2). The disease severity was quantified by the Hoehn-Yahr scale in 12 studies [27, 32, 38, 39, 42, 44, 45, 47, 49, 50, 52, 53] and the mean value re- ported in the studies was between 1.3 and 2.8. Risk of bias Thirteen studies had a PEDro score of 9 and 10, repre- senting excellent quality [26, 32, 38, 39, 42–49], four studies had a score of 8 [50–53], and one study had a score of 7 [27]. Sixteen studies were designed double- blinded. Two studies were single-blinded but had other- wise good quality, as shown by a score of 8 [52, 53]. Three studies reported dropouts without conduction of an intention-to-treat analysis [27, 50, 51]. In all studies, the allocation procedure was not described (Table 1). Five studies delivered data showing that the participants could not discriminate sham from real tDCS [39, 44, 48–50], while the other studies did not monitor discrim- ination between real and sham stimulation by the partic- ipants. Inter-rater agreement with respect to the PEDro scale ratings was calculated using Cohen’s kappa coeffi- cient, and the resulting κ value was 0.82. The effect of motor area stimulation on gait The effect of prefrontal area stimulation on gait In eight studies, tDCS was applied over the dorsolateral prefrontal cortex (DLPFC) to improve gait [39, 42–45, 47, 51, 52]. In four studies that reported an improve- ment of gait by the intervention, the anodal electrode Page 5 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 Table 1 PEDro quality assessment of the included studies Item Benninger et al. 2010 [50] Bueno et al. 2019 [42] Chang et al. 2017 [43] costa- Ribeiro et al. 2017 [26] Criminger et al. 2018 [51] da Silva et al. 2018 [27] Dagan et al. 2018 [44] Kaski et al. 2014 [54] Lattari et al. 2016 [45] Lu et al. 2018 [46] Manenti et al. 2014 [47] Mishra et al. 2021 [39] Schabrun et al. 2016 [32] Swank et al. 2016 [52] Valentino et al. 2014 [53] von Papen et al. 2014 [48] Workman et al. 2020 [49] Yotnuengnit et al. 2017 [38] 1 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y 2 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y 3 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y 4 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y 5 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y 6 N N N N N N Y N N N N N N N N N N N 7 Y Y Y Y Y Y Y Y Y Y Y Y Y N N Y Y Y 8 Y Y Y Y Y N Y Y Y Y Y Y Y Y Y Y Y Y 9 N Y Y Y N N Y Y Y Y Y Y Y Y Y Y Y Y 10 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y 11 Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y total 8 9 9 9 8 7 10 9 9 9 9 9 9 8 8 9 9 9 1-Eligibility criteria were specified. 2-Subjects were randomly allocated to groups. 3-Allocation was concealed. The effect of motor area stimulation on gait 4-The groups were similar at baseline regarding the most important prognostic indicators. 5-There was blinding of all subjects. 6-There was blinding of all therapists who administered the therapy. 7-There was blinding of all assessors who measured at least one key outcome. 8-Measures of at least one key outcome were obtained from more than 85% of the subjects initially allocated to groups. 9-All subjects for whom outcome measures were available received the treatment or control condition as allocated or, where this was not the case, data for at least one key outcome was analyzed by “intention to treat”. 10-The results of between-group statistical comparisons are reported for at least one key outcome. 11-The study provides both point measures and measures of variability for at least one key outcome. Table 1 PEDro quality assessment of the included studies Pol et al. Translational Neurodegeneration (2021) 10:22 Page 6 of 19 Table 2 Characteristics of the included studies Included Studies Trial design n Age (years) Sex (m, f) Hoehn-Yahr Mean duration of disease in years Medication dosage (LED) in mg/day Benninger et al. 2010 [50] Parallel groups; double blind 25 63.9 (8.7) 16, 9 ~ 2.8 ~ 10 ~ 1155 Bueno et al. 2019 [42] Cross-over; double blind 20 64.45 (8.98) 12, 8 2.25 (0.63) 7.8 (5.32) NR Chang et al. 2017 [43] Parallel groups; double blind 32 ~ 63 20, 12 NR 9.3 ~ 817 Costa-Ribeiro et al. 2017 [26] Parallel groups; double blind 22 ~ 62 15, 7 NR ~ 6.2 ~ 815 Criminger et al. 2018 [51] Cross-over; double blind 16 68.13 (9.76) 12, 4 NR 8.69 (9.7) NR da Silva et al. 2018 [27] Parallel groups; double blind 21 66 10, 7 ~ 2.5 ~ 5.5 NR Dagan et al. 2018 [44] Cross-over; double blind 20 68.8 (6.8) 17, 3 2.5 (0.6) NR 554.7 (401.1) Kaski et al. 2014 [54] Cross-over in two parallel groups; double blind 16 NR NR NR NR NR Lattari et al. 2016 [45] Cross-over; double blind 17 69.18 (9.98) 13, 4 2.35 (1.1) 7.06 (2.7) 748.29 (343.80) Lu et al. 2018 [46] Cross-over; double blind 10 66.3 (9.9) 7, 3 NR NR 761.0 (362.2) Manenti et al. 2014 [47] Cross-over; double blind 10 67.1 (7.2) 6, 4 1.3 (1.1) 8.1 (3.5) 749.2 (445.5) Mishra et al. 2021 [39] Cross-over; double blind 20 63.9 (8.7) 14, 6 ~ 2 NR NR Schabrun et al. The effect of motor area stimulation on gait 2016 [32] Parallel groups; double blind 16 ~ 67 10, 6 ~ 2 ~ 5.75 ~ 626 Swank et al. 2016 [52] Cross-over; single blind 10 68.7 (10.2) 8, 2 ~ 2 7.9 (7.1) NR Valentino et al. 2014 [53] Cross-over; single blind 10 72.3 (3.6) 5, 5 2.8 (0.5) 11 (4.9) NR von Papen et al. 2014 [48] Cross-over; double blind 10 63 (9) 3, 7 NR 7 (6) 794 (360) Workman et al. 2020 [49] Cross-over; double blind 7 72.4 (6.4) 5, 2 1.9 (0.4) 4.3 (2.5) 889.8 (497.7) Yotnuengnit et al. 2017 [38] Parallel groups; double blind 60 65.0 33, 20 ~ 2.5 ~ 7.9 ~ 863 LED Levodopa equivalent dosage; NR not reported. Mean (SD) Trial design n Age (years) Sex (m, f) Hoehn-Yahr Mean duration of disease in years Medication dosage (LED) in mg/day LED Levodopa equivalent dosage; NR not reported. Mean (SD) was placed over F3 or F4 [39, 43, 45, 47]. These studies assessed the effect of tDCS applied on five consecutive days [43] or in a single session [39, 45, 47] at intensity of 1 mA [43] or 2 mA [39, 45, 47]. The stimulation dur- ation was 30 min [39], 20 min [43, 45] or 7 min [47]. While in three studies the participants received the intervention during the medication-on state [39, 43, 47], the other study did not report the state of medication during intervention [45]. In three studies, tDCS was used as a stand-alone treatment [39, 45, 47], while in one study the effects of isolated tDCS and a combination of tDCS and repetitive transcranial magnetic stimulation (rTMS) were compared [43]. The outcome measures in- cluded TUG [43, 45, 47], gait speed [39], FoG [43], and dynamic gait index [45]. The gait assessment was con- ducted in the medication-on phase in three studies [39, 43, 47], while the other study did not report the state of medication during assessment [45]. All the included studies targeting the DLPFC reported an immediate ef- fect of tDCS. Furthermore, in the study by Mishra and Thrasher [39], gait was assessed during, as well as 15 min and 30 min after stimulation, and in another study the outcome measures were obtained one and five weeks after intervention [43]. Two studies did not report any follow-up assessment [45, 47] (Table 4). PD. The effect of motor area stimulation on gait In that study, single-session tDCS was used as a stand-alone treatment, and the intensity and duration of stimulation were 1.5 mA and 20 min, respectively (Table 4). was placed over F3 or F4 [39, 43, 45, 47]. These studies assessed the effect of tDCS applied on five consecutive days [43] or in a single session [39, 45, 47] at intensity of 1 mA [43] or 2 mA [39, 45, 47]. The stimulation dur- ation was 30 min [39], 20 min [43, 45] or 7 min [47]. While in three studies the participants received the intervention during the medication-on state [39, 43, 47], the other study did not report the state of medication during intervention [45]. In three studies, tDCS was used as a stand-alone treatment [39, 45, 47], while in one study the effects of isolated tDCS and a combination of tDCS and repetitive transcranial magnetic stimulation (rTMS) were compared [43]. The outcome measures in- cluded TUG [43, 45, 47], gait speed [39], FoG [43], and dynamic gait index [45]. The gait assessment was con- ducted in the medication-on phase in three studies [39, 43, 47], while the other study did not report the state of medication during assessment [45]. All the included studies targeting the DLPFC reported an immediate ef- fect of tDCS. Furthermore, in the study by Mishra and Thrasher [39], gait was assessed during, as well as 15 min and 30 min after stimulation, and in another study the outcome measures were obtained one and five weeks after intervention [43]. Two studies did not report any follow-up assessment [45, 47] (Table 4). In contrast, in the remaining three studies, DLPFC stimulation by anodal tDCS did not improve gait in PD. The intensity and duration of tDCS applied in these studies were 2 mA and 20 min respectively [42, 51, 52]. Criminger et al. paired tDCS with simultan- eous stationary bicycling or a golf video game, and the stimulation intensity was 1 mA [51]. Swank et al. [52] assessed the single-session effect of isolated tDCS. The intensity and duration of tDCS were 2 mA and 20 min respectively. In contrast to the studies with positive findings [39, 43, 45, 47] that placed the return electrode over the contralateral supraorbital area, they used dual site stimulation in which both anodal and cathodal electrodes were placed over the DLPFC (Table 4). The effect of tDCS over the cerebellum The effect of tDCS over the cerebellum Workman et al. [49] reported that the cerebellar tDCS at an intensity of 2 mA or 4 mA with a stimu- lation duration of 20 min did not improve gait in PD. However, 4-mA tDCS improved balance immediately after intervention, as assessed by the Berg balance scale (Table 5). Dagan et al. [44] reported that the dual-site tDCS of the DLPFC and M1 improved the gait speed in Page 7 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 Pol et al. Translational Neurodegeneration Table 3 Intervention protocols and results in studies that targeted the motor area Study Polarity of target electrode Electrode size (cm2) Intensity (mA) Current density (mA/cm2) Current density (mA/cm2) Duration (min) Number of sessions Anatomical target (target electrode placement) Target electrode placement Return electrode placement Benninger et al. 2010 [50] Anode T = 97.5 R = 50 (two 25 cm2) 2 0.02 0.02 20 8 sessions (3 times per week) M1, SMA (8 mm anterior to Cz or forehead above eyebrows) 8 mm anterior to Cz or forehead above eyebrows Mastoid Costa-Ribeiro et al. 2017 [26] Anode T = R = 5 × 7 2 0.06 0.06 13 10 sessions (3 times per week) M1, SMA (2 cm anterior to the Cz) 2 cm anterior to the Cz Supraorbital area of the contralateral hemisphere of the more affected side da Silva et al. 2018 [27] Anode T = R = 5 × 7 2 0.06 0.06 15 1 M1, SMA (1.8 cm anterior to Cz) 1.8 cm anterior to Cz Supraorbital area ipsilateral to the most affected side Kaski et al. 2014 [54] Anode T = 10 × 4 R = 4 × 4 2 0.05 0.05 15 1 M1, SMA (10% to 20% anterior to Cz) 10% to 20% anterior to Cz Inion Lu et al. 2018 [46] Anode T = medium butterfly 2.0 cc, 8.1 cm2 R = 8.5 × 6 1 0.12 0.12 10 1 M1, SMA (1.8 cm anterior to Cz) 1.8 cm anterior to Cz Centrally on the forehead Schabrun et al. 2016 [32] Anode T = R = 5 × 7 2 0.06 0.06 20 9 sessions (3 days per week) M1 (C3) Lt M1 Contralateral supraorbital Valentino et al. The effect of tDCS over the cerebellum 2014 [53] Anode T = R = 5 × 7 2 0.06 0.06 20 5 consecutive days M1 (C4) Right M1 Contralateral supraorbital von Papen et al. 2014 [48] Anode and cathode T = R = 5 × 7 1 0.03 0.03 10 1 M1 (Hotspot of first dorsal interosseus muscle) Hotspot of first dorsal interosseus muscle Contralateral frontal pole Yotnuengnit et al. 2018 [38] Anode T = R = 5 × 7 2 0.06 0.06 30 6 sessions (3 days per week) M1 (Cz) Cz Supraorbital T target electrode, R reference electrode; M1 primary motor cortex; SMA supplementary motor area; NR not reported; ↑: positive effect; →: no effect; TUG timed up and go; NR not reported; FoG freezing of gait; tDCS transcranial direct current stimulation Page 8 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 Table 3 Intervention protocols and results in studies that targeted the motor area (Continued) Study State of medication during intervention Combined intervention Online/ offline tDCS Adverse effects Outcome measurements Time points of assessment State of medication during assessment Results Conclusion Effect size (type) Benninger et al. 2010 [50] On – Offline Tingling 10-m walk test Before, 24 h, 1 and 3 months after the last tDCS intervention session On and off Significant decrease of walking time in off-medication state ↑ NR Costa-Ribeiro et al. 2017 [26] On Gait training associated with cues Offline No 10-m walk test; TUG; cadence; stride length Before, immediately and 1 month after intervention On and off 10-m walk test and TUG in 1 month after intervention ↑ NR da Silva et al. 2018 [27] On Group-based exercise program Offline No Gait kinematic analysis: stride length, cadence, duration, speed Before, after On Gait cadence decreased ↑ 0.87 (Cohen’s d) Kaski et al. 2014 [54] NR Physical training focused on improving gait and balance or no combined intervention Online NR 6-min walk; TUG; gait velocity, stride length, Before, after NR Gait velocity, stride length, TUG and 6-min walk test improved in the group that received both tDCS and physical training ↑ 0.5 (Cohen’s d) Lu et al. 2018 [46] NR – Offline No Center of pressure movement and force onsets in gait initiation (FoG) Before, immediately and every 12 min. For a total of 1 h after intervention Off No significant change → NR Schabrun et al. The effect of tDCS over the cerebellum 2016 [32] On Dual task gait training with cues Online Tingling Speed, step length, cadence, TUG 1 week before, 1 and 12 weeks after On No significant difference → NR Valentino et al. 2014 [53] NR – Offline No Stand Walk Sit test (FoG) Before, immediately after the 1st session, immediately, 2 days, 2 weeks and 4 weeks after 5th session NR Improvement in Stand Walk Sit test ↑ NR von Papen et al. 2014 [48] NR Transcranial magnetic stimulation with the frequency of 1 Hz Offline NR Number of steps, step and stride length, cadence, double support time Before, immediately and 30 min after stimulation On Improvement of number of steps, step and stride length, cadence, and double support time after anodal tDCS immediately and 30 min after stimulation ↑ NR Yotnuengnit et al. 2018 [38] NR Physical therapy focused on improving gait or no combined intervention Offline Burning sensation Walking speed, step length, step width, and cadence Before, immediately, 2, 4, and 8 weeks after intervention NR Similar positive outcomes in all intervention groups lasted for 8 weeks ↑ NR Page 9 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 Table 4 Intervention protocols and results in studies that targeted the DLPFC Study Polarity of target electrode Electrode size (cm2) Intensity (mA) Current density (mA/cm2) Current density (mA/cm2) Duration (min) Number of sessions Anatomical target (target electrode placement) Target electrode placement Return electrode placement Bueno et al. 2019 [42] Anode T = R = 5 × 7 2 0.06 0.06 20 1 DLPFC (F3) F3 Fp2 Chang et al. 2017 [43] Anode T = R = 5 × 5 1 0.04 0.04 20 5 consecutive days DLPFC (F3) F3 Contralateral supraorbital Criminger et al. 2018 [51] Anode, cathode T = R = 3 × 5 2 0.13 0.13 20 1 DLPFC (F3,F4) F3, F4 F3, F4 Dagan et al. 2018 [44] Anode T = R = 3 (Pi- electrodes) Max = 1.5 0.33 0.33 20 1 M1, DLPFC (AF4, CP1, F3, FC1, FC5, Cz) AF4, CP1, F3, FC1, FC5, Cz NR Lattari et al. 2016 [45] Anode T = R = 5 × 7 2 0.06 0.06 20 1 DLPFC (F3) F3 Contralateral supraorbital Manenti et al. 2014 [47] Anode T = R = 5 × 7 2 0.06 0.06 7 1 DLPFC (F3 or F4) F3 or F4 Contralateral supraorbital Mishra et al. The effect of tDCS over the cerebellum 2021 [39] Anode T = R = 5 × 7 2 0.06 0.06 30 1 DLPFC (F3) F3 Contralateral supraorbital Swank et al. 2016 [52] Anode NR 2 – 20 1 DLPFC (F3) F3 F4 T target electrode, R reference electrode; M1 primary motor cortex; F3 left DLPFC; F4 right DLPFC; DLPFC dorsolateral prefrontal cortex; Fp2 right supraorbital area; NR not reported; ↑: positive effect; →: no effect; TUG timed up and go; NR not reported; ST single task; DT dual task; FoG freezing of gait; tDCS transcranial direct current stimulation Page 10 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 Table 4 Intervention protocols and results in studies that targeted the DLPFC (Continued) Study State of medication during intervention Combined intervention Online/ offline tDCS Adverse effects Outcome measurements Time points of assessment State of medication during assessment Results Conclusion Effect size (type) Bueno et al. 2019 [42] NR – Offline NR TUG; Video gait analysis for time to cover a particular distance, gait speed, number of steps Before, after On No improvement in TUG time and data from video gait analysis → NR Chang et al. 2017 [43] On Repetitive transcranial magnetic stimulation with 10 Hz frequency Online Headache FoG questionnaire, modified Standing-Start 180° Turn Test (turning steps and turning time); TUG Before, immediately and 1 week after 5 sessions of intervention On Significant improvement in FoG and ambulatory function in both groups. No significant difference in the between-group analyses ↑ NR Criminger et al. 2018 [51] On Stationary bicycling; playing a video game of golf on Wii™ Online Headache TUG in ST and DT conditions Before, after On No significant effect of tDCS on TUG → NR Dagan et al. 2018 [44] NR – Offline NR FOG-provoking test; gait speed in 40 m walking, TUG Before, after NR TUG, gait speed in 40 m and FoG improved by only multitarget stimulation M1 alone: →; M1 and DLPFC: ↑ NR Lattari et al. 2016 [45] NR – Offline Tingling, itching Dynamic Gait Index; TUG before, after NR Improvement in dynamic gait index and TUG ↑ NR Manenti et al. 2014 [47] On – Offline No TUG Before, after On Decrease in TUG time comparing tDCS over F4 vs. sham tDCS ↑ NR Mishra et al. The effect of tDCS over the cerebellum 2021 [39] On – Online/ offline – Speed Before, during, immediately, 15 min, and 30 min after stimulation under ST and DT conditions On Improvement of gait under DT condition ↑ NR Swank et al. 2016 [52] On – Offline NR TUG under three conditions: alone, with motor task, with cognitive task Immediately after On No significant differences → 0.07 to 0.45 (Glass’ Δ) Page 11 of 19 Page 11 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 Pol et al. Translational Neurodegeneration Table 5 Intervention protocols and results in the study that targeted the cerebellum Study Polarity of target electrode Electrode size (cm2) Intensity (mA) Current density (mA/ cm2) Current density (mA/ cm2) Duration (min) Number of sessions Anatomical target (electrode placement) Target electrode placement Return electrode placement Workman et al. 2020 [49] Anode T = R = 5 × 7 2 and 4 0.06 and 0.11 0.06 and 0.11 20 1 Cerebellum (medial edge 1 cm below and 2 cm lateral to the inion over the cerebellar hemisphere contralateral to the more PD-affected side) Medial edge 1 cm below and 2 cm lateral to the inion over the cerebellar hemisphere contralateral to the more PD-affected side Upper arm or medial edge 1 cm below and 2 cm lateral to the inion over the cerebellar hemisphere ipsilateral to the more PD-affected side T target electrode; R reference electrode; NR not reported; →, no effect; TUG timed up and go; tDCS transcranial direct current stimulation Table 5 Intervention protocols and results in the study that targeted the cerebellum Study Polarity of target electrode Electrode size (cm2) Intensity (mA) Current density (mA/ cm2) Current density (mA/ cm2) Duration (min) Number of sessions Anatomical target (electrode placement) Target electrode placement Return electrode placement Workman et al. The effect of tDCS on FoG the results of the present review are in general accord- ance with systematic reviews suggesting the gait- improving effects of other non-invasive brain stimulation techniques for PD, and the positive effects of tDCS on gait in diseases other than PD. Nardone et al. [55] have suggested that rTMS targeting the M1 bilaterally de- creases motor symptoms in PD, and a meta-analysis by Li et al. [56] has shown a significant effect of tDCS in improving mobility in individuals after stroke. Four studies evaluated the effect of tDCS on FoG [43, 44, 46, 53]. In three studies, unilateral M1 stimulation [53], unilateral DLPFC stimulation [43] and M1 + DLPF C dual-site stimulation improved FoG in PD. The dur- ation of intervention was 20 min in all the three studies and the intensity was 1 mA [43], 1.5 mA [44], or 2 mA [53]. In two of these studies, tDCS was applied as a stand-alone intervention [44, 53], while in one study it was combined with rTMS [43]. While two studies did not report the state of medication during intervention and assessment [44, 53], in the other study both inter- vention and assessment were conducted in the medication-on state [43]. The outcome measures were conducted immediately after intervention [43, 44, 53] and 2 days [53], 1 week [43], or 2 and 4 weeks [53] after intervention. In contrast, Lu et al. [46] reported no im- provement of FoG by tDCS over the motor area. In that study, tDCS was applied as a stand-alone intervention at 1 mA intensity with 10 min duration, and FoG was how- ever measured in the medication-off state immediately and up to 1 h after intervention. The present review suggests that placing the anode electrode anterior to the vertex is a promising approach to improving gait in PD. Stimulation of this area with relatively large electrodes may affect both M1 and SMA bilaterally. Since the cortical gait regions are represented bilaterally in the brain [57], bilateral stimulation may be required to modulate cortical excitability and improve gait. For the leg area of the primary motor cortex and the SMA, the more vertical orientation and deeper ana- tomical position as compared to the hand area of the motor cortex, make it more challenging to apply tDCS at the lower limb representations and SMA than at the upper limb representations [58]. The effect of tDCS on FoG However, transcranial magnetic stimulation showed that the anode tDCS an- terior to the vertex can change the excitability of the leg motor and premotor regions, which suggests that tDCS over the leg motor and premotor regions can be used to improve locomotor control in PD patients [59]. Reported side effects of tDCS In six studies, participants reported no side effects [26, 27, 39, 46, 47, 53]. Seven studies reported mild and non- lasting side effects, including headache [43], tingling [32, 45, 49, 50], itching [45, 49], and burning sensations [38, 49]. Five studies reported no data about side effects [42, 44, 48, 52, 54]. In contrast to studies that showed a positive effect of motor area tDCS on gait in PD, Lu et al. [46] reported that bilateral anodal tDCS over the M1 did not improve FoG. In their study, tDCS was, however, applied only for 10 min at an intensity of 1 mA, and outcomes were mea- sured during the medication-off state. Likewise, in two studies with unilateral rather than bilateral application of tDCS over the M1, no improvement of gait was re- ported [32, 44]. de Paz et al. [28] concluded in a system- atic review that further research is required to identify the optimal stimulation targets for gait rehabilitation in neurological diseases. The effect of tDCS over the cerebellum Translational Neurodegeneration (2021) 10:22 The effect of tDCS over the cerebellum 2020 [49] Anode T = R = 5 × 7 2 and 4 0.06 and 0.11 0.06 and 0.11 20 1 Cerebellum (medial edge 1 cm below and 2 cm lateral to the inion over the cerebellar hemisphere contralateral to the more PD-affected side) Medial edge 1 cm below and 2 cm lateral to the inion over the cerebellar hemisphere contralateral to the more PD-affected side Upper arm or medial edge 1 cm below and 2 cm lateral to the inion over the cerebellar hemisphere ipsilateral to the more PD-affected side T target electrode; R reference electrode; NR not reported; →, no effect; TUG timed up and go; tDCS transcranial direct current stimulation Page 12 of 19 Page 12 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 Pol et al. Translational Neurodegeneration Table 5 Intervention protocols and results in the study that targeted the cerebellum (Continued) Study State of medication during intervention Combined intervention Online/ offline tDCS Adverse effects Outcome measurements Timepoints of assessment State of medication during assessment Results Conclusion Effect size (typr) Workman et al. 2020 [49] NR – Offline Burning sensation, itching, tingling, pins/needles 25 ft. walk test; TUG; 6- min walk test, Berg Bal- ance Scale Before, after NR No significant effects on gait parameters but improvement in Berg Balance Scale → NR Table 5 Intervention protocols and results in the study that targeted the cerebellum (Continued) Study State of medication during intervention Combined intervention Online/ offline tDCS Adverse effects Outcome measurements Timepoints of assessment State of medication during assessment Results Conclusion Effect size (typr) Workman et al. 2020 [49] NR – Offline Burning sensation, itching, tingling, pins/needles 25 ft. walk test; TUG; 6- min walk test, Berg Bal- ance Scale Before, after NR No significant effects on gait parameters but improvement in Berg Balance Scale → NR Table 5 Intervention protocols and results in the study that targeted the cerebellum (Continued) Study State of medication during intervention Combined intervention Online/ offline tDCS Adverse effects Outcome measurements Timepoints of assessment State of medication during assessment Results Conclusion Effect size (typr) Workman et al. 2020 [49] NR – Offline Burning sensation, itching, tingling, pins/needles 25 ft. walk test; TUG; 6- min walk test, Berg Bal- ance Scale Before, after NR No significant effects on gait parameters but improvement in Berg Balance Scale → NR Page 13 of 19 Page 13 of 19 Pol et al. Discussion These mixed findings indicate the importance of stimulation parameters (e.g., electrode placement, stimulation intensity, duration, repetition) that need to be adapted in order to improve treatment efficacy (e.g., [68]). y y p q In the studies included in this review, both single-session [27, 44, 45, 47, 48, 54] and multiple-session intervention ap- proaches [26, 38, 50, 53] improved gait in PD. Since the fa- cilitatory effect of single-session tDCS on M1 excitability can last for about one hour after intervention [73], the single-session approaches might be well suited for screen- ing of immediate intervention effects. In multiple-session approaches, the after-effects last for at least 3 months [50] in one study, 2 months [38] in one study, 1 month in three studies [38, 50, 53] and 2 weeks [38, 53] in two studies. One study has shown that the real and sham tDCS combined with visually cued gait training have similar positive effects immediately after the intervention, but at 1 month after intervention, these effects are only preserved in the real tDCS group [26]. The remote effects of tDCS on training- induced gait improvements may be due to the stabilization of training-induced plasticity, which could result in a long- lasting preservation of respective motor memories. The prolonged effects of repeated tDCS have also been reported for motor learning in healthy humans [76]. However, as the follow-up evaluations have not covered extensive durations after intervention, and comparisons of face-to-face inter- ventions with different session numbers are missing, the exact duration of effects and the superiority of specific stimulation protocols remain to be determined. In addition, since the number of sessions in multi-session approaches is limited in the available studies, it might be the case that more extended interventions can cause larger, and more stable effects. For the optimal electrode placement, in addition to anatomically defining areas, model-based optimization of electrode placement may be helpful, especially for target- ing surface-away areas, such as the target regions for gait improvement. The individual models also allow for personalization of intervention, thereby improving the efficacy of the intervention and reducing the between- subject heterogeneity of results. Based on available studies, the anode stimulation is a preferred target stimulation polarity. In this systemic re- view, all included studies have applied anodal tDCS, and two of them compared the effects of anodal and cathodal stimulation but reported no significant change of gait after cathodal tDCS [48, 51]. Discussion Furthermore, anodal tDCS with higher current density enhances the efficacy of tDCS to in- crease the excitability of the leg area of the motor cortex in healthy subjects [70]. Consistently, a study compar- ing the effects of anodal tDCS at different levels of intensity over the upper limb representations of M1 showed a trend of higher cortical excitability enhance- ments with increased current intensities [71, 72]. However, other studies targeting the DLPFC have not reported positive results [42, 51, 52]. In one study [51], tDCS was applied at a relatively low intensity (1 mA) and in combination with stationary bicycling or watching a golf video game, which are motor-related activities yet not as specific as the gait training exercise. In the study by Swank et al. [52], bipolar bilateral stimulation, in which both the anodal and the cathodal electrodes were placed over the DLPFC, might have resulted in the heteroge- neous effects on this area because of the up- and down- regulation of both hemispheres. In this connection, it is interesting to note that in the DLPFC studies with positive findings, a return electrode was positioned over the contralateral supraorbital area, rather than the DLPFC. Regarding the stimulation duration, while most studies applied tDCS for 20 min, positive findings have also been reported for shorter stimulation durations. Nitsche and Paulus [73] have shown that 13 min of tDCS is sufficient to elevate the motor-evoked potential amplitudes at ~ 1 h after intervention in young healthy adults, and another study has reported no significant differences in the ef- fects between 15 min and 30 min of anodal tDCS over the upper limb M1 on cortical excitability [71]. Although our review did not identify an association between the duration of stimulation and the efficacy in PD, studies in healthy humans have shown that longer stimulation du- rations can induce better effects [74, 75]. Therefore, a systematic evaluation of the impact of tDCS duration on its efficacy to treat PD symptoms would be required. In the only study that applied tDCS over the bilateral cerebellum, the score of the Berg balance scale was in- creased, but no significant change of gait was found [49]. This result is not surprising, as the cerebellum plays a strong role in balance control [67]. Discussion To the best of our knowledge, this is the first systematic review to evaluate the efficacy of tDCS in the treatment of gait symptoms in patients with PD. In general, the study results showed that tDCS over motor areas holds some promise, whereas the prefrontal stimulation was comparatively less explored, and the results of available studies were heterogeneous. The results of different studies were however at least partially heterogeneous, which might be caused by the differences in intervention protocol, state of patients during intervention, and as- sessment, etc. The heterogeneity of outcome measures used, and relevant heterogeneities with respect to stimu- lation intensity, duration, electrode position, and other factors between studies, which were often intermingled, make it difficult or impossible to track back differences between studies to a single factor at present. Given these limitations of the available data set, we decided to con- duct a narrative review. There is some evidence supporting for an effect of tDCS over the DLPFC on gait in PD [39, 43–45, 47]. In one study, a significant improvement in FoG after 10 Hz rTMS over the DLPFC was reported [60]. The general rationale for this stimulation approach is that the frontal areas are relevant for locomotion [61]. Anatomical stud- ies have shown that the DLPFC is an important part of the frontostriatal neural pathway that connects the frontal lobe regions with the striatum [62, 63]. In PD, the frontostriatal dysfunction is associated with signifi- cant deficits in executive functions, resulting in impaired walking [64]. A more recent study has shown higher DLPFC activity in PD patients compared to controls during both usual walking and walking while subtracting conditions, and impaired walking performance in PD A systematic review by Beretta et al. [24] has suggested that combined tDCS and motor intervention improves gait in PD. This review, however, did not include studies that used tDCS as a stand-alone treatment. Moreover, Page 14 of 19 Page 14 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 patients as compared to controls only during the walking while subtracting task [65]. In addition, dual-site stimu- lation of M1 and DLPFC has exerted positive effects on gait in PD [44], which is consistent with the observation that FoG is caused by impaired communication among the prefrontal cortex, motor cortex, and subcortical structures [44, 66]. perturbation in PD [69]. Discussion This means that the patients were moderately affected, and patients with severe symptoms were under-represented. Although the present review showed no association between age, dur- ation of disease, disease severity and gait improvement, future studies are needed to investigate the possible as- sociations between these factors and the tDCS effects. With respect to the impact of assessment methods on the outcome of interventions, most of the included stud- ies used the TUG as the main outcome parameter, while others reported spatiotemporal parameters of gait. A systematic review by Mollinedo et al. [84] has shown good reliability and validity of TUG in PD [84] and also a high sensitivity of this test for monitoring medication- dependent effects [85]. Measurements of the spatiotem- poral parameters of gait and balance can provide useful information on subtle effects that might not be identified by the TUG or other bed-side clinical tools. However, changes of these parameters may not provide informa- tion on clinically relevant effects for evey case. With re- spect to the reviewed studies, we did not find an association between the type of outcome assessment and the effect of tDCS, however, most of the studies differed in more than one intervention parameter. Therefore, the possible discernable effects of identical stimulation pro- tocols on different outcome measures should be ex- plored in future studies. With regard to the timing of combined intervention, it is so far unclear if tDCS during or before physical train- ing is better suited in PD. Both online [54] and offline [26, 27, 48] tDCS resulted in positive effects in the in- cluded studies. It has, however, been shown that online stimulation has better effects than offline stimulation on motor learning in healthy humans [79]. Therefore, sys- tematic face-to-face studies exploring the impact of on- line and offline stimulation on the effect of tDCS on gait in PD are needed to address this question. q In all studies that have reported the medication state during the intervention, tDCS was applied in the medication-on state. Dopamine modulates the motor cortex plasticity in M1. Hypo-dopaminergic states may prevent plasticity in PD patients and healthy humans [80, 81], thus it makes sense to conduct tDCS during the medication-on state. The state of medication is also im- portant for the validity of the assessment of outcome measures. Discussion Since the activity of pre- motor and primary motor cortical regions is reduced in PD [21], the excitability-enhancing anodal tDCS may be suitable for restoring the activity of these areas [9]. In this review, most included studied applied tDCS at the intensity of 2 mA. tDCS at 1 mA intensity might be too low to facilitate the leg motor cortex and improve gait in PD [46], at least in some patients. This assumption is con- firmed by the results of a study which showed that tDCS at 2 mA intensity performed better than 1-mA stimulation in modification of the postural response to an external The results of the reviewed studies further suggest that combining tDCS with conventional gait training [26, 54], Page 15 of 19 Page 15 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 Pol et al. Translational Neurodegeneration group-based exercise [27] or rTMS [48] may enhance the effect of intervention. Kaski et al. [54] have reported superior effects of combined tDCS and physical training, as compared to the isolated tDCS or the isolated phys- ical training, on gait rehabilitation in PD. Enhancing cor- tical excitability via tDCS in combination with other interventions might reinforce the positive effects of each, which can be translated into improved clinical outcomes, as compared with application of these interventions alone. The underlying mechanisms may be the synergis- tic induction of neuroplasticity, and enhanced motor network activation induced by these interventions and tDCS, as shown in healthy humans [77] and patients with motor deficits [78]. As mentioned above, this result mirrors those obtained in healthy humans by combining anodal motor cortex tDCS with motor learning [76]. In contrast, a systematic review by de Paz et al. [28] did provide conclusive results for an enhancing effect of ad- junctive stimulation with current tDCS methods in com- bination with gait exercises in patients with neurological disorders. studies was between 60 and 70 years, while only two studies reported a relatively higher mean age of > 70 years [49, 53]. For disease duration, the minima and maxima were 4.3 years [49] and 11 years [53], respect- ively. In the other studies, the mean disease duration was between 5 and 10 years. In 12 studies, the severity of the disease of the participants was evaluated by the Hoehn and Yahr scale, which had a mean value between 1.3 to 2.8 in these studies. Discussion In six studies with positive findings, the par- ticipants were in the on state during assessment [27, 38, 45, 47, 48, 53]. However, Costa-Ribeiro et al. [26] found a positive effect of tDCS under both on and off condi- tions of medication. More investigations are needed to clarify whether tDCS can improve gait also in the off phase. The levodopa equivalent dosage was similar across the studies included in this review. Since dopa- minergic medication can have non-linear dosage- dependent effects on plasticity in healthy humans [82, 83], it might be worthwile to assess the correlation be- tween levodopa equivalent dosage and the intervention effects in future studies. The loss of dopaminergic neurons in the substantia nigra pars compacta within the basal ganglia is a cause for gait impairments in PD [86]. In patients with PD, frontostriatal dysfunction has been associated with sig- nificant deficits in executive functions that are associated with difficulties in walking [64]. The cortical motor net- work and the movement-related activity of the SMA are altered in PD [17, 18], which contribute to the gait prob- lems in PD [19, 20]. The lasting effect of tDCS on gait shows that although PD is a neurodegenerative disease, there is a potential for compensation by neuromodula- tion techniques. Competing interests MAN is a member of the Scientific Advisory Boards of Neuroelectrics and Neurodevice. All other authors declare that they have no competing interests. Availability of data and materials Availability of data and materials Not applicable. Received: 23 March 2021 Accepted: 7 June 2021 Received: 23 March 2021 Accepted: 7 June 2021 Received: 23 March 2021 Accepted: 7 June 2021 Limitations and suggestions Besides the limitations already mentioned above, an im- portant common limitation of the reviewed studies is the small sample size, making it difficult to generalize re- sults. Future studies should include larger sample sizes. Furthermore, the substantial heterogeneity of study With respect to the demographic factors, the mean age of participants in the majority of the included Page 16 of 19 Page 16 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 protocols makes it difficult to draw firm conclusions about the best suited protocols to improve clinical symp- toms at present. Here, systematic evaluation of the im- pact of variations of specific aspects of tDCS protocols, which are known to affect the efficacy of the interven- tion, including the target area, the stimulation intensity, the session frequency, the number of sessions, set of combined vs stand-alone interventions, online vs offline stimulation in the case of combined interventions, asso- ciation of stimulation effects with dopaminergic medica- tion, is required to shape clinically useful interventions for future application. Additionally, most of the studies conducted so far have purely behavioral outcome param- eters. For future studies, it will be relevant to add physiological assessments to clarify the mechanisms underlying the respective effects. The quality assessment of the included studies was conducted with PEDro, which evaluates the minimum requirements of study quality, e.g. not taking sample size into account as qual- ity parameter. Most studies were proof of principle, and not designed to define clinically meaningful intervention protocols, which would require systematic evaluation of dosing and other intervention parameters. The effect size was only reported in a minority of studies and none of the papers delivered data about clinical meaningful- ness of the effects. Such data are crucial for future clin- ical implementation, thus should be reported in future studies. intervention needs to be explored directly. Furthermore, although the online tDCS has better effects than offline tDCS as shown for motor learning in healthy young adults, face-to-face studies in PD are lacking. The lon- gest follow up was 3 months, but many studies only cov- ered short timelines after intervention. More studies are thus needed to explore the duration of clinical effects of the intervention. Consent for publication Not applicable. Consent for publication Not applicable. The present review suggests that tDCS is a promising intervention approach to improving gait in PD. While applying anodal tDCS over the motor areas has shown a positive effect on gait in the majority of studies, stimula- tion over other areas like the DLPFC might be less promising. In addition, the small sample size and the heterogeneity of intervention protocols and outcome measures make it difficult to identify the best suited intervention protocols based on the current data, and to come to clear conclusions about the clinical usefulness of this intervention at present. Stimulation intensity, state of medication during intervention and assessment, and online versus offline tDCS in combination with trad- itional gait rehabilitation techniques are main aspects of variability of study protocols, which deserve further in- vestigation. Although higher stimulation intensity has been shown to be more efficient in improving motor learning in healthy subjects, such a dosage-dependent ef- fect needs to be tested directly in PD. The intervention has been conducted in the medication-on state in the studies as far as reported. This makes sense, because the plasticity-related effects of non-invasive brain stimula- tion are reduced in hypo-dopaminergic states; however, the impact of state of medication on the results of Abbreviations k ’ d PD: Parkinson’s disease; tDCS: Transcranial direct current stimulation; SMA: Supplementary motor area; FoG: Freezing of gait; M1: Primary motor cortex; DLPFC: Dorsolateral prefrontal cortex; TUG: Timed up and go Authors’ contributions HB and MAN contributed to the conception, project design, and data interpretation. FP, HB and MAS performed literature research and drafted the manuscript. All authors helped to collect the data and performed statistical analyses and also contributed to data interpretation. All authors edited and approved the manuscript. Funding This work was supported by the Musculoskeletal Research Center, Isfahan University of Medical Sciences, Isfahan, Iran. Author details 1 1Musculoskeletal Research Center, Isfahan University of Medical Sciences, Isfahan, Iran. 2Department of Psychology and Neurosciences, Leibniz Research Centre for Working Environment and Human Factors, Dortmund, Germany. 3Department of Neurology, University Medical Hospital Bergmannsheil, Bochum, Germany. Declarations Ethics approval and consent to participate Not applicable. References https://doi.org/10.1111/j.1469-7793.2000.t01-1-00633.x. 10. Brunoni A, Nitsche M, Loo C. Transcranial Direct Current Stimulation in Neuropsychiatric Disorders. Springer; 2016. 29. Schulz R, Gerloff C, Hummel FC. Non-invasive brain stimulation in neurological diseases. Neuropharmacology. 2013;64:579–87. https://doi. org/10.1016/j.neuropharm.2012.05.016. 11. Stagg CJ, Nitsche MA. Physiological basis of transcranial direct current stimulation. Neuroscientist. 2011;17(1):37–53. https://doi.org/10.1177/1073 858410386614. 30. Ferrucci R, Mameli F, Ruggiero F, Priori A. Transcranial direct current stimulation as treatment for Parkinson’s disease and other movement disorders. Basal Ganglia. 2016;6(1):53–61. https://doi.org/10.1016/j.baga.201 5.12.002. 12. Nitsche M, Fricke K, Henschke U, Schlitterlau A, Liebetanz D, Lang N, et al. Pharmacological modulation of cortical excitability shifts induced by transcranial direct current stimulation in humans. J Physiol. 2003;553(1):293– 301. https://doi.org/10.1113/jphysiol.2003.049916. 31. Costa-Ribeiro A, Maux A, Bosford T, Tenório Y, Marques D, Carneiro M, et al. Dopamine-independent effects of combining transcranial direct current stimulation with cued gait training on cortical excitability and functional mobility in Parkinson's disease. J Rehabil Med. 2016;48(9):819–23. https://doi. org/10.2340/16501977-2134. 13. Nitsche MA, Seeber A, Frommann K, Klein CC, Rochford C, Nitsche MS, et al. Modulating parameters of excitability during and after transcranial direct current stimulation of the human motor cortex. J Physiol. 2005;568(1):291– 303. https://doi.org/10.1113/jphysiol.2005.092429. 32. Schabrun SM, Lamont RM, Brauer SG. Transcranial direct current stimulation to enhance dual-task gait training in Parkinson's disease: a pilot RCT. PLoS One. 2016;11(6). https://doi.org/10.1371/journal.pone.0158497. 14. Pereira JB, Junqué C, Bartrés-Faz D, Martí MJ, Sala-Llonch R, Compta Y, et al. Modulation of verbal fluency networks by transcranial direct current stimulation (tDCS) in Parkinson's disease. Brain Stimul. 2013;6(1):16–24. https://doi.org/10.1016/j.brs.2012.01.006. 33. Polanía R, Paulus W, Nitsche MA. Modulating cortico-striatal and thalamo- cortical functional connectivity with transcranial direct current stimulation. Hum Brain Mapp. 2012;33(10):2499–508. https://doi.org/10.1002/hbm.21380. 15. Obeso JA, Marin C, Rodriguez-Oroz C, Blesa J, Benitez-Temiño B, Mena- Segovia J, et al. The basal ganglia in Parkinson's disease: current concepts and unexplained observations. Ann Neurol. 2008;64:S30–46. 34. Dominey PF, Inui T. Cortico-striatal function in sentence comprehension: insights from neurophysiology and modeling. Cortex. 2009;45(8):1012–8. https://doi.org/10.1016/j.cortex.2009.03.007. 16. Wider C, Wszolek ZK. Etiology and pathophysiology of frontotemporal dementia, Parkinson disease and Alzheimer disease: lessons from genetic studies. Neurodegener Dis. 2008;5(3-4):122–5. https://doi.org/10.1159/ 000113680. 35. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med. 2009;151(4):264–9. https://doi.org/10.7326/0003-4819-151-4-2009081 80-00135. 17. Sabatini U, Boulanouar K, Fabre N, Martin F, Carel C, Colonnese C, et al. Cortical motor reorganization in akinetic patients with Parkinson's disease: a functional MRI study. Brain. References 1. Canning CG, Paul SS, Nieuwboer A. Prevention of falls in Parkinson's disease: a review of fall risk factors and the role of physical interventions. Neurodegener Dis Manag. 2014;4(3):203–21. 2. Jankovic J. Parkinson’s disease: clinical features and diagnosis. J Neurol Neurosurg Psychiatry. 2008;79(4):368–76. 3. Takakusaki K, Tomita N, Yano M. Substrates for normal gait and pathophysiology of gait disturbances with respect to the basal ganglia dysfunction. J Neurol. 2008;255(S4):19–29. https://doi.org/10.1007/s00415- 008-4004-7. 4. Takakusaki K. Functional neuroanatomy for posture and gait control. J Mov Disord. 2017;10(1):1–17. https://doi.org/10.14802/jmd.16062. 5. Rizos A, Martinez-Martin P, Odin P, Antonini A, Kessel B, Kozul TK, et al. Characterizing motor and non-motor aspects of early-morning off periods in Parkinson's disease: an international multicenter study. Parkinsonism Page 17 of 19 Page 17 of 19 Page 17 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 Pol et al. Translational Neurodegeneration Relat Disord. 2014;20(11):1231–5. https://doi.org/10.1016/j.parkreldis.2014. 09.013. Relat Disord. 2014;20(11):1231–5. https://doi.org/10.1016/j.parkreldis.2014. 09.013. 25. Kim YW, Shin IS, Im Moon H, Lee SC, Yoon SY. Effects of non-invasive brain stimulation on freezing of gait in parkinsonism: a systematic review with meta-analysis. Parkinsonism Relat Disord. 2019;64:82–9. 6. Nutt JG, Bloem BR, Giladi N, Hallett M, Horak FB, Nieuwboer A. Freezing of gait: moving forward on a mysterious clinical phenomenon. Lancet Neurol. 2011;10(8):734–44. https://doi.org/10.1016/S1474-4422(11)70143-0. 26. Costa-Ribeiro A, Maux A, Bosford T, Aoki Y, Castro R, Baltar A, et al. Transcranial direct current stimulation associated with gait training in Parkinson’s disease: a pilot randomized clinical trial. Dev Neurorehabil. 2017; 20(3):121–8. https://doi.org/10.3109/17518423.2015.1131755. 7. Fox SH, Katzenschlager R, Lim SY, Ravina B, Seppi K, Coelho M, et al. The Movement Disorder Society evidence-based medicine review update: treatments for the motor symptoms of Parkinson's disease. Mov Disord. 2011;26(S3):S2–S41. https://doi.org/10.1002/mds.23829. 27. da Silva DCL, Lemos T, de Sá FA, Horsczaruk CHR, Pedron CA, de Carvalho RE, et al. Effects of acute transcranial direct current stimulation on gait kinematics of individuals with Parkinson disease. To Geriatr Rehabi. 2018; 34(4):262–8. https://doi.org/10.1097/TGR.0000000000000203. p g 8. Baizabal-Carvallo JF, Jankovic J. Movement disorders induced by deep brain stimulation. Parkinsonism Relat Disord. 2016;25:1–9. https://doi.org/10.1016/j. parkreldis.2016.01.014. 28. de Paz RH, Serrano-Muñoz D, Pérez-Nombela S, Bravo-Esteban E, Avendaño- Coy J, Gómez-Soriano J. Combining transcranial direct-current stimulation with gait training in patients with neurological disorders: a systematic review. J Neuroengineering Rehab. 2019;16(1):114. https://doi.org/10.1186/ s12984-019-0591-z. 9. Nitsche M, Paulus W. Excitability changes induced in the human motor cortex by weak transcranial direct current stimulation. J Physiol. 2000;527(3): 633–9. References 2000;123(2):394–403. https://doi.org/10.1093/bra in/123.2.394. 36. Maher CG, Sherrington C, Herbert RD, Moseley AM, Elkins M. Reliability of the PEDro scale for rating quality of randomized controlled trials. Phys Ther. 2003;83(8):713–21. https://doi.org/10.1093/ptj/83.8.713. 18. Jahanshahi M, Jenkins IH, Brown RG, Marsden CD, Passingham RE, Brooks DJ. Self-initiated versus externally triggered movements: I. An investigation using measurement of regional cerebral blood flow with PET and movement-related potentials in normal and Parkinson's disease subjects. Brain. 1995;118(4):913–33. https://doi.org/10.1093/brain/118.4.913. 37. de Morton NA. The PEDro scale is a valid measure of the methodological quality of clinical trials: a demographic study. Aust J Physiother. 2009;55(2): 129–33. https://doi.org/10.1016/S0004-9514(09)70043-1. 38. Yotnuengnit P, Bhidayasiri R, Donkhan R, Chaluaysrimuang J, Piravej K. Effects of transcranial direct current stimulation plus physical therapy on gait in patients with Parkinson disease: a randomized controlled trial. Am J Phys Med Rehabil. 2018;97(1):7–15. https://doi.org/10.1097/PHM. 0000000000000783. 19. Fling BW, Cohen RG, Mancini M, Nutt JG, Fair DA, Horak FB. Asymmetric pedunculopontine network connectivity in parkinsonian patients with freezing of gait. Brain. 2013;136(8):2405–18. https://doi. org/10.1093/brain/awt172. 0000000000000783. Page 18 of 19 Page 18 of 19 Page 18 of 19 Page 18 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 Pol et al. Translational Neurodegeneration (2021) 10:22 44. Dagan M, Herman T, Harrison R, Zhou J, Giladi N, Ruffini G, et al. Multitarget transcranial direct current stimulation for freezing of gait in Parkinson's disease. Mov Disord. 2018;33(4):642–6. https://doi.org/10.1002/mds.27300. 63. Salehinejad MA, Ghanavati E, Rashid MHA, Nitsche MA. Hot and cold executive functions in the brain: A prefrontal-cingular network. Brain Neurosci Adv. 2021;5:23982128211007769. https://doi.org/10.1177/2398212 8211007769. 45. Lattari E, Costa SS, Campos C, de Oliveira AJ, Machado S, Neto GAM. Can transcranial direct current stimulation on the dorsolateral prefrontal cortex improves balance and functional mobility in Parkinson’s disease? Neurosci Lett. 2016;636:165–9. 64. 64. Maidan I, Nieuwhof F, Bernad-Elazari H, Reelick MF, Bloem BR, Giladi N, et al. The role of the frontal lobe in complex walking among patients with Parkinson’s disease and healthy older adults: an fNIRS study. Neurorehabil Neural Repair. 2016;30(10):963–71. https://doi.org/10.1177/1 545968316650426. 46. Lu C, Amundsen Huffmaster SL, Tuite PJ, MacKinnon CD. The effects of anodal tDCS over the supplementary motor area on gait initiation in Parkinson's disease with freezing of gait: a pilot study. J Neurol. 2018;265(9): 2023–32. https://doi.org/10.1007/s00415-018-8953-1. 65. Ranchet M, Hoang I, Cheminon M, Derollepot R, Devos H, Perrey S, et al. Changes in prefrontal cortical activity during walking and cognitive functions among patients with Parkinson's disease. Front Neurol. 2020;11: 601686. https://doi.org/10.3389/fneur.2020.601686. 47. Manenti R, Brambilla M, Rosini S, Orizio I, Ferrari C, Borroni B, et al. Time up and go task performance improves after transcranial direct current stimulation in patient affected by Parkinson's disease. Neurosci Lett. 2014; 580:74–7. https://doi.org/10.1016/j.neulet.2014.07.052. 66. Shine JM, Matar E, Ward PB, Frank MJ, Moustafa AA, Pearson M, et al. Freezing of gait in Parkinson’s disease is associated with functional decoupling between the cognitive control network and the basal ganglia. Brain. 2013;136(12):3671–81. https://doi.org/10.1093/brain/awt272. 48. Von Papen M, Fisse M, Sarfeld AS, Fink GR, Nowak DA. The effects of 1 Hz rTMS preconditioned by tDCS on gait kinematics in Parkinson's disease. J Neural Transm. 2014;121(7):743–54. https://doi.org/10.1007/ s00702-014-1178-2. 67. França C, de Andrade DC, Teixeira MJ, Galhardoni R, Silva V, Barbosa ER, et al. Effects of cerebellar neuromodulation in movement disorders: a systematic review. Brain Stimul. 2018;11(2):249–60. 49. Workman CD, Fietsam AC, Uc EY, Rudroff T. Cerebellar transcranial direct current stimulation in people with parkinson’s disease: a pilot study. Brain Sci. 0000000000000783. 2020;10(2):96. 68. Salehinejad MA, Nejati V, Mosayebi-Samani M, Mohammadi A, Wischnewski M, Kuo MF, et al. Transcranial direct current stimulation in ADHD: a systematic review of efficacy, safety, and protocol-induced electrical field modeling results. Neurosci Bull. 2020;36(10):1191–212. 50. Benninger DH, Lomarev M, Lopez G, Wassermann EM, Li X, Considine E, et al. Transcranial direct current stimulation for the treatment of Parkinson's disease. J Neurol Neurosurg Psychiatry. 2010;81(10):1105–11. https://doi. org/10.1136/jnnp.2009.202556. 69. Beretta VS, Vitório R, Nóbrega-Sousa P, Conceição NR, Orcioli-Silva D, Pereira MP, et al. Effect of different intensities of transcranial direct current stimulation on postural response to external perturbation in patients with Parkinson's disease. Neurorehabil Neural Repair. 2020;34(11):1009–19. https:// doi.org/10.1177/1545968320962513. 51. Criminger C, Swank C, Almutairi S, Mehta J. Transcranial direct current stimulation plus concurrent activity may influence task prioritization during walking in people with Parkinson's disease - initial findings. J Parkinsonism Restless Legs Syndrome. 2018;8:25–32. https://doi.org/1 0.2147/JPRLS.S161740. 70. Foerster Á, Dutta A, Kuo MF, Paulus W, Nitsche MA. Effects of anodal transcranial direct current stimulation over lower limb primary motor cortex on motor learning in healthy individuals. Eur J Neurosci. 2018;47(7):779–89. https://doi.org/10.1111/ejn.13866. 52. Swank C, Mehta J, Criminger C. Transcranial direct current stimulation lessens dual task cost in people with Parkinson’s disease. Neurosci Lett. 2016;626:1–5. https://doi.org/10.1016/j.neulet.2016.05.010. 71. Agboada D, Samani MM, Jamil A, Kuo MF, Nitsche MA. expanding the parameter space of anodal transcranial direct current stimulation of the primary motor cortex. Sci Rep. 2019;9:1–11. 53. Valentino F, Cosentino G, Brighina F, Pozzi NG, Sandrini G, Fierro B, et al. Transcranial direct current stimulation for treatment of freezing of gait: a cross-over study. Mov Disord. 2014;29(8):1064–9. https://doi.org/10.1002/ mds.25897. 72. Mosayebi-Samani M, Jamil A, Salvador R, Ruffini G, Haueisen J, Nitsche MA. The impact of individual electrical fields and anatomical factors on the neurophysiological outcomes of tDCS: a TMS-MEP and MRI study. Brain Stimul. 2021;14(2):316–26. https://doi.org/10.1016/j.brs.2021.01.016. 54. Kaski D, Dominguez RO, Allum JH, Islam AF, Bronstein AM. Combining physical training with transcranial direct current stimulation to improve gait in Parkinson's disease: a pilot randomized controlled study. Clin Rehabil. 2014;28(11):1115–24. https://doi.org/10.1177/0269215514534277. 73. Nitsche MA, Paulus W. Sustained excitability elevations induced by transcranial DC motor cortex stimulation in humans. Neurology. 2001;57(10): 1899–901. https://doi.org/10.1212/WNL.57.10.1899. 55. Nardone R, Versace V, Brigo F, Golaszewski S, Carnicelli L, Saltuari L, et al. Transcranial magnetic stimulation and gait disturbances in Parkinson's disease: a systematic review. Neurophysiol Clin. 2020;50(3):213–25. https:// doi.org/10.1016/j.neucli.2020.05.002. 74. 0000000000000783. Monte-Silva K, Kuo MF, Hessenthaler S, Fresnoza S, Liebetanz D, Paulus W, et al. Induction of late LTP-like plasticity in the human motor cortex by repeated non-invasive brain stimulation. Brain Stimul. 2013;6(3):424–32. 56. Li Y, Fan J, Yang J, He C, Li S. Effects of transcranial direct current stimulation on walking ability after stroke: a systematic review and meta- analysis. Restor Neurol Neurosci. 2018;36(1):59–71. https://doi.org/10.3233/ RNN-170770. 75. Agboada D, Mosayebi-Samani M, Kuo MF, Nitsche MA. Induction of long- term potentiation-like plasticity in the primary motor cortex with repeated anodal transcranial direct current stimulation - better effects with intensified protocols? Brain Stimul. 2020;13(4):987–97. https://doi.org/10.1016/j.brs.2020. 04.009. 57. Wennberg AMV, Savica R, Mielke MM. Association between various brain pathologies and gait disturbance. Dement Geriatr Cogn Disord. 2017;43(3-4): 128–43. https://doi.org/10.1159/000456541. 76. Reis J, Schambra HM, Cohen LG, Buch ER, Fritsch B, Zarahn E, et al. Noninvasive cortical stimulation enhances motor skill acquisition over multiple days through an effect on consolidation. Proc Natl Acad Sci U S A. 2009;106(5):1590–5. https://doi.org/10.1073/pnas.0805413106. 58. Jeffery DT, Norton JA, Roy FD, Gorassini MA. Effects of transcranial direct current stimulation on the excitability of the leg motor cortex. Exp Brain Res. 2007;182(2):281–7. https://doi.org/10.1007/s00221-007-1093-y. 77. Polanía R, Nitsche MA, Paulus W. Modulating functional connectivity patterns and topological functional organization of the human brain with transcranial direct current stimulation. Hum Brain Mapp. 2011;32(8):1236–49. https://doi.org/10.1002/hbm.21104. 59. Kaski D, Quadir S, Patel M, Yousif N, Bronstein AM. Enhanced locomotor adaptation aftereffect in the “broken escalator” phenomenon using anodal tDCS. J Neurophysiol. 2012;107(9):2493–505. https://doi.org/10.1152/jn. 00223.2011. 78. Allman C, Amadi U, Winkler AM, Wilkins L, Filippini N, Kischka U, et al. Ipsilesional anodal tDCS enhances the functional benefits of rehabilitation in patients after stroke. Sci Transl Med. 2016;8:330re331. 60. Lee SY, Kim MS, Chang WH, Cho JW, Youn JY, Kim YH. Effects of repetitive transcranial magnetic stimulation on freezing of gait in patients with parkinsonism. Restor Neurol Neurosci. 2014;32(6):743–53. https://doi.org/1 0.3233/RNN-140397. 79. Stagg C, Jayaram G, Pastor D, Kincses Z, Matthews P, Johansen-Berg H. Polarity and timing-dependent effects of transcranial direct current stimulation in explicit motor learning. Neuropsychologia. 2011;49(5):800–4. https://doi.org/10.1016/j.neuropsychologia.2011.02.009. 61. Malouin F, Richards CL, Jackson PL, Dumas F, Doyon J. Brain activations during motor imagery of locomotor-related tasks: a PET study. Hum Brain Mapp. 2003;19(1):47–62. https://doi.org/10.1002/hbm.10103. 80. Nitsche MA, Lampe C, Antal A, Liebetanz D, Lang N, Tergau F, et al. Dopaminergic modulation of long-lasting direct current-induced cortical excitability changes in the human motor cortex. 0000000000000783. 39. Mishra RK, Thrasher AT. Transcranial direct current stimulation of dorsolateral prefrontal cortex improves dual-task gait performance in patients with Parkinson’s disease: a double blind, sham-controlled study. Gait Posture. 2021;84:11–6. https://doi.org/10.1016/j.gaitpost.2020.11.012. 20. Fling BW, Cohen RG, Mancini M, Carpenter SD, Fair DA, Nutt JG, et al. Functional reorganization of the locomotor network in Parkinson patients with freezing of gait. PLoS One. 2014;9(6):e100291. https://doi.org/10.1371/ journal.pone.0100291. 40. Hadoush H, Al-Jarrah M, Khalil H, Al-Sharman A, Al-Ghazawi S. Bilateral anodal transcranial direct current stimulation effect on balance and fearing of fall in patient with Parkinson's disease. Neurorehabilitation. 2018;42(1):63– 8. https://doi.org/10.3233/NRE-172212. 21. Playford E, Jenkins I, Passingham R, Nutt J, Frackowiak R, Brooks D. Impaired mesial frontal and putamen activation in Parkinson's disease: a positron emission tomography study. Ann Neurol. 1992;32(2):151–61. https://doi. org/10.1002/ana.410320206. 22. Peppe A, Chiavalon C, Pasqualetti P, Crovato D, Caltagirone C. Does gait analysis quantify motor rehabilitation efficacy in Parkinson's disease patients? Gait Posture. 2007;26(3):452–62. https://doi.org/10.1016/j.gaitpost.2 006.11.207. 41. Ricci M, Di Lazzaro G, Pisani A, Scalise S, Alwardat M, Salimei C, et al. Wearable electronics assess the effectiveness of transcranial direct current stimulation on balance and gait in Parkinson’s disease patients. Sensors. 2019;19(24):5465. https://doi.org/10.3390/s19245465. 23. Tanaka T, Takano Y, Tanaka S, Hironaka N, Kobayashi K, Hanakawa T, et al. Transcranial direct-current stimulation increases extracellular dopamine levels in the rat striatum. Front Syst Neurosci. 2013;7:6. 24. Beretta VS, Conceição NR, Nóbrega-Sousa P, Orcioli-Silva D, Dantas LKBF, Gobbi LTB, et al. Transcranial direct current stimulation combined with physical or cognitive training in people with Parkinson’s disease: a systematic review. J Neuroengineering Rehab. 2020;17:1–15. 42. Bueno MEB, do Nascimento Neto LI, Terra MB, Barboza NM, Okano AH, Smaili SM. Effectiveness of acute transcranial direct current stimulation on non-motor and motor symptoms in Parkinson's disease. Neurosci Lett. 2019;696:46–51. 23. Tanaka T, Takano Y, Tanaka S, Hironaka N, Kobayashi K, Hanakawa T, et al. Transcranial direct-current stimulation increases extracellular dopamine levels in the rat striatum. Front Syst Neurosci. 2013;7:6. 43. Chang WH, Kim MS, Park E, Cho JW, Youn J, Kim YK, et al. Effect of dual- mode and dual-site noninvasive brain stimulation on freezing of gait in patients with Parkinson disease. Arch Phys Med Rehabil. 2017;98(7):1283–90. https://doi.org/10.1016/j.apmr.2017.01.011. 24. Beretta VS, Conceição NR, Nóbrega-Sousa P, Orcioli-Silva D, Dantas LKBF, Gobbi LTB, et al. Transcranial direct current stimulation combined with physical or cognitive training in people with Parkinson’s disease: a systematic review. J Neuroengineering Rehab. 2020;17:1–15. 0000000000000783. Eur J Neurosci. 2006;23(6): 1651–7. https://doi.org/10.1111/j.1460-9568.2006.04676.x. 62. Hone-Blanchet A, Edden RA, Fecteau S. Online effects of transcranial direct current stimulation in real time on human prefrontal and striatal metabolites. Biol Psychiatry. 2016;80(6):432–8. https://doi.org/10.1016/j. biopsych.2015.11.008. Page 19 of 19 Page 19 of 19 Pol et al. Translational Neurodegeneration (2021) 10:22 Pol et al. Translational Neurodegeneration (2021) 10:22 81. Ueki Y, Mima T, Kotb MA, Sawada H, Saiki H, Ikeda A, et al. Altered plasticity of the human motor cortex in Parkinson's disease. Ann Neurol. 2006;59(1): 60–71. https://doi.org/10.1002/ana.20692. 82. Fresnoza S, Stiksrud E, Klinker F, Liebetanz D, Paulus W, Kuo MF, et al. Dosage-dependent effect of dopamine D2 receptor activation on motor cortex plasticity in humans. J Neurosci. 2014;34(32):10701–9. https://doi. org/10.1523/JNEUROSCI.0832-14.2014. 83. Monte-Silva K, Liebetanz D, Grundey J, Paulus W, Nitsche MA. Dosage- dependent non-linear effect of L-dopa on human motor cortex plasticity. J Physiol. 2010;588(18):3415–24. https://doi.org/10.1113/jphysiol.2010.190181. 84. Mollinedo I, Ma Cancela J. Evaluation of the psychometric properties and clinical applications of the timed up and go test in Parkinson disease: a systematic review. J Exerc Rehabil. 2020;16(4):302–12. https://doi.org/10.12 965/jer.2040532.266. 85. Morris S, Morris ME, Iansek R. Reliability of measurements obtained with the timed “up & go” test in people with Parkinson disease. Phys Ther. 2001; 81(2):810–8. https://doi.org/10.1093/ptj/81.2.810. 86. Muthukrishnan N, Abbas JJ, Shill HA, Krishnamurthi N. Cueing paradigms to improve gait and posture in Parkinson's disease: a narrative review. Sensors (Basel). 2019;19(24):5468. https://doi.org/10.3390/s19245468.
https://openalex.org/W3120575352	https://www.shs-conferences.org/articles/shsconf/pdf/2021/03/shsconf_glob20_08022.pdf	English	null	Energy saving as a necessary condition for the development of a green economy in the context of globalization and transformation of society	SHS web of conferences	2,021	cc-by	6,646	Abstract. Research background: The world is actively transforming socio-economic and production processes. Ensuring economic growth today is accompanied by an increase in living standards and environmental degradation, the exhaustion of natural resources, imbalance in the biosphere, and climate change, which leads to poor human health and limits the possibilities for further development. All this is due to the need to transform both the provision of technological progress for economic development, and the favouring of the natural environment in the context of globalization. Purpose of the article: The purpose of the article is to study energy conservation as a necessary condition for the development of a "green" Purpose of the article: The purpose of the article is to study energy conservation as a necessary condition for the development of a "green" economy in the context of globalization and transformation of society. y p g economy in the context of globalization and transformation of society. Methods: The research is based on general scientific methods of empirical research (observation, measurement, experiment, modelling), analysis and synthesis, analogy, systematization, as well as methods of structural-logical, statistical analysis. Findings & Value added: The article systematizes approaches to defining the concept of a green economy, reveals the relationship between the categories of politics and society, and elements of the green economy. The efficiency of the use of fuel and energy resources is determined by the energy intensity of GDP. In this regard, the predicted dynamics of energy intensity of GDP by regions of the world, energy consumption per capita in the world and groups of countries are analysed, measures of state support in the field of energy conservation and energy efficiency are substantiated. Keywords: energy conservation; energy efficiency; green economy; transformation. JEL Classification: F63; Q43; Q48 SHS Web of Conferences 92, 0 (2021) Globalization and its Socio-Economic Consequences 2020 8022 SHS Web of Conferences 92, 0 (2021) Globalization and its Socio-Economic Consequences 2020 8022 SHS Web of Conferences 92, 0 (2021) Globalization and its Socio-Economic Consequences 2020 8022 SHS Web of Conferences 92, 0 (2021) 8022 https://doi.org/10.1051/shsconf/20219208022 Energy saving as a necessary condition for the development of a green economy in the context of globalization and transformation of society Yulia Vertakova1,, Olga Kryzhanovskaya1 1South-West State University, Department of Regional Economics and Management, 305040, 94, 50 let Oktyabrya str., Kursk, Russia ia Vertakova , Olga Kryzhanovskaya uth-West State University, Department of Regional Economics and Management, 305040, 94, 50 Oktyabrya str., Kursk, Russia Corresponding author: vertakova7@yandex.ru 1 Introduction Ensuring economic growth today is associated with environmental pollution and deterioration of natural resources, imbalance in the biosphere, climate change, which leads to a deterioration of human health and limits development opportunities. This means that solving the important problem of improving the well-being of the population does not provide the required quality of life. All this determines the essence of modernization as ensuring technological progress for economic development and maintaining a favorable natural environment (environmental safety, which becomes decisive for economic growth and human existence itself). This task is voiced in the world as the implementation of the principle of "decoupling" (meeting growing needs while minimizing the depletion of natural capital), which implies a decrease in energy intensity and nature intensity in general economic growth, widespread use of renewable energy sources, modernization of production based on innovation. This direction determines the priorities of the country's economic development today. According to the UNEP reports (United Nations Environment Program) [1], a green economy is defined as an economy that enhances human well-being and social justice. Thus, it is necessary to study the prospects for the development of a green economy, as well as the possibility of improving the quality of life and modernizing existing production. p y p g q y g g p In the context of globalization and transformation, the whole world faces many new “challenges”. Related to this is the need to modernize the economy, including innovative development and energy efficiency. The main goal of all measures aimed at modernizing the economy is to improve the living conditions of every person today and to ensure favorable conditions for future generations. One of the newest models of economic development as a response to the challenge of natural and economic global transformation was formulated in 1992 in Rio de Janeiro at the World Conference on Environmental Development. It is called sustainable development. The essence of sustainable development is to ensure such economic growth, which makes it possible to harmonize the relationship "man-nature (environment)" and preserve the natural environment for current and future generations. Many countries are actively developing long-term programs within the framework of the concept of sustainable development. At the same time, an urgent task is to identify the main problems associated with the development of cities, energy, water, food and ecosystems, as well as the formation of principles and methods of state regulation of these processes. * Corresponding author: vertakova7@yandex.ru © The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (http://creativecommons.org/licenses/by/4.0/). https://doi.org/10.1051/shsconf/20219208022 SHS Web of Conferences 92, 0 (2021) 8022 Globalization and its Socio-Economic Consequences 2020 Globalization and its Socio-Economic Consequences 2020 2 Methods The United Nations Environment Program “Towards a Green Economy” synthesis report for policymakers sees a green economy as an economy that enhances human well-being and social justice, while significantly reducing environmental risks and impoverishment. It is a low carbon, resource efficient and socially inclusive economy. In the annual UNEP-2009 report, a green economy is “a system of economic activities related to the production, distribution and consumption of goods and services that leads to improved human well-being in the long term, without exposing future generations to significant environmental risks or environmental deficits. The South African Government Strategy notes that a green economy is one that includes new economic activities. It should be the starting point for broad participation of the black majority in the economy, and should also meet the needs of women, youth, and entrepreneurs and open up opportunities for entrepreneurship in the social economy. The second session of the UN Preparatory Committee for the Conference on Sustainable Development on March 7-8, 2011 sees the green economy as a concept that brings together a set of measures to attract investment in environmentally significant industries that contribute to sustainable development and poverty eradication. Thus, a green economy is described as an economy in which economic growth and environmental responsibility mutually reinforce each other, while supporting progress in social development. The terms «green economy», «green growth» and «low-carbon development» are often used interchangeably and are applied in different contexts to different industries, resources, areas, and even concepts. The main driving force behind the development of the latter was the development of an integrated and holistic approach to integrating environmental issues into the spheres of economic policy and planning. The relationship between different categories of politics and society, as well as elements of a green economy (green growth, green taxes and accounts, green innovation, a green new deal) are shown in Fig. 1. Based on the above, it can be concluded that the green economy not only values natural capital, but also invests in it. The green economy is replacing fossil fuels with clean energy and low-carbon technologies, reducing climate impacts while creating decent jobs and reducing import dependency. New technologies that drive energy and resource efficiency are opening up opportunities for growth in new directions, compensating job losses in the brown economy. 1 Introduction This task is defined as ensuring sustainable development based on the principles of the green economy [2]. g p p p g y [ ] The problem of the emergence and development of the green economy dates back to the 70s XX century. Scientists all over the world wrote about the need for a new format of the economy in the context of the development of world industry. One of the first was Frances Moore Lappé, an American researcher and author in the field of food and democratic policy ("Diet for a Small Planet"), where it was stated that world hunger is caused not by food shortages, but by ineffective food policies, as well as by inefficient distribution of money and energy resources. In 2012 she also published an article “Our Challenge - Developing an Eco- Mind” in the scientific journal “Green economy in action”, in which she pointed out the importance of developing eco-thinking among people around the world, because the green economy simply cannot survive in the face of growing goods - and energy consumption [3]. Jeremy Rifkin in his book "The Third Industrial Revolution" says that now fundamental economic changes are taking place in conjunction with the introduction of elements of the green economy [4]. Molly Scott Kato, British green politician, academic, environmental and social activist, argues that the world as we know it needs a new economy. Climate change, financial crisis and uncontrolled globalization are all major problems, rooted in the dominant economic system. The green economy argues that society must be embedded in the ecosystem, and markets and the economy as a whole must respond to rapidly emerging and 2 2 SHS Web of Conferences 92, 0 (2021) 8022 SHS Web of Conferences 92, 0 (2021) Globalization and its Socio-Economic Consequences 2020 8022 https://doi.org/10.1051/shsconf/20219208022 SHS Web of Conferences 92, 0 (2021) Globalization and its Socio-Economic Consequences 2020 8022 changing social and environmental priorities [5]. Among the Russian scientists studying the problems of the green economy, as well as the problems of energy conservation as a necessary condition for the development of the green economy, one can name Vertakova Yu. [6], Plotnikov V. [7], Babich T. [8], Zlobina N., Merkulova E., Muratova O. [9], Kondrakov O.,etc. changing social and environmental priorities [5]. 1 Introduction Among the Russian scientists studying the problems of the green economy, as well as the problems of energy conservation as a necessary condition for the development of the green economy, one can name Vertakova Yu. [6], Plotnikov V. [7], Babich T. [8], Zlobina N., Merkulova E., Muratova O. [9], Kondrakov O.,etc. 3 Results and discussions Energy problems have always been at the center of increased public attention. Energy production in the world is constantly growing, doubling every 5-7 years. Modern energy is 80% fuel that means it uses fossil carbon fuels. Thermal waste from such energy is considered the most dangerous for the biosphere and humanity. The energy sector must ensure the safe use of traditional types of resources and increase energy efficiency. The 2011 United Nations General Assembly launched the “Sustainable Energy for All” initiative. This initiative is addressed to business, governments, investors, communities, and academia and aims to achieve three main goals by 2030: ensuring universal access to modern energy services; reducing the intensity of global energy consumption by 40%; increasing the share of renewable energy sources in the world up to 30%. Fig. 1. Elements of the Green Economy Source: Compiled by the authors At the end of 2012 a resolution was adopted declaring 2014–2024 a Decade of Sustainable Energy for All. The UN General Assembly has developed recommendations to governments to create favorable conditions for the promotion and use of new and renewable energy sources and more efficient energy consumption. The energy sector can become a major one in the transition to a green economy. The Covid-19 pandemic is a global crisis. As a result of restrictions imposed in many countries to slow the spread of the virus, the share of energy consumption has changed significantly (Table 1). In the first quarter of 2020, global energy demand declined by 3.8%, with most of the impacts being felt in March when restrictive measures were introduced in Europe, North America and other countries. Global demand for coal was hit hardest, falling nearly 8% from Fig. 1. Elements of the Green Economy Source: Compiled by the authors Fig. 1. Elements of the Green Economy Source: Compiled by the authors At the end of 2012 a resolution was adopted declaring 2014–2024 a Decade of Sustainable Energy for All. The UN General Assembly has developed recommendations to governments to create favorable conditions for the promotion and use of new and renewable energy sources and more efficient energy consumption. The energy sector can become a major one in the transition to a green economy. The Covid-19 pandemic is a global crisis. As a result of restrictions imposed in many countries to slow the spread of the virus, the share of energy consumption has changed significantly (Table 1). 2 Methods Improving resource efficiency - the efficiency of using both electricity and raw materials - is manifested everywhere, including in improving the waste management system, strengthening the role of public transport, green building and reducing food waste along the entire food production and consumption chain. The research is based on general scientific methods of empirical research (observation, measurement, experiment, modelling), analysis and synthesis, analogy, systematization, as well as methods of structural-logical, statistical analysis. 3 3 https://doi.org/10.1051/shsconf/20219208022 SHS Web of Conferences 92, 0 (2021) 8022 Globalization and its Socio-Economic Consequences 2020 3 Results and discussions Demand for various types of energy in the world 2015-2020† World 2015 2016 2017 2018 2019 2020 Total (billion $) 2063 1928 1912 1905 1891 1520 Supply (by type) 1793 1634 1632 1624 1611 1273 Fossil fuels (fuel and electricity supply) 1150 972 978 975 976 699 Renewable energy sources 317 321 319 317 319 288 Power grids 296 306 298 294 273 248 Others 31 35 37 37 43 39 Fuel (total) 1009 835 850 854 854 595 Fossil fuel 1000 826 841 845 846 588 Oil 553 441 462 477 470 325 Gas 338 293 295 290 286 186 Coal 109 92 84 78 90 76 Liquid biofuel and biogas 9 9 9 9 8 7 Energy (total) 784 799 782 769 757 678 Coal 75 68 63 60 57 50 Gas and oil 74 78 74 70 74 61 Nuclear energy 29 33 34 33 39 35 Renewable energy sources 308 312 310 308 311 281 very significant change, given that the last sharp drop in global gas demand was recorded in 2009, when it fell by 2% [10]. very significant change, given that the last sharp drop in global gas demand was recorded in 2009, when it fell by 2% [10]. The baseline forecast for 2020 is a large-scale global recession caused by long-term restrictions on mobility and socioeconomic activity. With the gradual reopening of currently isolated economies, the recovery is U-shaped and accompanied by a significant permanent loss of economic activity. The global gross domestic product (GDP) is projected to decline by 6% in 2020, the highest percentage in 70 years and the highest ever recorded in absolute terms. The impact of Covid 19 on energy demand in 2020 will be more than seven times greater than the impact of the 2008 financial crisis on global energy demand. Global CO2 emissions are expected to decline by 8%, or nearly 2.6 gigatons (GT), which is equal to 10 years ago level. This annual reduction will be the largest ever; six times more than the previous record of 0.4 GT reduction in 2009 caused by the global financial crisis, and double the cumulative sum of all previous reductions since the end of World War II. †IEA - International Energy Agency https://www.iea.org/ 3 Results and discussions In the first quarter of 2020, global energy demand declined by 3.8%, with most of the impacts being felt in March when restrictive measures were introduced in Europe, North America and other countries. Global demand for coal was hit hardest, falling nearly 8% from the first quarter of 2019. Oil demand was also hit hard, falling nearly 5% in the first quarter, mainly due to declines in mobility and aviation, which account for nearly 60% of global oil demand. By the end of March, global road traffic activity was almost 50% below the 2019 average, and aviation was 60% below. The demand for electricity was significantly reduced (by 20%) as a result of restrictive measures, which negatively affected the balance of electricity (the share of supplied renewable energy sources increased significantly). The demand fell for all other sources: gas and nuclear energy, except for renewable sources (the latter became the only object of growth in demand). However, the pandemic had the least impact on gas - the drop in demand for this type of fuel was the smallest - about 2%. The International Energy Agency expects a further drop in global gas demand to 4-6%. This is a 4 https://doi.org/10.1051/shsconf/20219208022 SHS Web of Conferences 92, 0 (2021) 8022 Globalization and its Socio-Economic Consequences 2020 very significant change, given that the last sharp drop in global gas demand was recorded in 2009, when it fell by 2% [10]. Table 1. 3 Results and discussions However, as in previous crises, the rebound in emissions may be greater than the recession, unless the wave of investment to restart the economy is directed towards cleaner and more sustainable energy infrastructure [11]. Thus, the pandemic has made “its own” adjustments to the global economy, challenging all energy sectors. One can only hope that its consequences will not be as dire as the World Economic Agency predicts. Therefore, the energy component of the final product began to play a dominant role, as a result of which the problem of efficient use of energy has become one of the top priorities. The efficiency of using fuel and energy resources (FER) is determined by the energy intensity of GDP, which is used to assess the energy efficiency of national economies. Energy intensity of GDP determines the unit costs of fuel and energy resources per unit of manufactured products - it is a generalized indicator of the level of consumption of energy resources per unit of GDP. Usually, the energy intensity of GDP is considered as the ratio of the gross consumption of fuel and energy resources to the volume of GDP. This means that the higher the energy intensity in a country, the lower the energy efficiency.. 5 SHS Web of Conferences 92, 0 (2021) 8022 SHS Web of Conferences 92, 0 (2021) 8022 https://doi.org/10.1051/shsconf/20219208022 Globalization and its Socio-Economic Consequences 2020 Globalization and its Socio-Economic Consequences 2020 Figure 2 shows the forecast dynamics of GDP by regions of the world. The figure shows the principle of the development of scientific and technological progress, which consists in a gradual reduction in the cost of new technologies, as well as in maintaining the existing trends in reducing the energy intensity of the GDP of countries and regions with a tendency towards their convergence by the end of the forecast period. The absence of particularly significant technological revolutions and breakthroughs is assumed. Energy intensity is projected to decrease by 44% for the period from 2014 to 2040. For the world as a whole, the projected energy intensity of GDP for 2040 is 0.09 toe / thousand dollars [12]. Another equally important task of the world energy sector is to reduce the level of energy consumption. ‡ Forecast of energy development https://www.eriras.ru/files/forecast_2040.pdf in the world and Russia until 2040 ‡ Forecast of energy development in the world and Russia until 2040 https://www.eriras.ru/files/forecast_2040.pdf § Forecast of energy development https://www.eriras.ru/files/forecast_2040.pdf 3 Results and discussions A sustained effort will be required from politicians and their electorate to rethink and redefine traditional measures of wealth, prosperity and well-being. However, the biggest risk today is the risk of maintaining the status quo. These issues are institutionalized in various countries of the world (Table 3). Table 2. Primary Energy Consumption by Regions of the World, Base Scenario§ Consumption of primary energy resources, million toe Energy consumption growth rates,% 2010 2012 2020 2025 2030 2035 2040 2010- 2020 2010- 2030 2010- 2040 North America 2699 2738 2832 2901 2947 2966 2964 0,5 0,4 0,3 USA 2261 2273 2340 2385 2409 2411 2395 0,3 0,3 0,2 Europe 2020 1956 1983 2003 2014 2018 2016- -0,2 0,0 0,0 EU-28 1814 1731 1741 1743 1737 1725 1708 -0,4 -0,2 -0,2 Developed Asia 918 900 910 913 908 897 880 -0,1 -0,1 -0,1 Japan 509 462 445 426 405 384 362 -1,3 -1,1 -1,1 CIS 1047 1094 1150 1201 1239 1271 13080 0,9 0,8 0,7 Russia 718 752 789 822 845 866 892 0,9 0,8 0,7 Developing Asia 4187 5021 5744 6402 7008 7577 8094 3,2 2,6 2,2 China 2676 3309 3775 4157 4473 4751 4985 3,5 2,6 2,1 India 727 835 977 1137 1313 1497 1681 3,0 3,0 2,8 South and Central America 637 695 767 840 912 981 1047 1,9 1,8 1,7 Brazil 269 295 330 366 403 439 476 2,1 2,0 1,9 Middle East 698 766 856 941 1020 1097 1172 2,1 1,9 1,7 Africa 704 789 887 993 1103 1217 1334 2,3 2,3 2,2 World 12911 13970 15130 16194 17150 18024 18815 1,6 1,4 1,3 Table 2. Primary Energy Consumption by Regions of the World, Base Scenario§ In general, according to the table, it is seen that in many regions of the world a decrease in the volume of consumed primary energy is predicted. This demonstrates the effectiveness of the measures taken to develop the energy saving sphere. Thus, the strategic goal of energy conservation is to reduce energy intensity and increase the energy efficiency of the national economy to the level of developed countries of the world. A prerequisite for this is the use of approaches to energy conservation approved in advanced countries. The energy conservation policy, proceeding consistently, allows freeing up significant resources in the national economy, since the measures taken to implement it require relatively small costs. 3 Results and discussions Forecasting energy consumption is based on the methodology, which consists in the mutual agreement of the dynamics of consumption by countries, obtained by the "demographic method" (in terms of population and per capita energy consumption, which has multidirectional dynamics in the OECD countries, which will demonstrate a decrease in per capita energy consumption, while China and others developing countries, on the contrary, will increase this indicator (Fig. 2) using the “economic method” (based on GDP per capita and GDP energy intensity). Fig. 2. Energy consumption per capita by world and country groups, Base scenario‡ The forecast of primary energy consumption in the world obtained in this way in 2010– 2040 shows an increase of 46% (or an average of 1.3% annually). The distribution of energy consumption in the world is changing noticeably: with the growth of the population in developing countries, there is an increasingly active shift there of economic growth and energy consumption centers. And developed countries, due to active energy conservation, will increase their consumption by only 4.6% by 2040, while all this growth will take place until 2030, and then the growth in energy demand in the OECD will practically stop [12]. Fig. 2. Energy consumption per capita by world and country groups, Base scenario‡ Fig. 2. Energy consumption per capita by world and country groups, Base scenario‡ The forecast of primary energy consumption in the world obtained in this way in 2010– 2040 shows an increase of 46% (or an average of 1.3% annually). The distribution of energy consumption in the world is changing noticeably: with the growth of the population in developing countries, there is an increasingly active shift there of economic growth and energy consumption centers. And developed countries, due to active energy conservation, will increase their consumption by only 4.6% by 2040, while all this growth will take place until 2030, and then the growth in energy demand in the OECD will practically stop [12]. 6 https://doi.org/10.1051/shsconf/20219208022 SHS Web of Conferences 92, 0 (2021) 8022 Globalization and its Socio-Economic Consequences 2020 Table 2. 3 Results and discussions Primary Energy Consumption by Regions of the World, Base Scenario§ Consumption of primary energy resources, million toe Energy consumption growth rates,% 2010 2012 2020 2025 2030 2035 2040 2010- 2020 2010- 2030 2010- 2040 North America 2699 2738 2832 2901 2947 2966 2964 0,5 0,4 0,3 USA 2261 2273 2340 2385 2409 2411 2395 0,3 0,3 0,2 Europe 2020 1956 1983 2003 2014 2018 2016- -0,2 0,0 0,0 EU-28 1814 1731 1741 1743 1737 1725 1708 -0,4 -0,2 -0,2 Developed Asia 918 900 910 913 908 897 880 -0,1 -0,1 -0,1 Japan 509 462 445 426 405 384 362 -1,3 -1,1 -1,1 CIS 1047 1094 1150 1201 1239 1271 13080 0,9 0,8 0,7 Russia 718 752 789 822 845 866 892 0,9 0,8 0,7 Developing Asia 4187 5021 5744 6402 7008 7577 8094 3,2 2,6 2,2 China 2676 3309 3775 4157 4473 4751 4985 3,5 2,6 2,1 India 727 835 977 1137 1313 1497 1681 3,0 3,0 2,8 South and Central America 637 695 767 840 912 981 1047 1,9 1,8 1,7 Brazil 269 295 330 366 403 439 476 2,1 2,0 1,9 Middle East 698 766 856 941 1020 1097 1172 2,1 1,9 1,7 Africa 704 789 887 993 1103 1217 1334 2,3 2,3 2,2 World 12911 13970 15130 16194 17150 18024 18815 1,6 1,4 1,3 In general, according to the table, it is seen that in many regions of the world a decrease in the volume of consumed primary energy is predicted. This demonstrates the effectiveness of the measures taken to develop the energy saving sphere. Thus, the strategic goal of energy conservation is to reduce energy intensity and increase the energy efficiency of the national economy to the level of developed countries of the world. A prerequisite for this is the use of approaches to energy conservation approved in advanced countries. The energy conservation policy, proceeding consistently, allows freeing up significant resources in the national economy, since the measures taken to implement it require relatively small costs. For a successful transition to a green economy, every country needs to create an enabling environment and adequate funding, however both are achievable. Green economy can deliver the same growth and employment as brown economy in the medium to long term, while delivering more environmental and social benefits [13]. Thus, the transition to a green economy will require the concerted efforts of world leaders, civil society and leading companies. § Forecast of energy development in the world and Russia until 2040 https://www.eriras.ru/files/forecast_2040.pdf in the world and Russia until 2040 3 Results and discussions For a successful transition to a green economy, every country needs to create an enabling environment and adequate funding, however both are achievable. Green economy can deliver the same growth and employment as brown economy in the medium to long term, while delivering more environmental and social benefits [13]. Thus, the transition to a green economy will require the concerted efforts of world leaders, civil society and leading companies. A sustained effort will be required from politicians and their electorate to rethink and redefine traditional measures of wealth, prosperity and well-being. However, the biggest risk today is the risk of maintaining the status quo. These issues are institutionalized in various countries of the world (Table 3). in the world and Russia until 2040 7 SHS Web of Conferences 92, 0 (2021) 8022 https://doi.org/10.1051/shsconf/20219208022 Globalization and its Socio-Economic Consequences 2020 Table 3. Examples of institutional consolidation of the green economy principles Country (group of countries) Measures EU In 2008, the EU Renewable Energy Directive was adopted. It provides for an increase in the share of renewable energy sources in the total volume of electricity production by 2020 from the current 7.8% to 20%; USA The American Recovery and Reinvestment Act of 2009 provides for an increase in this indicator from 3.1% in 2007 to 10% in 2012 and up to 25% in 2020. Brazil 2/3 of renewable energy comes from biomass, which is used to produce ethanol, which covers 40% of the demand for motor fuel, and 1/3 from hydroelectric power plants, which are the basis of the national electricity industry (almost 76% of electricity production). Brazil's National Policy on Climate Change (Law 12.187 / 2009) envisages further accelerated development of low-carbon energy and economy in the period up to 2030, including an increase in ethanol production and export. India Taxes on cleantech components have been abolished and the intention to set fines for delays in solar installation has been officially announced. China The government has achieved production from renewable sources (excluding hydroelectric power plants) 3% of electricity in 2010, and by 2020 the goal is to bring this figure to 8%. In accordance with the XII Five-Year Development Plan of the country, the energy intensity of GDP should decrease in 2020 - by 40- 45%, and in 2050 - by 60-68%. Table 3. Examples of institutional consolidation of the green economy principles 4 Conclusion The Law "On Rational Use of Energy" includes 3 principles: The Law "On Rational Use of Energy" includes 3 principles: - ensuring energy security, which includes measures to save resources, diversify the sup - ensuring energy security, which includes measures to save resources, diversify the supply of imported energy resources and develop relations with the main exporting countries, increase self-sufficiency in energy, create strategic reserves of oil, oil products and natural gas; g - ensuring environmental protection, including measures to reduce greenhouse gas emissions, increase the use of alternative energy sources, develop and introduce resource- saving and energy-efficient technologies; - ensuring environmental protection, including measures to reduce greenhouse gas emissions, increase the use of alternative energy sources, develop and introduce resource- saving and energy-efficient technologies; implementation of energy policy based on market mechanisms g gy g ; implementation of energy policy based on market mechanisms. g gy g ; implementation of energy policy based on market mechanisms. By setting a 30% increase in energy efficiency by 2030 over 2006, the Japan government is committed to ensuring a modern energy supply / demand structure in a mar with the high prices expected by the government in the medium to long term. In April 20 By setting a 30% increase in energy efficiency by 2030 over 2006, the Japanese government is committed to ensuring a modern energy supply / demand structure in a market with the high prices expected by the government in the medium to long term. In April 2009, the Government of Japan, relying on the strategy of economic development and measures to y g gy y y , p government is committed to ensuring a modern energy supply / demand structure in a market with the high prices expected by the government in the medium to long term. In April 2009, the Government of Japan, relying on the strategy of economic development and measures to overcome the economic crisis, approved a concept to reduce CO2 emissions. The specific goal of the concept is to increase the share of renewable sources in energy consumption by 2 times and achieve the highest indicator in the world - 20%. the Government of Japan, relying on the strategy of economic development and measures to overcome the economic crisis, approved a concept to reduce CO2 emissions. 4 Conclusion - attract investments in developments to improve energy efficiency with high implementation risks; - attract investments in developments to improve energy efficiency with high implementation risks; 8 SHS Web of Conferences 92, 0 (2021) 8022 https://doi.org/10.1051/shsconf/20219208022 Globalization and its Socio-Economic Consequences 2020 Globalization and its Socio-Economic Consequences 2020 Globalization and its Socio-Economic Consequences 2020 - promote the introduction of clean energy technologies and practices; - promote the introduction of clean energy technologies and practices; - promote the introduction of clean energy technologies and practices - promote the production of energy efficient machinery and equipment with low environmental pollution; - promote the production of energy efficient machinery and equipment with low environmental pollution; - reduce the cost of building energy efficient houses; - help the low-income groups to reduce energy costs; - maintain the reliability of energy transport communications. At the same time, activities in the directions chosen by the state should be beneficial to the population and business due to state preferences; the state should provide detailed information to the population and business about the goals and priorities of increasing energy efficiency, as well as about the conditions for obtaining state support in activities in priority areas. In Japan, after the first oil crisis, energy conservation measures were taken, which led to a 35% decrease in the energy intensity of the gross national product by 1985. However, then, over 7 years, energy consumption increased by an average of 3.1% per year. Therefore, the Japanese government revised the Energy Rational Use Law in 1993. Under the revised law, the Ministry of Economy, Trade and Industry of Japan (METI) is to establish and declare the basic principles of policy aimed at comprehensively promoting the rational use of energy, and the main consumers should make efforts to rationalize the use of energy in accordance with this policy. 4 Conclusion There are very clear differences in approaches to energy conservation in different countries, associated with the peculiarities of the national mentality, cultural preferences and prevailing stereotypes of behavior. However, an important common feature of the developed countries is the policy focus on achieving energy savings at the energy use stage. p y g gy g gy g The US economy is 2.5 times more energy efficient than the Russian economy. According to some experts, the production of manufactured goods in America consumes 9 times less energy than in Russia. Currently, the level of energy consumed in the country for the production of goods and services in the amount of $ 1, has decreased by more than 50% compared to 1970. The American energy efficiency achievement is the result of years of energy conservation efforts. A feature of the US energy efficiency policy is the widespread use of various measures of financial incentives and the evasion of the adoption of all kinds of codes and regulations. That means that the main efforts are directed not at coercion, but at interest. Recognizing the opportunities offered by energy efficiency, more than 60 leading organizations representing various stakeholders across the country joined forces in 2006 to develop a National Energy Efficiency Action Plan. Many of these stakeholders are precisely the groups that can provide energy savings — power and gas utilities, state regulators, and other organizations. The action plan identifies the main barriers to investment in energy efficiency. It outlines five key recommendations for achieving the goal of cost-effective energy efficiency. The plan also provides the foundations of the “Outlook 2025” strategy, which provides for actions to achieve and evaluate progressive development towards the set goal. The main goals of the state policy to improve energy efficiency: state policy to improve energy efficiency: state policy to improve energy efficiency: d US d d il i - reduce US dependence on oil imports; - develop and introduce energy-saving technologies for public buildings, residential buildings, transport, energy and industrial sectors. - develop and introduce energy-saving technologies for public buildings, residential buildings, transport, energy and industrial sectors. 4 Conclusion The specific goal of the concept is to increase the share of renewable sources in energy consumption by 2 times and achieve the highest indicator in the world - 20%. The European Union is a major driving force in promoting energy efficiency strategies and combating global climate change. EU takes a comprehensive approach to shaping the legal framework for energy efficiency [14]. The measures, forms and methods of state support in the field of energy saving and energy efficiency applied abroad and provided for by the Russian Energy Saving Program include the following measures: The measures, forms and methods of state support in the field of energy saving and energy efficiency applied abroad and provided for by the Russian Energy Saving Program include the following measures: - Information support of state programs and individual measures to stimulate energy saving - Tax incentives for energy saving - Financial support of industrial enterprises in the implementation of energy saving projects - Application of voluntary agreements between the state and the industrial sector in the field of energy efficiency [15, 16]. Thus, the problem of energy saving is urgent and requires a solution. Only coordinated efforts at the federal, regional and local levels will make it possible to implement one of the Thus, the problem of energy saving is urgent and requires a solution. Only coordinated efforts at the federal, regional and local levels will make it possible to implement one of the 9 9 SHS Web of Conferences 92, 0 (2021) 8022 https://doi.org/10.1051/shsconf/20219208022 Globalization and its Socio-Economic Consequences 2020 strategically important tasks of the state - to increase energy efficiency and reduce the energy intensity of GDP. strategically important tasks of the state - to increase energy efficiency and reduce the energy intensity of GDP. The study was supported by the grant of the President of the Russian Federation for state support of the leading scientific schools of the Russian Federation NSh-2702.2020.6 "Conceptual Foundations of a New Paradigm of Economic Development in the Era of Technological and Social Transformation." leading scientific schools of the Russian Federation NSh-2702.2020.6 "Conceptual Foundations of a New Paradigm of Economic Development in the Era of Technological and Social Transformation." 15. Ferar, G.S., Rastvortsev, A.F., Blagadyreva, A.M. (2012). Methodological approaches to the formation and implementation of regional environmental policy. Issues of state and municipal management, 1, 27-36. 16. Nadanyiova, M., Gajanova, L., Majerova, J. (2020). Green Marketing as a Part of the Socially Responsible Brand's Communication from the Aspect of Generational Stratification. Sustainability, 12(17), Art. No. 7118. References 1. United Nations Environment Program. United Nations. Retrieved from: https://www.un.org/ru/ga/unep/ 2. Kozhevnikova, T.M., Terakopov, S.G. (2013). «Green economy» as one of the areas of sustainable development. Socio-economic phenomena and processes, 3(49), 78-82. 3. Moore Lappé, F. (2012). Our Challenge - Developing an Eco-Mind. Green economy in action: Articles and Excerpts that Illustrate Green Economy and Sustainable Development Efforts. Retrieved from: https://www.un.org/waterforlifedecade/pdf/green_economy_in_action_eng.pdf 4. Rifkin, J. (2009). The Third Industrial Revolution. Retrieved from: https://archive.org/details/greeneconomicsin00moll 5. Scott Cato, M. (2009). Green Economics: An Introduction to Theory, Policy and Practice. Retrieved from: https://archive.org/details/greeneconomicsin00moll 6. Vertakova, Y.V., Plotnikov V.A. (2019). Assessment of the economic activity greening level and the green economy development directions. In IOP Conference Series: Earth and Environmental Science, 392, Art. No. 012078. 7. Vertakova, Y.V., Plotnikov V.A. (2019). The integrated approach to sustainable development: the case of energy efficiency and solid waste management. International Journal of Energy Economics and Policy, 9(4), 194-201. 8. Vertakova, Y., Plotnikov, V., Babich, T. (2020). Conceptual framework of state economic policy in the technological and social transformation conditions. In E3S Web of Conferences. Topical Problems of Green Architecture, Civil and Environmental Engineering, 164, Art. No. 11016. 9. Zlobina, N., Merkulova, E., Muratova, O., Kondrakov, O., Vertakova, Y. (2019). Impact of energy economy development on the region’s population life quality. In E3S Web of Conferences, 110, Art. No. 02106. 10. IEA – International Energy Agency. Retrieved from: https://www.iea.org/ 11. Global Energy Review. (2020). The impacts of the Covid-19 crisis on global energy demand and CO2 emissions. Retrieved from: https://www.iea.org/reports/global- energy-review-2020 12. Forecast of energy development in the world and Russia until 2040. Retrieved from: https://www.eriras.ru/files/forecast_2040.pdf 13. Ignatieva, A.A. (2011). Green Economy: A Practical Vector of Sustainable Development or a Political Compromise? Russia in the World Around, 1, 28-60. 14. Development of a Regional Action Plan and Recommendations on the Formation and Implementation of a Policy in the Field of Energy Efficiency and Energy Saving in the CIS Countries. Retrieved from: https://www.unece.org/fileadmin/DAM/ energy/se/pdfs/ee21/EE21_Subregional_projects/RegionalActionPlanRus_Jan_2014_F inal.pdf 10 SHS Web of Conferences 92, 0 (2021) 8022 https://doi.org/10.1051/shsconf/20219208022 Globalization and its Socio-Economic Consequences 2020 15. Ferar, G.S., Rastvortsev, A.F., Blagadyreva, A.M. (2012). Methodological approaches to the formation and implementation of regional environmental policy. Issues of state and municipal management, 1, 27-36. 16. Nadanyiova, M., Gajanova, L., Majerova, J. (2020). Green Marketing as a Part of the Socially Responsible Brand's Communication from the Aspect of Generational Stratification. Sustainability, 12(17), Art. No. 7118. 16. https://doi.org/10.1051/shsconf/20219208022 References Nadanyiova, M., Gajanova, L., Majerova, J. (2020). Green Marketing as a Part of the Socially Responsible Brand's Communication from the Aspect of Generational Stratification. Sustainability, 12(17), Art. No. 7118. 11
https://openalex.org/W2606314801	http://periodicos.uem.br/ojs/index.php/BSocParanMat/article/download/33821/18880	English	null	Existence of solutions for a class of strongly coupled $p(x)-$ laplacian system	Boletim da Sociedade Paranaense de Matemática	2,018	cc-by	4,863	∗This work was funded by the Deanship of Scientiﬁc Research at the University of Dammam under the reference 2012017. 2010 Mathematics Subject Classiﬁcation: 35J65, 35J20, 35J70. Submitted October 21, 2016. Published January 07, 2017 (3s.) v. 36 4 (2018): 183–195. ISSN-00378712 in press doi:10.5269/bspm.v36i4.33821 (3s.) v. 36 4 (2018): 183–195. ISSN-00378712 in press doi:10.5269/bspm.v36i4.33821 Bol. Soc. Paran. Mat. c ⃝SPM –ISSN-2175-1188 on line SPM: www.spm.uem.br/bspm E.A. Al Zahrani, M.A. Mourou and K.Saoudi abstract: The present work is concerned with the study of a strongly coupled nonlinear elliptic system on the whole space RN involving the p(x)-laplacien op- erator. We employ variational methods and the theory of the variable exponent Sobolev spaces, in order to establish some suﬃcient conditions for the existence of non-trivial solutions. Key Words: p(x)-Laplace operator, Generalized Lebesgue-Sobolev spaces, Strongly system, Variational methods. Contents 1 Introduction 2 Preliminary Results 3 Main Results 2 Preliminary Results 3 Main Results 4 Proof of Theorem 3.1 4 Proof of Theorem 3.1 Existence of Solutions For a Class of Strongly Coupled p(x)-laplacian System ∗ E.A. Al Zahrani, M.A. Mourou and K.Saoudi Typeset by BSP Mstyle. c ⃝Soc. Paran. de Mat. E.A. Al Zahrani, M.A. Mourou and K.Saoudi E.A. Al Zahrani, M.A. Mourou and K.Saoudi The operator ∆p(x)u := div(\|∇u\|p(x)−2∇u) is called p(x)−Laplace where p is a continuous non-constant function. This diﬀerential operator is a natural general- ization of the p-Laplace operator ∆pu := div(\|∇u\|p−2∇u), where p > 1 is a real constant. However, the p(x)−Laplace operator possesses more complicated non- linearity than p-Laplace operator, due to the fact that ∆p(x) is not homogeneous. This fact implies some diﬀculties; for example, we can not use the Lagrange Mul- tiplier Theorem in many problems involving this operator. The study of diﬀerential and partial diﬀerential involving variable exponent conditions is a new and an interesting topic. The interest in studying such prob- lems was stimulated by their applications in elastic mechanics, ﬂuid dynamics, electrorheological ﬂuids, image processing, ﬂow in porous media, calculus of varia- tions, nonlinear elasticity theory, heterogeneous porous media models (see Acerbi- Mingione [1] , Diening [5], Ruˇzi˘cka [15], Zhikov [17]) etc.... These physical problems were facilitated by the development of Lebesgue and Sobolev spaces with variable exponent. In literature, elliptic systems with standard and nonstandard growth conditions have been studied by many authors. Let us brieﬂy recall the literature concern- ing related elliptic systems. In [3,2] the authors show the existence of nontrivial solutions for the following p−Laplacian problem:              −∆pu(x) = a(x)\|u(x)\|p−2u + b(x)\|u\|α\|v\|βv + f in RN, −∆qv(x) = c(x)\|u\|α\|v\|βv + d(x)\|u(x)\|q−2u + g) in RN, lim \|x\|→∞u(x) = lim \|x\|→∞v(x) = 0, (u, v) > 0 in RN. (1.2) (1.2) where the p-Laplace operator ∆pu := div(\|∇u\|p−2∇u), with p > 1, α, β > 0, p, q > 1, and is a real f, g are given functions. In [3], the author’s obtain nec- essary and suﬃcient conditions on the coeﬃcients for having a maximum principle for system (1.2). Then using the method of sup and super solutions, they prove the existence of positive solutions under some conditions on the functions f and g. In [2], the authors apply the theory of monotone operators to obtain the nontrivial solutions of the system (1.2). y the t (1.2). 1. Introduction In this paper, we study the existence of nontrivial weak solutions for the fol- lowing (p, q)-gradient elliptic system:           −∆p(x)u(x) + a(x)\|u(x)\|p(x)−2u = f(x, u, v) in RN, −∆q(x)v(x) + b(x)\|v(x)\|q(x)−2v = g(x, u, v) in RN, (1.1) (1.1) (u, v) ∈W 1,p(x)(RN) × W 1,q(x) RN . Here p, q : Ω→R two functions of class C(Ω) such that 1 < p(x), q(x) < N (N ≥2) for all x ∈RN and the coeﬃcients a, b, are variables. The real-valued functions f, g are given functions and ∆p(x)u is the p(x)-Laplacian operator de- ﬁned by ∆p(x)u := div(\|∇u\|p(x)−2∇u). 183 Typeset by BSP Mstyle. c ⃝Soc. Paran. de Mat. Typeset by BSP Mstyle. c ⃝Soc. Paran. de Mat. 183 184 E.A. Al Zahrani, M.A. Mourou and K.Saoudi In Khafagy-Serag [10] deal with the following problem: −∆p,P u = a(x)\|u\|p−2u + b(x)\|u\|α\|v\|βv + f in Ω, −∆Q,qv = c(x)\|u\|α\|v\|βu + d(x)\|v\|q−2v + g in Ω, u = v = 0 on ∂Ω, u = v = 0 on ∂Ω, where the degenerate p-Laplacian deﬁned as ∆p,P u = div[P(x)\|∇u\|p−2∇u]. Using an approximation method, they apply the Schauder’s Fixed Point Theorem to get the nontrivial solutions of the system. Moreover, they gives necessary and suﬃcient strongly p(x)-Laplace system 185 strongly p(x)-Laplace system 185 185 strongly p(x)-Laplace system conditions for having the maximum principle for this system. conditions for having the maximum principle for this system. In Djellit-Youbi-Tas [6] show the existence of nontrivial solutions for the following p(x)-Laplacian system:    −∆p(x)u = ∂F ∂u (x, u, v) in RN −∆q(x)v = ∂F ∂v (x, u, v) in RN (1.3) (1.3) Here p, q : Ω→R two functions of class C(Ω) such that 1 < p(x), q(x) < N (N ≥2) for all x ∈RN. However, the function F belongs to C(RN × R2). Introducing some natural growth hypotheses on the right-hand side of the system which will ensure the semi-continuous and coercivity for the corresponding Euler-Lagrange functional of the system, the authors use critical point theory to obtain the existence of non- trivial weak solution of the system (1.3). In Ogras-Mashiyev-Avci-Yucedag [13] using a weak version of the Palais-Smale condition, that is, Cerami condition, they apply the mountain pass theorem to get the nontrivial solutions of the system (1.3). In Xu-An [16] study the following elliptic systems of gradient type with non- standard growth conditions −∆p(x)u + \|u\|p(x)−2u = ∂F ∂u (x, u, v) in RN −∆p(x)v + \|v\|q(x)−2v = ∂F ∂v (x, u, v) in RN. The potential function F needs to satisfy Caratheodory conditions. Using critical point theory, they establish existence and multiplicity of solutions in sub-linear and super-linear cases. Inspired by the above-mentioned papers, we deal with the existence of nontriv- ial solutions for system (1.1). We know that in the study of p(x)-Laplace equations in RN, a main diﬃculty arises from the lack of compactness. In this paper we will overcome this diﬃculty by establishing some growth conditions and regularity on the nonlinearities f and g, which will ensure the mountain pass geometry and Ce- rami condition for the corresponding Euler-Lagrange functional. E.A. Al Zahrani, M.A. Mourou and K.Saoudi By the mountain pass theorem, the basic results on the existence of solutions of system (1.1) will be presented. The outline of this paper is as follows. In section 2, we will recall some basic facts about the variable exponent Lebesgue and Sobolev spaces which we will use later. Our main results are stated in Section 3. Proofs of our results will be presented in section 4. 186 E.A. Al Zahrani, M.A. Mourou and K.Saoudi 2. Preliminary Results To deal with the p(x)-Laplacian problem, we need introduce some functional spaces Lp(·)(Ω), W 1,p(·)(Ω), W 1,p(·) 0 (Ω) and properties of the p(x)-Laplacian which we will use later. Denote by S(Ω) be the set of all measurable real-valued functions deﬁned in Ω. Note that two measurable functions are considered as the same element of S(Ω) when they are equal almost everywhere. Set L∞ + (Ω) = {h; h ∈L∞(Ω), ess inf h(x) > 1 for all x ∈Ω¯} L∞ + (Ω) = {h; h ∈L∞(Ω), ess inf h(x) > 1 for all x ∈Ω¯} For any h ∈L∞ + (Ω) we deﬁne h+ = ess sup x∈Ω h(x) > 1 and h−= ess inf x∈Ωh(x) > 1. Let Let Lp(·)(Ω) = u ∈S(Ω) : Z Ω \|u(x)\|p(x)dx < ∞ , with the norm \|u\|p(·) = \|u\|Lp(·)(Ω) = inf λ > 0 : Z Ω \|u(x) λ \|p(x)dx ≤1 . The space (Lp(·)(Ω), \| · \|p(·)) becomes a Banach space. We call it variable exponent Lebesgue space. Moreover, this space is a separable, reﬂexive and uniform convex Banach space; see [9, Theorems 1.6, 1.10, 1.14]. p ; [ , , , The variable exponent Sobolev space W 1,p(·)(Ω) = u ∈Lp(·)(Ω) : \|∇u\| ∈Lp(·)(Ω) , can be equipped with the norm can be equipped with the norm ∥u∥= \|u\|p(·) + \|∇u\|p(·), ∀u ∈W 1,p(·)(Ω). ∥u∥= \|u\|p(·) + \|∇u\|p(·), ∀u ∈W 1,p(·)(Ω). Note that W 1,p(·) 0 (Ω) is the closure of C∞ 0 (Ω) in W 1,p(·)(Ω) under the norm ∥u∥= \|∇u\|p(·). The spaces W 1,p(·)(Ω) and W 1,p(·) 0 (Ω) are separable, reﬂexive and uniform convex Banach spaces (see [9, Theorem 2.1]). The inclusion between Lebesgue spaces also generalizes naturally: if 0 < \|Ω\| < ∞and p1, p2 are variable exponents so that p1(x) ≤p2(x) almost everywhere in Ωthen there exists the continuous embedding Lp2(x)(Ω) ֒→Lp1(x)(Ω). We denote by Lq(x)(Ω) the conjugate space of Lp(x)(Ω), where 1 q(x) + 1 p(x) = 1. For u ∈Lp(x)(Ω) and v ∈Lq(x)(Ω), the H¨older type inequality Z Ω u(x)v(x)dx ≤ 1 p−+ 1 q− \|u\|p(x)\|v\|q(x), (2.1) (2.1) holds true. holds true. 187 strongly p(x)-Laplace system An important role in manipulating the generalized Lebesgue spaces is played by the modular of the Lp(x)(Ω) space, which is the mapping ρp(x) : Lp(x)(Ω) →R deﬁned by ρp(x)(u) = Z Ω \|u\|p(x)dx. If (un), u ∈Lp(x)(Ω) and p+ < ∞. 2. Preliminary Results Then the following relations hold true. ∥u∥Lp(x) > 1 ⇒∥u∥p− Lp(x) ≤ρp(x)(u) ≤∥u∥p+ Lp(x), (2.2) ∥u∥Lp(x) < 1 ⇒∥u∥p+ Lp(x) ≤ρp(x)(u) ≤∥u∥p− Lp(x), (2.3) ∥un −u∥Lp(x) →0 if and if ρp(x)(un −u) →0. (2.4) (2.2) (2.3) (2.4) (2.4) The following result generalizes the well-known Sobolev embedding theorem. Theorem 2.1 ([8,11]). Let Ω⊂RN be an open bounded domain with Lipschitz boundary and assume that p ∈C(¯Ω) with p(x) > 1 for each x ∈¯Ω. If r ∈ C(¯Ω) and p(x) ≤r(x) ≤p∗(x) for all x ∈Ω, then there exists a continuous embedding W 1,p(x)(Ω) ֒→Lr(x)(Ω). Also, the embedding is compact r(x) < p∗(x) almost everywhere in Ωwhere p∗(x) = ( Np(x) N−p(x) if p(x) < N, +∞ if p(x) ≥N. 3. Main Results Before stating our main results, we make the following assumptions throughout this paper: (B1) a(x), b(x) ∈L∞ loc(Ω) and there exist a0, b0 > 0 such that a(x) ≥a0, b(x) ≥b0 ∀x ∈RN and a(x) →∞, b(x) →∞as \|x\| →∞. (B1) a(x), b(x) ∈L∞ loc(Ω) and there exist a0, b0 > 0 such that a(x) ≥a0, b(x) ≥b0 ∀x ∈RN and a(x) →∞, b(x) →∞as \|x\| →∞. (F1) f(x, w), g(x, w) ∈C1(RN × R2, R), f(x, 0, 0) = 0, g(x, 0, 0) = 0, ∀x ∈RN. Moreover, there exists a function F(x, w) ∈C1(RN × R2, R) such that (F1) f(x, w), g(x, w) ∈C1(RN × R2, R), f(x, 0, 0) = 0, g(x, 0, 0) = 0, ∀x ∈RN. Moreover, there exists a function F(x, w) ∈C1(RN × R2, R) such that ∂F ∂u = f(x, w), ∂F ∂v = g(x, w), ∀x ∈RN, w = (u, v) ∈R2 (F2) There exist a constant µ > max (p+, q+)such that 0 < µF(x, w) ≤w.∇F(x, w) 0 < µF(x, w) ≤w.∇F(x, w) (F3) For p∗= Np− N−p−, q∗= Nq− N−q−and p+ < Np− N−p−, q+ < Nq− N−q−, there exist a1, a2, b1, b2 such that (F3) For p∗= Np− N−p−, q∗= Nq− N−q−and p+ < Np− N−p−, q+ < Nq− N−q−, there exist a1, a2, b1, b2 such that \|∇f(x, w)\| ≤ a1(x) \|w\|p1−2 + a2(x) \|w\|p2−1 \|∇f(x, w)\| ≤ a1(x) \|w\|p1−2 + a2(x) \|w\|p2−1 \|∇g(x, w)\| ≤ b1(x) \|w\|q1 −2 + b2(x) \|w\|q2 −1 E.A. Al Zahrani, M.A. 3. Main Results Mourou and K.Saoudi 188 ai(x) ∈Lαi(RN) ∩Lβi(RN), bi(x) ∈Lγi(RN) ∩Lδi(RN), i = 1, 2 αi = p∗− p∗−−(pi −1), γi = q∗− q∗−−(qi −1), βi = p∗−q∗− p∗−q∗−−p∗(pi−2) −q∗, δi = p∗−q∗− p∗−q∗−−q∗−(q1i−2) −p∗ 2 < p1, q1 < min(p+ −1, q+ −1), max(p+ −1, q+ −1) < p2, αi = p∗− p∗−−(pi −1), γi = q∗− q∗−−(qi −1), βi = p∗−q∗− p∗−q∗−−p∗(pi−2) −q∗, δi = p∗−q∗− p∗−q∗−−q∗−(q1i−2) −p∗ 2 < p1, q1 < min(p+ −1, q+ −1), max(p+ −1, q+ −1) < p2, βi = p∗−q∗− p∗−q∗−−p∗(pi−2) −q∗, δi = p∗−q∗− p∗−q∗−−q∗−(q1i−2) −p∗ 2 < p1, q1 < min(p+ −1, q+ −1), max(p+ −1, q+ −1) < p2, q2 < min(p∗+ −1, q∗+ −1) Now we denote by E the product space D1,p(x) × D1,q(x), deﬁned as the com- pletion of C∞ 0 (RN) with respect to the norm ∥(u, v)∥= \|∇u\|p(x) + \|u\|(a(x),p(x)) + \|∇v\|q(x) + \|v\|(b(x),q(x)) We remark that condition (B1) implies that E ⊂W 1,p(x)(Ω) × W 1,q(x)(Ω). Set J(u, v) = Z \|∇u\|p(x) + a(x) \|u\|p(x) dx + Z \|∇v\|q(x) + b(x) \|v\|q(x) dx. Then, for all w ∈E, the following relations hold Then, for all w ∈E, the following relations hold ∥(u, v)∥ > 1 ⇒∥(u, v)∥min(p−,q−) ≤J(u, v) ≤∥(u, v)∥max(p+,q+) ∥(u, v)∥ < 1 ⇒∥(u, v)∥max(p+,q+) ≤J(u, v) ≤∥(u, v)∥min(p−,q−) We say that (u, v) ∈E is a weak solution of problem (1.1) if \|∇u\|p(x)−2 ∇u∇Φ + \|∇v\|q(x)−2 ∇v∇Ψ + Z a(x) \|u\|p(x)−2 uΦ + Z b(x) \|v\|q(x)−2 vΨ = Z f(x, u, v)Φ + Z g(x, u, v)Ψ, for all (Φ, Ψ) ∈E. ( ) The main result of this paper is given by the following theorem: Theorem 3.1. Assume conditions (B1) and (F1)-(F3) are fulﬁlled. Then prob- lem (1.1) has a non trivial weak solution. Theorem 3.1. Assume conditions (B1) and (F1)-(F3) are fulﬁlled. Then prob- lem (1.1) has a non trivial weak solution. We point out the fact that the result of Theorem 3.1 extends the results from [12] ,[14] where similar equations are studied in the case of p−laplacian operator. strongly p(x)-Laplace system strongly p(x)-Laplace system 189 F(x, u, v) ≤ c1[a1(x) \|u\|p1 + \|v\|p1−1 \|u\| + a2(x) \|u\|p2 + \|v\|p2−1 \|u\| 4. Proof of Theorem 3.1 The energy functional corresponding to problem (1.1) is deﬁned as I : E →R, I(u, v) = Z 1 p(x) \|∇u\|p(x) + a(x) \|u\|p(x) + Z 1 q(x) \|∇v(x)\|q(x) + b(x) \|v(x)\|q(x) − Z F(x, u, v) Similar arguments as those used in [7] assure that I ∈C1(E, R) with Similar arguments as those used in [7] assure that I ∈C1(E, R) with I‵(u, v) (Φ, Ψ) = Z \|∇u\|p(x)−2 ∇u∇Φ + Z \|∇v\|q(x)−2 ∇v∇Ψ + Z a(x) \|u\|p(x)−2 u(x)Φ(x) + Z b(x) \|v\|q(x)−2 vΨ − Z f(x, u, v)Φ − Z g(x, u, v)Ψ, for all (Φ, Ψ) ∈E. for all (Φ, Ψ) ∈E. Th b th t iti l i t f th f ti l I k l Thus, we observe that any critical points of the functional I are a weak solu- tions for problem (1.1). Our idea is to prove Theorem 3.1 by applying the Mountain pass theorem (see [4]). With that end in view, we prove some auxiliary results which show that the functional I(u, v) has a mountain pass geometry. Lemma 4.1. If (B1) and (F1)-(F3) holds, then there exist τ > 0 and δ > 0 such that for all (u, v) ∈E with ∥(u, v)∥= τ I(u, v) ≥δ > 0 I(u, v) ≥δ > 0 Proof: From (F2), it is easy to see that Proof: From (F2), it is easy to see that Proof: From (F2), it is easy to see that F(x, w) ≥min \|s\|=1 F(x, s). \|w\|µ , ∀x ∈RN and \|w\| ≥1, w ∈R2 (4.1) (4.1) and and 0 < F(x, w) < max \|s\|=1 F(x, s). \|w\|µ , ∀x ∈RN and 0 < \|w\| ≤1 (4.2) Using (F3),we have F(x, u, v) = u Z 0 ∂F ∂s (x, s, v) ds + F(x, 0, v) = u Z 0 ∂F ∂s (x, s, v) ds + v Z 0 ∂F ∂s (x, 0, s) ds + F(x, 0, 0) E.A. Al Zahrani, M.A. 4. Proof of Theorem 3.1 Mourou and K.Saoudi 190 and F(x, u, v) ≤ c1[a1(x) \|u\|p1 + \|v\|p1−1 \|u\| + a2(x) \|u\|p2 + \|v\|p2−1 \|u\| + b1(x) \|v\|q1 + b2(x) \|v\|q2] + b1(x) \|v\|q1 + b2(x) \|v\|q2] So max\|w\|=1 F(x, w) ≤C in view of (F3) and since max(p+ −1, q+ −1) < p2, q2 < min(p∗+ −1, q∗+ −1), + b1(x) \|v\|q1 + b2(x) \|v\|q2] So max\|w\|=1 F(x, w) ≤C in view of (F3) and since max(p+ −1, q+ −1) < p2, q2 < min(p∗+ −1, q∗+ −1), So max\|w\|=1 F(x, w) ≤C in view of (F3) and since ( + 1 + 1) i ( ∗+ 1 ∗+ 1) So max\|w\|=1 F(x, w) ≤C in view of (F3) and since max(p+ −1, q+ −1) < p2, q2 < min(p∗+ −1, q∗+ −1), we have lim \|w\|→∞ F(x, w) \|w\| Np−− N−p− = 0, (4.3) lim \|w\|→∞ F(x, w) \|w\| Nq− N−q− = 0, (4.3) It follows, that lim \|w\|→0 F(x, w) \|w\|p+ = 0, uniformly for x ∈RN (4.4) lim \|w\|→0 F(x, w) \|w\|q+ = 0, uniformly for x ∈RN (4.4) Thus, we obtain using condition (F2) and (F3), that ∀ε > 0, ∃Cǫ > 0, such that N − N − F(x, w) ≤ǫ \|w\|max(p+,q+) + Cǫ \|w\| max( Np N−p−, Nq N−q−) , ∥w∥< 1 (4.5) Using (4.5), we have (4.5) Using (4.5), we have I(w) ≥ 1 p+ J1(u) + 1 q+ J2(v) − Z F(x, W) dx ≥ 1 p+ J1(u) + 1 q+ J2(v) −ǫ \|W\|max(p+,q+) −Cǫ \|W\|max( Np−− N−p−, Nq−− N−q−) ≥ 1 max(p+, q+) ∥W∥−ǫc1 ∥W∥max(p+,q+) −Cǫ ∥W∥ max( Np−− N−p−, Nq−− N−q−) ≥ δ > 0, for some ﬁxed ǫ > 0, and δ, ∥W∥suﬃciently small. The proof of the Lemma 4.1 is now completed. ✷ for some ﬁxed ǫ > 0, and δ, ∥W∥suﬃciently small. The proof of the Lemma 4.1 is now completed. ✷ Lemma 4.2. Assume conditions (B1) and (F1)-(F3) holds. Then there exists e ∈E with ∥e∥> τ (τ is given in Lemma 4.1 such that I(e) < 0). 191 strongly p(x)-Laplace system Proof: Denote h(t) = F(x, tw) tµ , ∀t > 0. 4. Proof of Theorem 3.1 Then, using (F3), we get Then, using (F3), we get h′(t) = 1 tµ+1 [tu(f(x, w) + tvg(x, tw) −µF(x, tw)] ≥0, ∀t > 0 Thus, we deduce that for any t ≥1, F(x, tw) ≥tµF(x, w) Choosing w ∈E, with ∥w∥> 1 and R F(x, w)dx > 0 ﬁxed and t > 1,we have Thus, we deduce that for any t ≥1, F(x, tw) ≥tµF(x, w) Choosing w ∈E, with ∥w∥> 1 and R F(x, w)dx > 0 ﬁxed and t > 1,we have Thus, we deduce that for any t ≥1, F(x, tw) ≥tµF(x, w) Choosing w ∈E, with ∥w∥> 1 and R F(x, w)dx > 0 ﬁxed a I(tw) = Z 1 p(x)(\|∇tu\|p(x) + a(x) \|tu\|p(x)) + Z 1 q(x)(\|∇tv(x)\|q(x) + b(x) \|tv(x)\|q(x)) − Z F(x, tu, tv)dx = Z tp(x) p(x) (\|∇u\|p(x) + a(x) \|u\|p(x)) + Z tq(x) q(x) \|∇v(x)\|q(x) + b(x) \|v(x)\|q(x) −tµ Z F(x, u, v)dx ≤ tmax(p+,q+)(∥u∥p+ p− + ∥v∥q+ q ) −tµ Z F(x, u, v)dx w) = Z 1 p(x)(\|∇tu\|p(x) + a(x) \|tu\|p(x)) + Z 1 q(x)(\|∇tv(x)\|q(x) + b(x) \|tv(x)\|q(x)) − Z F(x, tu, tv)dx = Z tp(x) p(x) (\|∇u\|p(x) + a(x) \|u\|p(x)) + Z tq(x) q(x) \|∇v(x)\|q(x) Z Since µ > max(p+, q+), therefore I(tw) →−∞,when t →∞, which concludes our Lemma 4.2. ✷ Proof: [Proof of Theorem 3.1] We set Γ := {γ ∈C([0, 1], E); γ(0) = 0, γ(1) = e} where e ∈E is determined by Lemma 4.2 and c = inf γ∈Γ max t∈[0,1] I(γ(t)) According to Lemma 4.2, we know that ∥e∥> τ, so every path γ ∈Γ intersects the sphere ∥w∥= τ. Then, Lemma 4.1 implies c ≥inf∥u∥=t I(u) ≥δ with constants δ > 0. Thus c > 0. Hence, using the Mountain-pass theorem (see e.g.,[4]) we obtain a sequence (wn)n ⊂E such that I(wn) →c, I‵(wn) →0 (4.6) (4.6) We claim that (wn)n is bounded in E. Arguing by contradiction and passing to a subsequence, we have ∥wn∥→∞. Using (2.2), it follows that for n large enough, we have 1 c + 1 + ∥wn∥≥I(wn) −1 µ I‵(wn), wn E.A. Al Zahrani, M.A. 4. Proof of Theorem 3.1 Mourou and K.Saoudi 192 Using the above inequality, we have Using the above inequality, we have Using the above inequality, we have c + 1 + ∥wn∥ ≥ 1 p+ −1 µ J1(un) + 1 q+ −1 µ J2(vn) − Z F(x, wn) −1 µ[ Z f(x, un, vn) + Z g(x, un, vn)]dx By (F3) we have Z [F(x, wn) −1 µ(f(x, wn)un + g(x, wn)vn)] ≤0. The above inequality combined with relations (2.2), (2.3) yields c + 1 + ∥wn∥ ≥ ( 1 p+ −1 µ)J1(un) + ( 1 q+ −1 µ)J2(un) (4.7) ≥ ( 1 p+ −1 µ) ∥un∥p + ( 1 q+ −1 µ) ∥vn∥q− ≥ min( 1 p+ −1 µ, 1 q+ −1 µ)(∥un∥p + ∥vn∥q− ) (4.7) Now dividing by ∥un∥, ∥vn∥in (4.7) and passing to the limit as n →∞, we obtain a contradiction. So, up to a subsequence (un, vn)n converges weakly in E to some (u, v) ∈E. If Ωis a bounded domain, then there exists a com- pact embedding E(Ω) ֒→L Np N−p−(Ω) × L Nq− N−q−(Ω) . Then (un, vn) →(u, v) in L Np N−p−(Ω) × L Nq− N−q−(Ω) , for all Ωbounded domains in RN. Claim : ⟨I′(un, vn), (Φ, Ψ)⟩→⟨I′(u, v), (Φ, Ψ)⟩ ∀(Φ, Ψ) ∈C∞ 0 (RN) (4.8) (4.8) Assuming this Claim, using (4.6), (u, v) is a weak solution of the problem (1.1) since C∞ 0 is dense in E. Finally, let us prove the Claim. To do this, let (Φ, Ψ) ∈ C∞ 0 (RN) be ﬁxed. Firstly, we prove that lim n→∞[ Z RN f(x, un, vn)Φ + Z g(x, un, vn)Ψ] = Z RN f(x, u, v)Φdx + Z RN g(x, u, v)Ψdx A simple calculation implies Z (f(x, un, vn) −f(x, u, v)) Φdx ≤ Z \|(f(x, un, vn) −f(x, u, v))\| . \|Φ\| dx ≤∥Φ∥L∞ Z (f(x, un, vn) −f(x, u, v)) (un −u) × \|un −u\| dx ≤∥Φ∥L∞ Z fu(x, u∗ n, vn) \|un −u\| , 193 strongly p(x)-Laplace system where u∗ n ∈[un, u] or [u, un] . Similarly where u∗ n ∈[un, u] or [u, un] . Similarly Z (g(x, un, vn) −g(x, u, v)) Ψdx ≤∥Ψ∥L∞ Z gv(x, un, v∗ n) \|vn −v\| Z (g(x, un, vn) −g(x, u, v)) Ψdx ≤∥Ψ∥L∞ Z gv(x, un, v∗ n) \|vn −v\| where v∗ n ∈[vn, v] or [v, vn] . 4. Proof of Theorem 3.1 Now using condition (F2), we obtain where v∗ n ∈[vn, v] or [v, vn] . Now using condition (F2), we obtain Z (f(x, un, vn) −f(x, u, v)) Φdx + Z (g(x, un, vn) −g(x, u, v)) Ψdx ≤ max (∥Φ∥∞, ∥Ψ∥∞) [ ∥a1(x)∥α1 ∥u∗ n∥ p1−1 p∗−∥a2(x)∥α2 ∥vn∥p2−1 q∗− ∥un −u∥p∗−+ max (∥Φ∥∞, ∥Ψ∥∞) [ ∥a1(x)∥α1 ∥u∗ n∥ p1−1 p∗−∥a2(x)∥α2 ∥vn∥p2−1 q∗− ∥un −u∥p∗−+ ∥b1(x)∥β1 ∥un∥ q1 −1 p∗−+ ∥b2(x)∥β2 ∥v∗ n∥q2−1 q∗− ∥vn −v∥q∗−] ∥b1(x)∥β1 ∥un∥ q1 −1 p∗−+ ∥b2(x)∥β2 ∥v∗ n∥q2−1 q∗− ∥vn −v∥q∗−] Taking into account that un →u in L Np N−p−(Ω) and vn →v in L Nq− N−q−(Ω) , and for all n ≥1, there exist λn(x) ∈[0, 1] such that u∗ n = λn(x)un(x) + [1 −λn(x)] u(x), we deduce that Z \|u∗ n −u\|s dx = Z \|λn(x)\|s \|un −u\|s dx →0 as n →∞. It results that Z \|u∗ n\|s → Z \|u\|s dx as n →∞ Similarly Z Ω \|v∗ n\|s → Z Ω \|v\|s dx as n →∞ From the above considerations, we obtain From the above considerations, we obtain Z (f(x, un, vn) −f(x, u, v)) Φdx + Z (g(x, un, vn) −g(x, u, v)) Ψdx →0, as n →∞ as n →∞ as n →∞ Since C∞ 0 RN is dense in E, the above relation implies Since C∞ 0 RN is dense in E, the above relation implies lim n→∞ Z (f(x, un, vn) −f(x, u, v)) Φdx + Z (g(x, un, vn) −g(x, u, v)) Ψdx = 0, as n →∞ lim n→∞ Z (f(x, un, vn) −f(x, u, v)) Φdx + Z (g(x, un, vn) −g(x, u, v)) Ψdx = 0, as n →∞ E.A. Al Zahrani, M.A. 4. Proof of Theorem 3.1 Mourou and K.Saoudi 194 Next, since (un, vn) ⇀(u, v) in E, it follows that Next, since (un, vn) ⇀(u, v) in E, it follows that Next, since (un, vn) ⇀(u, v) in E, it follows that im →∞ Z f(x, u, v) (un −u) dx + Z g(x, u, v) (vn −v) dx = 0 Thus, actually, we ﬁnd Thus, actually, we ﬁnd lim n→∞ Z f(x, un, vn) (un −u) dx + Z (g(x, un, vn)(vn −v) dx = 0 (4.9) (4.9) On the other hand, we have On the other hand, we have lim n→∞⟨I′(un, vn), (un −u) (vn −v)⟩= 0 lim n→∞⟨I′(un, vn), (un −u) (vn −v)⟩= 0 (4.10) (4.10) Combining (4.9) with (4.10), to deduce that Combining (4.9) with (4.10), to deduce that Combining (4.9) with (4.10), to deduce that lim n→∞ Z \|∇un\|p(x)−2 ∇un∇(un −u) + \|∇vn\|q(x)−2 ∇vn∇(vn −v) + Z a(x) \|un\|p(x)−2 un (x) (un (x) −u (x)) + Z c(x) \|vn\|q(x)−2 vn (x) (vn (x) −v (x)) = 0 lim n→∞ Z \|∇un\|p(x)−2 ∇un∇(un −u) + \|∇vn\|q(x)−2 ∇vn∇(vn −v) Since relation (4.10) holds true and (un, vn) ⇀(u, v) in E. By [7, Lemma 3.1], we deduce that (un, vn) →(u, v) in E. Th i I C1(E R) l d h ( ) deduce that (un, vn) →(u, v) in E. ( ) ( ) Then since I ∈C1(E, R),we conclude that I′ (un, vn) →I′ (u, v) , as n →∞ (4.11) I′ (un, vn) →I′ (u, v) , as n →∞ (4.11) Relations (4.6) and (4.11) show that I′ (u, v) = 0 and thus (u, v) is a weak solution for (Pλ), Moreover, by relation (10), it follows that I (u, v) > 0 and (u, v) is a nontrivial. ✷ Acknowledgments The authors would like to thank the anonymous referees for their carefully reading this paper and their useful comments. 4. A. Ambrosetti A. and Rabinowitz P.H., Dual variatinal methods in critical point theory and applications, J. Funct. Anal. 14, 349-381, (1973). References 1. Acerbi E. and Mingione G., Regularity results for a class of functionals with nonstandard growth, Arch. Rational Mech. Anal. 156, 121-140, (2001). 2. Al Zahrani E. A. and Serag H., Existence of Weak Solutions for some Nonlinear Systems on RN, Electronic J. Diﬀ. Eqns. Vol.2006, No. 69, 1-10, (2006). 2. Al Zahrani E. A. and Serag H., Existence of Weak Solutions for some Nonlinear Systems on RN, Electronic J. Diﬀ. Eqns. Vol.2006, No. 69, 1-10, (2006). 3. Al Zahrani E. A. and Serag H., Maximum principle and Existence of Positive Solution for Nonlinear Systems on RN, Electronic J. Diﬀ. Eqns., Vol.2005, No. 85, 1-12, (2005). 4. A. Ambrosetti A. and Rabinowitz P.H., Dual variatinal methods in critical point theory and applications, J. Funct. Anal. 14, 349-381, (1973). 195 strongly p(x)-Laplace system 5. Diening L., Theoritical and Numerical Results for Electroheological Fluids, Ph.D. thesis, University of Frieburg, Germany, (2002). 6. Djellit A., Youbi Z. and Tas S., Existence of solutions for a class of elliptic systems in RN involving the p(x)-Laplacian, Electronic J. Diﬀ. Eqns., Vol. 2012, No. 131, 1?10, (2012). 7. Fan X. and Han X., Existence and multiplicity of solutions for p(x)−laplacian equations in RN, Nonlinear Anal. 59, 173-188, (2004). 8. Fan X., Shen J. and D. Zhao D., Sobolev Embedding Theorems for Spaces W k,p(x), J. Math. Anal. Appl. 262, 749-760, (2001). 9. Fan X. and Zhang D., On the Spaces Lp(x)(Ω) and W m,p(x)(Ω), J. Math. Anal. Appl. 263, 424-446, (2001). 10. Khafagy S. A. and Serag H., On maximum principle and existence of positive weak solutions for nn nonlinear elliptic systems involving degenerated p-Laplacian operators, Turk. J. Math. 34, 59-71, (2010). 11. Kovacik O. and Rakosnik J., On spaces Lp(x)(Ω) and W 1,p(x), Czechoslovak Math. J. 41, 592-618, (1991). 12. Mihailescu M. and Radulescu V., A multiplicity result for a nonlinear degenerate problem arising in the theory of electroheological ﬂuids, Proc. Roy. Soc. London Ser. A 462, 2625-2641, (2006). 13. Ogras S., Mashiyev R. A., Avci M. and Yucedag Z., Existence of solution for a class of elliptic systems in RN involving the (p(x), q(x))-Laplacian, J. Inequal. Appl. Art. Id 612938, 16p, (2008). 14. Rabinowitz P. H., On a class of nonlinear Schrodinger equations, Zeit. Angew. Math. Phys. (ZAMP) 43, 271–291, (1992). 15. Ruzicka M., Electroheological Fluids: Modeling and Mathematical Theory, Springer- Verlag, Berlin, (2002). 16. Xu X. References and An Y., Existence and multiplicity of solutions for elliptic systems with nonstan- dard growth conditions in RN Nonlinear Anal. 68, 956-968, (2008). 17. Zhikov V., Averaging of functionals in the calculus of variations and elasticity, Math. USS- RIzv. 29, 33-66, (1987). E.A. Al Zahrani, M.A. Mourou and K.Saoudi, Department of Mathematics, E-mail address: ealzahrani@uod.edu.sa E-mail address: ealzahrani@uod.edu.sa E-mail address: mohamed ali.mourou@yahoo. E-mail address: kasaoudi@gmail.com
https://openalex.org/W2137412556	https://www.icesi.edu.co/revistas/index.php/estudios_gerenciales/article/download/1793/PDF	es		Beneficios y perjuicios de la estrategia de imitaciÃ³n	Estudios Gerenciales	2,014	cc-by	8,451	Estudios Gerenciales 30 (2014) 145–152 ESTUDIOS GERENCIALES www.elsevier.es/estudios gerenciales Artículo Beneﬁcios y perjuicios de la estrategia de imitación Ana M. Arboleda ∗ Profesora, Departamento de Mercadeo y Negocios Internacionales, Universidad Icesi, Cali, Colombia información del artículo r e s u m e n Historia del artículo: Recibido el 28 de febrero de 2013 Aceptado el 16 de enero de 2014 On-line el 8 de abril de 2014 Este artículo analiza la estrategia de imitación resaltando sus potenciales beneﬁcios y perjuicios. Utilizando una metodología exploratoria, se usan 15 casos en los que se argumenta competencia desleal por imitación. La discusión ofrece un paralelo entre las razones que tiene una empresa para imitar y para no hacerlo. Se concluye que la imitación es una estrategia viable siempre que la marca se reconozca como atributo distintivo, al tiempo que las marcas seguidoras ganan participación de mercado y comunican atributos acordes a la categoría. Por el contrario, si la imitación afecta negativamente el esfuerzo de una marca por ser distintiva, vulnera la decisión del consumidor al generar confusión y desmotiva el esfuerzo empresarial por la innovación. © 2013 Universidad ICESI. Publicado por Elsevier España, S.L. Todos los derechos reservados. Códigos JEL: M31 M38 Palabras clave: Imitación Entorno competitivo Confusión del consumidor Beneﬁts and detriments of imitation strategy a b s t r a c t JEL classiﬁcation: M31 M38 Keywords: Imitation Competitive environment Consumer confusion This article discusses imitation strategy, highlighting its potential beneﬁts and detriments. By using an exploratory method, the article uses 15 cases that argue unfair competition due to imitation. The discussion offers a parallel among reasons for a company to imitate or not. To conclude, imitation is a feasible strategy as long as the brand is regarded as a distinctive attribute, and at the same time, following brands improve its market share and communicate attributes that are consistent to the product category. On the other hand, if imitation negatively affects brand distinctiveness, then it may cause consumer confusion and reduce the organizational incentive to innovate. © 2013 Universidad ICESI. Published by Elsevier España, S.L. All rights reserved. Vantagens e desvantagens da estratégia de imitação r e s u m o Classiﬁcações JEL: M31 M38 Palavras-chave: Imitação Meio competitivo Confusão do consumidor Este artigo analisa a estratégia de imitação destacando as suas potenciais vantagens e desvantagens. Utilizando uma metodologia exploratória, utilizam-se 15 casos nos quais se refere concorrência desleal por imitação. A discussão oferece um paralelo entre as razões que uma empresa tem para imitar e as razões para não o fazer. Conclui-se que a imitação é uma estratégia viável sempre que a marca se reconheça como atributo distintivo, ao mesmo tempo que as marcas seguidoras ganham participação no mercado e comunicam atributos de acordo com a categoria. Por outro lado, se a imitação afecta negativamente o esforço de uma marca por ser distintiva, vulnera a decisão do consumidor ao criar confusão e desmotiva o esforço empresarial pela inovação. © 2013 Universidad ICESI. Publicado por Elsevier España, S.L. Todos os direitos reservados. ∗ Autora para correspondencia: Calle 18 # 122-135. Departamento de Mercadeo y Negocios Internacionales, Universidad Icesi, Cali, Colombia. Correo electrónico: amarboleda@icesi.edu.co 0123-5923/$ – see front matter © 2013 Universidad ICESI. Publicado por Elsevier España, S.L. Todos los derechos reservados. http://dx.doi.org/10.1016/j.estger.2014.01.016 146 A.M. Arboleda / Estudios Gerenciales 30 (2014) 145–152 1. Introducción La estrategia de innovación y la estrategia de imitación ocurren como un proceso dinámico relativo al contexto en el que se encuentre una empresa. La imitación es una alternativa que ha demostrado a través de la historia permitir a empresas de países en desarrollo incrementar rápidamente su capacidad competitiva (Kale y Little, 2007; Madsen, Islam y Ang, 2010). El objetivo principal de este artículo es ofrecer una reﬂexión acerca de las prácticas de imitación desde el punto de vista estratégico resaltando sus potenciales beneﬁcios y perjuicios. Es importante comprender la estrategia organizacional que subyace tras la decisión de imitar un producto ya que es una acción competitiva que muestra la natural rivalidad entre organizaciones (Chen, 1996) y puede tener consecuencias favorables o desfavorables tanto para la empresa como para la sociedad en general. Así, este artículo analiza la imitación desde el punto de vista de la estrategia organizacional apoyándose en los parámetros establecidos por la legislación colombiana para dicha práctica. La legislación aprueba el proceso de imitación como una práctica competitiva y considera que el ejercicio de la imitación no implica per se un acto de competencia desleal. La imitación es una estrategia viable siempre y cuando su uso no conlleve la confusión del consumidor y se respete la propiedad de marca amparada por la ley (Congreso de Colombia, 1996; Munar-Cadena, 2005). Es decir, la legislación protege al consumidor de actos que representen un engaño vulnerando su bienestar y asimismo protege a las organizaciones por sus esfuerzos en el posicionamiento de una marca y en investigación y desarrollo de nuevos productos. Al mismo tiempo, la legislación colombiana no prohíbe la imitación argumentando la necesidad de libre competencia. La libre competencia motiva condiciones y acciones de rivalidad entre las empresas, que, como consecuencia, permiten ofrecer al consumidor una mayor diversidad de productos de calidad a precios competitivos. Se reconoce la rivalidad entre empresas o marcas como un acto natural y sano en el proceso de libre competencia, siendo el elemento que garantiza a largo plazo el funcionamiento más eﬁciente de los entes que participan en el mercado, por ejemplo productores, distribuidores, comercializadores e importadores (Chen, 1996). Dadas las condiciones de libre competencia, el consumidor ﬁnal tendrá la libertad de elegir entre diferentes opciones y será él quien demande mejores niveles de calidad, precio y servicio. De esta manera, la libre competencia que avala la legislación con miras a garantizar el desarrollo económico, tiene 2 condiciones: el respeto por la competencia y el respeto por el consumidor. Se deﬁne competencia desleal como «todo acto o hecho que se realice en el mercado con ﬁnes concurrenciales, cuando resulte contrario a las sanas costumbres mercantiles, al principio de la buena fe comercial, a los usos honestos en materia industrial o comercial, o bien cuando esté encaminado a afectar o afecte la libertad de decisión del comprador o consumidor, o el funcionamiento concurrencial del mercado” (Congreso de Colombia, 1996, pp. 1-2). Bajo los parámetros establecidos por la ley, este documento revisa el concepto de imitación como una acción estratégica a través de la cual las organizaciones retoman el aprendizaje de una categoría de productos para soportar sus propios desarrollos. La organización imitadora aprende del líder aspectos que son observables en el mercado, como por ejemplo, el empaque y características físicas del producto. El proceso de observar la competencia y subsecuentes acciones de imitación forman parte de la dinámica competitiva de las organizaciones y son importantes para motivar su crecimiento y productividad (Kale y Little, 2007; Madsen et al., 2010). El crecimiento organizacional también implica que a largo plazo la imitación se complementa con la innovación siendo una dinámica importante en el crecimiento económico de un país. Ejemplos de la dinámica entre imitación e innovación se pueden observar en el proceso de desarrollo de Japón (Yamamura, Sonobe y Otsuka, 2005), Corea (Leonard, 1997), China (Chen, 2009; Dobson y Safarian, 2008) e India (Kale y Little, 2007). Teniendo en cuenta las características emergentes de los países iberoamericanos, este artículo se centra en el proceso de imitación, siendo una etapa que puede parecer más atractiva para empresarios que buscan mecanismos para competir de manera rápida y eﬁciente con el mercado internacional. La imitación parecería ser una estrategia pertinente en la medida en que permite a las empresas ponerse al día en términos competitivos y ofrecer productos tan buenos como aquellos que ya han alcanzado el liderazgo en la categoría. Este artículo muestra casos en los que la estrategia de imitación puede tener consecuencias positivas para la sociedad y la competitividad organizacional y otros casos en los que la estrategia es sancionada al ser considerada como competencia desleal. La comparación de casos exitosos y casos erróneos por el uso de la imitación busca evidenciar el alcance de la estrategia en términos competitivos ya que por un lado fortalece la oferta del mercado, pero por otro lado vulnera la protección a la marca y al consumidor. Este artículo está organizado de la siguiente manera: el marco teórico explica la imitación como parte del ciclo de vida del producto y cómo esta estrategia antecede a la innovación en el caso de países en vías de desarrollo. En la segunda sección, se utiliza la metodología exploratoria con base en una búsqueda documental acudiendo a fallos emitidos por la jurisprudencia colombiana, para ﬁnalmente reportar 15 casos que se han presentado en el entorno de competencia desleal. A partir de los casos seleccionados por su relevancia, la discusión en la tercera sección se centra en la reﬂexión a nivel estratégico que analiza las razones que puede tener una empresa para imitar y para no hacerlo. Finalmente, se concluye que la imitación es una estrategia viable siempre que se respete la marca como un atributo distintivo que tiene la capacidad de motivar la competitividad empresarial. 2. La imitación como una acción competitiva El siguiente marco teórico introduce el concepto de imitación en el proceso de evolución de un producto. En términos generales, la imitación sigue a la innovación como consecuencia de la madurez y el nivel de competencia entre las empresas de una categoría (Mukoyama, 2003). La competitividad se reﬁere al esfuerzo que hacen las empresas por obtener ganancias a partir de la oferta de productos al consumidor en igualdad de condiciones que los demás oferentes. Una empresa competitiva buscará diferenciar su marca de las demás a través de mejores características asociadas al producto o al posicionamiento de la marca. Teniendo en cuenta las condiciones propias de un mercado competitivo, el marco teórico explica a) la dinámica natural entre la innovación y la imitación y b) el proceso de imitación para el caso de países en desarrollo. 2.1. La imitación en el ciclo de vida del producto El ciclo de vida de un producto normalmente evoluciona desde su creación hacia la mejora de características de calidad y variedad en la medida en que aumenta la demanda y llega a ser comercializado por varias empresas (Grant, 2008). Cuando las características tecnológicas de un producto logran ser imitadas y existe claramente en el mercado un amplio número de consumidores, nuevas empresas competidoras estarán interesadas en acceder a la comercialización de dicho producto. Es decir, la evolución natural de un producto va desde su innovación hasta su imitación en un mercado de consumo masivo (Mukoyama, 2003). Las ﬁrmas que entran en la etapa de la innovación asumen los costos de investigación y desarrollo, así como los de publicidad; estas ﬁrmas se conocen como pioneras y son las que inﬂuyen en la creación de nuevos productos. A.M. Arboleda / Estudios Gerenciales 30 (2014) 145–152 Las ﬁrmas que entran a competir más adelante, cuando hay un claro desarrollo de producto y una mayor demanda del mismo, se conocen como seguidoras; estas son las organizaciones que aprenden de las líderes a partir de la imitación y adaptación de sus productos (Lieberman y Montgomery, 1998). Independientemente de cuál es la empresa pionera, en la medida en que la demanda del producto aumenta habrá una difusión de los procesos de producción y comercialización. De esta manera, el proceso de imitación comienza una vez la empresa pionera logra el reconocimiento del producto en el mercado y las características del producto pueden ser adquiridas e imitadas. Al observar 2 productos con iguales características físicas y funcionales, la ventaja competitiva que tiene la empresa innovadora disminuye y el principal diferencial comienza a ser el precio. Con el ﬁn de evitar una guerra de precios, las empresas que compiten en una categoría buscarán mejoras incrementales en el producto o en el servicio asociado de manera que les permita diferenciar su marca. Por tanto, la madurez del mercado se maniﬁesta en el máximo potencial de producción de las marcas que compiten en una categoría y el máximo número de usuarios de dicha categoría. Un alto número de marcas compitiendo en una misma categoría está asociado a la capacidad que tienen las empresas para imitar las características del producto líder (Romeo, 1977). De manera complementaria, la madurez del mercado también se asocia a un mayor énfasis por la protección de los derechos de propiedad intelectual (Braguinsky, Gabdrakhmanov y Ohyama, 2007). Ya que la innovación disminuye y la imitación aumenta, las empresas hacen un mayor esfuerzo por establecerse en un nicho de mercado con características de producto asociadas a la marca y protegidas bajo el registro de marca. Bajo condiciones competitivas las empresas no cuentan con protecciones diferentes al registro de patente y protección a la marca para contrarrestar las acciones de otras empresas. Una empresa competitiva tiene 2 alternativas para que su marca sobreviva en el mercado. Una opción es cambiar rápidamente las características de su producto para diferenciarse de los imitadores y una segunda opción es incurrir en acciones legales protegiendo su producto de la imitación (Braguinsky et al., 2007). La primera opción haría más difícil consolidar la imagen del producto en la mente del consumidor; además, dada la capacidad de imitación de los competidores en un mercado maduro la continua inversión en innovación no resulta ser rentable (Braguinsky et al., 2007). Por tanto, la segunda alternativa, que busca mayor protección a la marca, resulta ser una mejor estrategia para limitar el número de competidores en el mercado y para apaciguar la tentación de imitar. Resumiendo, la imitación es un síntoma de un mercado maduro con un alto nivel de competencia entre las empresas del sector. En el desarrollo de un producto, naturalmente ocurre primero la innovación y posteriormente la imitación. En esta segunda etapa, las empresas líderes buscan argumentos legales que les permitan evitar la imitación mientras que las empresas seguidoras buscan aprender e imitar las características del producto líder. La transición de la innovación a la imitación es el proceso natural en la evolución de un producto en su industria, sin embargo, lo rápido que ocurra este proceso depende de características competitivas propias de su mercado (Yamamura et al., 2005). 2.2. La imitación como una estrategia de crecimiento La tendencia hacia la imitación o la innovación es una diferencia clara al comparar países según el índice de competitividad de la industria. De acuerdo con diferentes reportes en competitividad, los países en desarrollo tienden a reportar índices inferiores a los países desarrollados; esto se debe a mediciones como el capital disponible para los negocios, el producto interno bruto, el nivel de tecnología, la innovación, entre otros (Lall, 2001; Önsel et al., 2008). 147 Teniendo en cuenta las capacidades y características propias de las organizaciones en países en desarrollo, la dinámica entre la innovación y la imitación ocurre de manera inversa (Chen, 2009; Dobson y Safarian, 2008; Kale y Little, 2007; Yamamura et al., 2005). Las empresas en países en desarrollo no cuentan con los recursos en términos de capital, conocimiento y tamaño de la demanda como para realizar la inversión inicial que implica la innovación. Dadas estas capacidades, las políticas nacionales pueden estar motivadas a proteger la industria limitando la entrada de productos y organizaciones extranjeras. En el momento en que dichas organizaciones se enfrentan a la apertura de importaciones y a la posibilidad de satisfacer una demanda externa, deben mejorar sus estándares de producción y comercialización para asemejarse a la oferta internacional. La apertura de importaciones a países en desarrollo se asocia al crecimiento del país a través del uso de tecnología y la imitación. Por un lado, los procesos productivos pueden ser más eﬁcientes gracias a la importación y el uso de nuevas tecnologías. Por otro lado, la apertura abre el mercado al desarrollo de «nuevos» productos a través de la imitación (Connolly, 2003). Es decir, la imitación es la forma como las empresas de países en desarrollo pueden alcanzar rápidamente estándares internacionales; es una manera de compensar las debilidades tecnológicas y en fundamentos cientíﬁcos para el desarrollo de nuevos productos que pueden tener las organizaciones de países en desarrollo (Chen, 2009). Organizaciones en economías que han estado protegidas y han tenido bajos estándares de competitividad tienen un gran incentivo para alcanzar los niveles de organizaciones desarrolladas (Shinkle y McCann, 2013). Lo que muchas empresas en países en vías de desarrollo hacen es imitar productos que han sido desarrollados internacionalmente y que muchas veces llegan al país a través de importaciones o gracias a la comercialización que realizan a nivel local las empresas multinacionales (Yamamura et al., 2005). Las organizaciones en países que buscan alcanzar mayores niveles de apertura se destacan por apoyarse en la imitación y mostrar bajos niveles de innovación (Shinkle y McCann, 2013). Adicionalmente, conociendo las características del mercado local, la imitación puede ser más fácil para una empresa local en comparación con el conocimiento del contexto del que carece la empresa pionera (internacional). Esto se debe a que el nivel de incertidumbre de la demanda por un producto es alto como para que el líder internacional decida inicialmente hacer una inversión en investigación y desarrollo para el mercado local (Lieberman y Asaba, 2006). Al mismo tiempo, un alto grado de imitación desincentiva a las empresas a realizar inversiones para la creación de productos nuevos, lo cual posteriormente signiﬁca menores tasas de crecimiento para las empresas (Braguinsky et al., 2007). En este punto, es importante proteger la propiedad industrial y la inversión en investigación y el desarrollo de productos realizada por la empresa privada. Estos nuevos productos son potencialmente la ventaja competitiva que fortalece el crecimiento organizacional. En resumen, la innovación y la imitación forman parte de un proceso que ocurre gracias a la evolución de los productos en el mercado (Teece, Pisano y Shuen, 1997), pero este proceso tiene lugar de manera inversa en el caso de empresas de países en desarrollo. La imitación es una primera instancia que permite a la organización formar un capital y fortalecer su capacidad para competir. Sin embargo, a largo plazo la imitación puede frenar el crecimiento ya que inhibe la creación de productos con valor diferencial. Así, una legislación que promueve la libre competencia motivará la imitación como un acto competitivo siempre y cuando se acoja al respeto por la propiedad industrial y por el consumidor (SIC, 2013). Teniendo en cuenta este marco teórico, el siguiente estudio ejempliﬁca cómo la imitación motiva el crecimiento del mercado, fortalece el número y la distintividad de la marca y ofrece más alternativas al consumidor ﬁnal. Por otro lado, la imitación no 148 A.M. Arboleda / Estudios Gerenciales 30 (2014) 145–152 Tabla 1 Casos según consecuencia de la imitación Líder Seguidor Categoría Fortalecimiento de la oferta Actimel Cola (referente a la categoría) Acid-Ness Vick VapoRub Cristal Yox Big Cola Acid-Mantle AlibRub Agua pura Yogur funcional Gaseosa Crema para la piel Ungüento tópico Agua embotellada Protección a la marca Sultana Nopión Kokorikosaurios Oma Bocatto Santana Chenpión Nuggetsaurios Coloma Rokotto Galleta Champú antipiojos Alimento base pollo Café Helado Competencia desleal Bon Bon Bum Ponky Trolli Full Throttle Finess Bin Bum Pompy Trully Full Cola Men‘s Fitness Chupeta Ponqué Caramelo de goma Refresco energético Yogur/ropa masculina Fuente: elaboración propia. es aceptable en casos en los que crea competencia desleal, afecta negativamente al consumidor siendo engañosa o creando confusión y debilitando las marcas. La competencia desleal es el término que utiliza la jurisprudencia para señalar que la imitación no se hace respetando la propiedad industrial. Este señalamiento y vigilancia es importante porque el respeto a la propiedad (por ejemplo, marca, patente) es lo que motiva a largo plazo el crecimiento de las organizaciones, y con ello el crecimiento de un país. 3. Metodología Este estudio utiliza una metodología exploratoria para realizar la búsqueda de casos que han sido clasiﬁcados en la jurisprudencia colombiana como competencia desleal. Como fuente de información se utilizó la página web del Noticiero Oﬁcial (2013), revisas y periódicos colombianos en los que se reportan noticias del entorno legal: Dinero, Portafolio y La República. De estas fuentes se extrajeron fallos emitidos por el Consejo de Estado o la Superintendencia de Industria y Comercio (SIC). Los casos seleccionados se eligieron por ser relevantes para diferentes industrias y por representar de manera clara las decisiones jurisprudenciales en cuanto a competencia desleal por imitación. Adicionalmente, al utilizar casos reportados por fuentes oﬁciales como casos de competencia desleal se busca eliminar el sesgo del investigador al juzgar por sí mismo si la semejanza entre 2 competidores de una categoría representa o no una acción de imitación. Así pues, se seleccionó un total de 15 casos que se organizan conceptualmente de acuerdo con las implicaciones que tienen sobre el fortalecimiento de la oferta y la protección a la marca (tabla 1). Primero, el fortalecimiento de la oferta, se ilustra a través de 5 casos; estos primeros casos representan la situación en la que la estrategia de imitación motiva la competencia. Segundo, los casos de protección de marca pueden ser situaciones que desempeñan un papel a favor o en contra de la competencia. Teniendo esto en cuenta, se incluyen 5 casos en los que la competencia no genera confusión y otros 5 en los que sí ocurre confusión y constituyen competencia desleal. 3.1. Fortalecimiento de la oferta Permitir que existan diferentes marcas en una misma categoría de productos estimula la competencia, evitando que una sola marca ejerza una posición de poder en el mercado. A continuación se presentan 5 casos en los que se muestra cómo es pertinente la Figura 1. Actimel-Yox. Fuente: MedTempus (2007); Fundamerani (2014). existencia de diferentes marcas de productos semejantes siempre y cuando estas no generen confusión en el consumidor. El mercado de los lácteos ha venido creciendo gracias a los productos funcionales; crecimiento que ha ocurrido gracias a la disputa entre Danone y Alpina. Danone es el líder internacional en productos lácteos, siendo Actimel una de sus marcas que ofrece el beneﬁcio de crear defensas en el organismo. Antes de la entrada de Danone en Colombia, Alpina lanzó Yox, un producto con beneﬁcios y empaque semejante (ﬁg. 1). Sin embargo, Danone ha perdido sus acciones de nulidad tramitadas ante el Consejo de Estado solicitando la cancelación de registro de marca de su competidor. En contra de las demandas de Danone, la Corte Suprema de Justicia no ha contemplado el retiro de la marca Yox y considera pertinente la presencia de ambas marcas en el mercado (Portafolio, 2009). Por otro lado, la categoría de gaseosas se amplía gracias a la entrada de la marca peruana Big Cola. La sociedad Posada Tobón S.A. argumentaba confusión por parte de Big Cola, pero el Consejo de Estado falló a favor de Big Cola entendiendo que la marca no genera confusión en el consumidor y la expresión Cola corresponde con el vocabulario general en este tipo de bebidas (Portafolio, 2012d). Asimismo, la demanda impuesta por Bayer a Tecnoquímicas por la reproducción de aspectos característicos de la presentación comercial de Acid-Ness a través de la marca Acid-Mantle fue desestimada por la SIC, reconociendo que las semejanzas entre productos no constituyen un acto desleal de confusión (Varela-Sánchez, 2012). Un caso semejante es el de Vick VapoRub de Proter & Gamble (P&G) y AlibRub de Tecnoquímicas. El demandante, P&G, solicitaba ante la SIC que Tecnoquímicas retirara del mercado su producto teniendo en cuenta que este estaba utilizando la misma apariencia (ﬁg. 2). Tal como en el caso anterior, la SIC favoreció a la empresa demandada teniendo en cuenta que la reproducción de signos distintivos de un producto (por ejemplo, colores o forma en el empaque) no constituye competencia desleal (Varela-Sánchez, 2012). La competencia desleal ocurre en el caso en que se imiten presentaciones mercantiles, es decir, aspectos propios de una marca. Sin embargo, si las palabras que conforman la marca son descriptivas del producto pueden ser utilizadas por otras empresas. Por ejemplo, Colombina compite en el mercado de agua embotellada con la marca Agua Pura, registro avalado por la SIC a pesar de la A.M. Arboleda / Estudios Gerenciales 30 (2014) 145–152 149 Figura 2. Vick VapoRub-AlivRub. Fuente: Revista Dinero, 2012. demanda instaurada por Postobón (que comercializa agua Cristal, Oasis, Vitality y H20). La SIC respaldó la marca Agua Pura reconociendo la debilidad de la marca ya que estos términos descriptivos y genéricos pueden ser utilizados libremente por marcas de la categoría; los derechos son dados sobre el conjunto de elementos que representan la marca (Mancera, 2013). 3.2. Protección a la marca La marca es un signo distintivo por excelencia. Para un consumidor la marca es el elemento que le permite reconocer las características y calidad del producto y diferenciarlo frente a otros de la categoría. Para una empresa la marca garantiza el valor y los derechos sobre este activo intangible (SIC, 2013). Asimismo, dado el uso de la estrategia de imitación, la marca se constituye en el factor que permite la diferencia y por tanto previene la confusión del consumidor (Arboleda y Alonso, 2010). A continuación se describen 5 casos en los que el Consejo de Estado o la SIC apoyan el carácter distintivo de la marca, fallando en contra de registros que puedan generar confusión en el consumidor. Posteriormente, se presentan 5 casos en los que el tribunal avala registros de marca que no generen confusión. En la categoría de bombones, el Consejo de Estado apoyó la demanda de Colombina con su marca Bon Bon Bum e impidió el registro de la marca competidora Bin Bum; el demandante argumentaba similitud fonética que generaría confusión en el consumidor (Portafolio, 2011). De manera semejante, teniendo en cuenta la similitud fonética, el Consejo de Estado declaró la nulidad del registro de la marca Pompy (comercializado por Meals de Colombia) por su semejanza con la marca Ponky, marca de Colombina (Portafolio, 2012c). Un caso consistente es el de Trolli (de Procaps) y Trully (de Comestibles Aldor). La demanda de Procaps sobre Aldor fue respaldada por la SIC, negando el registro de la marca Trully al ser un nombre que no es consecuente con el principio de distintividad de las marcas (Bedoya, 2013). Adicionalmente, una marca que sea semejante (o igual) puede ser negada aunque se encuentre acompañada de una palabra genérica que se introduzca con la expectativa de hacerla distintiva. Por ejemplo, la SIC negó el registro de Full Cola ya que a pesar de que Cola es un vocablo general, la palabra Full ya es utilizada por otra marca de bebidas de la categoría: Full Throttle de CocaCola (Santamaria, 2012). Por otro lado, Finess vs. Fitness es un ejemplo de cómo se busca proteger la marca y la confusión del consumidor incluso cuando las marcas que se asemejan no corresponden a una misma categoría. El Consejo de Estado ordenó la cancelación del registro de la marca Men’s Fitness, marca de ropa masculina y calzado, por su semejanza con la marca de yogur Finesse. A pesar de que corresponden a categorías totalmente diferentes, el tribunal considera que el Figura 3. Oma y Coloma. Fuente: FamiliasUnidas, 2013. alto posicionamiento que tiene el yogur Finesse de Alpina podría llevar al consumidor a pensar que esta marca de ropa pertenece a la organización Alpina (Portafolio, 2008a). Contrario a lo observado en estos 5 ejemplos, existen otros casos en los que no se encontraron similitudes sustanciales entre las marcas que puedan llevar al consumidor a confundirse y por tanto el regulador avaló la presencia de ambas en el mercado. En la categoría de galletas se tiene el caso de Sultana y Santana. El Consejo de Estado consideró que estas 2 marcas pueden coexistir y que sus nombres no generan confusión en el consumidor por el signiﬁcado de las palabras y porque su estructura fonética es diferente (Portafolio, 2012a). Bajo el mismo argumento, el Consejo de Estado rechazó la demanda de Tecnoquímicas en defensa de su marca Nopión frente al registro de marca aprobado para la marca Chenpión (Ostau de Lafont Pianeta, 2011). La competencia directa entre las marcas Nopión y Chenpión en la categoría de champú antipiojos se consideró pertinente. En el caso de alimentos elaborados a base de pollo, el Consejo de Estado aprobó la coexistencia de las marcas Kokorikosaurios y Nuggetsaurios ya que no hay ninguna similitud ortográﬁca o fonética. Además, está claro para el consumidor que ambas marcas utilizan un vocablo de fantasía para nombrar el producto con base en pollo (Portafolio, 2012b). Asimismo, el Consejo de Estado, aprobó la coexistencia de las marcas de café Oma y Coloma (ﬁg. 3) sin perjuicio de causar confusión en el consumidor (Portafolio, 2008b). Finalmente, la demanda de Bocato (de Meals de Colombia S.A.) contra Rokotto (de la sociedad Alicorp S.A.), argumentando confusión por semejanza entre las marcas, fue rechazada por el Consejo de Estado, el cual avala la existencia de ambas. En los pleitos en los que se argumenta competencia desleal por confusión en el consumidor dada la semejanza entre las marcas, lo que prima es el carácter distintivo de la marca como aspecto que le permite al consumidor reconocer el fabricante y los atributos del producto al momento de la toma de decisión. La diferencia en la estructura fonética y las características simbólicas de los términos utilizados para la marca son argumentos claves para demostrar que una marca es realmente distintiva y desestimar la posibilidad de confusión. 4. Discusión: los efectos contrarios de la imitación De acuerdo con los 15 casos observados está claro que hay unas situaciones donde la imitación es una estrategia pertinente que 150 A.M. Arboleda / Estudios Gerenciales 30 (2014) 145–152 motiva el fortalecimiento de la categoría y beneﬁcia al consumidor. Sin embargo, en otros casos la imitación se constituye en un acto desleal perjudicando la fundamentación de la libre competencia así como al consumidor. 4.1. Por qué imitar: beneﬁcios de la imitación Existen 2 grandes teorías que explican la imitación: las teorías basadas en la rivalidad y las teorías basadas en la información. Los casos analizados permiten representar ambas teorías como se discutirá a continuación. 4.1.1. Participación de mercado Las teorías basadas en la rivalidad entre empresas interpretan la imitación como una respuesta diseñada para mitigar la rivalidad competitiva de las otras empresas (Yamamura et al., 2005). Esto signiﬁca que las empresas imitan a otras esperando mantener su posición relativa o neutralizar la acción agresiva de sus rivales. Las empresas pueden actuar como seguidoras, no porque este sea su principal interés en el mercado, sino por ser una acción que debilita la participación de su competidora. Este tipo de imitación implica que las empresas competidoras son relativamente semejantes en su dominio sobre el mercado. Es decir, una empresa reacciona con una imitación frente a la estrategia innovadora de otra si considera que la pionera es un rival que puede afectar negativamente su participación de mercado. Este es el caso de Danone y Alpina. Las empresas compiten en varios segmentos lácteos a nivel internacional. El debate entre Actimel y Yox es un pleito que representa su rivalidad por el liderazgo. Adicionalmente, que en el mercado exista la posibilidad de una intensa competencia es un incentivo para que las empresas asuman comportamientos paralelos de imitación (Lieberman y Asaba, 2006). Las empresas dentro de una misma industria tienden a comportarse de manera semejante; a largo plazo no es conveniente para el grupo de empresas tener estrategias muy diferentes porque esto reduce el poder de la industria como un todo y por lo tanto su rentabilidad. En otras palabras, empresas de una misma industria tienden a adoptar comportamientos y estrategias semejantes, lo que les permite permanecer tácitamente aliadas restringiendo la entrada de nuevas empresas (Lieberman y Asaba, 2006). Por ejemplo, BigCola es una marca nueva de una empresa que no había estado compitiendo en el mercado colombiano de agua embotellada. Lo mismo ocurre en el caso de Agua Pura. Ante el lanzamiento de una marca nueva, la reacción natural de las empresas ya posicionadas es tratar de impedir el ingreso de un nuevo competidor que entra a dividir la participación de mercado de una categoría ya madura. La imitación es una manifestación de la rivalidad entre empresas compitiendo por ganar participación de mercado. En este sentido, la imitación tiene consecuencias positivas para la sociedad, no en términos de calidad sino de cantidad, al extender la oferta del mercado (Sohn, 2008). 4.1.2. Comunicación de atributos La imitación por información ocurre cuando es ideal para una empresa observar las acciones de aquellas más desarrolladas; las empresas seguidoras aprenden de la innovadora acerca de las características con las que se debe desarrollar un producto. De acuerdo con la teoría basada en la información, la imitación ocurre porque la empresa que imita espera comunicar al consumidor que su producto tiene características semejantes a las ya conocidas en el líder. Por esta razón, los comportamientos observados en el producto líder serán valorados por las empresas seguidoras asumiendo que las pioneras informan (a través de su producto) acerca de las características necesarias para participar en el mercado (Bikhchandani, Hirshleifer y Welch, 1992, 1998). Los casos analizados, cuya estrategia competitiva es aceptada de acuerdo con el fallo de la corte, se pueden explicar a partir de esta teoría. Por ejemplo, los consumidores de AlibRub van a entender cuáles son los atributos del producto al relacionarlo con el líder de la categoría, Vick VapoRub. Lo mismo ocurre con la estrategia de Acid-Mantle, Nuggetsaurios, Coloma, Rokotto, Chenpión y Santana. Estas marcas en su conjunto (nombre, empaque, colores) permiten evocar los atributos del líder respetando el carácter distintivo de la marca líder y seguidora. Es importante notar que esta competencia ocurre en categorías maduras donde la entrada de nuevos competidores no signiﬁca necesariamente un aumento en la demanda, pero sí mayores opciones para el consumidor. Ya que el mercado no está en un momento de crecimiento, no está claro que la empresa líder tenga realmente una ventaja competitiva sobre la seguidora (imitadora). Los estudios muestran que en muchos casos no es así y que las características de la marca líder es aprovechada por las empresas imitadoras (Levitt y March, 1988; Lieberman y Asaba, 2006). La teoría basada en la información resulta ser una buena explicación a la imitación en el caso de empresas en países en desarrollo. Estas empresas cuentan con menos recursos, de capital o conocimiento, que son necesarios para innovar; adicionalmente, en estos países, si la demanda por el producto es pequeña y la empresa se limita al mercado local. El incurrir en costos de innovación puede implicar pérdida de valor, contrario a lo que podría esperarse de la innovación. Además, la imitación o la adaptación de productos ya conocidos permite a las empresas minimizar riesgos relacionados con la innovación y entrar en nuevos mercados, riesgos que las empresas no están preparadas para asumir (Lieberman y Asaba, 2006). Para resumir, el proceso de imitación es una estrategia competitiva que permite a las empresas defender su participación en un mercado especíﬁco. Además, la estrategia de imitación permite a las empresas aprender de lo que las líderes ya han construido, comunicando al consumidor atributos semejantes a aquellos que son reconocidos en la marca líder. 4.2. Por qué no imitar: perjuicios de la imitación Esta sección discute cómo la imitación puede ser perjudicial desde 2 perspectivas complementarias entre sí: competencia desleal y competitividad en el mercado. A corto plazo, la imitación es una estrategia vigilada por la ley, pero a largo plazo la imitación es sancionada por el mercado. 4.2.1. Competencia desleal En la práctica, las organizaciones prestan especial atención a las acciones de sus competidores, cuestionando aquellos comportamientos que potencialmente pueden enmarcarse como conductas desleales o puedan signiﬁcar una transgresión de los derechos de propiedad industrial. Así, la organización de la marca líder buscará proteger su marca. Al mismo tiempo, la organización imitadora debe garantizar que, aunque utilice características ya reconocidas como ideales para un tipo de producto, el uso de estos atributos no sea una imitación exacta o genere confusión en el consumidor. La imitación tiene unos límites deﬁnidos en la ley y no acatar dichos límites genera sanciones para quienes ejecutan dichos actos. Las sanciones se reﬁeren a la cancelación del registro de marca, imposibilidad de comercializar dicha marca y retribución económica a la parte afectada. Los 5 casos analizados como fallos negativos a la imitación argumentan competencia desleal evidenciando una clara similitud entre la marca seguidora y la marca líder (tabla 1; competencia desleal). El problema observado consistentemente en estos casos es que la similitud entre las marcas no permite al consumidor diferenciarlas. Dadas las condiciones de los productos en mercados A.M. Arboleda / Estudios Gerenciales 30 (2014) 145–152 maduros, las características en la apariencia y funcionalidad del producto son evidentes y relativamente fáciles de imitar. Por tanto, bajo este contexto el único aspecto que es realmente diferente y valioso es la marca. Por esta razón, la madurez del mercado se asocia con una alta inversión en registros y actividades legales que vigilan el patrimonio de la marca (Braguinsky et al., 2007). En términos de consumidor, la confusión entre las marcas conlleva una evaluación negativa y sentimientos de frustración (Arboleda, 2011; van Horen y Pieters, 2012). En términos sociales, un alto grado de imitación puede afectar negativamente la competitividad empresarial como se discute a continuación. 4.2.2. Competitividad La innovación es lo que permite a una empresa crear un valor diferencial y una mayor rentabilidad. Sin embargo, si no existe protección sobre la inversión en innovación las empresas tendrán un menor incentivo para innovar (Aghion, Harris, Howitt y Vickers, 2001; Segerstrom, 1991). Esta es una decisión que afecta a las empresas líderes pero también a la sociedad en general en términos de la calidad de productos disponibles en el mercado (Sohn, 2008). Para lograr una ventaja competitiva una ﬁrma debe garantizar ser realmente diferente a las demás. La innovación permite a una empresa distanciarse de lo que ofrecen las demás adquiriendo una ventaja competitiva (Peteraf, 1993). Al innovar es importante que la empresa aprenda procedimientos, adquiera un know-how, desarrolle tecnologías o capacidades humanas que no estén fácilmente disponibles en el mercado o que no sean asequibles por otras empresas (Parthasarathy, Chenglei y Aris, 2011). Estas barreras de entrada permitirán a una empresa que su innovación sea realmente una característica diferencial y que conlleve a una mayor rentabilidad (Peteraf, 1993). Si la exclusividad sobre el producto o la característica diferencial se pierden, gracias a la imitación que hacen otros, la empresa no tendrá una ventaja competitiva en su mercado y por tanto no tendrá un argumento para crear valor y mejorar la rentabilidad. En una empresa, para garantizar el retorno en la inversión de su innovación es esencial que existan barreras fuertes sobre la disponibilidad de recursos (de producción o conocimiento). Cuanto más tiempo pueda la empresa conservar estas barreras más tiempo podrá sostener una ventaja competitiva y más ganancias obtendrá del producto innovador (Parthasarathy et al., 2011). El objetivo de las barreras de entrada es hacer la imitación más costosa para las empresas seguidoras y garantizar una ventaja competitiva duradera a la empresa pionera. Las barreras de entrada se pueden crear a través de los procesos de fabricación o generando diﬁcultades en la imitación dadas las características propias del producto. Cuando esto no es posible, las barreras se establecen a través de argumentos legales que se amparan en la propiedad industrial y derechos de autor (Fan, Gillan y Yu, 2013). Sin embargo, la protección dada por la ley vela por el bienestar social y por lo tanto procurará, por un lado, que la imitación genere mayor competitividad entre las empresas al aumentar la oferta del mercado, y por otro, que la imitación no afecte la competitividad en términos de innovación. 5. Conclusiones y limitaciones La imitación tiene potenciales beneﬁcios y perjuicios para las empresas, el consumidor y la sociedad en general. Desde el punto de vista positivo, la imitación aumenta la oferta de productos disponibles. Desde el punto de vista negativo, la imitación debilita el interés por innovar (Sohn, 2008) y vulnera la conﬁanza de los consumidores en las marcas al generar confusión. Los 15 casos observados representan ambos puntos de vista, el positivo y el negativo, mostrando 151 que la estrategia de imitación es viable siempre que sea un soporte para el crecimiento del mercado (Iwai, 1984) y redunde en beneﬁcio para el consumidor. La coexistencia de productos semejantes cuyas marcas sean distintivas respetan al consumidor, previniendo la confusión, y aumentan la oferta de productos de una misma categoría, estimulando la competitividad del mercado. La imitación es un proceso natural en mercados donde las características del producto son evidentes y por tanto fáciles de imitar. Restringir la imitación en un merado maduro no es coherente con la búsqueda de competitividad. En estas condiciones, el consumidor espera el mayor número posible de oferentes en el mercado y asimismo existe una gran demanda por productos de la categoría. Sin embargo, dada la semejanza entre productos de una categoría, la marca debe sobresalir como un atributo único y distintivo. Consecuentemente, la imitación detallada y exacta es un acto de competencia desleal que daña la propiedad que ha construido la marca y afecta al consumidor, llevando a un consumo erróneo y contra su voluntad. La estrategia de imitación debe contar con ambos aspectos, una regulación clara y un mercado competitivo. Por un lado, la regulación gubernamental debe proteger los derechos del consumidor y las marcas que invierten en investigación y desarrollo, lo cual garantiza la motivación para innovar a largo plazo (Fan et al., 2013). Por otro lado, la competencia organizacional asegura que haya un proceso dinámico en el ciclo de innovación e imitación, garantizando que la empresa que desarrolla un producto puede recibir una retribución por su inversión inicial en investigación y desarrollo, y también que otras organizaciones pueden acceder al mercado de dicho producto expandiendo al máximo su demanda. La presencia de diversos oferentes en un mismo mercado obliga a repensar el producto a través de la creación de nuevos productos o modiﬁcaciones del producto inicial. Este artículo realiza un análisis de la práctica de imitación desde el punto de vista estratégico a través de casos que se han fallado teniendo en cuenta la jurisprudencia colombiana de acuerdo con el concepto de competencia desleal por imitación. Sin embargo, la preocupación por el uso de la estrategia de imitación es común a cualquier país interesado en fortalecer las características competitivas de sus empresas y por lo tanto dicho análisis sería pertinente en otros países. Adicionalmente, el tipo de análisis realizado en el artículo permite profundizar en los argumentos a favor o en contra de la imitación, pero no demuestra hasta qué punto la práctica de imitación se falla en mayor proporción como una acción desleal o se avala como una acción competitiva. Un futuro análisis cuantitativo permitiría deﬁnir si la estrategia de imitación está realmente representando para la sociedad una ventaja al aumentar la oferta de productos y promover la competencia, o por el contrario, la imitación expresa un acto en el que se busca confundir al consumidor y apropiarse de manera indebida de la propiedad construida y protegida por una marca líder. Es posible que las consecuencias positivas de la imitación sean mayores que las negativas en términos sociales (Sohn, 2008) y especialmente en el caso de países que estén buscando mayores niveles de competitividad a nivel nacional e internacional (Mukoyama, 2003; Shinkle y McCann, 2013). Finalmente, no es un objetivo de este estudio ilustrar al lector acerca de la legislación que discute el concepto de competencia desleal. El artículo incorpora dicha legislación ya que esta determina que la práctica de imitación sea posible y los 15 fallos representan las consecuencias de dicha práctica. Sin embargo, para mayor profundidad con relación a la competencia desleal es recomendable remitirse directamente a la legislación referenciada. Conﬂicto de intereses La autora declara no tener ningún conﬂicto de intereses. 152 A.M. Arboleda / Estudios Gerenciales 30 (2014) 145–152 Bibliografía Aghion, P., Harris, C., Howitt, P. y Vickers, J. (2001). Competition, imitation and growth with step-by-step innovation. The Review of Economic Studies, 68(3), 467–492. Arboleda, A. M. (2011). Actitudes del consumidor después de elegir un producto equivocado. Diversitas: Perspectivas en Psicología, 7(2), 211–224. Arboleda, A. M. y Alonso, J. C. (2010). Determinantes de una elección errada al escoger medicamentos de venta libre. Innovar, 20(38), 175–190. Bedoya, A. M. (2013). Marcas y competencia desleal, un terreno para duelo de titanes. La República [acceso 21 Jun 2013]. Disponible en: http://www.larepublica.com. co/asuntos-legales/los-10-pleitos-de-marca-que-m%C3%A1s-sonaron-en2012-en-la-sic 29673 Bikhchandani, S., Hirshleifer, D. y Welch, I. (1992). A theory of fads, fashion, custom, and cultural change as informational cascades. The Journal of Political Economy, 100(5), 992–1026. Bikhchandani, S., Hirshleifer, D. y Welch, I. (1998). Learning from the behavior of others: Conformity, fads, and informational cascades. The Journal of Economic Perspectives, 12(3), 151–170. Braguinsky, S., Gabdrakhmanov, S. y Ohyama, A. (2007). A theory of competitive industry dynamics with innovation and imitation. Review of Economic Dynamics, 10(4), 729–760. Chen, M. (1996). Competitor analysis and interﬁrm rivalry: Toward a theoretical integration. Academy of Management Review, 21(1), 100–134. Chen, Z. (2009). A brief history of China’s Y-10: Imitation versus innovation. Technology in Society, 31(4), 414–418. Congreso de Colombia. (1996). Ley 256: Por la cual se dictan normas sobre competencia desleal. Diario Oﬁcial [acceso 1 Jul 2012]. Disponible en: http:// www.secretariasenado.gov.co/senado/basedoc/ley/1996/ley 0256 1996.html Connolly, M. (2003). The dual nature of trade: measuring its impact on imitation and growth. Journal of Development Economics, 72(1), 31–55. Dobson, W. y Safarian, A. E. (2008). The transition from imitation to innovation: An enquiry into China’s evolving institutions and ﬁrm capabilities. Journal of Asian Economics, 19(4), 301–311. FamiliasUnidas (2013). Café en Colombia. FamiliasUnidas.com.co [acceso 1 Oct 2013]. Disponible en: http://www.co.all.biz/caf-bgg1085569 Fan, J. P. H., Gillan, S. L. y Yu, X. (2013). Innovation or imitation?: The role of intellectual property rights protections. Journal of Multinational Financial Management, 23(3), 208–234. Fundamerani. 2014. Fundación Internacional de Pedagogía Conceptual: Supertienda [acceso 26 Mar 2014]. Disponible en: http://www.fundamerani.org/ supertienda/inicio/index.php?route=product/product&path=35&product id= 325 Grant, R. M. (2008). Contemporary strategy analysis (6th ed). Oxford: Blackwell Publishing Ltd. Iwai, K. (1984). Schumpeterian dynamics: An evolutionary model of innovation and imitation. Journal of Economic Behavior & Organization, 5(2), 159–190. Kale, D. y Little, S. (2007). From imitation to innovation: The evolution of R&D capabilities and learning processes in the Indian pharmaceutical industry. Technology Analysis & Strategic Management, 19(5), 589–609. Lall, S. (2001). Competitiveness indices and developing countries: An economic evaluation of the global competitiveness report. World Development, 29(9), 1501–1525. Leonard, J. W. (1997). Imitation to innovation: The dynamics of Korea’s technological learning. Academy of Management Executive, 11(4), 102–103. Levitt, B. y March, J. G. (1988). Organizational learning. Annual Review of Sociology, 14(1), 319–340. Lieberman, M. y Asaba, B. S. (2006). Why do ﬁrms imitate each other? Academy of Management. The Academy of Management Review, 31(2), 366–385. Lieberman, M. y Montgomery, D. B. (1998). First-movers (dis)advantage: retrospective and link with the resource-based view. Strategic Management Journal, 19(12), 1111–1125. Madsen, J. B., Islam, M. R. y Ang, J. B. (2010). Catching up to the technology frontier: The dichotomy between innovation and. Canadian Journal of Economics, 43(4), 1389–1411. Mancera, A. d. P. (2013). Marca Agua Pura Colombina será reemplazada por signo genérico. La República [acceso 21 Ene 2013]. Disponible en: http://www. larepublica.co/asuntos-legales/marca-agua-pura-colombina-ser%C3%A1reemplazada-por-signo-gen%C3%A9rico 28908 MedTempus (2007). El bulo por correo de Actimel hace tomar medidas a Danone. MedTempus Blog de medicina de la salud [acceso 1 Oct 2013]. Disponible en: http://medtempus.com/archives/el-bulo-por-correo-de-actimel-hace-tomarmedidas-a-danone/ Mukoyama, T. (2003). Innovation, imitation, and growth with cumulative technology. Journal of Monetary Economics, 50(2), 361–380. Munar-Cadena, P. O. (2005). Ref. Expediente 4018. Bogotá, Distrito Capital. Corte Suprema de Justicia y Sala de Casación Civil [acceso 15 Jul 2012]. Disponible en: http://190.147.213.68:8080/HOMEPAGE/CORTE SUPREMA DE JUSTICIA/SENT CIVIL/2005/4018.htm Noticiero Oﬁcial. (2013). Noticiero Oﬁcial [acceso 15 Dic 2012]. Disponible en: de www.noticieroﬁcial.com Önsel, Ş., Ülengin, F., Ulusoy, G., Aktaş, E., Kabak, Ö. y Topcu, Y. İ. (2008). A new perspective on the competitiveness of nations. Socio-Economic Planning Sciences, 42(4), 221–246. Ostau de Lafont Pianeta, R. E. (2011). Bogotá: Consejo de Estado. Sala de lo Contencioso Administrativo [acceso 21 Ene 2013]. Disponible en: http://www. google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0CCgQFjAA& url=http%3A%2F%2F190.24.134.67%2FSENTENCIAS%2FSECCION%2520PRIMERA %2F2011%2F11001-03-24-000-2004-00102-01.doc&ei=V-ooUrS8NYXo8wS0mYDADQ&usg=AFQjCNHtJp-Yaael4iWF9GZZBX9j3mfGHA&bvm=bv.51773540, d.eWU Parthasarathy, R., Chenglei, H. y Aris, S. (2011). Impact of dynamic capability on innovation. Value creation and industry leadership. The IPU Journal of Knowledge Management, 9(3), 59–73. Peteraf, M. A. (1993). The cornerstones of competitive advantage: A resource-based view. Strategic Management Journal, 14(3), 179–191. Portafolio. (2008a, septiembre 8). Consejo de Estado ordenó a Superindustria cancelar registro de marca Men’s Fitness. Portafolio Economía [acceso 12 Oct 2012]. Disponible en: http://www.portafolio.co/archivo/documento/CMS-4522856 Portafolio. (2008b, noviembre 13). Marcas Oma y Coloma sí pueden coexistir en mercado pues no existiría confusión, sentenció Consejo de Estado. Portafolio [acceso 20 Ene 2013]. Disponible en: http://www.portafolio.co/archivo/ documento/CMS-4665241 Portafolio. (2009, febrero 9). Lío de Danone y Alpina llega al Consejo de Estado. Portafolio [acceso 20 Ene 2013]. Disponible en: http://www.portafolio.co/detalle archivo/MAM-3309162 Portafolio. (2011, abril 25). Consejo de Estado avala a Bon Bon Bum. Portafolio [acceso 20 Ene 2013]. Disponible en: http://www.portafolio.co/negocios/ consejo-estado-avala-bon-bon-bum Portafolio. (2012a, junio 4). Sultana y Santana sí pueden coexistir en el mercado. Portafolio [acceso 20 Ene 2013]. Disponible en: http://www.portafolio. co/detalle archivo/DR-50239 Portafolio. (2012b, agosto 21). Marcas Kokorikosaurios y Nuggetsaurios sí pueden coexistir. Portafolio [acceso 20 Ene 2013]. Disponible en: http://www.portafolio. co/negocios/kokorisaurios-y-nuggetsaurios-si-pueden-coexistir Portafolio. (2012c, octubre 11). Colombina le gana pleito a la ﬁrma Meals por el uso de la marca Ponky. Portafolio [acceso 20 Ene 2013]. Disponible en: http:// www.portafolio.co/detalle archivo/DR-67173 Portafolio. (2012d, noviembre 13). Big Cola gana otro pleito por registro de su marca. Portafolio [acceso 20 Ene 2013]. Disponible en: http://www.portafolio.co/ negocios/big-cola-gana-otro-pleito-juridico-registro-su-marca Revista Dinero (2012). Vick Vaporub pierde pleito jurídico. Revista Dinero [acceso 28 Ene 2013]. Disponible en: http://www.dinero.com/negocios/ articulo/vick-vaporub-pierde-pleito-juridico/156278 Romeo, A. A. (1977). The rate of imitation of a capital-embodied process innovation. Economica, 44(173), 63–69. Santamaria, E. (2012). Niegan registro Full Cola para resguardar marcas de éxito. La República [acceso 20 Ene 2013]. Disponible en: http://america-retail.com/ industria-y-mercado/niegan-registro-full-cola-para-resguardar-marcas-decompania-exito Segerstrom, P. S. (1991). Innovation, imitation, and economic growth. The Journal of Political Economy, 99(4), 807–827. Shinkle, G. A. y McCann, B. T. (2013). New product deployment: The moderating inﬂuence of economic institutional context. Strategic Management Journal. John Wiley & Sons, Ltd [acceso 6 Sep 2013]. Disponible en: http:// dx.doi.org/10.1002/smj.2132 SIC. (2013). Superintendencia de industria y comercio [acceso 20 Ene 2013]. Disponible en: http://www.sic.gov.co/es/competencia-desleal;jsessionid=oXmJB-Bl7xTwIHbHz-8oQR-.undeﬁned Sohn, S. J. (2008). The two contrary effects of imitation. Economic Modelling, 25(1), 75–82. Teece, D. J., Pisano, G. y Shuen, A. (1997). Dynamic capabilities and strategic management. Strategic Management Journal, 18(7), 509–533. Van Horen, F. y Pieters, R. (2012). When high-similarity copycats lose and moderatesimilarity copycats gain: The impact of comparative evaluation. Journal of Marketing Research (JMR), 49(1), 83–91. Varela-Sánchez, A. L. (2012). Sentencia N.o . 3288. Bogotá: Superintendencia de Industria y Comercio [acceso 20 Ene 2013]. Disponible en: http:// www.sic.gov.co/es/c/document library/get ﬁle?uuid=2e11ecff-6921-479f86e7-158b31ecb2fa&groupId=10157 Yamamura, E., Sonobe, T. y Otsuka, K. (2005). Time path in innovation, imitation, and growth: The case of the motorcycle industry in postwar Japan. Journal of Evolutionary Economics, 15(2), 169–186.
https://openalex.org/W2991272044	https://europepmc.org/articles/pmc7007410?pdf=render	English	null	Are Asian foods as “fattening” as western-styled fast foods?	European journal of clinical nutrition	2,019	cc-by	2,834	* Christiani Jeyakumar Henry jeya_henry@sics.a-star.edu.sg Abstract In Asia, the consumption of western-styled fast foods is widely perceived as the cause of the rise in obesity and chronic disease. Twenty-ﬁve of the most popular local Asian foods were compared for energy, total fat, saturated fat, sodium, and cholesterol with twenty-nine western-styled fast foods. The comparative analysis showed no signiﬁcant difference in energy (p = 0.150) and total fat (p = 0.346) between the two food categories. These ﬁndings suggest that many local Asian foods contribute as much energy and total fat in a single meal as western-styled fast foods. Local Asian foods had greater amounts of sodium (p < 0.001), saturated fat (p = 0.007), and cholesterol (p = 0.009) than western-styled fast foods. The persistent presumption that the consumption of western-styled fast foods is the cause of obesity in Asia needs to be challenged. This observation that local Asian foods are as energy dense as western-styled fast foods, will enable us to redress the necessary strategies to address the Asian diet-health debate. 1 Clinical Nutrition Research Centre (CNRC), Singapore Institute for Clinical Sciences (SICS), Agency for Science, Technology and Research (A*STAR) Singapore and National University Health System, 30 Medical Drive, Singapore 117609, Singapore 2 Department of Biochemistry, Yong Loo Lin School of Medicine, National University of Singapore, S14 Level 5, Science Drive 2, Singapore 117543, Singapore Received: 28 August 2019 / Revised: 10 November 2019 / Accepted: 17 November 2019 / Published online: 29 November 2019 © The Author(s) 2019. This article is published with open access Received: 28 August 2019 / Revised: 10 November 2019 / Accepted: 17 November 2019 / Published online: 29 November 2019 © The Author(s) 2019. This article is published with open access European Journal of Clinical Nutrition (2020) 74:348–350 https://doi.org/10.1038/s41430-019-0537-3 European Journal of Clinical Nutrition (2020) 74:348–350 https://doi.org/10.1038/s41430-019-0537-3 BRIEF COMMUNICATION Local Asian foods Twenty-nine popular western-styled fast foods from McDonalds’, Pizza Hut, and KFC were chosen and their energy value, total fat, saturated fat, sodium, and cho- lesterol were computed using food composition tables and their nutrient content obtained from the representative fast food websites. Twenty-ﬁve of the most commonly consumed local Asian foods from hawker centres and food courts in Singapore were chosen for this analysis. The choice of foods selected was based on interviews/questionnaires on the most widely consumed foods by the three major ethnic groups in Sin- gapore, namely the Chinese, Malays, and Indians [6]. The description of the local Asian foods has been provided in Table 1 (Supplementary ﬁle). One of the most commonly consumed rice accompaniments is chicken and this is reﬂected in the choice of our local Asian foods. In addition, due to portion control, portion sizes are tightly controlled between local vendors. Our previous work also demon- strated that most local foods purchased from various loca- tions had similar energy content [7, 8]. Sodium, total fat, saturated fats, and cholesterol were computed using the Materials and methods “Energy and Nutrient Composition of Food” database by the Health Promotion Board, Singapore [9]. Introduction Chinese: Roasted chicken rice, steamed chicken rice, roasted chicken rice (skinless), braised chicken rice, braised chicken rice (skinless), fried kway teow, beef hor fun, fried seafood hor fun, char siew fried rice, char siew rice, Malay: Nasi lemak with chicken wing, nasi lemak with fried egg, mee siam, mee soto, mee goreng, Lontong with sayur lodeh, Indian: Thosia, masala, roti prata (plain), chicken murtabak, chicken briyani, mutton briyani, vegetable briyani, mee goreng (mamak style). Western fast foods. McDonalds extra value meals Fig. 1 Average (a) energy (kJ), (b) fat (g), (c) saturated fat (g), (d) sodium (mg) and (e) cholesterol (mg) of 25 local Asian foods and 29 western fast foods (mean ± SEM) (Detailed breakdown of foods ana- lyzed is shown below; nutritional values of individual foods are pro- vided in Supplementary ﬁle). Foods analyzed: Local Asian foods. Chinese: Roasted chicken rice, steamed chicken rice, roasted chicken rice (skinless), braised chicken rice, braised chicken rice (skinless), fried kway teow, beef hor fun, fried seafood hor fun, char siew fried rice, char siew rice, Malay: Nasi lemak with chicken wing, nasi lemak with fried egg, mee siam, mee soto, mee goreng, Lontong with sayur lodeh, Indian: Thosia, masala, roti prata (plain), chicken murtabak, chicken briyani, mutton briyani, vegetable briyani, mee goreng (mamak style). Western fast foods. McDonalds extra value meals (Medium fries and Small Coke): Big Mac, Cheeseburger, McChicken, McSpicy, Fillet O Fish, Grilled Chicken McWrap. KFC (Singapore data): 2/3 pcs chicken drumstick + 1 regular whipped potato + 1 regular Coleslaw + 1 regular Pepsi, Shrooms ﬁllet burger/zinger/BBQ Pockett + 1 regular fries + 1 regular Pepsi. Pizza Hut (HPB database): Chicken curry pizza, thin crispy pepperoni pizza, thin crispy supreme pizza, thin crispy cheese pizza, thin crispy super supreme pizza, cheese pan pizza, pepperoni pan pizza, supreme pan pizza, super supreme 9” regular pan pizza, veggie lover’s 10” regular crispy thin pizza, veggie lover’s 12” large pan pizza, veggie lover’s 9” regular pan pizza, shrooms 10” regular crispy thin pizza, chic delite 10” regular crispy thin pizza, hawaiian 12” large pan pizza, hawaiian 9” regular pizza, ocean catch pizza, and meat galore 10” regular crispy thin pizza Introduction accessible in Singapore, an examination of the frequency of fast food consumption shows a counter-intuitive pattern. In a population-based survey, 37% were nonconsumers, 43% were occasional consumers (less than once a week but more than once a month) and 20% consumed fast foods at least once a week [4]. A similar low frequency of western-styled fast food consumption was also reported in Malaysia, Indonesia, and the Philippines [5]. This indicates that the penetration and con- sumption of western-styled fast foods in this region still remains low. In contrast, the consumption of local foods remains a major source of nutrient intake in Asia [5]. This makes local Asian foods a signiﬁcant source of energy, sodium, total fat, saturated fat and cholesterol compared with western-styled fast foods. In order to test this hypothesis, we collated a selection of widely consumed local Asian foods and compared their energy, sodium, total fat, saturated fats, and cholesterol values with popular western-styled fast foods (Table 2a, b, Supplementary Files) [6]. There is widespread presumption that the increase in obesity, cardiovascular disease, and hypertension in Asia is driven by the overconsumption of western-styled fast foods [1]. In par- allel, there continues to remain an urban myth that local Asian foods are both healthy and nutritious. We wish to report that both these perceptions are false. This observation is based on a comparative analysis of the energy, total fat, saturated fat, cholesterol, and sodium content of the most commonly con- sumed foods in Asia (Singapore) with western-styled fast foods. Singapore is a microcosm of dietary diversity representing Chinese, Indian, and Malay cuisines. These three cuisines encompass the dietary habits of ~4.5 billion people in Asia [2]. Surrounded by a plethora of local food eateries within easy walking distance of most homes and workplaces, Singapor- eans eat out regularly [3]. Despite fast food chains being easily Supplementary information The online version of this article (https:// doi.org/10.1038/s41430-019-0537-3) contains supplementary material, which is available to authorized users. Supplementary information The online version of this article (https:// doi.org/10.1038/s41430-019-0537-3) contains supplementary material, which is available to authorized users. Are Asian foods as “fattening” as western-styled fast foods? 349 Materials and methods “Energy and Nutrient Composition of Food” database by 1 1 2 2 3 Energy (kJ) 0. 500. 1000. 1500. 2000. 2500. 3000. . .00 .00 .00 .00 .00 . Introduction 00 00 23 Local 324 foodds p=00.150 2 4 6 8 10 12 14 16 18 Sodium (mg) 1 Fast 0 0.00 200.00 400.00 600.00 800.00 000.00 200.00 400.00 600.00 800.00 1888 t Food 0 0 0 0 0 0 0 0 0 0 ds Lo 144 ocal fo A 41 oods p 1 1 1 1 1 2 2 2 Total Fat (g) p<0.0 2.00 4.00 6.00 8.00 0.00 2.00 4.00 6.00 8.00 0.00 2.00 0.00 4.00 001 75 Fast F Loc 58 Foods 21 cal foo s D ods p=0.3 Cholesterol (mg) F 346 0. 20. 40. 60. 80. 100. 120. 140. 1 Fast F . .00 .00 .00 .00 .00 .00 . 00 00 8 Foodss 1 Loca B 108 l foodds p==0.009 0. 2. 4. 6. 8. 10. 12. Saturated Fat (g) Fas 9 . .00 .00 .00 .00 .00 . 00 00 37 st Foo 7 ods Local E 9 l foodds p==0.00 Fas 07 6 st Fooods C Fig. 1 Average (a) energy (kJ), (b) fat (g), (c) saturated fat (g), (d) sodium (mg) and (e) cholesterol (mg) of 25 local Asian foods and 29 western fast foods (mean ± SEM) (Detailed breakdown of foods ana- lyzed is shown below; nutritional values of individual foods are pro- vided in Supplementary ﬁle). Foods analyzed: Local Asian foods. Chinese: Roasted chicken rice, steamed chicken rice, roasted chicken rice (skinless), braised chicken rice, braised chicken rice (skinless), fried kway teow, beef hor fun, fried seafood hor fun, char siew fried rice, char siew rice, Malay: Nasi lemak with chicken wing, nasi lemak with fried egg, mee siam, mee soto, mee goreng, Lontong with sayur lodeh, Indian: Thosia, masala, roti prata (plain), chicken murtabak, chicken briyani, mutton briyani, vegetable briyani, mee goreng (mamak style). Western fast foods. McDonalds extra value meals (Medium fries and Small Coke): Big Mac, Cheeseburger, McChicken, McSpicy, Fillet O Fish, Grilled Chicken McWrap. KFC (Singapore data): 2/3 pcs chicken drumstick + 1 regular whipped potato + 1 regular Coleslaw + 1 regular Pepsi, Shrooms ﬁllet burger/zinger/BBQ Pockett + 1 regular fries + 1 regular Pepsi. Introduction Pizza Hut (HPB database): Chicken curry pizza, thin crispy pepperoni pizza, thin crispy supreme pizza, thin crispy cheese pizza, thin crispy super supreme pizza, cheese pan pizza, pepperoni pan pizza, supreme pan pizza, super supreme 9” regular pan pizza, veggie lover’s 10” regular crispy thin pizza, veggie lover’s 12” large pan pizza, veggie lover’s 9” regular pan pizza, shrooms 10” regular crispy thin pizza, chic delite 10” regular crispy thin pizza, hawaiian 12” large pan pizza, hawaiian 9” regular pizza, ocean catch pizza, and meat galore 10” regular crispy thin pizza B 0. 2. 4. 6. 8. 10. 12. Saturated Fat (g) . .00 .00 .00 .00 .00 . 00 00 Local 9 l foodds p==0.00 Fas 07 6 st Fooods C 1 1 2 2 3 Energy (kJ) 0. 500. 1000. 1500. 2000. 2500. 3000. . .00 .00 .00 .00 .00 . 00 00 23 Local 324 foodds p=00.150 1 Fast 0 1888 t Foodds A 1 1 1 1 1 2 2 2 Total Fat (g) 2.00 4.00 6.00 8.00 0.00 2.00 4.00 6.00 8.00 0.00 2.00 0.00 4.00 Loc 21 cal fooods p=0.3 F 346 1 Fast F 8 Foodss B 0. 2. 4. 6. 8. 10. 12. Saturated Fat (g) . .00 .00 .00 .00 .00 . 00 00 Local 9 l foodds p==0.00 Fas 07 6 st Fooods C B C 2 4 6 8 10 12 14 16 18 Sodium (mg) 0.00 200.00 400.00 600.00 800.00 000.00 200.00 400.00 600.00 800.00 0 0 0 0 0 0 0 0 0 0 Lo 144 ocal fo 41 oods pp<0.0001 75 Fast F 58 Foodss D Cholesterol (mg) 0. 20. 40. 60. 80. 100. 120. 140. . .00 .00 .00 .00 .00 .00 . 00 00 1 Loca 108 l foodds p==0.009 Fas 9 37 st Foo 7 ods E E Fig. 1 Average (a) energy (kJ), (b) fat (g), (c) saturated fat (g), (d) sodium (mg) and (e) cholesterol (mg) of 25 local Asian foods and 29 western fast foods (mean ± SEM) (Detailed breakdown of foods ana- lyzed is shown below; nutritional values of individual foods are pro- vided in Supplementary ﬁle). Foods analyzed: Local Asian foods. References 1. Guldan GS. Asian children’s obesogenic diets—time to change this part of the energy balance equation? Res Sports Med. 2010;18:5–15. Admittedly, we do not know everything about the rela- tionship between diet and human health. However, in the age of evidence-based science, this communication is intended to provide new insights into the diet-health debate in Asia. Our results highlight the need to reexamine the notion that the consumption of western-styled fast foods alone is the bane of our ill health in Asia. It further demonstrates that Asian foods have unhealthy levels of energy, total fat, saturated fat, sodium and cholesterol. The continuous allegation that consumption of western-styled fast food as the cause of obesity in Asia needs to be reex- amined. Asian foods are as high in energy content, saturated fat, sodium and cholesterol as western-styled fast foods. This new observation will facilitate crafting an alternative framework needed to improve the health of the Asian community. 2. Liu C, Schänzel H. Introduction to tourism education and Asia. Tourism education and Asia: Springer Nature Singapore Pte Ltd., 2019;p. 3–11. 3. Story M, Kaphingst KM, Robinson-O’Brien R, Glanz K. Creating healthy food and eating environments: policy and environmental approaches. Annu Rev Public Health. 2008;29:253–72. 4. Whitton C, Ma Y, Bastian AC, Chan MF, Chew L. Fast-food consumers in Singapore: demographic proﬁle, diet quality and weight status. Public Health Nutr. 2014;17:1805–13. 5. Naidoo N, van Dam RM, Ng S, Tan CS, Chen S, Lim JY, et al. Determinants of eating at local and western fast-food venues in an urban Asian population: a mixed methods approach. Int J Behav Nutr Phys Act. 2017;14:69. 6. Chen L-W, Low YL, Fok D, Han WM, Chong YS, Gluckman P, et al. Dietary changes during pregnancy and the postpartum period in Singaporean Chinese, Malay and Indian women: the GUSTO birth cohort study. Public Health Nutr. 2014;17:1930–8. 7. Quek RYC, Jen GH, Henry CJ. Energy density of ethnic cuisines in Singaporean hawker centres: a comparative study of Chinese, Malay and Indian foods. Malays J Nutr. 2019;25:171–84. 8. Lau E, Goh HJ, Quek R, Lim SW, Henry J. Rapid estimation of the energy content of composite foods: the application of the Calorie Answer. Asia Pac J Clin Nutr. 2016;25:18. Author contributions CJH contributed to conceptualization of the study design, formal analysis, and writing. BK contributed to literature search, data curation, analysis and writing. Calculation and statistical analysis The energy value, total fat, saturated fat, sodium, and cho- lesterol of local Asian foods and western-styled fast foods were collated based on the serving size of each food (Sup- plementary Tables 2a, b). Unpaired t test (unequal variance) was used to test for the difference in mean between local Asian foods and western-styled fast foods. The alpha (α) level for all statistical analyses in this study was set at 0.05. Values 350 C. J. Henry et al. Conﬂict of interest The authors declare that they have no conﬂict of interest. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. Results and discussion The energy value, total fat, saturated fat, sodium, and cholesterol of local Asian foods and western-styled fast foods are displayed in Fig. 1. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/. There were no signiﬁcant differences in energy content (p = 0.150) and total fat (p = 0.346) between local Asian foods and western-styled fast foods. Indeed, several Asian local foods contributed as much energy and total fat in a single meal as western-styled fast foods. Local foods had signiﬁcantly higher saturated fats (p = 0.007), sodium (p < 0.001) and cholesterol content (p = 0.009) than western- styled fast foods. The higher saturated fat and cholesterol content of Asian foods was predominantly due to animal fats such as lard, fatty meats (pork, beef, and mutton) and skin of poultry (chicken). The higher sodium levels in local Asian foods may be attributed to the use of seasonings such as soya sauce and MSG (monosodium glutamate). Compliance with ethical standards were reported as mean ± standard error of mean. All statistical analyses were performed using Statistical Package for the Social Science (SPSS version 24). Conﬂict of interest The authors declare that they have no conﬂict of interest. Conﬂict of interest The authors declare that they have no conﬂict of interest. References RYCQ contributed to lit- erature search, data collection, analysis, and writing. 9. Health Promotion Board. Food composition guide Singapore. Singapore: Cicada Design Pte Ltd; 2003.
https://openalex.org/W3210407595	https://www.nature.com/articles/s41467-021-26454-x.pdf	English	null	Mechanics-driven nuclear localization of YAP can be reversed by N-cadherin ligation in mesenchymal stem cells	Nature communications	2,021	cc-by	13,766	ARTICLE 21-26454-x Mechanics-driven nuclear localization of YAP can be reversed by N-cadherin ligation in mesenchymal stem cells Cheng Zhang 1,2,11, Hongyuan Zhu 1,2,11, Xinru Ren1,2, Bin Gao3, Bo Cheng1,2, Shaobao Liu4, Baoyong Sha5, Zhaoqing Li1,2, Zheng Zhang1,2, Yi Lv6, Haohua Wang6, Hui Guo7, Tian Jian Lu4,8, Feng Xu1,2, Guy M. Genin 1,2,9,10 & Min Lin 1,2✉ Cheng Zhang 1,2,11, Hongyuan Zhu 1,2,11, Xinru Ren1,2, Bin Gao3, Bo Cheng1,2, Shaobao Liu4, Baoyong Sha5, Zhaoqing Li1,2, Zheng Zhang1,2, Yi Lv6, Haohua Wang6, Hui Guo7, Tian Jian Lu4,8, Feng Xu1,2, Guy M. Genin 1,2,9,10 & Min Lin 1,2✉ Mesenchymal stem cells adopt differentiation pathways based upon cumulative effects of mechanosensing. A cell’s mechanical microenvironment changes substantially over the course of development, beginning from the early stages in which cells are typically sur- rounded by other cells and continuing through later stages in which cells are typically sur- rounded by extracellular matrix. How cells erase the memory of some of these mechanical microenvironments while locking in memory of others is unknown. Here, we develop a material and culture system for modifying and measuring the degree to which cells retain cumulative effects of mechanosensing. Using this system, we discover that effects of the RGD adhesive motif of ﬁbronectin (representative of extracellular matrix), known to impart what is often termed “mechanical memory” in mesenchymal stem cells via nuclear YAP localization, are erased by the HAVDI adhesive motif of the N-cadherin (representative of cell-cell contacts). These effects can be explained by a motor clutch model that relates cellular traction force, nuclear deformation, and resulting nuclear YAP re-localization. Results demonstrate that controlled storage and removal of proteins associated with mechanical memory in mesenchymal stem cells is possible through deﬁned and programmable material systems. 1 The Key Laboratory of Biomedical Information Engineering of Ministry of Education, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an 710049, People’s Republic of China. 2 Bioinspired Engineering and Biomechanics Center (BEBC), Xi’an Jiaotong University, Xi’an 710049, People’s Republic of China. 3 Department of Endocrinology, Second Afﬁliated Hospital of Air Force Military Medical University, Xi’an 710038, People’s Republic of China. 4 State Key Laboratory of Mechanics and Control of Mechanical Structures, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, People’s Republic of China. 5 School of Basic Medical Science, Xi’an Medical University, Xi’an 710021, People’s Republic of China. 1 The Key Laboratory of Biomedical Information Engineering of Ministry of Education, School of Life Science and Technology, Xi’an Jiaotong University, Xi’an 710049, People’s Republic of China. 2 Bioinspired Engineering and Biomechanics Center (BEBC), Xi’an Jiaotong University, Xi’an 710049, People’s Republic of China. 3 Department of Endocrinology, Second Afﬁliated Hospital of Air Force Military Medical University, Xi’an 710038, People’s Republic of China. 4 State Key Laboratory of Mechanics and Control of Mechanical Structures, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, People’s Republic of China. 5 School of Basic Medical Science, Xi’an Medical University, Xi’an 710021, People’s Republic of China. 6 National Local Joint Engineering Research Center for Precision Surgery & Regenerative Medicine, Shaanxi Provincial Center for Regenerative Medicine and Surgical Engineering, The First Afﬁliated Hospital of Xi’an Jiaotong University, Xian, People’s Republic of China. 7 Department of Medical Oncology, The First Afﬁliated Hospital of Xi’an Jiaotong University, Xi’an 710061 Shaanxi, People’s Republic of China. 8 MOE Key Laboratory of Multifunctional Materials and Structures, Xi’an Jiaotong University, Xi’an 710049, People’s Republic of China. 9 Department of Mechanical Engineering & Materials Science, Washington University in St. Louis, St. Louis 63130 MO, USA. 10 NSF Science and Technology Center for Engineering Mechanobiology, Washington University in St. Louis, St. Louis 63130 MO, USA. 11These authors contributed equally: Cheng Zhang, Hongyuan Zhu. ✉email: minlin@xjtu.edu.cn NATURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications ARTICLE Mechanics-driven nuclear localization of YAP can be reversed by N-cadherin ligation in mesenchymal stem cells 6 National Local Joint Engineering Research Center for Precision Surgery & Regenerative Medicine, Shaanxi Provincial Center for Regenerative Medicine and Surgical Engineering, The First Afﬁliated Hospital of Xi’an Jiaotong University, Xian, People’s Republic of China. 7 Department of Medical Oncology, The First Afﬁliated Hospital of Xi’an Jiaotong University, Xi’an 710061 Shaanxi, People’s Republic of China. 8 MOE Key Laboratory of Multifunctional Materials and Structures, Xi’an Jiaotong University, Xi’an 710049, People’s Republic of China. 9 Department of Mechanical Engineering & Materials Science, Washington University in St. Louis, St. Louis 63130 MO, USA. 10 NSF Science and Technology Center for Engineering Mechanobiology, Washington University in St. Louis, St. Louis 63130 MO, USA. 11These authors contributed equally: Cheng Zhang, Hongyuan Zhu. ✉email: minlin@xjtu.edu.cn 1 TURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038 ARTICLE NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x H H uman mesenchymal stem cells (hMSCs) differentiate into a range of cell types depending in part upon the mechanical cues they received1–4. These cues can vary dramatically during development5, healing6, cancer7 and regen- erative therapies8, due to stochastic variations, stress concentra- tions, and cell culture protocols. For example, in regenerative applications9,10, MSC populations are typically expanded on stiff (~3 GPa11) plastic substrates for relatively long intervals and must forget these mechanical signals prior to transplantation into compliant (~kPa) in vivo microenvironments. Understanding how MSCs learn to forget their previous mechanical micro- environment variations holds potential for improving MSC therapies. in response to mechanosensing, and apply this to control the differentiation of hMSCs. Results and discussion Physiological responses to cell-cell interactions can be mimicked by HAVDI immobilized on a hydrogel. To elucidate how N-cadherin affects YAP localization, we developed a dual- peptide functionalized PEG hydrogel analogous to the hyaluronic acid (HA) based hydrogel of Cosgrove et al.39. The PEG hydrogel system presented either HAVDI/RGD (from N-cadherin, repre- senting neighboring cells), or non-functional scrambled HAVDI sequence (Scram) combined with RGD (Scram/RGD) (Fig. 1). 8-arm PEG maleimide (PEG-MAL) modiﬁed with mono- thiolated peptides was subsequently crosslinked by 8-arm PEG thiol (PEG-SH) via Michael addition) (Fig. 1a)41. As with ana- logous HA hydrogels, PEG hydrogels modiﬁed with 1 mM RGD and 1 mM non-functional scrambled HAVDI sequence (Scram/ RGD) established only integrin adhesions, while PEG hydrogels modiﬁed with 1 mM RGD and 1 mM HAVDI enabled both integrin and N-cadherin mediated interactions (HAVDI/RGD) (Fig. 1b and Supplementary Fig. 1). Mechanics-driven nuclear localization of YAP can be reversed by N-cadherin ligation in mesenchymal stem cells Concentrations of HAVDI peptides followed Cosgrove et al.39, and were within the range of E-cadherin density at adherens junctions in monolayer epithelial cells of developing Drosophila embryos42. The grafted peptide was observed to distribute homogeneously on the surface of hydrogels (Supplementary Fig. 1). The binding efﬁciency of peptide to the PEG backbone was measured at over 86% (Sup- plementary Fig. 2). Through the binding efﬁciency, we calculated the surface concentrations of peptides, which fell into a reason- able range (Supplementary Text). The opposite problem – the ways that MSCs learn and remember their mechanical environments – has been long studied in isolated MSCs. A key indicator of the degree to which MSCs perceive and transduce mechanobiological signals and commit to a lineage-based upon these is the action of the YAP/TAZ tran- scriptional co-regulators12–14. Nuclear compartmentalization of YAP/TAZ in MSCs cultured on relatively stiff substrates activates signaling pathways that promote osteogenesis15–17, while cyto- plasmic retention of YAP/TAZ in MSCs cultured on relatively compliant substrates promotes adipogenesis18–20. Sufﬁciently long (~10 d) exposure of MSCs to stiff substrates causes nuclear accumulation of YAP/TAZ that is retained to inﬂuence long-term cell fate, a phenomenon that is widely termed “mechanical memory”11,21. For example, MSCs retain their osteogenic differ- entiation course when re-exposed to compliant substrates, with the nuclear accumulation of YAP/TAZ becoming apparently irreversible11. Several pathways have been identiﬁed for this per- sistence, in addition to the cytoplasm-nucleus shuttling of YAP, including a role for cytoplasmic miR-21 whose presence extends the persistence of the effects of stiff (100 kPa) substrates in MSCs subsequently cultured on soft (15 kPa) hydrogels, and whose knockdown effectively eliminates this persistence22. g pp y Hydrogel stiffness could be tuned independently from 2-41 kPa by varying relative concentrations of PEG-SH and PEG-MAL (Fig. 1c). Modiﬁcation with HAVDI/RGD or Scram/RGD did not affect Young’s moduli of the hydrogel over the range of concentrations of PEG-SH and PEG-MAL used in this study (Fig. 1c). Young’s moduli were obtained from load-indentation curves (Supplementary Fig. 3). When we seeded cells on the hydrogels with different stiffness and peptide, cell area increased with increasing substrate stiffness, in accordance with previous studies43,44; no signiﬁcant difference was observed between the Scram/RGD and HAVDI/RGD groups (Supplementary Fig. 4). Results indicated that the HAVDI peptide did not affect cell spreading, and differences in YAP n/c ratio observed between Scram/RGD and HAVDI/RGD hydrogels did not arise from difference in cell area. Mechanics-driven nuclear localization of YAP can be reversed by N-cadherin ligation in mesenchymal stem cells y p However, cells, including differentiating MSCs, rarely exist in isolation in vivo14,23. Two questions thus arise: how do MSCs know to take cues from their extracellular matrix (ECM) rather than from their compliant neighbors, and can MSCs use infor- mation from their neighbors to learn when to forget previous mechanical information? Cell-cell contact affects mechanosensa- tion, with conﬂuent cells that have reached a stable density showing cytoplasmic re-localization of YAP/TAZ24–26. Cell-ECM connections (focal adhesions) and cell-cell connections (cadherin junctions) share the same cytoskeletal force network and have many other similarities27–29. Sufﬁciently high adhesion or con- tractile force in this cytoskeletal network causes YAP to translo- cate to the nucleus30,31. However, there is a trade-off: cell-ECM adhesion weakens cell-cell adhesion, and vice versa32, with FAK activation33 from strong integrin-ECM connections34,35 down- regulating VE-cadherin-36,37 and N-cadherin-28,38 mediated intercellular connections. Adhesion ligation experiments reveal that N-cadherin connections compete with integrin connections to mediate force and determine YAP localization, MSC mechanosensing, and MSC fate39,40. N-cadherin clearly affects YAP localization and MSC behavior, but has not been studied in the context of “mechanical memory”. We began by using this system to establish whether HAVDI presentation faithfully replicates the N-cadherin clustering and β- catenin recruitment associated with N-cadherin-mediated cell- cell interactions. These two mechanosensing proteins accumulate as force transmission between cells matures adherens junctions via an α-catenin/vinculin dependent pathway45. To study this, we cultured hMSCs at prescribed cell densities and levels of conﬂuence on Scram/RGD or HAVDI/RGD hydrogels with a stiffness of 20 kPa for 3 d, then ﬁxed and stained the hMSCs for N-cadherin and β-catenin on their basal planes (Fig. 2a). Sparse cells that had no contact with their neighbors showed neither N-cadherin clustering nor β-catenin recruitment when cultured on Scram/RGD hydrogels (Fig. 2b). However, N-cadherin clustering and β-catenin co-localization were rescued in sparse cells cultured upon HAVDI/RGD hydrogels (Fig. 2c). Likewise, conﬂuent cells, which formed adherens junctions with their neighbors, showed clustering of N-cadherin as well as recruitment and co-localization of β-catenin on both the HAVDI/ RGD and Scram/RGD hydrogels (Fig. 2d). To check whether In this work, we hypothesize that N-cadherin in cell-cell interactions could reverse these effects (“erase” mechanical memory) in MSCs by restoring YAP to the cytoplasm. We thus develop a material system with independently or jointly presented RGD peptides for cell-ECM adhesion and HAVDI peptides for cell-cell adhesion. NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454- RGD peptide, 'GCGYGRGDSSPG' HAVDI peptide, 'HAVDIGGGC' HAVDI peptide (Scram), 'AGVGDHIGC' 8-arm PEG Integrin RGD Scram Integrin N-cadherin RGD HAVDI a b c hexaglycerol core structure 8-arm PEG thiol (PEG-SH) hexaglycerol core structure 8-arm PEG maleimide (PEG-MAL) R peptides (containing SH) Michael addition R peptides modified PEG-MAL PEG-MAL Michael addition PEG-SH Hydrogels Fig. 1 Peptide-functionalized PEG gels that mimic integrin and cadherin adhesion. a Mono-thiolated peptides functionalized PEG backbone (8-arm PEG maleimide, PEG-MAL) and subsequent crosslinking with 8-arm PEG thiol (PEG-SH) via Michael addition reaction. b Schematic of PEG-MAL hydrogels modiﬁed with HAVDI and RGD peptides acting as N-cadherin and ﬁbronectin adhesive domains, respectively. Scram/RGD hydrogel substrate allows only integrin/RGD interactions, while HAVDI/RGD substrate allows both integrin/RGD and N-cadherin/HAVDI interactions. c Young’s moduli of hydrogels synthesized at different PEGSH concentrations without peptide (No peptide) or with peptide (Scram/RGD or HAVDI/RGD). Results indicated that modiﬁcation of peptides did not alter the stiffness of PEG hydrogels in this system. The data represented the mean ± s.e.m., from left to right, n = 60, 66, 47, 31, 60, 30, 30, 89, 50, 50, 60, 46, 65 points respectively, adopted from 3 independent samples for each group (p values were obtained using one-way ANOVA f ll d b T k ’ t h t t ± ) S d t id d S D t ﬁl a hexaglycerol core structure 8-arm PEG thiol (PEG-SH) hexaglycerol core structure 8-arm PEG maleimide (PEG-MAL) R peptides (containing SH) Michael addition R peptides modified PEG-MAL PEG-MAL Michael addition PEG-SH Hydrogels a hexaglycerol core structure 8-arm PEG maleimide (PEG-MAL) R peptides (containing SH) Michael addition R peptides modified PEG-MAL PEG-MAL a hexaglycerol core structure 8-arm PEG thiol (PEG-SH) Michael addition PEG-SH Hydrogels RGD peptide, 'GCGYGRGDSSPG' HAVDI peptide, 'HAVDIGGGC' HAVDI peptide (Scram), 'AGVGDHIGC' 8-arm PEG Integrin RGD Scram Integrin N-cadherin RGD HAVDI b c b c Fig. 1 Peptide-functionalized PEG gels that mimic integrin and cadherin adhesion. a Mono-thiolated peptides functionalized PEG backbone (8-arm PEG maleimide, PEG-MAL) and subsequent crosslinking with 8-arm PEG thiol (PEG-SH) via Michael addition reaction. b Schematic of PEG-MAL hydrogels modiﬁed with HAVDI and RGD peptides acting as N-cadherin and ﬁbronectin adhesive domains, respectively. Scram/RGD hydrogel substrate allows only integrin/RGD interactions, while HAVDI/RGD substrate allows both integrin/RGD and N-cadherin/HAVDI interactions. c Young’s moduli of hydrogels synthesized at different PEGSH concentrations without peptide (No peptide) or with peptide (Scram/RGD or HAVDI/RGD). Mechanics-driven nuclear localization of YAP can be reversed by N-cadherin ligation in mesenchymal stem cells We harness this to query how these adhesive interactions interact to affect persistent cellular changes in hMSCs 2 NATURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x ARTICLE NATURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x a HAVDI ligation caused sparse cells to adopt the N-cadherin and β-catenin patterning of conﬂuent cells, as evident from confocal images of DAPI (blue), N-cadherin (green) and β-catenin (purple) staining in hMSCs cultured on either sparse (non-conﬂuent) or conﬂuent conditions on Scram/ RGD or HAVDI/RGD hydrogels of Young’s modulus 20 kPa. The white arrow highlights real cell-cell adhesions between neighboring hMSCs; the yellow arrow highlights N-cadherin/HAVDI binding for an isolated hMSC on a HAVDI/RGD hydrogel. Scale bar: 20 μm. b–d Representative subcellular distributions of N-cadherin and β-catenin in three different cases corresponding to the numbered areas shown in a, showing that HAVDI ligation caused sparse cells to adopt the N-cadherin and β-catenin patterning of conﬂuent cells. Scale bars: 5 μm. b Isolated cells on Scram/RGD hydrogels, which do not mimic cell-cell adhesions, did not show substantial N-cadherin clustering or β-catenin recruitment. c Isolated cells cultured on HAVDI/RGD hydrogel, a DAPI N- nire h d a c - nin eta c Merge Sparse Confluent Scram/RGD HAVDI/RGD Scram/RGD HAVDI/RGD b c d N-cadherin -catenin 0 1 2 3 0 1,000 2,000 3,000 Distance ( m) Fluorescent intensity (a.u.) 0 1 2 3 0 200 400 1 0 1 2 3 0 1,000 2,000 3,000 Distance ( m) Fluorescent intensity (a.u.) 2 N-cadherin -catenin 0 1 2 3 0 1,000 2,000 3,000 Distance ( m) Fluorescent intensity (a.u.) 3 3 2 1 b a - nin eta c Merge d N-cadherin -catenin 0 1 2 3 0 Distance ( m) i 0 1 2 3 0 1,000 2,000 3,000 Distance ( m) Fluorescent intensity (a.u.) 3 e f D G R /I D V A H D G R / m arc S 2 kPa 11 kPa 20 kPa 41 kPa DAPI YAP 3 2 1 d f e f D G R /I D V A H D G R / m arc S 2 kPa 11 kPa 20 kPa 41 kPa DAPI YAP f DAPI YAP Fig. 2 N-cadherin clustering and β-catenin recruitment in mimetic PEG hydrogels and stiffness dependent YAP nuclear localization via N-cadherin signaling. a HAVDI ligation caused sparse cells to adopt the N-cadherin and β-catenin patterning of conﬂuent cells, as evident from confocal images of DAPI (blue), N-cadherin (green) and β-catenin (purple) staining in hMSCs cultured on either sparse (non-conﬂuent) or conﬂuent conditions on Scram/ RGD or HAVDI/RGD hydrogels of Young’s modulus 20 kPa. NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x The white arrow highlights real cell-cell adhesions between neighboring hMSCs; the yellow arrow highlights N-cadherin/HAVDI binding for an isolated hMSC on a HAVDI/RGD hydrogel. Scale bar: 20 μm. b–d Representative subcellular distributions of N-cadherin and β-catenin in three different cases corresponding to the numbered areas shown in a, showing that HAVDI ligation caused sparse cells to adopt the N-cadherin and β-catenin patterning of conﬂuent cells. Scale bars: 5 μm. b Isolated cells on Scram/RGD hydrogels, which do not mimic cell-cell adhesions, did not show substantial N-cadherin clustering or β-catenin recruitment. c Isolated cells cultured on HAVDI/RGD hydrogel, which mimicked cell-cell adhesions, showed N-cadherin clustering or β-catenin recruitment that was indistinguishable from that in d conﬂuent cells on Scram/RGD hydrogels. Yellow lines in b–d indicate the pixel regions of interest used to generate intensity proﬁles of N-cadherin (green) and β-catenin (purple). e Confocal images of YAP immunostaining on hMSCs cultured on 2, 11, 20 and 41 kPa Scram/RGD or HAVDI/RGD hydrogels for 3 d. Scale bars, 20 μm. f Quantiﬁcation of YAP n/c ratios on Scram/RGD or HAVDI/RGD hydrogels of increasing stiffness with or without blebbistatin (myosin inhibitor) as shown in e (from left to right n= 82, 87, 79, 74, 126, 71, 60, 65, 71, 68, 53, 35, 105, 82, 39, 63 cells examined over 14, 13, 16, 18, 17, 16, 27, 19, 15, 13, 19, 24, 9, 9, 23, 27 images respectively, p values were obtained using one-way ANOVA followed by Tukey’s post hoc test, mean ± s.e.m.). Source data are provided as a Source Data ﬁle. dherin clustering and β-catenin recruitment in mimetic PEG hydrogels and stiffness dependent YAP nuclear local kPa). These results suggested that actomyosin inhibition shielded the mechanosensing of substrates stiffness by cells48–50. In addition, differences of YAP n/c ratio on substrates with different peptides could be attributed to different actomyosin contractility, because myosin inactivation could eliminate the observed HAVDI effect. changed YAP nuclear-to-cytoplasmic (n/c) ratios on Scram/RGD hydrogels, indicating that 0.25 mM RGD was sufﬁcient to induce a maximal YAP n/c ratio (Supplementary Fig. 7). Cells on HAVDI/ RGD hydrogels showed lower YAP n/c ratios, with the differential becoming more pronounced with increasing HAVDI concentration. These results suggested that the increasing RGD concentration (≥0.25 mM) did not affect YAP nuclear translocation measurably, while increasing HAVDI concentration inhibited it. NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x Results indicated that modiﬁcation of peptides did not alter the stiffness of PEG hydrogels in this system. The data represented the mean ± s.e.m., from left to right, n = 60, 66, 47, 31, 60, 30, 30, 89, 50, 50, 60, 46, 65 points respectively, adopted from 3 independent samples for each group (p values were obtained using one-way ANOVA followed by Tukey’s post hoc test, mean ± s.e.m.). Source data are provided as a Source Data ﬁle. arising in cells cultured on HAVDI/RGD hydrogels served to generate force from substrates triggering recruitment and co- localization β-catenin forming force bearing proteins chain. N-cadherin co-localized with β-catenin, the percentage of N-cadherin area that was also positive for β-catenin was quantiﬁed from ﬂuorescence images using a custom MATLAB (R2014a) script that implemented standard cross-correlation techniques from the literature39,46 (Supplementary Fig. 5). For both sparse and conﬂuent cells, signiﬁcantly more co-localization (i.e., N-cadherin area positive for β-catenin) was observed for cells cultured on HAVDI/RGD hydrogels than for cells cultured on Scram/RGD hydrogels, indicating that substrates with immobilized HAVDI induced both N-cadherin clustering and β-catenin recruitment. HAVDI ligation reverses YAP nuclear localization in hMSCs. HAVDI presentation is known to affect nuclear localization of YAP39. Immunostaining that replicated this result in our PEG system showed nucleus-to-cytoplasm (n/c) ratios of YAP sig- niﬁcantly attenuated by HAVDI presentation over an inter- mediate range of substrate moduli (10~20 kPa), but unaltered for compliant (~1 kPa) or stiff (~40 kPa) substrates (Fig. 2e–f). Consistent with the observations of others15,39,47–50, our myosin inactivation experiments showed YAP signaling to be highly dependent on the contractile state of the cell (Fig. 2f). After myosin inactivation using blebbistatin, we found that all YAP n/c ratios dropped to the same level for both the Scram/RGD and HAVDI/RGD hydrogels, independent of substrate moduli (2-41 β Treatment with 1 mM soluble HAVDI(-C) peptide, designed to bind with membrane N-cadherin receptors while avoiding conjugation to the PEG-Mal backbone, blocked recruitment and co-localization of N-cadherin and β-catenin in cells cultured on both Scram/RGD and HAVDI/RGD hydrogels (Supplementary Fig. 6). These results suggested that HAVDI/N-cadherin ligation TURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x a DAPI N- nire h d a c - nin eta c Merge Sparse Confluent Scram/RGD HAVDI/RGD Scram/RGD HAVDI/RGD b c d N-cadherin -catenin 0 1 2 3 0 1,000 2,000 3,000 Distance ( m) Fluorescent intensity (a.u.) 0 1 2 3 0 200 400 1 0 1 2 3 0 1,000 2,000 3,000 Distance ( m) Fluorescent intensity (a.u.) 2 N-cadherin -catenin 0 1 2 3 0 1,000 2,000 3,000 Distance ( m) Fluorescent intensity (a.u.) 3 e f D G R /I D V A H D G R / m arc S 2 kPa 11 kPa 20 kPa 41 kPa DAPI YAP 3 2 1 Fig. 2 N-cadherin clustering and β-catenin recruitment in mimetic PEG hydrogels and stiffness dependent YAP nuclear localization via N-cadherin signaling. a HAVDI ligation caused sparse cells to adopt the N-cadherin and β-catenin patterning of conﬂuent cells, as evident from confocal images of DAPI (blue), N-cadherin (green) and β-catenin (purple) staining in hMSCs cultured on either sparse (non-conﬂuent) or conﬂuent conditions on Scram/ RGD or HAVDI/RGD hydrogels of Young’s modulus 20 kPa. The white arrow highlights real cell-cell adhesions between neighboring hMSCs; the yellow arrow highlights N-cadherin/HAVDI binding for an isolated hMSC on a HAVDI/RGD hydrogel. Scale bar: 20 μm. b–d Representative subcellular distributions of N-cadherin and β-catenin in three different cases corresponding to the numbered areas shown in a, showing that HAVDI ligation caused sparse cells to adopt the N-cadherin and β-catenin patterning of conﬂuent cells. Scale bars: 5 μm. b Isolated cells on Scram/RGD hydrogels, which do not a DAPI N- nire h d a c - nin eta c Merge Sparse Confluent Scram/RGD HAVDI/RGD Scram/RGD HAVDI/RGD b c d N-cadherin -catenin 0 1 2 3 0 1,000 2,000 3,000 Distance ( m) Fluorescent intensity (a.u.) 0 1 2 3 0 200 400 1 0 1 2 3 0 1,000 2,000 3,000 Distance ( m) Fluorescent intensity (a.u.) 2 N-cadherin -catenin 0 1 2 3 0 1,000 2,000 3,000 Distance ( m) Fluorescent intensity (a.u.) 3 e f D G R /I D V A H D G R / m arc S 2 kPa 11 kPa 20 kPa 41 kPa DAPI YAP 3 2 1 Fig. 2 N-cadherin clustering and β-catenin recruitment in mimetic PEG hydrogels and stiffness dependent YAP nuclear localization via N-cadherin signaling. ARTICLE found that nuclear accumulation of YAP reached a steady state within 6 h51, and Elosegui-Artola et al. found that YAP n/c ratio changed within several minutes after nuclear force is applied31. To exclude the possible effects of mechanical stimuli generated by cell passaging, we conducted experiments where cells were transferred from TCP to TCP. The YAP n/c ratio showed no difference between groups that had undergone passaging (DT3 + DT0.25 and DT3 + DT3 in Supplementary Fig. 9) and those that had not (DT3.25 and DT6 in Supplementary Fig. 9), indicating that mechanical stimuli associated with transferring substrates had no measurable effect on the MSC mechanosensing or memory. To verify that the observed RUNX2 nuclear localization was indicative of osteogenesis, the late osteogenic markers alkaline phosphatase (ALP) and osteocalcin (OCN)52,53 were tracked in hMSCs that were exposed to TCP for 1-7 d in growth medium, and that were then transferred onto soft (20 kPa) Scram/RGD or HAVDI/RGD hydrogels for 7 d in osteogenic differentiation medium before being ﬁxed and stained (Fig. 4a). Consistent with the results of Yang, et al.11, ALP and OCN expression increased with increasing culture time on TCP for hMSCs transferred to Scram/RGD hydrogels (Fig. 4d–g). However, HAVDI ligation signiﬁcantly attenuated both ALP and OCN levels, verifying that HAVDI ligation diminished osteogenic commitment. These ﬁndings demonstrated HAVDI ligation as a tool for attenuating nuclear localization of YAP and sustaining stem cell plasticity during in vitro expansion. In hMSCs cultured for 3 d on TCP (DT3) and subsequently transferred to soft HAVDI/RGD hydrogels for 1 d (DT3 + So1), the YAP n/c ratio recovered to the level of “soft control” group. Further increasing the culture time on soft HAVDI/RGD hydrogels to 3 d did not change the YAP n/c ratio. This suggested that YAP nuclear localization was again fully reversed within 1 d (Fig. 3d). In contrast, transferring to Scram/RGD hydrogels led to a higher YAP n/c ratio than that observed in the “soft control” group, suggesting that YAP nuclear localization was only partially reversible under these conditions (Fig. 3d). After 7 d on TCP (DT7), YAP nuclear localization was partially reversible HAVDI dependent motor-clutch dynamics affect the mechanical sensation and nuclear translocation of YAP. Growing evidence implicates a motor-clutch machine of myosin, actin ﬁlaments, and focal adhesion complexes54,55 in sensation of extracellular matrix (ECM) stiffness56–58. ARTICLE This was increasingly persistent in response to increasing mechanical dosing on TCP, but could be partially or entirely reversed via HAVDI ligation that mimicked N-cadherin signaling in cell-cell interactions. HAVDI ligation can be used to control stem cell fate. To demonstrate therapeutically useful reversal of YAP nuclear loca- lization in hMSCs, we explored whether the seemingly irrever- sible, osteogenic commitment of hMSCs11 could be reversed by HAVDI ligation. Yang, et al., demonstrated that priming of hMSCs for 10 d on stiff substrates resulted in nuclear localization of the osteogenic marker RUNX2 even after in-situ softening of the substratum for 10 d11. We found similar results for cells cultured in growth medium for 1-7 d on TCP (~3 GPa) then transferred to soft (20 kPa) Scram/RGD hydrogels for 3 d in osteogenic medium (Fig. 4a), with nuclear localization of RUNX2 increasing with increasing culture time on TCP (Fig. 4b, c). However, in cells that were instead transferred from TCP to soft (20 kPa) HAVDI/RGD hydrogels, a negligible increase of RUNX2 nuclear localization was observed following stiff priming of up to 3 d, and elevation of RUNX2 levels was greatly attenuated for stiff priming of 7 d. We next tested the hypothesis that YAP nuclear localization, known to arise from priming of cells on stiff tissue culture plastic (TCP, ~3 GPa), could be reversed by N-cadherin signaling from HAVDI/RGD substrates. For this, hMSCs were cultured on TCP for 1-10 d, and then transferred to soft (20 kPa) HAVDI/RGD or Scram/RGD hydrogels for 3 or 10 d (Fig. 3a). As seen by others11, the YAP n/c ratio in hMSCs increased with increasing culture time on TCP (Fig. 3b, “no transfer”). When hMSCs were transferred to Scram/RGD or HAVDI/RGD hydrogels after 1 d on TCP substrate (DT1), the YAP n/c ratio returned to the levels seen in cells that had never been exposed to TCP (“soft control” group), which suggested that YAP nuclear localization was fully reversible over 1 d (DT1 + So1) (Fig. 3c). Longer culture time on soft gels (DT1 + So3) did not change the YAP n/c ratio, suggesting that the kinetics of YAP nuclear localization reached a steady state within 1 d. This observation was in accordance with previous studies that after substrate softening, the YAP n/c ratio reduced to baseline levels within 1 d and maintained these levels for up to 10 d21. Cui et al. ARTICLE ARTICLE NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x in cells transferred to either Scram/RGD or HAVDI/RGD hydrogels; the YAP n/c ratio of cells transferred to HAVDI/ RGD gels dropped more than those of cells transferred to Scram/ RGD gels (Supplementary Fig. 10). combinations (Scram/RGD for 1 mM Scram and 1 mM RGD peptides, HAVDI/RGD for 1 mM HAVDI and 1 mM RGD peptides) to test the hypothesis that N-cadherin ligation can reverse nucleus-to-cytoplasm shuttling of YAP. Before studying this reversal of nucleus-to-cytoplasm shuttling of YAP, we repeated earlier results showing that sparse hMSCs developed signiﬁcantly higher YAP nuclear localization when cultured on Scram/RGD hydrogels than when cultured on HAVDI/RGD hydrogels (Supplementary Fig. 8a, upper-left; Supple- mentary Fig. 8b, left); blocking N-cadherin with soluble HAVDI(-C) peptide (1 mM) restored the YAP n/c ratio in sparse hMSCs cultured on HAVDI/RGD substrates to the level seen in hMSCs cultured on Scram/RGD substrates (Supplementary Fig. 8a, lower-left; Supple- mentary Fig. 8b, left), consistent with the observations of Cosgrove et al.39. YAP was excluded from the nucleus in hMSCs in conﬂuent cells cultured on either Scram/RGD or HAVDI/RGD substrates (Supplementary Fig. 8a, upper-right; Supplementary Fig. 8b, right), consistent with previous studies15,18,25 and suggesting that cell-cell interactions inhibited nuclear localization of YAP. Blocking N-cadherin with soluble HAVDI(-C) peptide abrogated this inhibition in conﬂuent cells as well (Supplementary Fig. 8a, lower- right; Supplementary Fig. 8b, right), indicating that N-cadherin mechanoresponsiveness controlled YAP exclusion associated with both real and mimicked cell-cell interactions. In hMSCs cultured for 10 d on TCP (DT10) and then transferred to Scram/RGD hydrogels, the YAP n/c ratio remained the much higher levels in cells in the “soft control” group with 1-10 d culture time. However, transferring to HAVDI/RGD hydrogels led to a drop of the YAP n/c ratio within 1 d to a level that remained constant for up to 10 d culture time. This conﬁrmed that 10 d of culture on TCP made YAP nuclear localization irreversible in cells transferred to soft Scram/RGD hydrogels, as observed in previous studies11,21, and revealed that YAP nuclear localization was again partially reversible in hMSCs transferred to HAVDI/RGD hydrogels (Fig. 3e). Increasing HAVDI concentration increased the magnitude of the drop in YAP n/c ratio (Supplementary Fig. 11). Thus, hMSCs integrated and stored nuclear YAP when exposed to stiff TCP. NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x To quantify how peptide concentrations affect nuclear localiza- tion of YAP, we varied the HAVDI/Scram/RGD peptide concen- tration ratios on hydrogels while keeping the total effective ligand density constant. Increasing RGD presentation resulted in un- g We therefore chose two hydrogels with a ﬁxed, intermediate stiffness (Young’s modulus = 20 kPa) and two different peptide 4 NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x a b No transfer +(S/R)So3 +(S/R)So10 +(H/R)So3 +(H/R)So10 DT0 (Cont rol) DT1 DT3 DT10 S l b 20 DT0 (Control) a b increased traction on the substrate also deform the nucleus and can initiate nuclear-cytoplasmic shuttling of YAP and down- stream signaling pathways31,59. We hypothesized that N-cadherin affected mechanical sensation and the YAP n/c ratio (RNC) by inhibiting the rate kon of integrin binding, and hence reducing integrin clustering, tractions, and nuclear deformation. To test this hypothesis about the effects of HAVDI ligation, we applied a motor-clutch model (model details in Supplementary Methods and Supplementary Table 1)56–58,60. Brieﬂy, in the motor-clutch model (Fig. 5a), actin ﬁlaments attached to ECM through dynamic bonds at integrin clutches with binding rate (kon) and unbinding rate (koff), while myosin motors contracted the actin ﬁlaments with traction force (Ftrac) at velocity (vf). The actin ﬁlament deformed the nucleus and substrate Scale bar: 20 DAPI YAP c d e 6 NATURE COMMUNICATIONS \| (2021)12 6229 \| htt //d i /10 1038/ 41467 021 26454 \| t / t i ti increased traction on the substrate also deform the nucleus and can initiate nuclear-cytoplasmic shuttling of YAP and down- stream signaling pathways31,59. We hypothesized that N-cadherin affected mechanical sensation and the YAP n/c ratio (RNC) by inhibiting the rate kon of integrin binding, and hence reducing integrin clustering, tractions, and nuclear deformation. To test this hypothesis about the effects of HAVDI ligation, we applied a motor-clutch model (model details in Supplementary Metho and Supplementary Table 1)56–58,60. Brieﬂy, in the motor-clutch model (Fig. 5a), actin ﬁlame attached to ECM through dynamic bonds at integrin clutches w binding rate (kon) and unbinding rate (koff), while myosin mot contracted the actin ﬁlaments with traction force (Ftrac) at veloc (vf). The actin ﬁlament deformed the nucleus and substr Scale bar: 20 DAPI YAP c d e c Scale bar: 20 DAPI YAP c d e d e e motor-clutch model (model details in Supplementary Methods and Supplementary Table 1)56–58,60. motor-clutch model (model details in Supplementary Methods and Supplementary Table 1)56–58,60. increased traction on the substrate also deform the nucleus and can initiate nuclear-cytoplasmic shuttling of YAP and down- stream signaling pathways31,59. We hypothesized that N-cadherin affected mechanical sensation and the YAP n/c ratio (RNC) by inhibiting the rate kon of integrin binding, and hence reducing integrin clustering, tractions, and nuclear deformation. ARTICLE The forces associated with actomyosin contraction increase with substrate stiffness because of increased resistance to contraction, increased engagement of the motor-clutch mechanism, and increased development of polarized stress ﬁbers. The forces associated with TURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications 5 ed traction on the substrate also deform the nucleus and tiate nuclear-cytoplasmic shuttling of YAP and down- motor-clutch model (model details in Supplementary M and Supplementary Table 1)56–58,60. a b No transfer +(S/R)So3 +(S/R)So10 +(H/R)So3 +(H/R)So10 DT0 (Cont rol) DT1 DT3 DT10 Scale bar: 20 DAPI YAP DT0 (Control) c d e ICLE NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-2 ARTICLE NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x Fig. 3 HAVDI ligation reversed YAP nuclear localization in hMSCs. a Schematic of the experimental protocol for assessing the role of HAVDI on reversing YAP nuclear localization in hMSCs. YAP accumulation in the nucleus was measured in hMSCs cultured on TCP (dark purple) in growth medium (pink) for 1, 3, 7, or 10 d, denoted DT1, DT3, DT7, and DT10, respectively. Reversal of YAP nuclear accumulation was assessed in hMSCs that were cultured on TCP in growth medium for these same time intervals, but then transferred (light yellow, day 0) to either soft (20 kPa) Scram/RGD (orange) or soft (20 kPa) HAVDI/RGD (blue) hydrogels for 1-10 d in growth medium before collection and analysis (gray, day 1–10). These conditions were denoted (DT1 + So1, 3), (DT3 + So1, 3), (DT7 + So1, 3, 10), or (DT10 + So1, 3, 5, 7, 10). Control cases included 3 d of cell culture on soft Scram/RGD (orange) or HAVDI/RGD (blue) hydrogels (So3), denoted “soft control (S/R)” and “soft control (H/R)”, respectively. b Representative immunostaining images of hMSC nuclei (blue) and YAP (green) after speciﬁed treatments on stiff and soft substrates as shown in a. Scale bars, 20 µm. S/R is Scram/RGD, H/R is HAVDI/RGD. c–e Quantiﬁcation of YAP nuclear/cytoplasmic (n/c) ratios in hMSCs after speciﬁed treatments on stiff and soft substrates as shown in a. S/ R (orange) and H/R (blue) represent all conditions related to soft Scram/RGD and HAVDI/RGD hydrogels, respectively. Soft control (S/R) and soft control (H/R) are baselines of YAP n/c ratios after 3 d culture for the S/R and H/R groups, respectively. YAP n/c ratio increased with culture time on TCP. c After 1 d of culture on TCP (purple, DT1), YAP localized to the nucleus (from left to right n = 108, 79, 70, 71, 108, 120, 100, 68 cells examined over 5, 11, 23, 15, 5, 26, 13, 13 images respectively, p values were obtained using one-way ANOVA followed by Tukey’s post hoc test, mean ± s.e.m.). Following transfer of hMSCs to either soft Scram/RGD (orange) or soft HAVDI/RGD (blue) substrates and culture for 1 d or 3 days (DT1 + So1, 3), the YAP n/c ratios decreased separately to the baseline of soft control S/R or H/R levels, showing a fully reversible YAP nuclear accumulation. NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x 5f), an effect attributed to the enlargement of nuclear pores in deformed nuclei that promoted YAP import rate without changing the export rate31. Thus, RNC served as a proxy for nuclear deformation in our model. simultaneously. The nucleus was regarded as viscoelastic, with elastic deformation that recovered transiently with reduced traction force and inelastic deformation that accumulated loading (culture) time on TCP. As expected, cell traction forces increased with substrate stiffness (Esub) due to the tension-mediated recruitment of integrin, which resulted in a higher kon on stiffer substrates57. The range of kon depended upon the integrin density on the membrane60. N-cadherin could downregulate traction force and focal adhesion length over an intermediate range of Esub (5–20 kPa)39, due to lower integrin density and lower kon60. In our model, an Arrhenius relationship was used to relate kon and Esub to the presentation of HAVDI (details in Supplementary Methods) (Fig. 5b). The different kon Esub relationships on Scram/RGD and HAVDI/RGD hydrogels resulted in double sigmoidal relation- ships between Ftrac and Esub (Fig. 5c). The stiffness threshold for elevating traction force was thus higher on HAVDI/RGD substrates than on Scram/RGD substrates. Our predictions of RNC matched trends observed in our experiments, and were consistent with the model that HAVDI and stiffness co-determined kon, and that this effect gave rise to the observed double sigmoidal relationships between RNC and Esub on substrates with and without HAVDI, showing attenuated YAP nuclear localization (Fig. 5g). After this veriﬁcation of the reasonability of our model for kon, the model was used to reproduce the phenomena observed in Fig. 3. Consistent with other observations of nuclear deformation, the nucleus was treated as viscoelastic, with deformation having both elastic (reversible) and plastic (irreversible) components64–66. Our model predicted the reductions in RNC over time following transferal of MSCs from TCP to soft hydrogels, and attributed these trends to reductions in tractions associated with release of elastic deformation of the nucleus. The plastic deformation of the nucleus captured the observed increase in RNC with increasing culture time on TCP (Fig. 5h). In cells transferred from TCP to soft gels, the elastic deformation recovered while the plastic deformation remained, reducing RNC over time. RNC in cells transferred to HAVDI/RGD substrates was lower than in those transferred to Scram/RGD substrates because HAVDI reduced traction and thus facilitated elastic recovery. NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x d Following 3 d of culture on TCP (DT3), the YAP n/c ratio remained above the baselines of soft control in hMSCs transferred to Scram/RGD substrates for 1 d or 3 d (S/R: DT3 + So1, 3), but returned to the baselines in hMSCs transferred to HAVDI/RGD substrates for 1 d or 3 d (H/R: DT3 + So1, 3) (from left to right n = 108, 85, 93, 77, 71, 108, 85, 68, 72, 68 cells examined over 5, 6, 17, 15, 15, 5, 6, 19, 6, 13 images respectively, p values were obtained using one-way ANOVA followed by Tukey’s post hoc test, mean ± s.e.m.). This showed that, following 3 d of culture on TCP, YAP nuclear localization was partially reversible response without HAVDI (S/R), but fully reversible with HAVDI (H/R). e Following 10 d of culture on TCP (DT10), the YAP n/c ratio remained elevated and unchanged in hMSCs transferred to Scram/RGD substrates for up to 10 d (S/R: DT10 + So1, 3, 5, 7, 10), but decreased towards baseline levels within 1 d and maintained the levels in hMSCs transferred to HAVDI/RGD substrates for up to 10 d (H/R: DT10 + So1, 3, 5, 7, 10) (from left to right n = 108, 85, 66, 76, 74, 66, 78, 84, 84, 71, 108, 85, 66, 76, 70, 75, 102, 70, 78, 68 cells examined over 5, 6, 9, 11, 16, 21, 21, 23, 26, 15, 5, 6, 9, 11, 28, 21, 37, 20, 8, 13 images respectively, p values were obtained using one-way ANOVA followed by Tukey’s post hoc test, mean ± s.e.m.). This showed that, following 10 d of culture on TCP, YAP nuclear accumulation was permanent (irreversible) over the experimental timescale without HAVDI (S/R), but partially reversible with HAVDI (H/R). Source data are provided as a Source Data ﬁle. rate of translocation of cytoplasmic YAP to the nucleus31. In our model, HAVDI/RGD and Scram/RGD substrates yielded differ- ent tractions, nuclear deformation, pore sizes and hence RNC (Fig. 5d). We hypothesized that actomyosin contractility affects nuclear deformation, which in turn affects the YAP n/c ratio. We thus characterized the nuclear ﬂattening (λN, nuclear length/ height), a measure of nuclear deformation evidenced from our confocal images, and conﬁrmed that it increased monotonically with increasing Ftrac (Fig. 5e, Supplementary Fig. 14). As predicted, nuclear ﬂattening was proportional to RNC (Fig. NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x To test this hypothesis about the effects of HAVDI ligation, we applied a Brieﬂy, in the motor-clutch model (Fig. 5a), actin ﬁlaments attached to ECM through dynamic bonds at integrin clutches with binding rate (kon) and unbinding rate (koff), while myosin motors contracted the actin ﬁlaments with traction force (Ftrac) at velocity (vf). The actin ﬁlament deformed the nucleus and substrate Brieﬂy, in the motor-clutch model (Fig. 5a), actin ﬁlaments attached to ECM through dynamic bonds at integrin clutches with binding rate (kon) and unbinding rate (koff), while myosin motors contracted the actin ﬁlaments with traction force (Ftrac) at velocity (vf). The actin ﬁlament deformed the nucleus and substrate NATURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications 6 ARTICLE NATURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x ulture system could be explained through motor-clutch ECM adhesions. This is important for regenerative th b c a d e f g / m arc S RGD /I D V A H RGD So7 DT1+So7 DT3+So7 DT7+So7 ALP D G R / m arc S D G R /I D V A H So3 DT1+So3 DT3+So3 DT7+So3 RUNX2 D G R / m arc S D G R /I D V A H So7 DT1+So7 DT3+So7 DT7+So7 OCN So7 DT1+So7 DT3+So7 DT7+So7 0 20 40 60 80 100 OCN-positive cells (%) Scram/RGD HAVDI/RGD p=0.026 p=0.034 p=0.038 p=0.011 So7 DT1+So7 DT3+So7 DT7+So7 0 20 40 60 80 100 ALP-positive cells (%) Scram/RGD HAVDI/RGD p=0.032 p=0.018 p=0.0037 p=0.0012 c a a b c d e f g / m a GD So7 DT1+So7 DT3+So7 DT7+So7 D G R / m arc S D G R /I D V A H So3 DT1+So3 DT3+So3 DT7+So3 RUNX2 D G R / m arc S D G R /I D V A H So7 DT1+So7 DT3+So7 DT7+So7 OCN So7 DT1+So7 DT3+So7 DT7+So7 0 20 40 60 80 100 OCN-positive cells (%) Scram/RGD HAVDI/RGD p=0.026 p=0.034 p=0.038 p=0.011 60 80 100 e cells (%) Scram/RGD HAVDI/RGD p=0.0012 c c c b D G R / m arc S D G R /I D V A H So3 DT1+So3 DT3+So3 DT7+So3 b d e g So7 DT1+So7 DT3+So7 DT7+So7 0 20 40 60 80 100 OCN-positive cells (%) Scram/RGD HAVDI/RGD p=0.026 p=0.034 p=0.038 p=0.011 So7 DT1+So7 DT3+So7 DT7+So7 0 20 40 60 80 100 ALP-positive cells (%) Scram/RGD HAVDI/RGD p=0.032 p=0.018 p=0.0037 p=0.0012 e f g / m arc S RGD /I D V A H RGD So7 DT1+So7 DT3+So7 DT7+So7 ALP OCN f f g ECM adhesions. This is important for regenerative therapies10 involving in vitro expansion of MSCs on TCP and subsequent in vivo transplantation, which must overcome the effects of TCP exposure that bias MSCs toward osteogenic differentiation. Using PEG hydrogels with co-presentation of RGD and HAVDI peptides that we showed to mimic cell-ECM and cell-cell adhesions, we demonstrated that HAVDI signaling can be used in our culture system could be explained through motor-clutch dynamics. y MSCs in vivo receive mechanical cues from both cell-ECM and cell-cell adhesions, with the former promoting mechanosensing and the latter inhibiting it. NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x Results suggested that effects of HAVDI on mechanical sensation and nuclear translocation of YAP observed Stress ﬁbers alignments are well known to respond dynamically to the mechanics of the cellular microenvironment61–63. To validate model predictions, we ﬁrst measured the organization of F-actin (quantiﬁed by F-actin anisotropy). In accordance with prior observations39, F-actin anisotropy increased with substrate stiffness, and decreased signiﬁcantly in presence of HAVDI on substrates of intermediate stiffness (20 kPa) (Supplementary Fig. 12). We then performed traction force microscopy for cells cultured on substrates of prescribed peptide mixtures and stiffness. For cells seeded on Scram/RGD hydrogels, the traction increased with increasing hydrogel stiffness (Supplementary Fig. 13). For cells seeded on HAVDI/RGD hydrogels, consistent with observa- tions of Cosgrove et al.39 and our models, tractions were reduced by HAVDI presentation on substrates with an intermediate range of substrate moduli (10–20 kPa), but unaltered for compliant (1 kPa) or stiff (41 kPa) substrates (Fig. 5c). These results veriﬁed the prediction that MSC mechanosensing, and the peptide depen- dence thereof, required actomyosin contractility. The model predicted that RNC increased with increasing nuclear deformation, due to previously reported increases in the 7 ARTICLE NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x NATURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunicatio Methods P i Preparation of PEG hydrogels and peptide conjugation. PEG hydrogels were prepared from 8-arm PEG maleimide (hexaglycerol) (PEG-MAL, 10 kDa, JenKem Technology) backbone and 8-arm PEG thiol (hexaglycerol) (PEG-SH, 10 kDa, JenKem Technology) crosslinker. RGD peptides (GCGYGRGDSSPG) for ﬁbro- nectin, and HAVDI (HAVDIGGGC) or scrambled HAVDI control (AGVGD- HIGC) peptides for N-cadherin (Sangon Biotech, Shanghai) were covalently conjugated to the PEG-MAL backbone via Michael addition reactions between the cysteine residues on these peptides and the maleimide on the PEG-MAL backbone. PEG-MAL (5 mM) and peptides were dissolved in phosphate-buffered saline (PBS) for 1 h at 37 °C for peptide conjugation. For these studies, all peptides were used at a ﬁnal concentration of 1 mM in hydrogels (except for hydrogel control without any peptide conjugation). Cell culture. Isolated hMSCs (labeled as P1) from human bone marrow were frozen down in protein-free cryopreservation medium (Cyagen, HUXMX-07021) and then used for all experiments in this manuscript. Growth medium (Cyagen, HUXMA- 90011) was changed every 2-3 days. Hydrogels for cell culture were sterilized in 75% (v/v) aqueous ethanol for 3–4 h followed by ﬁve rinses with sterilized PBS as described75. Prior to cell seeding, hydrogels were prewetted with growth medium for 30 min. hMSCs were seeded at low density (1000 cells per cm2) to avoid cell-cell interactions, except as noted. Sparse or conﬂuent cells were seeded at 1000 or 20,000 cells per cm2. For studies without mechanical dosing on TCP, hMSCs were directly seeded on soft hydrogels after cell thawing. For studies with mechanical dosing on TCP, hMSCs were seeded on TCP at cell density from 1000 to 20,000 cells per cm2, depending on the culture time on TCP. For myosin inactivation studies, inhibition of myosin was achieved using 50 µM blebbistatin (Abcam, ab120425) for 30 min. For osteogenic differentiation studies, cells were cultured in an osteogenic medium (Cyagen, HUXMA-90021) to remove any confounding effect of chemical dosing and accelerate the process of osteogenic differentiation. For N-cadherin-blocking studies, 1 mM soluble HAVDI(-C) peptides (HAVDIGGG) were added to the growth media directly following hMSCs seeding, according to methods previously described39,76. The soluble peptides designed to bind with the N-cadherin receptor on the cell membrane can avoid conjugating to the PEG-MAL backbone by removing cysteine in the sequence of original HAVDI peptides. Hydrogel fabrication and characterization. Methods P i Hydrogels with a ﬁnal thickness of ~200 μm were formed after 30 min Michael addition reaction between dual peptide modiﬁed PEG-MAL and PEG-SH crosslinker on acrylated cover glasses (diameter ~14 mm). After multiple PBS washes (>4 times for 5 min each to remove the unreacted groups), the as-prepared hydrogels were used to characterize mechanical properties or for cell experiments. Rho-RGD (rhodamine-labeled RGD peptide) and FITC-HAVDI (FITC-labeled HAVDI peptide) were used to characterize the distributions and concentrations of RGD or HAVDI peptides in hydrogels and to determinate their coupling efﬁciency by measuring the ﬂuorescence spectroscopy at 588 nm and 520 nm, respectively. Brieﬂy, the prepared PEG hydrogels incorporated with 1 mM peptides (Rho-RGD or FITC-HAVDI) were rinsed with PBS ( > 4 times for 5 min each) to collect unbound peptides on a supernatant solution. The supernatant and the control solution (containing the initial peptide concentration of 1 mM) were both diluted with PBS to obtain the same volume. Their ﬂuores- cence emission intensity was measured to calculate the conjugation efﬁciency. Immunostaining and quantiﬁcation. hMSCs were ﬁxed in 4% paraformaldehyde (formaldehyde solution: PBS = 1:9) for 20 min, followed by 10 min permeabili- zation with 0.5% Triton X-100 in PBS, and 30 min blocking non-speciﬁc binding sites with 5% bovine serum albumin (BSA) in PBS. Primary antibodies were diluted in 1% BSA in PBS and added overnight at 4 °C. Antibodies and dilutions used in this study contained anti-YAP (1:100, rabbit, Cell Signaling no. 14074), anti-β- catenin (1:200, mouse, Cell Signaling no. 2677), anti-N-cadherin (1:200, rabbit, Cell Signaling no. 13116), anti-RUNX2 (1:1600, rabbit, Cell Signaling no. 12556), anti- OCN (1:200, rabbit, Thermo Fisher no. PA5-96529). After three PBS rinses, AlexaFluor-488[H + L] secondary antibodies (1:500, goat anti-rabbit, Cell Signal- ing no. 4412) and AlexaFluor-647[H + L] secondary antibodies (1:500, goat anti- mouse, Cell Signaling no. 4410) were added for 2 h at room temperature, followed by F-actin staining using Rhodamine Phalloidin (1:1,000; Invitrogen no. R415) incubated for 30 min. All immunostained samples were embedded in ProLong® Gold Antifade Reagent with DAPI (Cell Signaling no. 8961) and visualized with Olympus FV3000 confocal microscope at 10 × 0.4NA (1.243 μm/pixel) and 60 × 1.42NA (0.207 μm/pixel) oil immersion objective lenses. The amount of the N-cadherin and β-catenin in a region of mimetic or real cell-cell adhesion was quantiﬁed utilizing RGB Proﬁler plugin in Image J. ARTICLE ARTICLE Fig. 4 The HAVDI/RGD hydrogel can be used to reduce long-term retention of the effects of mechanotransduction and inhibit osteogenic differentiation of hMSCs. a Schematic of the experimental protocol for assessing osteogenic differentiation of hMSCs cultured on TCP for 1-7 d, followed by 3 d (for RUNX2) or 7 d (for OCN and ALP) culture on 20 kPa Scram/RGD or HAVDI/RGD hydrogels. hMSCs were cultured on TCP (dark purple) in growth medium (pink) for 1, 3, or 7 d and then transferred (light yellow, day 0) to soft Scram/RGD (orange) or HAVDI/RGD (blue) hydrogels for additional 3 d or 7 d of culture in osteogenic medium (light blue) before collection and analysis (gray, day 3 or day 7). RUNX2 expression was assessed in the conditions denoted So3 (soft control, without culture on TCP), DT1 + So3, DT3 + So3, or DT7 + So3. OCN and ALP expression were assessed in the conditions denoted So7 (soft control, without culture on TCP), DT1 + So7, DT3 + So7, or DT7 + So7. b Representative confocal images of cells stained for RUNX2, a marker of early osteogenic differentiation, in hMSCs following treatment as shown in a. Scale bars, 20 µm. c Analysis of confocal images showed a signiﬁcant increase in the RUNX2 n/c ratio with increasing exposure to TCP in hMSCs transferred to Scram/RGD hydrogels (from left to right n = 77, 49, 59, 73, 72, 44, 84, 92 cells examined over 7, 15, 19, 20, 12, 24, 23, 26 images respectively, p-values were obtained using one-way ANOVA followed by Tukey’s post hoc test, mean ± s.e.m.). However, effects of TCP were evident only after 7 days of exposure to TCP in hMSCs transferred to HAVDI/RGD hydrogels, and was greatly attenuated compared to Scram/RGD for 7 days of mechanical dosing on TCP. d–g Representative images of staining for late osteogenic differentiation markers OCN (d) and ALP (f) in hMSCs following the treatment shown in a. Scale bars, 20 µm. Quantiﬁcation of the percentage of OCN-positive cells (e) and ALP-positive cells (g) afﬁrmed that HAVDI ligation signiﬁcantly attenuated osteogenic differentiation after mechanical priming on TCPS (n = 3 experiments per group, p values were obtained using one-way ANOVA followed by Tukey’s post hoc test, mean ± s.e.m.). Source data are provided as a Source Data ﬁle. ARTICLE to reverse a nuclear translocation of YAP, as well as sustained mechanosensing effects in the form of RUNX2, ALP and OCN activity by altering the motor-clutch dynamics in MSCs. Taken together, results show that N-cadherin signaling arising from cell- cell adhesion reduces long-term retention of the effects of mechanotransduction in MSCs and can be used to engineer cell culture platforms to modulate “mechanical memory” in hMSCs. condition, the mean and standard error of mean were obtained from dozens of random points on 3 hydrogel replicates. We note that, unlike most ECM materials, which are hyperelastic and/or ﬁbrous in character72–74, the hydrogel used in this study was well ﬁt by the Hertz model. Although this simpliﬁes analysis, it is a limitation of the study and future work should address the role of ECM non- linearity on binding energetics and kinetics. condition, the mean and standard error of mean were obtained from dozens of random points on 3 hydrogel replicates. We note that, unlike most ECM materials, which are hyperelastic and/or ﬁbrous in character72–74, the hydrogel used in this study was well ﬁt by the Hertz model. Although this simpliﬁes analysis, it is a limitation of the study and future work should address the role of ECM non- linearity on binding energetics and kinetics. hMSC isolation. Human mesenchymal stem cells (hMSCs) were isolated from human bone marrow provided by commercial sources (Cyagen Biosciences). Brieﬂy, cells were obtained from donors by bone marrow aspiration, then mono- cyte density centrifugation was performed and selected for adherent culture. Standard analytical methods were used to screen cell growth and differentiation into fat and bone. NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x We showed that these two antagonistic factors also affect nuclear translocation of YAP, and found that N-cadherin signaling can reverse these effects arising from cell- NATURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications 8 NATURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications Methods P i Percentages of OCN positive cells were calculated by counting the number of OCN positive cells and then Mechanical characterization of hydrogels. Indentation experiments were per- formed to estimate the Young’s moduli of hydrogels. Following standard procedures67–69, Young’s modulus was assessed from load-displacement curves obtained using a commercial nanoindenter (Piuma Nanoindenter, Optics11, Amsterdam, N.L.) with a microsphere indentation tip (diameter: 2.5 µm, stiffness: 0.28 N/m). Load-displacement curves (representative data are shown in Supple- mentary Fig. 3) were converted to load-indentation curves and ﬁtted to the Hertz model, assuming a linear elastic and isotropic material response70: P ¼ 4 3 E 1 υ2 R1=2h3=2 ð1Þ ð1Þ where P is the load on the hydrogel, R is the radius of the spherical tip, h is the indentation depth, and E and υ are the Young’s modulus and Poisson’s ratio of the hydrogel, respectively. For isotropic materials, including PEG hydrogels, the compressive and tensile moduli often have the same amplitude, and vary together with changes in material parameters71. Thus, the Young’s modulus obtained by localized indentation could reﬂect the substrate stiffness sensed by cells. For each compressive and tensile moduli often have the same amplitude, and vary together with changes in material parameters71. Thus, the Young’s modulus obtained by localized indentation could reﬂect the substrate stiffness sensed by cells. For each 9 ARTICLE NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x the total number of cells based on DAPI staining. To exclude the data cells with cell-cell interactions, only the immunostaining results of having no contact with neighbors were counted for quantiﬁcation in our pt as noted the case of the conﬂuent cells. ng and quantiﬁcation. hMSCs were ﬁxed with 4% paraformaldehyde Phosphatase Color Development Kit (Beyotime, C3206) followin facturer’s instructions. Cell nuclei were then stained with DAPI t cell numbers. Percentages of ALP positive cells were calculated b number of blue-violet stained cells and then dividing by the tota YAP and RUNX2 n/c ratio quantiﬁcation. For sparse cells, the d e g h [YAP]C dN hN [YAP]N = = YAP YAP f Model Data c b a TCP-S/R TCP-H/R koff kon Esub Myosin F-actin RGD HAVDI vf Nucleus Ftrac ηN EN Integrin N-cadherin HAVDI/RGD substrate Integrin N-cadherin RGD HAVDI Nucleus Cytoplasm YAP Small pore Balanced transport Low Ftrac Low Ftrac F-actin Scram/RGD substrate Integrin RGD Scram High Ftrac High Ftrac YAP Large pore Enhanced import b c b a koff kon Esub Myosin F-actin RGD HAVDI vf Nucleus Ftrac ηN EN Integrin N-cadherin a c d e [YAP]C dN hN [YAP]N = = YAP YAP f HAVDI/RGD substrate Integrin N-cadherin RGD HAVDI Nucleus Cytoplasm YAP Small pore Balanced transport Low Ftrac Low Ftrac F-actin Scram/RGD substrate Integrin RGD Scram High Ftrac High Ftrac YAP Large pore Enhanced import d HAVDI/RGD substrate Integrin N-cadherin RGD HAVDI Nucleus Cytoplasm YAP Small pore Balanced transport Low Ftrac Low Ftrac F-actin Scram/RGD substrate Integrin RGD Scram High Ftrac High Ftrac YAP Large pore Enhanced import total number of cells based on DAPI staining. To exclude the data ls with cell-cell interactions, only the immunostaining results of ing no contact with neighbors were counted for quantiﬁcation in our s noted the case of the conﬂuent cells. and quantiﬁcation. hMSCs were ﬁxed with 4% paraformaldehyde lowed by washing with PBS, and stained with BCIP/NBT Alkaline Phosphatase Color Development Kit (Beyotime, C3206) following the manu- facturer’s instructions. Cell nuclei were then stained with DAPI to count the total cell numbers. Percentages of ALP positive cells were calculated by counting the number of blue-violet stained cells and then dividing by the total cell number. YAP and RUNX2 n/c ratio quantiﬁcation. ARTICLE ARTICLE Fig. 5 A modiﬁed motor-clutch model explains the role of HAVDI ligation in reversing YAP nuclear localization in MSCs. a In the motor-clutch model, integrin clutches connected to RGD on ECMs with binding rate (kon), and broke with ECM at force dependent unbinding rate (koff). Myosin motors pulled actin ﬁlaments with traction force (Ftrac) at velocity (vf). HAVDI binding (N-cadherin based adherens junctions) disturbed integrin clustering and decreased kon in integrin binding to RGD on ECMs. Decreased kon reduced the Ftrac in actin ﬁlaments, and hence deformation of the viscoelastic nucleus (spring stiffness EN and viscosity ηN). b The model for variation of kon with ECM stiffness (Esub) on HAVDI/RGD (H/R) and Scram/RGD (S/R) substrates. c Traction on H/R or S/R substrates with different stiffness Esub, showing that H/R increased the stiffness threshold for generating elevated traction. Symbols are measured from experiments (mean ± s.e.m. from Supplementary Fig. 13b), while lines are modeling results. d In the force-dependent YAP redistribution model, elevated traction resulted in increased nuclear deformation, enlarged nuclear pores, increased YAP import rate, and thus higher YAP n/c ratio (RNC) on HAVDI/RGD substrates than on Scram/RGD substrates. e The relationship between nuclear ﬂattening (λN) and actin contraction. Symbols are measured from experiments (mean ± s.e.m. from Supplementary Fig. 14b), while lines are modeling results. f RNC increased linearly with nuclear ﬂattening for all culture conditions. The line is prediction by our model. Hollow squares are data from reference31. The values of solid symbols on x- axis (λN) are from Supplementary Fig. 14b. The values of solid symbols on y-axis (RNC) are from Fig. 2f and Fig. 3c–e. The lateral and vertical error bars indicate s.e.m. Results indicated that RNC is an indicator of nuclear deformation. g Validation of the model for cells cultured on substrates of different stiffness. The model reproduced the di-sigmoidal curves in data from S/R and H/R groups (3 d for culture time). The values of dots on x-axis (stiffness) are from Fig. 1c, The values of dots on y-axis (RNC) are from Fig. 2f. The lateral and vertical error bars indicate s.e.m. h An explanation of mechanical memory in cells transferred from TCP (dark purple) to S/R (orange) or to H/R (blue) substrates, denoted TCP-S/R or TCP-H/R, respectively. RNC increased with culture time due to plastic deformation of the nucleus. Data availability Rsparse NC ¼ ðInucleus=AnucleusÞ ðIcell InucleusÞ=ðAcell AnucleusÞ ð2Þ Source data of the Figs. 1c, 2f, 3c-e, 4c, 4e, 4g, 5b, 5c, 5e, 5g, 5h; Supplementary Fig. 4, Supplementary Fig. 5, Supplementary Fig. 6b, Supplementary Fig. 7b, Supplementary Fig. 8b, Supplementary Fig. 9c, Supplementary Fig. 10b, Supplementary Fig. 11, Supplementary Fig. 12b, Supplementary Fig. 13b, Supplementary Fig. 14b-e are provided with this paper. All other relevant data supporting the key ﬁndings of this study are available within the article and its Supplementary Information ﬁles or from the corresponding author upon reasonable request. Source data are provided with this paper. ð2Þ where Anucleus is the area of the nucleus as measured by DAPI staining, Acell is the overall area of the cell as delineated by F-actin staining, and Icell is the total ﬂuorescence intensity in the overall cell Intensities and areas were measured using where Anucleus is the area of the nucleus as measured by DAPI staining, Acell is the overall area of the cell as delineated by F-actin staining, and Icell is the total ﬂuorescence intensity in the overall cell. Intensities and areas were measured using Image J (1.52p). Images of cells with saturated ﬂuorescence were excluded. ﬂ ll l ll l f ll h hb cell ﬂuorescence intensity in the overall cell. Intensities and areas were measured using Image J (1.52p). Images of cells with saturated ﬂuorescence were excluded. g p g For conﬂuent cells, a multicellular zone of cells contacting with neighbors was delineated, and the above procedures were repeated to obtain the average YAP n/c ratio, Rconfluent NC , over the region of conﬂuent cells: Code availability Source code for motor-clutch model is available for download at https://github.com/ zhuhy2021/Motor-clutch-model-including-HAVDI-effect (https://doi.org/10.5281/ zenodo.5537041)80. Rconfluent NC ¼ ∑ N i¼1 Ii nucleus ∑ N i¼1 Ai nucleus Izone ∑ N i¼1 Ii nucleus Azone ∑ N i¼1 Ai nucleus ð3Þ Rconfluent NC ¼ ∑ N i¼1 Ii nucleus ∑ N i¼1 Ai nucleus Izone ∑ N i¼1 Ii nucleus Azone ∑ N i¼1 Ai nucleus ð3Þ ð3Þ Received: 6 November 2020; Accepted: 1 October 2021; where Ai nucleus is the area of the nucleus i in the zone containing N nuclei, Ii nucleus is the total ﬂuorescence of nucleus i, Izone is the total ﬂuorescence intensity of the zone, and Azone is the area of the zone. Nuclear deformation analysis. z-stacks of 0.5 µm steps were captured, nuclear x-y projections (major and minor axis of nucleus) and x-z projections were recon- structed using Bitplane Imaris (7.2.3) software on the DAPI channel to measure the nuclear length (major axis) and nuclear height, respectively. Nuclear ﬂattening was calculated as the ratio of nuclear length to height31. ARTICLE Differences between TCP-S/R and TCP-H/R groups could be attributed to the differences in elastic recovery of the nucleus modulated by N-cadherins. Source data are provided as a Source Data ﬁle. YAP or RUNX2 was calculated following techniques used by others21,77–79 as the ratio of the total ﬂuorescence intensity in the nucleus, Inucleus, to the total ﬂuor- escence in the remainder of the cell, weighted by the areas of the nucleus and the remainder of the cell: Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article. NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x For sparse cells, the nucleus and cytoplasm were identiﬁed by F-actin and DAPI staining, and the n/c ratio Rsparse NC for d e g h [YAP]C dN hN [YAP]N = = YAP YAP f Model Data TCP-S/R TCP-H/R HAVDI/RGD substrate Integrin N-cadherin RGD HAVDI Nucleus Cytoplasm YAP Small pore Balanced transport Low Ftrac Low Ftrac F-actin Scram/RGD substrate Integrin RGD Scram High Ftrac High Ftrac YAP Large pore Enhanced import NATURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications d HAVDI/RGD substrate Integrin N-cadherin RGD HAVDI Nucleus Cytoplasm YAP Small pore Balanced transport Low Ftrac Low Ftrac F-actin e f d e [YAP]C dN hN [YAP]N = = YAP YAP f f trac trac Scram/RGD substrate Integrin RGD Scram High Ftrac High Ftrac YAP Large pore Enhanced import g h Model Data TCP-S/R TCP-H/R h h g dividing by the total number of cells based on DAPI staining. To exclude the data of adjacent cells with cell-cell interactions, only the immunostaining results of single cells having no contact with neighbors were counted for quantiﬁcation in our study, except as noted the case of the conﬂuent cells. Phosphatase Color Development Kit (Beyotime, C3206) following the manu- facturer’s instructions. Cell nuclei were then stained with DAPI to count the total cell numbers. Percentages of ALP positive cells were calculated by counting the number of blue-violet stained cells and then dividing by the total cell number. YAP and RUNX2 n/c ratio quantiﬁcation. For sparse cells, the nucleus and cytoplasm were identiﬁed by F-actin and DAPI staining, and the n/c ratio Rsparse NC for YAP and RUNX2 n/c ratio quantiﬁcation. For sparse cells, the nucleus and cytoplasm were identiﬁed by F-actin and DAPI staining, and the n/c ratio Rsparse NC for ALP staining and quantiﬁcation. hMSCs were ﬁxed with 4% paraformaldehyde for 20 min, followed by washing with PBS, and stained with BCIP/NBT Alkaline 10 NATURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications References 1. Engler, A. J., Sen, S., Sweeney, H. L. & Discher, D. E. Matrix elasticity directs stem cell lineage speciﬁcation. Cell 126, 677–689 (2006). 1. Engler, A. J., Sen, S., Sweeney, H. L. & Discher, D. E. Matrix elasticity directs stem cell lineage speciﬁcation. Cell 126, 677–689 (2006). 2. Frith, J. E. et al. Mechanically-sensitive miRNAs bias human mesenchymal stem cell fate via mTOR signalling. Nat. Commun. 9, 257 (2018). 2. Frith, J. E. et al. Mechanically-sensitive miRNAs bias human mesenchymal stem cell fate via mTOR signalling. Nat. Commun. 9, 257 (2018). g g 3. Liu, L. et al. Mechanoresponsive stem cells to target cancer metastases through biophysical cues. Sci. Transl. Med 9, eaan2966 (2017). Implementation of motor-clutch model. The model was modiﬁed from a well- established stochastic motor-clutch model56–58,60. Brieﬂy, the model considered the connection between actin ﬁlament and substrate was realized through nc clutches (focal adhesion complex, including integrin and adapter proteins). nm myosin molecules with stall force (Fm) pulled the actin ﬁlament at a retrograde velocity vu. The clutches broke and reconnected according to characteristic on and off rates (kon and koff). kon was dependent upon the integrin density on the cell membrane, which changed according to the conformation (folding or unfolding) of the adapter protein talin. Previous studies have shown that kon grew with increasing substrate stiffness (Esub)60. Here, the rela- tionship between kon and HAVDI presentation was modeled using Arrhenius function. Simulation ran for 105 time steps (>1000 s) to ensure the reaching of a steady state. Details of the model can be found in Supplementary Methods and Supplementary Table 1. The calculation of the model is conducted by a custom code80. p y 4. Chaudhuri, O. et al. Hydrogels with tunable stress relaxation regulate stem cell fate and activity. Nat. Mater. 15, 326–334 (2015). 5. Vining, K. H. & Mooney, D. J. Mechanical forces direct stem cell behaviour in development and regeneration. Nat. Rev. Mol. Cell Bio 18, 728 (2017). 6. Grifﬁn, D. R., Weaver, W. M., Scumpia, P. O., Di Carlo, D. & Segura, T. Accelerated wound healing by injectable microporous gel scaffolds assembled 6. Grifﬁn, D. R., Weaver, W. M., Scumpia, P. O., Di Carlo, D. & Segura, T. Accelerated wound healing by injectable microporous gel scaffolds assembled 6. Grifﬁn, D. R., Weaver, W. M., Scumpia, P. O., Di Carlo, D. & Segura, T. NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x 13. Halder, G., Dupont, S. & Piccolo, S. Transduction of mechanical and cytoskeletal cues by YAP and TAZ. Nat. Rev. Mol. Cell Bio 13, 591–600 (2012). 44. Wei, Q. et al. BMP-2 signaling and mechanotransduction synergize to drive osteogenic differentiation via YAP/TAZ. Adv. Sci. 7, 1902931 (2020). osteogenic differentiation via YAP/TAZ. Adv. Sci. 7, 19029 g 45. Labernadie, A. et al. A mechanically active heterotypic E-cadherin/N-cadherin adhesion enables ﬁbroblasts to drive cancer cell invasion. Nat. Cell Biol. 19, 224–237 (2017). 14. Totaro, A., Panciera, T. & Piccolo, S. YAP/TAZ upstream signals and downstream responses. Nat. Cell Biol. 20, 888–899 (2018). 46. Pasapera, AnaM. et al. Rac1-dependent phosphorylation and focal adhesion recruitment of myosin IIA regulates migration and mechanosensing. Curr. Biol. 25, 175–186 (2015). 15. Aragona, M. et al. A mechanical checkpoint controls multicellular growth through YAP/TAZ regulation by actin-processing factors. Cell 154, 1047–1059 (2013). 16. Driscoll, TristanP., Cosgrove, BrianD., Heo, S.-J., Shurden, ZachE. & Mauck, RobertL. Cytoskeletal to nuclear strain transfer regulates YAP signaling in mesenchymal stem cells. Biophys. J. 108, 2783–2793 (2015). 47. Guvendiren, M. & Burdick, J. A. Stiffening hydrogels to probe short-and long- term cellular responses to dynamic mechanics. Nat. Commun. 3, 792 (2012). 48. Paknikar, A. K., Eltzner, B. & Köster, S. Direct characterization of cytoskeletal reorganization during blood platelet spreading. Prog. Biophys. Mol. Biol. 144, 166–176 (2019). 17. Zhong, W. et al. Mesenchymal stem cell and chondrocyte fates in a multishear microdevice are regulated by yes-associated protein. Stem Cells Dev. 22, 2083–2093 (2013). 49. Sunyer, R. et al. Collective cell durotaxis emerges from long-range intercellular force transmission. Science 353, 1157–1161 (2016). 18. Dupont, S. et al. Role of YAP/TAZ in mechanotransduction. Nature 474, 179 (2011). 50. Hou, Y. et al. Surface roughness and substrate stiffness synergize to drive cellular mechanoresponse. Nano Lett. 20, 748–757 (2020). 50. Hou, Y. et al. Surface roughness and substrate stiffness synergize to drive ll l h N L tt 20 748 757 (2020) 19. Tang, Y. et al. MT1-MMP-dependent control of skeletal stem cell commitment via a β1-integrin/YAP/TAZ signaling axis. Dev. Cell 25, 402–416 (2013). cellular mechanoresponse. Nano Lett. 20, 748–757 (2020). 51. Cui, Y. et al. Cyclic stretching of soft substrates induces spreading and growth. Nat. Commun. 6, 6333 (2015). 52. Valenti, M. T., Dalle Carbonare, L. & Mottes, M. Osteogenic differentiation in healthy and pathological conditions. Int J. Mol. Sci. 18, 41 (2017). 20. NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x Force loading explains spatial sensing of ligands by cells. Nature 552, 219–224 (2017). 28. Mui, K. L., Chen, C. S. & Assoian, R. K. The mechanical regulation of integrin–cadherin crosstalk organizes cells, signaling and forces. J. Cell Sci. 129, 1093 (2016). g integrin–cadherin crosstalk organizes cells, signaling and forces. J. Cell Sci. 129, 1093 (2016). 61. McEvoy, E., Deshpande, V. S. & McGarry, P. Transient active force generation and stress ﬁbre remodelling in cells under cyclic loading. Biomech. Model Mechanobiol. 18, 921–937 (2019). 29. Petridou, N. I., Spiró, Z. & Heisenberg, C.-P. Multiscale force sensing in development. Nat. Cell Biol. 19, 581–588 (2017). 62. Sears, C. & Kaunas, R. The many ways adherent cells respond to applied stretch. J. Biomech. 49, 1347–1354 (2016). 30. Tajik, A. et al. Transcription upregulation via force-induced direct stretching of chromatin. Nat. Mater. 15, 1287–1296 (2016). 31. Elosegui-Artola, A. et al. Force triggers YAP nuclear entry by regulating transport across nuclear pores. Cell 171, 1397–1410 (2017). 63. Elson, E. L. & Genin, G. M. The role of mechanics in actin stress ﬁber kinetics. Exp. Cell Res 319, 2490–2500 (2013). p p 32. De Pascalis, C. & Etienne-Manneville, S. Single and collective cell migration: the mechanics of adhesions. Mol. Biol. Cell 28, 1833–1846 (2017). 64. Kim, J.-K. et al. Nuclear lamin A/C harnesses the perinuclear apical actin cables to protect nuclear morphology. Nat. Commun. 8, 2123 (2017). p p gy 65. Swift, J. et al. Nuclear lamin-A scales with tissue stiffness and enhances matrix-directed differentiation. Science 341, 975 (2013). 33. Seong, J. et al. Distinct biophysical mechanisms of focal adhesion kinase mechanoactivation by different extracellular matrix proteins. Proc. Natl Acad. Sci. USA 110, 19372–19377 (2013). 66. Pajerowski, J. D., Dahl, K. N., Zhong, F. L., Sammak, P. J. & Discher, D. E. Physical plasticity of the nucleus in stem cell differentiation. Proc. Natl Acad. Sci. USA 104, 15619–15624 (2007). 34. Kechagia, J. Z., Ivaska, J. & Roca-Cusachs, P. Integrins as biomechanical sensors of the microenvironment. Nat. Rev. Mol. Cell Bio 20, 457–473 (2019). 67. Oyen, M. L. Mechanical characterisation of hydrogel materials. Int. Mater. Rev. 59, 44–59 (2014). 35. Fraley, S. I. et al. A distinctive role for focal adhesion proteins in three- dimensional cell motility. Nat. Cell Biol. 12, 598–604 (2010). 68. Zhu, H. et al. Modeling the mechanics, kinetics, and network evolution of photopolymerized hydrogels. J. Mech. Phys. Solids 142, 104041 (2020). 36. NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x Chen, Z., Luo, Q., Lin, C., Kuang, D. & Song, G. Simulated microgravity inhibits osteogenic differentiation of mesenchymal stem cells via depolymerizing F-actin to impede TAZ nuclear translocation. Sci. Rep. 6, 30322 (2016). 53. Shen, W.-C. et al. Methylation and PTEN activation in dental pulp mesenchymal stem cells promotes osteogenesis and reduces oncogenesis. Na Commun. 10, 2226 (2019). 21. Killaars, A. R. et al. Extended exposure to stiff microenvironments leads to persistent chromatin remodeling in human mesenchymal stem cells. Adv. Sci. 6, 1801483 (2019). 54. Elosegui-Artola, A., Trepat, X. & Roca-Cusachs, P. Control of mechanotransduction by molecular clutch dynamics. Trends Cell Biol. 28, 356–367 (2018). 22. Li, C. X. et al. MicroRNA-21 preserves the ﬁbrotic mechanical memory of mesenchymal stem cells. Nat. Mater. 16, 379 (2016). 55. Case, L. B. & Waterman, C. M. Integration of actin dynamics and cell adhesion by a three-dimensional, mechanosensitive molecular clutch. Nat. Cell Biol. 17, 955–963 (2015). 23. Ladoux, B. & Mège, R.-M. Mechanobiology of collective cell behaviours. Nat. Rev. Mol. Cell Bio 18, 743 (2017). 24. Hsiao, C. et al. Human pluripotent stem cell culture density modulates YAP signaling. Biotechnol. J. 11, 662–675 (2016). 56. Chan, C. E. & Odde, D. J. Traction dynamics of ﬁlopodia on compliant substrates. Science 322, 1687–1691 (2008). g g 25. Totaro, A. et al. YAP/TAZ link cell mechanics to notch signalling to control epidermal stem cell fate. Nat. Commun. 8, 15206 (2017). 57. Elosegui-Artola, A. et al. Rigidity sensing and adaptation through regulation of integrin types. Nat. Mater. 13, 631–637 (2014). 58. Elosegui-Artola, A. et al. Mechanical regulation of a molecular clutch deﬁnes force transmission and transduction in response to matrix rigidity. Nat. Cell Biol. 18, 540 (2016). 26. Bosch-Fortea, M. & Martín-Belmonte, F. Mechanosensitive adhesion complexes in epithelial architecture and cancer onset. Curr. Opin. Cell Biol. 50, 42–49 (2018). 59. Dasgupta, I. & McCollum, D. Control of cellular responses to mechanical cues through YAP/TAZ regulation. J. Biol. Chem. 294, 17693–17706 (2019). 27. Han, M. K. L. & de Rooij, J. Converging and unique mechanisms of mechanotransduction at adhesion sites. Trends Cell Biol. 26, 612–623 (2016). 59. Dasgupta, I. & McCollum, D. Control of cellular responses to mechanical cues through YAP/TAZ regulation. J. Biol. Chem. 294, 17693–17706 (2019). 60. Oria, R. et al. Force loading explains spatial sensing of ligands by cells. Nature 552, 219–224 (2017). 60. Oria, R. et al. References Accelerated wound healing by injectable microporous gel scaffolds assembled from annealed building blocks. Nat. Mater. 14, 737–744 (2015). Accelerated wound healing by injectable microporous gel scaffolds assembled from annealed building blocks. Nat. Mater. 14, 737–744 (2015). g 7. Chaudhuri, P. K., Low, B. C. & Lim, C. T. Mechanobiology of tumor growth. Chem. Rev. 118, 6499–6515 (2018). 8. Trounson, A. & McDonald, C. Stem cell therapies in clinical trials: progress and challenges. Cell Stem Cell 17, 11–22 (2015). 9. Miao, C., Lei, M., Hu, W., Han, S. & Wang, Q. A brief review: the therapeutic potential of bone marrow mesenchymal stem cells in myocardial infarction. Stem Cell Res. Ther. 8, 242 (2017). Statistical analysis. All data for cell experiments were collected using a single cell line studied on at least three hydrogel replicates per condition, except where noted. The number of cells counted for each condition was indicated in each ﬁgure legend. Statistical comparisons were performed with Graphpad Prism (8.0.1) or Origin (2020) using one-way analysis of variance (ANOVA) with Tukey’s post hoc test for comparison of multiple groups. Differences were considered to be signiﬁcant when p values were below 0.05. Details of sample sizes and signiﬁcance levels were given in ﬁgure legends. In all ﬁgures, data were shown as mean ± standard error of the mean (s.e.m.) unless otherwise stated. 10. Madl, C. M., Heilshorn, S. C. & Blau, H. M. Bioengineering strategies to accelerate stem cell therapeutics. Nature 557, 335–342 (2018). 11. Yang, C., Tibbitt, M. W., Basta, L. & Anseth, K. S. Mechanical memory and dosing inﬂuence stem cell fate. Nat. Mater. 13, 645–652 (2014). 1. Yang, C., Tibbitt, M. W., Basta, L. & Anseth, K. S. Mechanic dosing inﬂuence stem cell fate. Nat. Mater. 13, 645–652 (201 12. Panciera, T., Azzolin, L., Cordenonsi, M. & Piccolo, S. Mechanobiology of YAP and TAZ in physiology and disease. Nat. Rev. Mol. Cell Bio 18, 758 (2017). 11 TURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications Additional information Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41467-021-26454-x. 77. Killaars, A. R., Walker, C. J. & Anseth, K. S. Nuclear mechanosensing controls MSC osteogenic potential through HDAC epigenetic remodeling. Proc. Natl Acad. Sci. USA 117, 21258–21266 (2020). Correspondence and requests for materials should be addressed to Min Lin. Correspondence and requests for materials should be addressed to Min Lin. 78. Nardone, G. et al. YAP regulates cell mechanics by controlling focal adhesion assembly. Nat. Commun. 8, 15321 (2017). Peer review information Nature Communications thanks Matthew Dalby and the other anonymous reviewer(s) for their contribution to the peer review this work. Peer review information Nature Communications thanks Matthew Dalby and the other anonymous reviewer(s) for their contribution to the peer review this work. 79. Chronopoulos, A. et al. Syndecan-4 tunes cell mechanics by activating the kindlin-integrin-RhoA pathway. Nat. Mater. 19, 669–678 (2020). 80. Zhang C., et al. Mechanics-driven nuclear localization of YAP can be reversed by N-cadherin ligation in mesenchymal stem cells, https://doi.org/10.5281/ zenodo.5537041 (2021). Reprints and permission information is available at http://www.nature.com/reprints Reprints and permission information is available at http://www.nature.com/reprints Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afﬁliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/ licenses/by/4.0/. This work was supported by the National Natural Science Foundation of China (12022206, 11772253), the Shaanxi Province Youth Talent Support Program, the Key Research and Development Program of Shaanxi Province (2021SF-061), the Young Talent Support Plan of Xi’an Jiaotong University, the Foundation of Xi’an Medical University (2018XNRC06) and the National Institutes of Health (R01AR077793). We thank Miss Hao at Instrument Analysis Center of Xi'an Jiaotong University for their assistance with CLSM analysis. Competing interests 75. Ding, B. et al. Tough and cell-compatible chitosan physical hydrogels for mouse bone mesenchymal stem cells in vitro. ACS Appl. Mater. Int. 8, 19739–19746 (2016). 76. Madl, C. M. et al. Maintenance of neural progenitor cell stemness in 3D hydrogels requires matrix remodelling. Nat. Mater. 16, 1233–1242 (2017). NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x Pulous, F. E., Grimsley-Myers, C. M., Kansal, S., Kowalczyk, A. P. & Petrich, B. G. Talin-dependent integrin activation regulates VE-cadherin localization and endothelial cell barrier function. Circulation Res. 124, 891–903 (2019). 69. Yue, L., Wang, S., Wulf, V. & Willner, I. Stiffness-switchable DNA-based constitutional dynamic network hydrogels for self-healing and matrix-guided controlled chemical processes. Nat. Commun. 10, 4774 (2019). 37. Yamamoto, H. et al. Integrin β1 controls VE-cadherin localization and blood vessel stability. Nat. Commun. 6, 6429 (2015). 38. Bertocchi, C. et al. Nanoscale architecture of cadherin-based cell adhesions. Nat. Cell Biol. 19, 28–37 (2017). 70. Amann, E., Wolff, P., Breel, E., van Griensven, M. & Balmayor, E. R. Hyaluronic acid facilitates chondrogenesis and matrix deposition of human adipose derived mesenchymal stem cells and human chondrocytes co- cultures. Acta Biomater. 52, 130–144 (2017). 39. Cosgrove, B. D. et al. N-cadherin adhesive interactions modulate matrix mechanosensing and fate commitment of mesenchymal stem cells. Nat. Mater. 15, 1297 (2016). 71. Nguyen, Q. T., Hwang, Y., Chen, A. C., Varghese, S. & Sah, R. L. Cartilage-like mechanical properties of poly (ethylene glycol)-diacrylate hydrogels. Biomaterials 33, 6682–6690 (2012). 40. Dalby, M. J., García, A. J. & Salmeron-Sanchez, M. Receptor control in mesenchymal stem cell engineering. Nat. Rev. Mater. 3, 17091 (2018). 72. Dowling, E. P., Ronan, W. & McGarry, J. P. Computational investigation of in situ chondrocyte deformation and actin cytoskeleton remodelling under physiological loading. Acta Biomater. 9, 5943–5955 (2013). 41. Zhu, H. et al. The relationship between thiol-acrylate photopolymerization kinetics and hydrogel mechanics: an improved model incorporating photobleaching and thiol-Michael addition. J. Mech. Behav. Biomed. 88, 160–169 (2018). 73. Hall, M. S. et al. Fibrous nonlinear elasticity enables positive mechanical feedback between cells and ECMs. Proc. Natl Acad. Sci. USA 113, 14043–14048 (2016). 42. Truong Quang, B.-A., Mani, M., Markova, O., Lecuit, T. & Lenne, P.-F. Principles of E-cadherin supramolecular organization In vivo. Curr. Biol. 23, 2197–2207 (2013). 74. Shakiba, D. et al. The balance between actomyosin contractility and microtubule polymerization regulates hierarchical protrusions that govern efﬁcient ﬁbroblast–collagen interactions. ACS Nano 14, 7868–7879 (2020). 43. Rape A. D., Zibinsky M., Murthy N., Kumar S. A synthetic hydrogel for the high-throughput study of cell-ECM interactions. Nat. Commun. 6, (2015). 12 NATURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunicatio ARTICLE NATURE COMMUNICATIONS \| https://doi.org/10.1038/s41467-021-26454-x Competing interests Acknowledgements g This work was supported by the National Natural Science Foundation of China (12022206, 11772253), the Shaanxi Province Youth Talent Support Program, the Key Research and Development Program of Shaanxi Province (2021SF-061), the Young Talent Support Plan of Xi’an Jiaotong University, the Foundation of Xi’an Medical University (2018XNRC06) and the National Institutes of Health (R01AR077793). We thank Miss Hao at Instrument Analysis Center of Xi'an Jiaotong University for their assistance with CLSM analysis. Author contributions C.Z., H.Y.Z., M.L. and G.M.G. designed the study, in consultation with F.X., B.C., S.B.L., Y.L., H.H.W., H.G. and T.J.L.; C.Z., H.Y.Z., X.R.R., B.G., Z.Z., Z.Q.L., B.C., S.B.L. B.Y.S performed the experiments, modeling, collected and analyzed the data. C.Z., H.Y.Z., M.L., G.M.G., T.J.L., F.X., Y.L., H.H.W. and H.G. prepared the manuscript. All authors discussed the experiments, modeling, read and commented on the manuscript. 13 NATURE COMMUNICATIONS \| (2021) 12:6229 \| https://doi.org/10.1038/s41467-021-26454-x \| www.nature.com/naturecommunications
https://openalex.org/W2033153539	https://portal.research.lu.se/files/2314013/5154216.pdf	English	null	Home and Health in the Third Age — Methodological Background and Descriptive Findings	International journal of environmental research and public health/International journal of environmental research and public health	2,014	cc-by	10,590	Citation for published version (APA): Kylén, M., Ekström, H., Haak, M., Elmståhl, S., & Iwarsson, S. (2014). Home and health in the third age - methodological background and descriptive findings. International Journal of Environmental Research and Public Health, 11(7), 7060-7080. https://doi.org/10.3390/ijerph110707060 Link to publication Citation for published version (APA): Kylén, M., Ekström, H., Haak, M., Elmståhl, S., & Iwarsson, S. (2014). Home and health in the third age - methodological background and descriptive findings. International Journal of Environmental Research and Public Health, 11(7), 7060-7080. https://doi.org/10.3390/ijerph110707060 Total number of authors: 5 5 Home and health in the third age - methodological background and descriptive findings. ylén, Maya; Ekström, Henrik; Haak, Maria; Elmståhl, Sölve; Iwarsson, Susanne Published in: International Journal of Environmental Research and Public Health DOI: 10.3390/ijerph110707060 General rights U l th p g g g g pp y Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. g q g • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal Read more about Creative commons licenses: https://creativecommons.org/licenses/ Read more about Creative commons licenses: https://creativecommons.org/licenses/ Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. LUND UNIVERSITY PO Box 117 221 00 Lund +46 46-222 00 00 Int. J. Environ. Res. Public Health 2014, 11, 7060-7080; doi:10.3390/ijerph110707060 International Journal of Environmental Research and Public Health ISSN 1660-4601 www.mdpi.com/journal/ijerph Article Home and Health in the Third Age — Methodological Background and Descriptive Findings Maya Kylén 1,, Henrik Ekström 1,2,†, Maria Haak 1,†, Sölve Elmståhl 1,2 and Susanne Iwarsson 1 1 Department of Health Sciences, Lund University, SE-22100, Sweden; E-Mails: Henrik.ekstrom@med.lu.se (H.E.); Maria.haak@med.lu.se (M.H.); Solve.elmstahl@med.lu.se (S.E.); Susanne.iwarsson@med.lu.se (S.I.) 2 Skåne University Hospital, SE-205 02, Sweden † These authors contributed equally to this work. Author to whom correspondence should be addressed; E-Mail: maya.kylen@med.lu.se; Tel.: +46-46-222-19-47. Received: 30 April 2014; in revised form: 30 June 2014 / Accepted: 1 July 2014 / Published: 11 July 2014 Abstract: Background: The understanding of the complex relationship between the home environment, well-being and daily functioning in the third age is currently weak. The aim of this paper is to present the methodological background of the Home and Health in the OPEN ACCESS Int. J. Environ. Res. Public Health 2014, 11, 7060-7080; doi:10.3390/ijerph110707060 Article Int. J. Environ. Res. Public Health 2014, 11 Int. J. Environ. Res. Public Health 2014, 11 Abbreviations ADL: Activities of Daily Living I-ADL: Instrumental Activities of Daily Living P-ADL: Personal Activities of Daily Living HE: Housing Enabler MAP: Magnitude of Accessibility Problems ENABLE-AGE: Enabling Autonomy, Participation, and Well-Being in Old Age MOH: Meaning of Home HCB: Housing Related Control Beliefs UIMH: Usability in My Home GDS: Geriatric Depression Scale PWQ: Psychological wellbeing HOOP: Housing Options for Older People SNAC: Swedish National Study on Aging and Care GÅS: Gott Åldrande i Skåne AD: Assistive Device FES-I: Falls Efficacy Scale-International FOF: Fear of Falling Home and Health in the Third Age — Methodological Background and Descriptive Findings 1 Department of Health Sciences, Lund University, SE-22100, Sweden; E-Mails: Henrik.ekstrom@med.lu.se (H.E.); Maria.haak@med.lu.se (M.H.); Solve.elmstahl@med.lu.se (S.E.); Susanne.iwarsson@med.lu.se (S.I.) † These authors contributed equally to this work. * Author to whom correspondence should be addressed; E-Mail: maya.kylen@med.lu.se; Tel.: +46-46-222-19-47. * Author to whom correspondence should be addressed; E-Mail: maya.kylen@med.lu.se; Tel.: +46-46-222-19-47. Received: 30 April 2014; in revised form: 30 June 2014 / Accepted: 1 July 2014 / Published: 11 July 2014 Abstract: Background: The understanding of the complex relationship between the home environment, well-being and daily functioning in the third age is currently weak. The aim of this paper is to present the methodological background of the Home and Health in the Third Age Study, and describe a sample of men and women in relation to their home and health situation. Methods and Design: The study sample included 371 people aged 67–70, living in ordinary housing in the south of Sweden. Structured interviews and observations were conducted to collect data about objective and perceived aspects of home and health. Results: The majority of the participants were in good health and had few functional limitations. Women had more functional limitations and reported more symptoms than men. Environmental barriers were found in every home investigated; the most were found in the kitchen and hygiene area. Environmental barriers were more common in multi-family than in one-family dwellings. Discussion: This study will increase our knowledge on home and health dynamics among people in the third age. The results have potential to contribute to societal planning related to housing provision, home care and social services for senior citizens. 7061 1. Introduction It is widely known that the world’s age composition is radically changing towards a higher proportion of older people than ever seen before [1]. For example, in Sweden 19.1% of the population are 65 years old or older, and this proportion is expected to increase to 25% by 2060 [2]. Thus, more than one out of five will be over the age of 65 by the year 2030. Since 1990, the life expectancy after retirement age has risen for 65-year-olds as well as for those who have reached the age of 80 [3]. While aging can be viewed from a chronological, biological, psychological or social perspective, chronological and biological aging are not equivalent for all individuals. Hence, different stages of the aging process can be defined in terms of the fourth and third age [3], in terms of population or personal characteristics [4]. For the present study we used a definition based on personal characteristics, where the fourth age is characterized by frailty, cognitive decline and functional loss, and the third age by independence, social engagement and good health. Commonly, the third age is represented by individuals recently retired from work, but the fourth and third ages are dynamic in the sense that they cannot be defined by specific age ranges. As the proportion of the population aged 65 years or older is increasing and the home environment is an important arena in order to support independence and well-being in old age [5], home and health dynamics play an increasingly important role for societal planning in terms of housing provision, health care and social services. There is a growing body of research with regard to home and health in 7062 Int. J. Environ. Res. Public Health 2014, 11 very old age, that is, studies involving individuals who have reached the fourth age [6–9]. However, research is lacking on the complex interaction between the home environment, well-being and daily functioning among people in the third age. The health development along the process of aging is the subject of an extensive amount of studies with some controversial and contradictory results [1,10]. That is, functional limitations and symptoms do escalate with age; hearing impairments, mobility restrictions, depression, fatigue and joint pain increase with age and are common among people 77 years and older [11]. 1. Introduction Even so, research indicates that there is a positive development in aspects of health such as independence in daily activities [3], which might be an effect of good housing standard and better technical equipment in the home. Moreover, housing adaptations and assistive devices (ADs) seem to compensate for deteriorating functional capacity as we age [3,10], but as yet few studies exist where home and health dynamics in the third age have been the focus of interest. In Sweden, the majority of people in the third age live in the same types of ordinary housing as the younger population [10]. In the 65- to 74-year-old age group, 65% live in one-family, whereas 36% live in multi-family dwellings, and it is only among those 85 years old and older that a shifting trend towards living in assistive living facilities is seen. Overall, as a result of policy changes in Sweden and other western countries that involve a shift from institutional care to community services, a growing number of older people remain in their ordinary homes well into old age [10,12]. The results of studies involving very old people show that the home environment plays a central role in supporting autonomy, social inclusion and well-being in the aging population [13,14], but little is known regarding the situation during earlier phases of the process of aging. While most studies on aging cover a multitude of information on the aging person, empirical aging research that takes a balanced person and environment view remains rare [15]. Theoretically, the relation of housing and health is closely linked to Lawton and Nahemow’s [16,17] ecological theory of aging. According to this theory, the interacting combination of an individual’s competence and the physical demands of the environment (person-environment fit—P-E-fit) is important for an individual’s level of functioning. Moreover, the docility hypothesis suggests that the lower the individual’s competence, the greater the impact of the environment on the individual’s ability to compensate for negative consequences. Though, it is important to note that when studying the relation between the aging person and the environment, the environment needs to be understood as a dynamic and context-bound phenomenon which encompasses a collection of objective as well as perceived meaning-related aspects such as emotions of a person in relation to his/her home [18]. 2.1. The SNAC/GÅS Project The Home and Health in the Third Age Study is a part of the Gott Åldrande i Skåne (GÅS) [Good Aging in Skåne] project which is one arm of the largest on-going longitudinal population-based sequential cohort study on aging in Sweden (Swedish National Study on Aging and Care) (SNAC) [21,22], started in 2001. In Skåne County, the SNAC/GÅS database currently comprises 2931 people 60 to 93 years old. A randomised population register selection for the age groups 60, 66, 72, 78, 81, 84, 87, 90, and 93 years has been made. For the present study we were restricted to using the SNAC/GÅS subsample including the age cohort 66 years at inclusion. The participants were recruited from five municipalities that differ in sizes and cover rural as well as urban and semi-urban areas; Malmö (urban), Eslöv, Hässleholm, Osby and Ystad (rural, semi-urban and urban). The cohorts are followed up in recurring evaluations every third (the older cohorts) or sixth (the younger cohorts) year [21] with the purpose to increase the knowledge on normal aging, identify predictors for chronic diseases and functional decline, as well as describe the need and use of health care. 1. Introduction As research on very old people has shown that it is not adequate to only measure objective aspects of housing such as physical environmental barriers, accessibility (an aspect of P-E fit) and housing standard, it is also necessary to account for perceived aspects of housing [9]. Already a decade ago, Gitlin [12] urgently called for a broader diversity of research regarding home environments that includes older people from different age cohorts with different levels of competencies and life experiences. For example, individuals born in the 1940s will have expectations and demands different from those of earlier generations. They work into a higher age, move more often, and have an overall active lifestyle [19]. In order to increase the knowledge on health trajectories related to housing, studies need to involve various cohorts of older people, with the potential to expose contrasts and shed new light on home and health dynamics along the process of aging. Thus, in order to complement the existing knowledge on home and health dynamics among 7063 Int. J. Environ. Res. Public Health 2014, 11 people in the fourth age (see e.g., [20]), the aim of this paper is to present the methodological background of the Home and Health in the Third Age Study, and describe a sample of men and women aged 67 to 70 in relation to their home and health situation. 2.2. The Home and Health in the Third Age Study Approximately two years after the ordinary SNAC/GÅS data collection, a cohort of 673 participants aged 67–70 years was selected from the SNAC/GÅS database and invited to take part in the Home and Health in the Third Age Study. The core methodology derives from instrumentation used to capture aspects of housing and health within the cross-national European project “Enabling Autonomy, Participation, and Well-Being in Old Age: The Home Environment as a Determinant for Healthy Aging” (ENABLE-AGE) [23]. 2.3. Recruitment Process A letter to all potential participants, comprising information about the study and asking them to participate was sent out by mail, asking them to return a letter of consent or decline. The individuals who consented to participate were contacted by a project administrator via telephone to book an appointment for data collection during a home visit. Due to the large number of participants, it was not possible to contact all within a set timeframe; it took anywhere between one week and two months until a participant was called. Difficulty in ADL Performance In addition to the ADL Staircase, a question on self-perceived difficulty in ADL was used to capture the heterogeneity within the group of participants that were rated as independent. Directly after a participant had been rated as independent in an ADL Staircase item, the project administrator asked whether he/she performed the specific task with or without difficulty [25]. 2.4.1. Descriptive Variables Socio-demographic descriptive variables included were age, sex, marital status, level of education, and type of housing. Age was calculated precisely by computing the difference between the date of Int. J. Environ. Res. Public Health 2014, 11 7064 interview and the date of birth. Marital status was dichotomized into married/cohabitant and unmarried/divorced/ widowed, with the intent to capture whether the participants lived alone or cohabitated. Level of education was divided into three categories: elementary school/less, secondary school, or one year more than secondary school/university degree. Project administrators trained for the data collection noted if the participants lived in a one or multi-family dwelling, and the participants were asked if they owned or rented their home. Type of housing was then divided into three categories: one-family house, rented or owned apartment in multi-family building. 2.4.2. Objective Aspects of Health 2.4.2. Objective Aspects of Health 2.4.3. Perceived Aspects of Health 2.4.3. Perceived Aspects of Health Functional Independence Perceived functional independence (PFI) was addressed by the question “All in all, how would you evaluate your own independence, i.e., in performing activities of daily living?”, scored from 0 (completely dependent) to 10 (completely independent); only the endpoints are defined. Activities in Daily Life (ADL) The ADL Staircase was used to assess dependence in activities of daily life (ADL). The instrument includes five items of personal activities of daily living (P-ADL; feeding, transfer, toileting, dressing, bathing) and four instrumental ADL items (I-ADL; cooking, transportation, shopping, cleaning). The instrument is administrated using a combination of interview and observation. The assessment is recorded on a three-point scale (independent, partly dependent and dependent), with dependence defined in terms of assistance from another person. The ADL Staircase is used to summarize an individual’s overall ADL ability where the degree of dependence is ranked from 0 (independent in all activities) to 9 (dependent in all activities). The instrument is reliable and valid for the assessment of older people’s functional ability [24]. Perceived Health and Mobility The question ‘‘In general would you say your health is…?” from the SF-36 questionnaire [30] was used to capture a global self-rating of perceived health. The scale has five response alternatives ranging from 1 (excellent) to 5 (poor). Using the same response alternatives, the participants were also asked, “How would you rate your physical mobility at the moment?” Symptoms A checklist consisting of 30 items was used to dichotomously (yes/no) assess the number of symptoms in seven different domains (depression, tension, gastrointestinal, musculoskeletal, metabolism, heart lung, head symptoms). The participants were asked to answer yes when they had experienced a symptom during the past three months [29]. In addition to the 30 items, three items (frequency in passing urine, incontinence, dental problems) introduced by the ENABLE-AGE consortium were included. Life Satisfaction Life satisfaction was assessed through a single study-specific question, “On the whole, what do you think about your life right now?” A five-point rating scale ranging from 1 (very good) to 5 (very bad) was presented to the participants. Falls A short version of the Falls Efficacy Scale-International (FES-I) [27] was used to assess fear of falling (FOF). The short FES-I contains seven items that exemplify different social and physical activities performed inside and outside the home (getting dressed/undressed, taking a bath/shower, getting in/out of a chair, going up/down stairs, reaching for something above your head/on the ground, walking up/down a slope, going out to a social event). The participants were asked to state their level of concern about falling when performing the given activity. In case the activity was not currently performed, the participants were asked to think about how afraid they would be if doing it. The instrument has four response alternatives ranging from 1 (not at all concerned) to 4 (very concerned of falling). The scores are added to a total score which can range from 7 (no concern about falling) to 28 (severe concern about falling) [28]. The participants were asked four additional questions regarding falls; “During the past year, how often have you fallen?” (never, once, and more than once) and if yes;” Where did you fall?”; “How did it happen?”, and, “Did you hurt yourself so that you needed medical care?” Int. J. Environ. Res. Public Health 2014, 11 Int. J. Environ. Res. Public Health 2014, 11 7065 Depressive Symptoms To capture depressive symptoms, the 15 item version of the Geriatric Depression Scale (GDS) [26] was used. The project administrator presented each item and asked the participants to answer yes or no based on how they felt over the past week. Five items indicate a depressive symptom when rated negatively while the remaining 10 items indicate a depressive symptom when rated positively. Each “depressive” answer equals 1 point, with possible scores ranging from 0 to 15. Chronbach’s alpha on our dataset indicates acceptable internal consistency; α = 0.77. Psychological Wellbeing The Ryff scales of Psychological Wellbeing (PWQ) [31] incorporate several different theoretical perspectives and measure positive psychological functioning. A short form with 19 items divided in two domains, autonomy (10 items) and purpose in life (nine items), was used. Statements were presented to the participants with the instruction to rate each statement on a scale ranging from 1 Int. J. Environ. Res. Public Health 2014, 11 7066 (strong disagreement) to 5 (strong agreement). Examples of statements in the two domains are, “I am not afraid to voice my opinions even when they are in opposition of most people” (autonomy), and “Many daily activities often seem trivial and unimportant to me” (purpose in life). Some items are negatively phrased and need to be reversed when computing a sum score so that higher scores on all items indicate higher well-being. Scores are computed for each domain; a high score indicates a higher feeling of mastery. A sum score of both domains gives an indication of overall psychological well-being. Chronbach’s alpha in our dataset indicates rather low but acceptable internal consistency for group comparisons [32]: purpose in life α = 0.65 and autonomy α = 0.71. Assistive Devices (ADs) Questions regarding use and need of assistive devices (ADs) were adopted from the ENABLE-AGE Project, subsequently categorized according to ISO classifications [33]. This section covers ADs for communication, such as optical (three items) and hearing (three items), mobility devices indoors (six items) and outdoors (seven items), personal care (six items) and other ADs (seven items) such as stair lift and adjustable bed. In total, the participants answered 32 predefined questions regarding ADs with four response alternatives (available, in use, not available but would be necessary, not available, would not be necessary). If the participants expressed a need for or used an AD not listed, the project administrator used an open-ended question to register the responses. Int. J. Environ. Res. Public Health 2014, 11 7067 Int. J. Environ. Res. Public Health 2014, 11 barriers), at individual or group level, can be calculated. Furthermore, participants were asked about how many years they had lived in the same home, how many rooms there were in their home, and how many people lived there. barriers), at individual or group level, can be calculated. Furthermore, participants were asked about how many years they had lived in the same home, how many rooms there were in their home, and how many people lived there. Domain 2: Usability The Usability in My Home (UIMH) questionnaire was used to capture to what degree the physical environment supports the performance of daily activities in the home [37,38]. We used a short version containing 10 items divided in two subscales targeting activity aspects (4 items), for example, “In terms of how you normally manage your cooking/heating of food or preparation of snacks, to what extent is the home environment suitable designed in relation to this?”, and physical environmental aspects (6 items) of usability, for example, “How usable do you feel that your home environment is in general?”. The items are rated on a five-point scale ranging from 1 (not at all suitable/usable) to 5 (fully suitable/usable); higher scores mean higher usability. Chronbach’s alpha in our dataset indicates acceptable internal consistency [32] in both domains, that is, activity aspects, α = 0.72, and physical environmental aspects, α = 0.79. Domain 1: Housing Satisfaction Housing satisfaction was assessed via the single question “Are you happy with the conditions of your home?”, adapted from the Housing Option for Older People (HOOP) Questionnaire (Sixsmith and Sixsmith, unpublished ENABLE AGE working paper). A five-point rating scale ranging from 1 (no, definitely not satisfied) to 5 (yes, definitively satisfied) was presented to the participants. 2.4.5. Perceived Aspects of Home Perceived aspects of home were captured using a four-domain model, operationalized and empirically tested as described by Oswald et al. [36]. 2.4.4. Objective Aspects of Home Number of environmental barriers and magnitude of accessibility problems were captured with the Housing Enabler (HE) instrument which is administrated in three steps. The instrument is based on extensive research [34] and has proven to be valid and reliable [34,35]. Step 1 (the personal component checklist) of the HE concerns functional limitations (12 items; difficulty in interpreting information, visual impairment, blindness, loss of hearing, poor balance, incoordination, limitations of stamina, difficulties in moving head, reduced upper extremity function, reduced fine motor skill, loss of upper extremity skills, reduced spine and/or lower extremity function) and dependence on mobility devices (two items; dependence on walking devices, wheelchair). All items are dichotomously assessed as present/not present. The assessment results in a sum score of number of functional limitations (range 0–12) and dependence on mobility devices (range 0–2). This functional profile (12 + 2 items) can also be used as an objective aspect of health variable. Step 2 (the environmental component checklist) is based on observation of the actual environment in a detailed rating of environmental barriers (161 items). Each environmental barrier inside the home (n = 87), at entrances (n = 46) and in the immediate exterior environment (n = 28) is dichotomously assessed as present/not present. The ratings of the environmental component are based on national standards for housing design. Step 3 (the P-E fit analysis) involves calculating a total score that quantifies the magnitude of accessibility problems (MAP) in a particular case, and predicts the load caused by a particular combination of functional limitations and environmental barriers. The higher the score, the greater the accessibility problems are. The total score is always 0 if the individual has no functional limitations/dependence on mobility devices, regardless of the presence of environmental barriers. In addition, a rank order of the environmental barriers that cause the most accessibility problems (known as weighted environmental Housing Adaptations Information on housing adaptations was gathered through five study-specific questions. The participants were asked whether they had knowledge about the housing adaptation grant provided by local municipalities (yes or no). Thereafter, with the same response alternatives, they were asked if there had been any adaptations made in the home. In cases where the participants answered yes, they were asked to provide information on the location of the adaptation as well as on how it had been financed. Moreover, the participants were asked if the adaptation had any positive or negative influences on ADL, using six response alternatives (it has become easier to perform my daily activities, I have been independent from help of others, I was able to remain living in the present dwelling, the changes had small/no effect, the situation has worsened, and other). Domain 4: Housing-Related Control Beliefs Control beliefs in relation to home were addressed using the 24-item Housing-related Control Beliefs Questionnaire (HCB) [39]. The HCB captures three domains: internal control (8 items, sum-score), external control-powerful others (8 items, sum-score), and external control-chance (8 items, sum-score). Internal control denotes that housing-related outcomes are dependent on own behavior “Everything in my home will stay the way it is no one is going to tell me what to do”. External control means that an external power such as another person is responsible or that things happen by luck, chance, or fate. External control-powerful others is captured through statements like “In order to do anything interesting outside of my home I have to rely on others” whereas external control-chance is determined through statements like “Having a nice place is all luck. You cannot influence it; you just have to accept it”. The participants were asked to rate each statement on a five-point rating scale ranging from 1 (I do not at all agree) to 5 (I agree very much); higher scores indicate higher perceived control in the domain of internal control whereas higher scores in the domains of external control indicate lower perceived control. In accordance with previous studies [8,9,36], the domain of internal control will be excluded from further analysis due to low internal consistency (α = 0.38). Also similar to the previous studies, the two domains of external control reached rather low levels: powerful others, α = 0.54 and chance α = 0.56. Applying the same strategy as in studies based on data collected within the ENABLE-AGE Project [7,9] after combining the two dimensions of external control the 16-item scale reached an acceptable level [32] of internal consistency α = 0.69. Int. J. Environ. Res. Public Health 2014, 11 Int. J. Environ. Res. Public Health 2014, 11 7068 Domain 3: Meaning of Home The 28-item Meaning of Home (MOH) questionnaire was used to gain knowledge of the individual’s subjective meanings in relation to home. The MOH was developed to be used with older people and captures four aspects of the meaning of home: behavioral (6 items), for example, “doing everyday tasks”, physical (seven items) “feeling that home has become a burden”, cognitive/emotional (10 items) “feeling safe” and social (five items) “being excluded from social and community life”. Each item is rated on a scale with 11 response alternatives ranging from 0 (strongly disagree) to 10 (strongly agree). Higher scores indicate a stronger bonding/attachment to home [36]. In accordance with previous studies [8,9,36], Chronbach’s alpha in our dataset indicates rather low but acceptable internal consistency for group comparisons [32]: physical aspects, α = 0.53; behavioral aspects, α = 0.59; cognitive/emotional aspects, α = 0.62; and social aspects, α = 0.62. 2.5. Questions Regarding Reliability In order to ensure the quality of the collected data the project administrator answered eight questions in order to evaluate the perceived reliability of the participant’s responses. This procedure was performed shortly after the home visit, without any contribution of the participant. It was noted if another person had been present during the interview, and if so, in what way the presence of this person might have influenced the responses given by the participants. The project administrator also registered the perceived communication ability of the participant (scored 0–10; higher = better) and her assessment of the reliability of the participant’s responses (very high reliability, high reliability, reliable, low reliability, very low reliability). Moreover, the status of the dwelling (neglected, normal, or well kept) and of the participant (neglected, average, well presented) as perceived by the project administrator were registered. 2.6. Data Collection After project-specific training provided by experienced scientists with profound knowledge on the ENABLE-AGE data collection format, two project administrators (experienced registered occupational therapists) collected the data. To be able to administer the HE instrument, the project administrators completed a four-day training course and additional recommended practical training [41]. Data were collected at home visits over a 9-month period (5 October, 2010 to 21 June, 2011). Each home visit lasted 2.0–2.5 hours. In cases where it was not possible to complete the data collection during one home visit, a second appointment was made for completion of the data collection within 10 days. Neighborhood Attachment Neighborhood attachment was captured through the single item “Are you rooted and feel a strong affinity to your residential area?”. The question has four response alternatives ranging from 1 (to a great extent) to 4 (not at all) [40]. All variables including the domains covered are presented in Table 1. Int. J. Environ. Res. Public Health 2014, 11 7069 Table 1. Instruments and domains covered. Table 1. Instruments and domains covered. Table 1. Instruments and domains covered. Instrument Domain Items, n Literature Reference Objective aspects of health Activities of daily life (ADL) Staircase Personal ADL 5 [24] Instrumental ADL 4 Perceived aspects of health Geriatric Depression Scale (GDS) Mood Disturbance 10 [26] Motivation Disturbance 5 Difficulty in ADL Activity performed with/without difficulty1 2 [25] Functional independence (PFI) Perceived functional independence 1 - Short FES-I Fear of falling 7 [27,28] Falls 4 - Symptom list Depression symptoms 5 [29] Tension symptoms 5 Gastrointestinal symptoms 8 Musculoskeletal symptoms 3 Metabolism symptoms 4 Heart-lung symptoms 3 Head symptoms 5 SF-36, global health Perceived global health 1 [30] Physical mobility Perceived physical mobility 1 - Life Satisfaction Life Satisfaction 1 - Psychological wellbeing (PWQ) Autonomy 10 [31] Purpose in life 9 Objective environmental aspects Housing Enabler (HE) Functional limitations/dependence on mobility devices 2 14 [34,35] Environmental barriers; Exterior surroundings 28 Environmental barriers; Entrance 46 Environmental barriers; Indoors 87 Other aspects of objective housing No. of rooms, no. of people, years of habitance 3 - Assistive Devices/Technical Aids Optical aids 3 [33] Hearing aids 3 Mobility devices, indoors 6 Mobility devices, outdoors 7 ADL devices 6 Other assistive devices 7 Perceived environmental aspects Housing Option for Older People (HOOP) Housing satisfaction3 1 Usability In My Home (UIMH) Activity 4 [37,38] Int. J. Environ. Res. Public Health 2014, 11 7070 Table 1. Cont. Instrument Domain Items, n Literature Reference Physical environmental aspects 6 Meaning of Home (MOH) Activity 6 [36] Physical 7 Cognitive/emotional 10 Social 5 Housing Related Control Beliefs (HCB) External control combined 16 [39] Housing adaptations 5 - 1 Number of items used with each participant depends on the results of the objective assessment of ADL according to the ADL Staircase [24]. 2 Used separately, the personal component of the Housing Enabler can also be used as a health variable. 3 Sixsmith, A.J and Sixsmith, J.A, unpublished ENABLE AGE working paper. Table 1. Cont. 2.8. Data Quality Control In order to monitor progress and data quality, meetings were arranged regularly with an experienced researcher (third author; M.H.) during the entire data collection period. A proof reading procedure was carried out to ensure the database accurately reflects the data collected. The proof reading included a sample of 40 randomly selected participants (>10%), and the acceptable error rate was set to not exceed 0.5%. Discrepancies found were noted on a log sheet and rectified in the database. The error rate was calculated to 0.18% (38 errors found among 40 individuals answering up to 520 variables), indicating that a 100% proof reading was not necessary. In addition, a validation of the data was performed by checking ranges, logical consistency and completeness. Missing or unclear data underwent a data cleaning process using data clarification forms. Changes applied to data in the database during the data cleaning process were noted on a log sheet. After completion of the data cleaning process the database was locked. Int. J. Environ. Res. Public Health 2014, 11 7071 Int. J. Environ. Res. Public Health 2014, 11 2.9. Data Analysis For the empirical part of this paper, data collected with the following instruments were used: the ADL Staircase [24], SF-36 perceived global health [30], Symptoms list [29], Geriatric Depression Scale [26] and the HE instrument, Steps 1 and 2 [41]. Depending on the instrument scale properties, descriptive statistics were used and the findings reported with means and standard deviation (for continuous normally distributed data), medians and quartiles (for categorical data and data deviating from normal distribution), and frequencies and percentages (to describe group proportions). The Pearson Chi-Square test was used to test differences between sub-groups. The Mann-Whitney test was used to test for differences between medians, while the Student T-test was used to test differences between means. All tests were two-sided and p-values < 0.05 were considered statistically significant. All analyses were computed by means of the SPSS software version 20 (IBM Corporation, Armonk, NY, USA). 2.7. Ethics The Home and Health in the Third Age Study was conducted in accordance with the Helsinki Declaration and was approved by the Ethical Board in Lund (2010/431). Informed consent was obtained from all participants and anonymity was ensured. This was reinforced verbally as well as by means of written information at the start of the home visit. Participants were informed that they could withdraw from the study if and whenever they wished. 3.1. Participants and Attrition Analysis In the target sample of 673 men and women there were nine deaths. Consequently, 664 individuals (314 men, 350 women) were invited to participate. In all, 371 (55.9%) agreed to participate. At the start of the data collection, the mean age for participants was 68.4 years (SD = 0.9). Among the 293 individuals that declined to participate, 283 said they were unwilling without giving any reason, and 10 stated that their health was too poor to allow participation. A larger proportion of men than of women declined, 155 (52.9%) vs. 138 (47.1%) (p = 0.010). There was no age difference between participants and non-participants at time of recruitment: mean age 68.1 years (SD = 0.9) and 68.2 years (SD = 0.9), respectively (p = 0.324). Geographical area (urban or rural) did not differ between non-participants and participants. Among the non-participants, 224 (76.5%) lived in an urban area (i.e., Malmö); among the participants the corresponding number was 287 (77.4%) (p = 0.783). 7072 Int. J. Environ. Res. Public Health 2014, 11 For one participant, the HE assessment was not completed, but the remaining 370 participants completed all of the data collection. For one participant, the HE assessment was not completed, but the remaining 370 participants completed all of the data collection. 3.3. Health and Home Aspects As presented in Table 3, the vast majority of the participants rated their health as good or very good and there were no significant differences between sub-groups. Women reported more symptoms (p = 0.001) and had more functional limitations than men (p = 0.002). Approximately half of the participants (50.4%) had one or more functional limitation. For the men the most common functional limitation was “difficulty in bending or kneeling” whereas “difficulty in reaching with arms” was the most common among the women. Seventeen participants were reliant on a walking device and one participant used a wheelchair. Out of the total sample, 10% were dependent in one or more I-ADL whereas no participant was dependent in any P-ADL. Table 3. Health variables and assistive devices for men and women in the study sample, N = 371. Variable Literature Reference Men n = 159 Women n = 212 Total N = 371 p-value Activities in daily life [24] Independence in I-ADL, n (%) 137 (86.2) 197 (92.9) 334 (90) 0.031 Independence in P-ADL, n (%) 159 (100) 212 (100) 371 (100) - Perceived health Mn (Sd) [30] 3.6 (0.95) 3.5 (1.05) 3.6 (1.01) 0.183 Symptoms (no.), Md (q1-q3) 5.0 (2.0-8.0) 6.5 (3.2-12.0) 6.0 (3.0-11.0) 0.001 Depressive symptoms Md (q1-q3) [26] 1.0 (0.0- 1.0) 1.0 (0.0-2.0) 1.0 (0.0- 2.0) 0.066 Functional profile [41] Functional limitations (n), Mn (Sd) 0.77 (1.07) 1.17 (1.47) 1.0 (1.33) 0.002 Dependence on mobility devices Reliance on walking aids, n (%) 4 (2.5) 13 (6.1) 17 (4.6) 0.099 Wheelchair user, n (%) 1 (0.6) - 1 (0.3) 0.248 Table 3. Health variables and assistive devices for men and women in the study sam Health variables and assistive devices for men and women in the study sample, N = 371. In every home environment investigated environmental barriers were identified. As presented in Table 4, environmental barriers were more common in multi-family than in one-family type of housing (p =< 0.001). There were no differences between the dwellings of men and women. The most prevalent environmental barriers were identified indoors in the kitchen and/or hygiene area (Table 5). In both multi-family and one-family housing “use requires hands” and “controls in high/low/inaccessible position” were identified as the most common environmental barriers. Int. J. Environ. Res. Public Health 2014, 11 7073 Int. J. Environ. Res. Public Health 2014, 11 3.2. Sample Characteristics Description of the study sample according to sex, marital status, geographical area, type of housing, age and level of education is provided in Table 2. There were slightly more women than men participating in the study (p = 0.006). Out of the complete sample 64.2% were cohabiting (p ≤ 0.001) and 59.3% were living in multi-family buildings (p = 0.003). The majority were living in an urban environment (p ≤ 0.001). The distribution of participants living in rural and urban districts of our sample reflects the actual demographic distribution in the county of Skåne. Table 2. Sample characteristics, N = 371. Characteristic N % p-value Sex Women 212 57.1 Men 159 42.9 0.006 Marital Status Married/cohabitating 238 64.2 Unmarried 28 7.5 Widow/widower 25 6.7 Divorced/separated 67 18.1 In a relationship 12 3.2 Missing 1 0.3 <0.001 Geographical Area Rural 41 11.1 Urban 330 88.9 <0.001 Type of Housing Apartment, owned in multi-family building 123 33.2 Apartment, rented in multi-family building 97 26.1 0.003 One-family 151 40.7 Age in Years 67 158 42.6 68 99 26.7 69 87 23.5 70 27 7.3 <0.001 Education Elementary school or less 139 37.5 Secondary school 124 33.4 One year more than secondary school or university degree 104 28.0 Missing 4 1.1 0.080 Table 2. Sample characteristics, N = 371. Table 2. Sample characteristics, N = 371. 3.3. Health and Home Aspects In the close exterior surroundings the most prevalent environmental barriers in the total sample were “irregular walking surface” (85.2%), “landscape furniture placed in the path of travel” (69.3%) and “narrow parking spaces” (67.1%). Table 4. Number of environmental barriers in relation to type of housing and sex, N = 371. Environmental Barriers/Housing Section Multi-Family Dwellings n = 220 One-Family Dwellings n= 151 p-value Men n = 158 Women n = 212 p-value Exterior surroundings, Mn (Sd) 11 (2.9) 7.5 (2.8) <0.001 9.3 (3.4) 9.8 (3.1) 0.186 Entrances, Mn (Sd) 17.8 (6.7) 8.9 (3.2) <0.001 13.6 (6.9) 14.6 (7.0) 0.235 Indoors, Mn (Sd) 43.0 (5.1) 48.7 (6.1) <0.001 45.2 (6.4) 45.4 (6.1) 0.745 Total , Mn (Sd) 71.7 (9.9) 65.2 (8.4) <0.001 68.2 (10.4) 69.7 (9.4) 0.135 Min-Max 44–95 45–86 0–91 45–95 mber of environmental barriers in relation to type of housing and sex, N = 371. Table 4. Number of environmental barriers in relation to type of housing and sex, N = Int. J. Environ. Res. Public Health 2014, 11 7074 Table 5. The 20 most frequent environmental barriers at entrances and indoors, in different types of housing, N = 371. 3.3. Health and Home Aspects Housing Section and Environmental Barrier 1 Multi-Family Dwellings n = 220 One-Family Dwellings n = 151 Total N = 371 p-value Entrances n (%) n (%) n (%) High thresholds and/or steps (more than 15 mm), (sitting-out place/balcony) 208 (94.5) 138 (91.4) 346 0.357 Kitchen, laundry room, utility kitchen No working surfaces with leg room 209 (95.0) 141 (96.4) 350 (94.3) 0.676 Turning motion of wrist required 217 (98.6) 149 (98.7) 366 (98.7) 0.524 Use requires hands 220 (100) 150 (99.3) 370 (99.7) - Use requires fingers (i.e., isolated grip, e.g., pinch and lateral grip) 219 (99.5) 145 (96.0) 364 (98.1) 0.031 Controls in high/inaccessible position (more than 1.1 m above the floor) 220 (100) 150 (99.3) 370 (99.7) - Controls in low position (less than 80 cm above the floor) 220 (100) 150 (99.3) 370 (99.7) - Hygiene area Turning motion of wrist required 209 (95.0) 141 (93.4) 350 (94.3) 0.875 Use requires hands 219 (99.5) 149 (98.7) 368 (99.2) 0.410 Controls in high/inaccessible position (more than 1.1 m above the floor) 218 (99.1) 148 (98.0) 366 (98.7) 0.803 Wash-basin placed at a height for use only when standing 204 (92.7) 139 (92.1) 343 (92.5) 0.836 Toilet 47 cm or lower 199 (90.5) 143 (94.7) 342 (92.2) 0.046 Mirror placed at a height for use only when standing 360 (97.0) 0.701 Storage cupboards, towel hooks, etc. placed high/low 207 (94.1) 132 (87.4) 339 (91.4) 0.083 Inappropriate design of wardrobes/ clothes cupboards 194 (88.2) 142 (94.0) 336 (90.6) 0.058 Turning motion of wrist required 217 (98.6) 147 (97.4) 364 (98.1) 0.986 Use requires hands 220 (100) 149 (98.7) 369 (99.5) 0.225 Use requires fingers 216 (98.2) 149 (98.7) 365 (98.4) 0.346 Controls in high/inaccessible position (more than 1.1 m above the floor) 217 (98.6) 147 (97.4) 364 (98.1) 0.634 Controls in low position (less than 80 cm above the floor) 220 (100) 150 (99.3) 370 (99.7) - 1 No barriers in the immediate outdoor environment were found among the top 20 most frequent environmental barriers [41]. 4. Discussion According to the aim of this paper we present basic home and health characteristics of the Home and Health in the Third Age Study sample. The results show that environmental barriers are more Int. J. Environ. Res. Public Health 2014, 11 7075 common in multi-family than in one-family dwellings. Considering that a large number of people in the third age lives in multi-family housing [10] and will continue to do so, attention should be paid to removing environmental barriers in the already existing housing stock as well as when planning for new buildings. Especially since the current trend in Sweden and other western countries favors community-based health care and social services provided in ordinary housing before assisted living and institutional care [12]. Overall, it was beneficial to build the present study on the core methodology developed and tested within the internationally acknowledged ENABLE-AGE Project [23]. The main aim of the ENABLE-AGE was to explore the home environment as a determinant for autonomy, participation, and well-being in very old age within a follow-up perspective. For the present project we made use of the instruments, data collection and data quality assurance procedures that, based on our previous research (see e.g., 23), have been proven as the most efficient and valuable for studies with a specific focus on home and health in old age. To date, 50 original papers have been published based on the quantitative and qualitative data collected within the aforementioned project, and the present study was thus built on a strong knowledge base and methodological experiences gained during one decade of research. As in line with previous findings [42], physical environmental barriers were identified in 100% of the homes assessed. This might be surprising, but since the HE environmental assessment [41] is very detailed and based on the present Swedish standards for housing design, virtually all housing units have some environmental barriers. Accordingly, when investigating the frequency of environmental barriers we found that among the 20 most common, four were identified as present in all of the 370 dwellings of the sample. Therefore, it is important to note that to understand the impact of the environmental barriers on aspects of health, in further studies based on the present sample analyses linking environmental barriers to the person’s functional limitations will be accomplished. 4. Discussion Studying a sample of people in the third age, the low prevalence of functional limitations was expected and in line with previous research [3]. However, earlier research has shown that the number of functional limitations and use of mobility devices will increase with advancing age [11]. As the number and profile of functional limitations and use of mobility devices are one component of accessibility [41], the magnitude of accessibility problems will also increase as people age [42]. As we have verified that the number of functional limitations is low among people in the third age, this study is as a starting point for future studies with the aim to identify turning points in home and health dynamics during the process of aging. It should be noted that the Home and Health in the Third Age Study addresses key questions on the home and health interactions among individuals in an earlier phase of aging than those of previous studies with a similar focus and design (see e.g., [9,43]). An important goal in health promotion is to create home environments that support healthy aging [23]. Consequently, from a societal planning perspective it is of great importance to consider aspects of home and health from an earlier stage of aging and into very old age. To the best of our knowledge, no major studies exist that include detailed high-quality survey data not only on aspects of health but also on aspects of home among people in the third age. Under the canopy of a comprehensive research program labeled “Home and Health along the Process of Aging”, we are in a strong position to be able to deliver comparative studies, comparing this sample with other datasets with similar data on very old individuals [23]. Having access to a rich database will allow us to accomplish in-depth studies on home and health dynamics. Furthermore, Int. J. Environ. Res. Public Health 2014, 11 Int. J. Environ. Res. Public Health 2014, 11 7076 we also have the possibility to compare this dataset with similar data on individuals aging with a chronic neurological disease [44] or a spinal cord injury [45] to shed new light on home and health dynamics along the process of aging. We will also be able to conduct longitudinal studies within the context of the SNAC/GÅS [21] utilizing data from previous and forthcoming data collection waves. 4. Discussion The fact that the project administrators assigned with the data collection task were responsible for contacting their respective participants presumably had a positive effect on the participation ratio. Still, the participation ratio was most likely somewhat negatively affected by the sometimes long waiting time between the invitation and the follow-up phone call. Furthermore, albeit resource intensive, face-to-face interviews administered at home visits have several advantages. In particular for participants with poorer health, continuity in the process of recruitment, fieldwork and individualized data collection at home visits are advantageous. Prior to the data collection, the project administrators obtained the necessary skills through training to use the instruments included in the survey provided by senior researchers with longstanding experiences of this kind of data collection. They were instructed to provide explicit information and give examples whenever a participant did not directly understand a question. That is, the project administrators made great efforts to tailor the data collection situation in an optimal way for each participant while still keeping up a high level of structure. Based on our extensive experiences from data collection with older people in similar projects, this procedure was used to make sure that the data collected is of the highest possible quality. Without any kind of influence on their responses as such, all the participants were given the same guidance to optimize their understanding of the question or item at hand. Moreover, the home visit format was necessary for the administration of the objective assessment of the home environment by means of the HE instrument [41]. A noteworthy challenge for reliable administration of data on physical environmental features is the dynamics of outdoor environments. For example, since the data collection was carried out over a 9-month period in a Nordic country, the seasonal variation posed certain challenges. During winter the climate in southern Sweden is rather cold, sometimes with ice and snow. Naturally, this influenced the administration of the outdoor environmental assessment since the project administrator was not always in a position to rate specific items as present or absent due to weather conditions, resulting in some unintentional but unavoidable missing data. Furthermore, since the proportion of participants assessed as being influenced by another person during the interview was as low as 1.3% and only 1.1% were assessed as communicating with very low reliability, we conclude that the influence of these responses on the results are negligible. 4. Discussion Regarding the psychometric aspects of the instruments used, they all have documented validity and reliability based on studies including individuals in the fourth age (see e.g., 15, 23). Based on earlier observations [8,9,36], for the present study we concentrated our efforts regarding psychometric properties to the instruments that target perceived aspects of housing. In accordance with our studies on samples of people in the fourth age and a recent study by Oswald et al. [46], these instruments demonstrated low internal scale consistency also when used with people in the third age. As to the other instruments used in the present study, based on the data collected with people in the third age now at hand we are in a strong position to be able to contribute to further methodological refinement and development, aiming for the optimization of quantitative assessments of aspects of home and health in different phases of the aging process. Based on our earlier studies, valid data treatment and Int. J. Environ. Res. Public Health 2014, 11 7077 analysis strategies have been established [9,36], and we are thus well equipped to deliver forthcoming studies based on the data collected. analysis strategies have been established [9,36], and we are thus well equipped to deliver forthcoming studies based on the data collected. 5. Conclusions The present study will generate a better understanding of home and health dynamics among people in the third age. Future results have the potential to contribute to and facilitate societal planning, particularly in terms of housing provision but also regarding home care and social services for senior citizens. Author Contributions The authors are justifiably credited with authorship, according to the authorship criteria. In more detail, Maya Kylén was one of the project administrators who prepared and administered the data collection, and she also drafted the manuscript for submission. Henrik Ekström managed the data collection in relation to the SNAC/GÅS database, had an active role in the quality control process, provided statistical and methodological support, and contributed to the development of the manuscript. Maria Haak contributed to the concept and design of the study, organized and supervised the data collection, contributed to the data quality control process, and critically revised the manuscript for submission. In his role as the PI for the SNAC/GÅS Project, Sölve Elmståhl decided on the definition of the sample for the study, participated in the concept and design process, and read and commented on the next-to-final version of the manuscript. Susanne Iwarsson is the PI of the Home and Health in the Third Age Study and is responsible for its concept and design, and she critically and repeatedly revised the manuscript for submission. All of the authors approved the final manuscript version. Acknowledgments We would like to extend our thanks to all participants and to Lizette Norin and Zinka Tucek for help with the data collection and administrative support. We would also like to express our gratitude to Björn Slaug for data management support, and to Steven Schmidt for input on data quality, revising the English text, and critical review of the final manuscript. The Home and Health in the Third Age Study is financed by the Swedish Research Council and the Ribbingska Foundation in Lund, Sweden. It is accomplished within the context of the Centre for Ageing and Supportive Environments (CASE) at Lund University, Sweden, financed by the Swedish Research Council for Health, Working Life and Welfare (Forte). Conflicts of Interest The last author (Susanne Iwarsson) is the shared copyright holder and owner of the HE instrument and software, provided as commercial products (see www.enabler.nu). The remaining authors declare no competing interests. Int. J. Environ. Res. Public Health 2014, 11 7078 References 1. Cauley, J.A. The demography of ageing. In The Epidemiology of Aging; Newman, A.B., Cauley, J.A., Eds.; Springer: London, England, 2012; pp. 3–15. 1. Cauley, J.A. The demography of ageing. In The Epidemiology of Aging; Newman, A.B., Cauley, J.A., Eds.; Springer: London, England, 2012; pp. 3–15. 2. Statistics Sweden, Forecast Institute. The Future Population of Sweden 2012–2060; Statistics Sweden: Örebro, Sweden, 2012. 2. Statistics Sweden, Forecast Institute. The Future Population of Sweden 2012–2060; Statistics Sweden: Örebro, Sweden, 2012. 3. Lennartsson, C.; Heimerson, I. Elderly People’s Health: Health in Sweden: The National Public Health Report 2012. Chapter 5. Scand. J. Public Health 2012, 40, 95–120. 3. Lennartsson, C.; Heimerson, I. Elderly People’s Health: Health in Sweden: The National Public Health Report 2012. Chapter 5. Scand. J. Public Health 2012, 40, 95–120. 4. Baltes, P.B.; Smith, J. New Frontiers in the Future of Aging: From Successful Aging of the Young Old to the Dilemmas of the Fourth Age. Gerontology 2003, 49, 123–135. 4. Baltes, P.B.; Smith, J. New Frontiers in the Future of Aging: From Successful Aging of the Young Old to the Dilemmas of the Fourth Age. Gerontology 2003, 49, 123–135. 5. Wahl, H.W.; Schilling, O.; Oswald, F.; Iwarsson, S. The Home Environment and Quality of Life-related Outcomes in Advanced Old Age: Findings of the enable-age project. Eur. J. Ageing 2009, 6, 101–111. 5. Wahl, H.W.; Schilling, O.; Oswald, F.; Iwarsson, S. The Home Environment and Quality of Life-related Outcomes in Advanced Old Age: Findings of the enable-age project. Eur. J. Ageing 2009, 6, 101–111. 6. Haak, M. Participation and Independence in Old Age: Aspects of Home and Neighborhood Environments. PhD Thesis, Lund University, Lund, Sweden, 2006. 7. Oswald, F.; Wahl, H.W.; Schilling, O.; Iwarsson, S. Housing-related Control Beliefs and Independence in Activities of Daily Living in Very Old Age. Scand. J. Occup. Ther. 2007, 14, 33–43. 8. Iwarsson, S.; Horstmann, V.; Slaug, B. Housing Matters in Very Old Age—Yet Differently due to Adl Dependence Level Differences. Scand. J. Occup. Ther. 2007, 14, 3–15. 9. Nygren, C.; Oswald, F.; Iwarsson, S.; Fange, A.; Sixsmith, J.; Schilling, O.; Sixsmith, A.; Szeman, Z.; Tomsone, S.; Wahl, H.W. Relationships between objective and perceived housing in very old age. Gerontologist 2007, 47, 85–95. 10. Jennbert, K. Developments in Elderly Policy in Sweden; Swedish Association of Local Authorities and Regions: Stockholm, Sweden, 2009. 11. Parker, M.G.; Ahacic, K.; Thorslund, M. 17. Scheidt, R.; Norris-Baker, C. The general ecological model revisited: Evolution, current status, continuing challanges. In Aging in Context: Socio-physical Environments Annual Review of Gerontology and Geriatrics; Wahl, H.-W., Scheidt, R.J., Windley, P.G., Eds.; Springer: New York, NY, USA, 2004; pp. 34–58. References Health changes among Swedish oldest old: Prevalence rates from 1992 and 2002 show increasing health problems. J. Gerontol. A-Biol. 2005, 60, 1351–1355. 12. Gitlin, L.N. Conducting research on home environments: Lessons learned and new directions. 11. Parker, M.G.; Ahacic, K.; Thorslund, M. Health changes among Swedish oldest old: Prevalence rates from 1992 and 2002 show increasing health problems. J. Gerontol. A-Biol. 2005, 60, 1351–1355. rates from 1992 and 2002 show increasing health problems. J. Gerontol. A-Biol. 2005, 60, 1351–1355. 12. Gitlin, L.N. Conducting research on home environments: Lessons learned and new directions. Gerontologist 2003, 43, 628–637. 12. Gitlin, L.N. Conducting research on home environments: Lessons learned and new directions. Gerontologist 2003, 43, 628–637. 13. Haak, M.; Fänge, A.; Iwarsson, S.; Ivanoff, S.D. Home as a signification of independence and autonomy: Experiences among very old Swedish people. Scand. J. Occup. Ther. 2007, 14, 16–24. 14. Dahlin-Ivanoff, S.; Haak, M.; Fänge, A.; Iwarsson, S. The multiple meaning of home as experienced by very old Swedish people. Scand. J. Occup. Ther. 2007, 14, 25–32. 15. Iwarsson, S.; Wahl, H.-W.; Nygren, C. Challenges of cross-national housing research with older persons: Lessons from the ENABLE-AGE project. Eur. J. Ageing 2004, 1, 79–88. 16. Lawton, M.P.; Nahemow, L. Ecology and the ageing process. In The Psychology of Adult Development and Ageing; Eisdorfer, C., Lawton, M., Eds.; American Psychological Association: Washington DC, USA, 1973; pp. 619–674. 17. Scheidt, R.; Norris-Baker, C. The general ecological model revisited: Evolution, current status, continuing challanges. In Aging in Context: Socio-physical Environments Annual Review of Gerontology and Geriatrics; Wahl, H.-W., Scheidt, R.J., Windley, P.G., Eds.; Springer: New York, NY, USA, 2004; pp. 34–58. Int. J. Environ. Res. Public Health 2014, 11 7079 18. Leith, K.H. “Home is Where the Heart is ... Or is it?” A phenomenological exploration of the meaning of home for older women in congregate housing. J. Aging Stud. 2006, 20, 317–333. 19. Andersson, E.; Abramsson, M. Changing residental mobility rates of older people in Sweden. Ageing Soc. 2012, 32, 963–982. 20. Oswald, F.; Wahl, H.W.; Schilling, O.; Nygren, C.; Fange, A.; Sixsmith, A.; Sixsmith, J.; Szeman, Z.; Tomsone, S.; Iwarsson, S. Relationships between housing and healthy aging in very old age. Gerontologist 2007, 47, 96–107. 21. Lagergren, M.; Fratiglioni, L.; Hallberg, I.R.; Berglund, J.; Elmstahl, S.; Hagberg, B.; Holst, G.; Rennemark, M.; Sjolund, B.M.; Thorslund, M.; et al. A longitudinal study integrating population, care and social services data. References The Swedish national study on aging and care (SNAC). Aging Clin. Exp. Res. 2004, 16, 158–168. 22. Ekstrom, H.; Elmstahl, S. Pain and fractures are independently related to lower walking speed and grip strength: Results from the population study “Good Ageing in Skane”. Acta. Orthop. 2006, 77, 902–911. 23. Iwarsson, S.; Wahl, H.W.; Nygren, C.; Oswald, F.; Sixsmith, A.; Sixsmith, J.; Szeman, Z.; Tomsone, S. Importance of the home environment for healthy aging: Conceptual and methodological background of the European ENABLE-AGE project. Gerontologist 2007, 47, 78–84. 24. Sonn, U.; Asberg, K.H. Assessment of activities of daily living in the elderly. A study of a population of 76-year-olds in Gothenburg, Sweden. Scand. J. Rehabil. Med. 1991, 23, 193–202. 25. Iwarsson, S.; Horstmann, V.; Sonn, U. Assessment of dependence in daily activities combined with a self-rating of difficulty. J. Rehabil. Med. 2009, 41, 150–156. 26. Sheikh, J.; Yesavage, J. Geriatric depression scale (GDS): Recent evidence and development of a shorter version. In Clinical Gerontology: A Guide to Assessment and Intervention; Brink, T., Ed.; Haworth Press: New York, NY, USA, 1986; pp. 165–173. 27. Yardley, L.; Beyer, N.; Hauer, K.; Kempen, G.; Piot-Ziegler, C.; Todd, C. Development and Initial Validation of the Falls Efficacy Scale-International (FES-I). Age Ageing 2005, 34, 614–619. 28. Kempen, G.I.; Yardley, L.; van Haastregt, J.C.; Zijlstra, G.A.; Beyer, N.; Hauer, K.; Todd, C. The short FES-I: A shortened version of the falls efficacy scale-international to assess fear of falling. Age Ageing 2008, 37, 45–50. 29. Tibblin, G.; Bengtsson, C.; Furunes, B.; Lapidus, L. Symptoms by age and sex. The population studies of men and women in Gothenburg, Sweden. Scand. J. Prim. Health Care. 1990, 8, 9–17. 30. Ware, J.E., Jr.; Sherbourne, C.D. The mos 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med. Care. 1992, 30, 473–483. 31. Ryff, C.D. Beyond ponce de leon and life satisfaction: New directions in quest of successful ageing. Int. J. Behav. Dev. 1989, 12, 35–55. 32. Arnold, B. Chapter 3: Measurement of quality of life in frail elderly. In The Concept and Measurement of Quality of Life in the Frail Elderly, 1st ed.; Birren, J.E., Lubben, J.E., Rowe, J.C., Deutchman, D.E., Eds.; Academic Press Inc.: San Diego, CA, USA, 1991; p. 56. 33. Lofqvist, C.; Nygren, C.; Szeman, Z.; Iwarsson, S. Assistive Devices Among Very Old People in Five European Countries. Scand. J. Occup. Ther. 2005, 12, 181–192. 34. © 2014 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/). References Iwarsson, S.; Haak, M.; Slaug, B. Current Developments of the Housing Enabler Methodology. Br. J. Occup. Ther. 2012, 75, 517–521. Int. J. Environ. Res. Public Health 2014, 11 7080 35. Helle, T.; Nygren, C.; Slaug, B.; Brandt, A.; Pikkarainen, A.; Hansen, A.; Pétursdórttir, E.; Iwarsson, S. The Nordic Housing Enabler: Inter-rater Reliability in Cross-nordic Occupational Therapy Practice. Scand. J. Occup. Ther. 2010, 17, 258–266. 36. Oswald, F.; Schilling, O.; Wahl, H.W.; Fange, A.; Sixsmith, J.; Iwarsson, S. Homeward Bound: Introducing a Four-domain Model of Perceived Housing in Very Old Age. J. Environ. Psychol. 2006, 26, 187–201. 37. Fange, A.; Iwarsson, S. Physical housing environment: Development of a Self-assessment Instrument. Can. J. Occup. Ther. 1999, 66, 250–260. 38. Fange, A.; Iwarsson, S. Accessibility and Usability in Housing: Construct Validity and Implications for Research and Practice. Disabil. Rehabil. 2003, 25, 1316–1325. 39. Oswald, F.; Wahl, H.-W.; Martin, M.; Mollenkopf, H. Toward Measuring Proactivity in Person-Environment Transactions in Late Adulthood: The Housing-related Control Beliefs Questionnaire. J. Hous. Elderly 2003, 17, 135–155. 40. Hanson, B.S.; Ostergren, P.O.; Elmstahl, S.; Isacsson, S.O.; Ranstam, J. Reliability and Validity Assessments of Measures of Social Networks, Social Support and Control—Results from the Malmo Shoulder and Neck Study. Scand. J. Soc. Med. 1997, 25, 249–257. 41. Iwarsson, S.; Slaug, B. Housing Enabler a Method for Rating/Screening and Analysing. Accessibility Problems in Housing, 2nd ed.; Veten & Skapen HB & Slaug Enabling Development: Lund, Sweden, 2010. 42. Iwarsson, S.; Wilson, G. Environmental Barriers, Functional Limitations, and Housing Satisfaction Among Older People in Sweden: A Longitudinal Perspective on Housing Accessibility. Technol. Disabil. 2006, 18, 57–66. 43. Tomsone, S.; Horstman, V.; Oswald, F.; Iwarsson, S. Aspects of Housing and Perceived Health Among ADL Independent and ADL Dependent Groups of Older People in Three National Samples. Aging Clin. Exp. Res. 2013, 25, 317–328. 44. Nilsson, M.H.; Iwarsson, S. Home and health in people ageing with Parkinson’s disease: Study protocol for a prospective longitudinal cohort survey study. BMC Neurol. 2013, 13, doi:10.1186/1471-2377-13-142. 45. Jörgensen, S.; Mårtensson, L.; Iwarsson, S.; Lexell, J. SASCIS—Swedish Ageing with a Spinal Cord Injury Study. Demographics, Injury Characteristics and Outcome. Presented at the 52nd Annual Scientific Meeting, Istanbul, Turkey, 28–30, October 2013. 46. Oswald, F.; Kaspar, R. On the Quantitative Assessment of Perceived Housing in Later Life. J. Hous. Elderly 2012, 26, 72–93. © 2014 by the authors; licensee MDPI, Basel, Switzerland. References This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/). © 2014 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/). © 2014 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/3.0/).
https://openalex.org/W2749508703	https://europepmc.org/articles/pmc7800797?pdf=render	English	null	Complete mitochondrial genome of the house bat <i>Pipistrellus abramus</i> (Mammalia: Chiroptera) from Korea	Mitochondrial DNA. Part B. Resources	2,017	cc-by	1,191	ABSTRACT ARTICLE HISTORY Received 20 July 2017 Accepted 7 August 2017 We determined and annotated the complete mitogenome of the house bat Pipistrellus abramus (Chiroptera: Vespertilionidae) from Korea. The complete mitogenome is a circular molecule of 17,236 bp in length, including 13 protein-coding genes, 2 ribosomal RNA genes, 22 transfer RNA genes, and 2 non-coding regions (L-strand replication origin and control region). The mitogenome is AT-biased, with a nucleotide composition of 33.7% A, 29.9% T, 23.2% C, and 13.2% G. The phylogenetic analysis revealed that the house bat P. abramus from Korea is well grouped with that from Japan and placed within the genus Pipistrellus clade, which has the noctule bat Nyctalus as sister clade. KEYWORDS Complete mitogenome; Korean house bat; Pipistrellus abramus The house bats of Pipistrellus abramus, known as a species of vesper bat, are wildly distributed across East Asia, from China and Taiwan into the Ussuri region (Russia and China), the Korean Peninsula, and Japan (Won 2004; Bates and Tsytsulina 2008). and 16S rRNA). The mitogenome is AT-biased, with a nucleotide composition of 33.7% A, 29.9% T, 23.2% C, and 13.2% G. Total length of the 22 tRNA genes is 1517 bp and their average length is 69 ± 3.3 bp, ranging from 59 bp (tRNASer(AGY)) to 74 bp (tRNAPhe, tRNAGln, and tRNALeu(UUR)). Lengths of the two rRNA genes and control region are 955 bp (12S rRNA), 1567 bp (16S rRNA), and 928 bp (control regions), respectively. Total length of 13 PCGs is 11,379 bp, with the exclusion of stop codons (30 bp), which encode 3793 amino acids. Mitochondrial PCGs of P. abramus use the three kinds of start codon. ATG is the most common start codon, which is used in 11 PCGs, but the start codons ATT and ATA are used only once in Nd3 and Nd5, respectively. The incomplete stop codons are used for ter- mination of five PCGs (TA for Nd1 and T for Nd2, Nd3, Nd4, and Cox3). AGA is used only once as a stop codon for Cytb. TAA is most common stop codon, which is used for termination of all the other seven PCGs. The replication origin OL is 35 bp in size and located between tRNAAsn and tRNACys within the WANCY tRNA cluster as seen in most vertebrates (Kim and Park 2012; Yoon et al. 2013; Nam et al. 2015; Rahman et al. 2016). CONTACT Y. C. Park parky@kangwon.ac.kr College of Forest and Environmental Science, Kangwon National University, Chuncheon 24341, Republic of Korea; J. Y. Cho jaecho@skku.edu Department of Genetic Engineering, Sungkyunkwan University, Chunchun-Dong 300, Suwon 440-746, Republic of Korea 2017 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, dis- tribution, and reproduction in any medium, provided the original work is properly cited. MITOCHONDRIAL DNA PART B: RESOURCES, 2017 VOL. 2, NO. 2, 540–541 https://doi.org/10.1080/23802359.2017.1365644 Complete mitochondrial genome of the house bat Pipistrellus abramus (Mammalia: Chiroptera) from Korea ng Kima , Hye Ri Kimb , Sang Jin Lima , Hyun Ju Kimc , Jae Youl Chod and Yu Ri Kimb , Sang Jin Lima , Hyun Ju Kimc , Jae Youl Chod and Yung Chul Parka aDivision of Forest Science, College of Forest and Environmental Sciences, Kangwon National University, Republic of Korea; bEcosystem Research Division, National Park Research Institute, Korea National Park Service, Wonju, Republic of Korea; cMicrobial Safety Team, National Institute of Agricultural Sciences, Rural Development Administration, Wanju, Republic of Korea; dDepartment of Genetic Engineering, Sungkyunkwan University, Suwon, Republic of Korea ABSTRACT We sequenced and annotated a mitogenome of P. abra- mus from Korea. A wing membrane tissue sample for gen- omic DNA extraction was collected from a bat individual caught around an agricultural region in Odaesan National Park (N37 42 42.8, E128 35 57.6), South Korea. The voucher specimen (VEPIAB-1) was deposited in the Wildlife and Fish Conservation Center of the Institute of Forest Science, Kangwon National University. Genomic DNA extraction, PCR, and gene annotation were conducted according to the previ- ous studies (Yoon et al. 2013; Jeon and Park 2015; Rahman et al. 2016). Previously published mitogenomes of Indian P. coromandra (KP688404) and Japanese P. abramus (NC_005436) were used as references for gene annotation and primer design for PCR amplification of the Korean P. abramus mitogenome. Phylogenetic tree was constructed using maximum-likelihood (ML) procedures implemented in MEGA6 (Tamura et al. 2013). The complete mitogenome (KX355640) of the Korean house bat P. abramus contains total 17,236 bp in length, which consists of a control region (one D-loop region) and a conserved set of 37 genes including 13 protein-coding genes (PCGs), 22 tRNA genes, and 2 ribosomal RNA genes (12S rRNA The phylogenetic analysis revealed that the house bat P. abramus from Korea is well grouped with that from Japan and placed within the genus Pipistrellus clade, which has the noctule bat Nyctalus as sister clade (Figure 1). MITOCHONDRIAL DNA PART B: RESOURCES 5 541 541 Figure 1. The phylogenetic relationship of P. abramus and its allied species inferred from maximum-likelihood analysis based on mitogenome sequences. The ML tree was generated using the GTR þ G þ I model, and the robustness of the tree was tested with 1000 bootstrap. The numbers on the branches indicate bootstrap values. Figure 1. The phylogenetic relationship of P. abramus and its allied species inferred from maximum-likelihood analysis based on mitogenome sequences. The ML tree was generated using the GTR þ G þ I model, and the robustness of the tree was tested with 1000 bootstrap. The numbers on the branches indicate bootstrap values. Jeon MG, Park YC. 2015. The complete mitogenome of the wood-feeding cockroach Cryptocercus kyebangensis (Blattodea: Cryptocercidae) and phylogenetic relations among cockroach families. Anim Cells Syst. 19:432–438. Disclosure statement The authors report no conflicts of interest. The authors alone are respon- sible for the content and writing of the paper. Kim HR, Park YC. 2012. The complete mitochondrial genome of the striped field mouse, Apodemus agrarius (Rodentia, Murinae) from Korea. Mitochondrial DNA. 23:145–147. ORCID Ji Young Kim http://orcid.org/0000-0001-7460-6353 Hye Ri Kim http://orcid.org/0000-0003-1697-9121 Sang Jin Lim http://orcid.org/0000-0001-9108-3048 Hyun Ju Kim http://orcid.org/0000-0001-9935-2484 Jae Youl Cho http://orcid.org/0000-0001-8141-9927 Yung Chul Park http://orcid.org/0000-0002-5466-2339 Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. 2013. MEGA6: molecular evolutionary genetics analysis version 6.0. Mol Biol Evol. 30:2725–2729. Won BO. 2004., Mammals of Korea (Hangugui poyudongmul). Seoul: Dongbang Media. Yoon KB, Cho JY, Park YC. 2013. Complete mitochondrial genome of the Korean ikonnikov's bat Myotis ikonnikovi (Chiroptera: Vespertilionidae)). Mitochondrial DNA. 26:274–275. Funding This work was carried out with support of ‘Cooperative Research Program for Agricultural Science & Technology Development (Project No. PJ0108592016)’ Rural Administration of Republic of Korea. Nam TW, Kim HR, Cho JY, Park YC. 2015. Complete mitochondrial gen- ome of a large-footed bat, Myotis macrodactylus (Vespertilionidae). Mitochondrial DNA. 26:661–662. Rahman MM, Yoon KB, Kim JY, Hussin MZ, Park YC. 2016. Complete mitochondrial genome sequence of the Indian pipi- strelle Pipistrellus coromandra (Vespertilioninae). Anim Cells Syst. 20:86–94. References Bates P, Tsytsulina K. 2008. Pipistrellus abramus. The IUCN Red List of Threatened Species 2008: e.T17320A6972924. http://dx.doi.org/10. 2305/IUCN.UK.2008.RLTS.T17320A6972924.en. [2017 Jul 16].
https://openalex.org/W4323288270	https://sciencepubco.com/index.php/IJAA/article/download/32117/16848	English	null	Light deflection angle in General Relativity via Daftardar-Jafari method	International journal of advanced astronomy	2,022	cc-by	7,802	1. Introduction The well-known effect of the light deflection in the gravitational fields of compact objects is one of the first and key predictions of Gen- eral Relativity (GR) [1], [2], [3]. Although the study of light deflection have historically been associated with the Solar System [4]-[7], in the recent decades, much more attention has been paid to the study of light deflection in the gravitational fields of various compact astro- physical objects [8]. physical objects [8]. Basically, several approximate approaches are used to determine the deflection of light in the gravitational field of black holes. One of them is the standard parameterized post-Newtonian approach which applicable for b >> M, where b is the impact parameter of the unper- turbed light ray and M is the mass of a black hole. Here and below, we use the units in which G = c = 1. As the next approach, one can note the standard weak-field approximate lens equation, which usually is called the classical lens equations [9]. However, these exact lens equations are also given in terms of elliptic integrals. Therefore, approximations of these exact solutions are also needed for a time- efficient data reduction. Several proposals for generalized lens equations have then appeared in the literature. One decisive advantage of the classical lens equation is its validity for arbitrarily small values of the impact distance b. A lens equation which allows an arbitrary large values of the deflection angle and used the deflection angle expression for the Schwarzschild metric is obtained in [10], [11]. As is known, weak gravitational lensing makes it possible to find the mass of astronomical objects without requiring information about their composition or dynamic states. Therefore, in recent years, many authors have proposed studies of gravitational lensing by various astrophysical objects using various methods. Let us mention just a few of the latest articles on this topic. For example, the equations of motion of the massive and massless particles in the Schwarzschild geometry is studied by using the Laplace-Adomian Decomposition Method in [12], that shows the obvious success of this method in obtaining series solutions to a wide range of strongly nonlinear differ- ential equations. Basically, several approximate approaches are used to determine the deflection of light in the gravitational field of black holes. 1. Introduction One of them is the standard parameterized post-Newtonian approach which applicable for b >> M, where b is the impact parameter of the unper- turbed light ray and M is the mass of a black hole. Here and below, we use the units in which G = c = 1. As the next approach, one can note the standard weak-field approximate lens equation, which usually is called the classical lens equations [9]. However, these exact lens equations are also given in terms of elliptic integrals. Therefore, approximations of these exact solutions are also needed for a time- efficient data reduction. Several proposals for generalized lens equations have then appeared in the literature. One decisive advantage of the classical lens equation is its validity for arbitrarily small values of the impact distance b. A lens equation which allows an arbitrary large values of the deflection angle and used the deflection angle expression for the Schwarzschild metric is obtained in [10], [11]. As is known, weak gravitational lensing makes it possible to find the mass of astronomical objects without requiring information about their composition or dynamic states. Therefore, in recent years, many authors have proposed studies of gravitational lensing by various astrophysical objects using various methods. Let us mention just a few of the latest articles on this topic. For example, the equations of motion of the massive and massless particles in the Schwarzschild geometry is studied by using the Laplace-Adomian Decomposition Method in [12], that shows the obvious success of this method in obtaining series solutions to a wide range of strongly nonlinear differ- ential equations. A new method for calculating the angle of small deviation using the Gauss-Bonnet theorem was proposed by the authors of [13] and has found wide application in studies of this kind. This method was also applied to the study of light rays in a plasma medium in a static and spherically symmetric gravitational field and to the study of time-like geodesics followed for test massive particles in a spacetime with the same symmetries in Ref. [14]. The calculation of the bending angle using the trajectory equation based on geometric optics is also provided in [15], [16]. Because wormholes also cause gravitational lensing, this effect has been recently investigated in several papers (see, for example, [17], [18] and references therein). Light deflection angle in General Relativity via Daftardar-Jafari method V. K. Shchigolev * Department of Theoretical Physics, Ulyanovsk State University, Ulyanovsk, 432000, Corresponding author E-mail: vkshch@yahoo.com V. K. Shchigolev Department of Theoretical Physics, Ulyanovsk State University, Ulyanovsk, 432000, *Corresponding author E-mail: vkshch@yahoo.com Abstract In this paper, the iterative method suggested by Daftardar and Jafari hereafter called Daftardar-Jafari method (DJM) is applied for study- ing the deflection of light in General Relativity. For this purpose, a brief review of the nonlinear geodesic equations in the spherical symmetry spacetime and the main ideas of DJM are given. As an illustrative example, the simple case of the Schwarzschild metric is considered for which the approximate solution to the null-geodesic equation and the deflection angle of light are obtained. We also com- pare the obtained result with some similar results presented earlier in the literature. Keywords: Daftardar-Jafari Iterative Method; Deflection of Light; General Relativity; Schwarzschild Metric. Copyright © V. K. Shchigolev. This is an open access article distributed under the Creative Commons A stricted use, distribution, and reproduction in any medium, provided the original work is properly cited. opyright © V. K. Shchigolev. This is an open access article distributed under the Creative Commons Attribution License, which permits un icted use, distribution, and reproduction in any medium, provided the original work is properly cited. International Journal of Advanced Astronomy, 10 (1) (2022) 4-10 International Journal of Advanced Astronomy, 10 (1) (2022) 4-10 1. Introduction The author of this article has recently proposed to use two well - known methods, such as the homotopy perturbation method and the variational iteration method [19, 20], to study the deflection of light and the perihelion precession in GR [21]-[24]. Here we use another efficient iterative method for approximating null - geodesics and finding the deflection angle of light. This iterative method has been proposed by Daftardar-Gejji and Jafari [25], and proved the effectiveness for solving many of the linear and nonlinear ordinary differen- tial equations, partial differential equations and integral equations [26], [27]. The proposed DJM is very effective and reliable, and the solution is obtained in the series form with easily computed components [28]. The main aim of this paper is to apply DJM in the Schwarzschild metric to solve the null-geodesic equation and to find the approximate value of the light deflection angle. In addition, we compare the obtained result with the results previously known from the literature. International Journal of Advanced Astronomy 5 2. Null-geodesic equation in a spherically symmetric spacetime ( ) dr L h r L E h r k d r f r r       = − +            (4) 2 2 2 2 2 2 ( ) ( ) . ( ) dr L h r L E h r k d r f r r       = − +            (4) As the coordinate r u / 1  is more convenient than r in studying the geodesic equations in the spherically symmetric gravitational fields, equation (4) can be converting to the following one: As the coordinate r u / 1  is more convenient than r in studying the geodesic equations in the spherically symmetric gravitational fields, equation (4) can be converting to the following one: ). ( ) ( ) ( 2 2 2 u h u L E u f u h d du −       =        Finally, differentiating this equation with respect to  , we get the second-order geodesic equation in the following form: ating this equation with respect to  , we get the second-order geodesic equation in the following form: ating this equation with respect to  , we get the second-order geodesic equation in the following form: Finally, differentiating this equation with respect to  , we get the second-order geodesic equation in the following form: .) ( 2 1 ) ( ) ( ) ( 2 2 2 2 2 2 du u dh u u uh u f u h du d L E d u d − −     =  .) ( 2 1 ) ( ) ( ) ( 2 2 2 2 2 2 du u dh u u uh u f u h du d L E d u d − −     =  (5) (5) Thereafter, we are going to apply DJM for studying the propagation of light in metric (1). But first, we need to recall the main idea of the DJM and its implementation in solving the second-order ordinary differential equations. 2. Null-geodesic equation in a spherically symmetric spacetime dt dr d f r h r r d d d     −       − − =             (2) 2 2 2 1 2 ( ) ( ) 0. dt dr d f r h r r d d d     −       − − =             (2) Since there are two conserved quantities along the geodesic in metric (1), the total energy ( ) dt d E f r  = and the angular momentum per unit mass   d d r L 2 = , we can substitute these constants into equation (2) to obtain Since there are two conserved quantities along the geodesic in metric (1), the total energy ( ) dt d E f r  = and the angular momentum per unit mass   d d r L 2 = , we can substitute these constants into equation (2) to obtain . ) ( ) ( ) ( 2 2 2 2       + − =       r L k r h E r f r h d dr  (3) . ) ( ) ( ) ( 2 2 2 2       + − =       r L k r h E r f r h d dr  (3) This equation contains only one unknown function ) ( r and can in principle be solved. However, the deflection of light are usually re- lated to the geodesics orbits, i.e. ) ( r . Therefore, with the help of   d d r L 2 = , one can rewrite equation (3) as follows This equation contains only one unknown function ) ( r and can in principle be solved. However, the deflection of light are usually re- lated to the geodesics orbits, i.e. ) ( r . Therefore, with the help of   d d r L 2 = , one can rewrite equation (3) as follows 2 2 2 2 2 2 ( ) ( ) . 2. Null-geodesic equation in a spherically symmetric spacetime According to General Relativity [1, 8], the line element of general static spherically symmetric spacetime can be represented by According to General Relativity [1, 8], the line element of general static spherically symmetric spacetime can be rep ). sin ( ) ( ) ( 2 2 2 2 2 2 2    d d r r h dr dt r f ds + + + − = (1) (1) The trajectories of photon in GR are usually considered as the null geodesics in spacetime. Therefore, we have to study the geodesics in the spherically symmetric spacetime (1) in the spherical coordinates ) , , , (    r t x = as ) (  x , where  is some affine parameter, and satisfies the geodesic equation: .0 2 2 =  +         d dx d dx d x d The geodesic trajectories can be obtained as the solutions of this equation. As is known, when deriving the geodesic equation, one can follow a simpler way if one takes into account the space-time symmetries of metric (1). First of all we should note that one component of the geodesic line can always be chosen as 2 / ) (  which means that we can The geodesic trajectories can be obtained as the solutions of this equation. As is known, when deriving the geodesic equation, one can follow a simpler way if one takes into account the space-time symmetries of metric (1). First of all, we should note that one component of the geodesic line can always be chosen as 2 / ) (    = , which means that we can always choose a geodesic lying in the equatorial plane of spherically symmetric space-time. Then, taking into account the null geodesics follow a simpler way if one takes into account the space-time symmetries of metric (1). First of all, we should note that one component of the geodesic line can always be chosen as 2 / ) (    = , which means that we can always choose a geodesic lying in the equatorial plane of spherically symmetric space-time. Then, taking into account the null geodesics condition 0 = ds in (1), we can obtain the following equation 2 2 2 1 2 ( ) ( ) 0. And And (12) . 1 i i u f u   = + = The m -term approximate solution of Eq. (7) is given by 1 1 0 ... − + + + = m u u u u . If N is a contraction, then the series 1 i iu  =  in (12) absolutely and uniformly converges to a solution of Eq. (6), which is unique, in view of the Banach fixed point theorem [25]. For more details about the convergence of DJM, we refer the reader to Ref. [28]. The m -term approximate solution of Eq. (7) is given by 1 1 0 ... − + + + = m u u u u . If N is a contraction, then the series 1 i iu  =  in (12) absolutely and uniformly converges to a solution of Eq. (6), which is unique, in view of the Banach fixed point theorem [25]. For more details about the convergence of DJM, we refer the reader to Ref. [28]. 3. Main ideas of DJM in brief In Ref. [25], the authors consider the general functional equation: (6) , ) ( f u N u + = , ) ( f u N u + = Where N a nonlinear operator from a Banach is space B B → and f is a known function. Suppose that the solution u of Eq. (6) has the series form: International Journal of Advanced Astronomy 6 . 0 i i u u   = = (7) (7) The nonlinear operator N can be decomposed as               −       + =           − = =  =  = j i j j i j i i i u N u N u N u N 1 0 0 0 0 0 ) ( (8) As it follows from (7) and (8), Eq. (6) is equivalent to As it follows from (7) and (8), Eq. (6) is equivalent to . ) ( 1 0 0 0 0 0               −       + + =     − = =  =  = j i j j i j i i i u N u N u N f u (9) One can suppose the following recurrence relation: One can suppose the following recurrence relation:    + + − + + = = = − + ), ... ( ) ... ( ), ( , 1 0 0 1 0 1 0 m m m u u N u u N u u N u f u    = = ), ( , 0 1 0 u N u f u (10)   = ), ( 0 1 u N u  + + − + + = − + ), ... ( ) ... ( 1 0 0 1 m m m u u N u u N u  + + − + + = − + ), ... ( ) ... ( 1 0 0 1 m m m u u N u u N u Where ,... 2 ,1 = m . Then Where ,... 2 ,1 = m . 4. Application of DJM to a second order differential equation description mainly follows to Ref. [26]. Consider some non-linear ordinary differential equation of the second order, Here our description mainly follows to Ref. [26]. Consider some non-linear ordinary differential equation of the sec Here our description mainly follows to Ref. [26]. Consider some non-linear ordinary differential equation of the second order, ), ( ~ ) ( ~ ) ( ) ( ) ( 2 1     f u N u k u k u = + +  +  (13) ), ( ~ ) ( ~ ) ( ) ( ) ( 2 1     f u N u k u k u = + +  +  (13) Where a prime stands for the derivative with respect to  , 2 1, k k are arbitrary constants, ) ( ~  f is a given continuous function, and ) ( ~ u N is a non-linear term. Besides, this equation must satisfy the initial condition: Where a prime stands for the derivative with respect to  , 2 1, k k are arbitrary cons ) ( ~ u N is a non-linear term. Besides, this equation must satisfy the initial condition: (14) Equation (13) can be written in an operator form as: ), ( ~ ) ( ~ ) ( ) ( ) ( 2 1       f u N u k u L k u L = + + + (15) ), ( ~ ) ( ~ ) ( ) ( ) ( 2 1       f u N u k u L k u L = + + + (15) ), ( ~ ) ( ~ ) ( ) ( 2 1     f u N u k u L k = + + + (15) ), ( ~ ) ( ~ ) ( ) ( ) ( 2 1       f u N u k u L k u L = + + + Where   d d L = and 2 2   d d L = . We assume that the inverse operators 1 −  L and 1 −  L exist and can be taken as follows (16) 3. Main ideas of DJM in brief Then (11) ,..., 2 ,1 ), ... ( ) ... ( 0 1 1 = + + = + + + m u u N u u m m And And ( ) . ))] ( ( ~ ) ( )[ ( ) ( ))] ( ( ~ ) ( [ ) ( ) ( 2 1 0 2 1 1 1 ds s u N s u k s s u k u N u k L u k L u N + − + − = + − − =  − −        ( ) . ))] ( ( ~ ) ( )[ ( ) ( ))] ( ( ~ ) ( [ ) ( ) ( 2 1 0 2 1 1 1 ds s u N s u k s s u k u N u k L u k L u N + − + − = + − − =  − −        (22) By using expressions (21) and (22) in equation (6), we can follow the procedure (10) in order to obtain solution (12) of ODE (13), pro- vided (14). Thereafter, we are going to apply DJM to the approximate solution of the null geodesic equation and finding the light deflec- tion angle in Schwarzschild spacetime. By using expressions (21) and (22) in equation (6), we can follow the procedure (10) in order to obtain solution (12) of ODE (13), pro- vided (14). Thereafter, we are going to apply DJM to the approximate solution of the null geodesic equation and finding the light deflec- tion angle in Schwarzschild spacetime. And And , )(.) ( (.) (.) 0 0 0 1 ds s dt ds L s − = =    −     (17) 7 International Journal of Advanced Astronomy Where we have used the Cauchy formula for repeated integration: Where we have used the Cauchy formula for repeated integration: , ) ( ) ( )! 1 ( 1 ... ) ( ... 1 0 1 2 0 0 0 1 1 ds s v s n ds ds ds s v n n n s s n − − − =    −    (18) , ) ( ) ( )! 1 ( 1 ... ) ( ... 1 0 1 2 0 0 0 1 1 ds s v s n ds ds ds s v n n n s s n − − − =    −    (18) (18) Then, applying the inverse operator L 1 to both sides of the equation (15) and taking into account the initial condition (14), we have ))], ( ( ~ ) ( [ ) ( ) ( ) ( 2 1 1 1 1          u N u k L u k L g B A k A + − − + + + = − − ))], ( ( ~ ) ( [ ) ( ) ( ) ( 2 1 1 1 1          u N u k L u k L g B A k A u + − − + + + = − − ))], ( ( ~ ) ( [ ) ( ) ( ) ( 2 1 1 1 1          u N u k L u k L g B A k A u + − − + + + = − − (19) ))], ( ( ~ ) ( [ ) ( ) ( ) ( 2 1 1 1 1          u N u k L u k L g B A k A u + − − + + + = − − (19) (19) Where . And ) ( ~ ) ( ) ( ~ ) ( 0 0 0 ds s f s dt t f ds g s − = =        . ) ( ~ ) ( ) ( ~ ) ( 0 0 0 ds s f s dt t f ds g s − = =        (20) (20) Therefore, by using equations (16)-(20), we can represent equation (14) in the form of equation (1) by setting Therefore, by using equations (16)-(20), we can represent equation (14) in the form of equation (1) by setting , ) ( ~ ) ( ) ( 0 1 ds s f s B A k A f − + + + =      (21) And ( ) ))] ( ( ~ ) ( )[ ( ) ( ))] ( ( ~ ) ( [ ) ( ) ( 1 1 d N k k N k L k L N + + +  − −  (22) , ) ( ~ ) ( ) ( 0 1 ds s f s B A k A f − + + + =      (21) 5. Light Deflection in Schwarzschild spacetime via DJM Therefore, comparing these equations with the corresponding equations (13), (14), we get the following equalities: International Journal of Advanced Astronomy 8 , 1 ,0 ,1 ,0 2 1 b B A k k = = = = (27) And (28) . 3 ) ( ~ ,0 ~ 2 u M N f − = =  According to equations (21), (22) and (16), (17), we get According to equations (21), (22) and (16), (17), we get  , ) ( 3 ) ( ) ( )) ( ( , ) ( 2 0 ds s Mu s u s u N b f − − − = =      (29) And, therefore, the functional equation (6) becomes as follows And, therefore, the functional equation (6) becomes as follows  . ) ( 3 ) ( ) ( ) ( 2 0 ds s Mu s u s b u − − − =     (30) So we can to use this equation to construct an approximate solution according to the procedure described by equation (10). The 0-term of the approximate solution (7), that is, the unperturbed motion, is described by b f u / ) ( ) ( 0    = = , according to equations (10) and (29). ( ) It is easy to verify that ) ( u , defined by equation (30), satisfies the initial conditions (26). Moreover, by differentiating this equation twice, we can verify that it is equivalent to the geodesic equation (25). Then, applying (10) to equation (30), one can obtain the following terms:   ), ... ( ) ... ( ) ( , ) ( 3 ) ( ) ( )) ( ( , ) ( 1 0 0 1 2 0 0 − + + + − + + = − − − = =  m m m u u N u u N u ds s Mu s u s u N b u       (31)   ), ... ( ) ... 5. Light Deflection in Schwarzschild spacetime via DJM ( ) ( ) ( 3 ) ( ) ( 1 2 1 0 1 0 ] [  u ds s u s u M s u s u − + − + (33) (33) As a result of calculating the integral in equation (33), the approximate solution 2 1 0 u u u u + + = can be obtained in the following form: s a result of calculating the integral in equation (33), the approximate solution 2 1 0 u u u u + + = can be obtained in the As a result of calculating the integral in equation (33), the approximate solution 2 1 0 u u u u + + = can be obtained in the following form: . 480 288 28 672 24 4 !5 !3 1 ) ( 4 10 3 9 7 3 2 8 6 4 2 5 3 b M b M b M b u           +       − +       + − +       + − = (34) (34) Note that this expansion is quite different from the one given in [Shchigolev4] and does not seem very useful, since the coefficients are expressed as power polynomials in  , although they usually contain trigonometric functions. However, some functions can be restored if we have enough terms in the expansion (12). Indeed, applying the power series of trigonometric functions, Note that this expansion is quite different from the one given in [Shchigolev4] and does not seem very useful, since the coefficients are expressed as power polynomials in  , although they usually contain trigonometric functions. However, some functions can be restored if we have enough terms in the expansion (12). 5. Light Deflection in Schwarzschild spacetime via DJM The following example demonstrates the use of DJM for the analytical computation of the deflection angle of light in the simplest spher- ically symmetric spacetimes (1). In the absence of mass ( 0 = M ), the obvious analytic solution for (5) is a straight line expressed in polar coordinates as , sin ) ( . . b u l s  = (23) Where b is a constant impact parameter. Obviously, the term 2 3 u M comes from the correction by GR. Therefore, we can consider (23) to be the null approximation for (5). To solve the problem of finding the subsequent approximations to the solution of equation (5) by the iterative method, we have to follow the procedure of DJM for solving the second order nonlinear differential equations described in Sec- tion 3. Let us consider the simplest case of metric (1), namely, the Schwarzschild spacetime describing the gravitational field of an un- charged non-rotating star. For the Schwarzschild solution, we have r M r h r f 2 1 ) ( ) ( − = = , or , 2 1 ) ( ) ( Mu u h u f − = = (24) Where M is the mass of a star. Therefore, equation (5) for the null geodesic can be written as . 3 2 2 2 u M u d u d = +  (25) In the absence of mass ( 0 = M ), the obvious analytic solution for (25) is a straight line expressed in polar coordinates (23). Thus, the term 2 3 u M comes from the correction of path by GR. Taking into account that a trajectory of light due to equation (25) is started as the straight line (23) at  , we have the following initial conditions for ) ( u : . 1 ) 0 ( ,0 ) 0 ( b u u =  = (26) . 1 ) 0 ( ,0 ) 0 ( b u u =  = (26) We are going to solve the Cauchy problem (25), (26) for the nonlinear differential equation of the second order with a certain approxima- tion using DJM. International Journal of Advanced Astronomy 5. Light Deflection in Schwarzschild spacetime via DJM ( ) ( , ) ( 3 ) ( ) ( )) ( ( , ) ( 1 0 0 1 2 0 0 − + + + − + + = − − − = =  m m m u u N u u N u ds s Mu s u s u N b u       (31) In the m -term approximate solution (7) of the following form: ) ( ) ( 1 1    i m i b u u  + = − = . Thus, due to equations (10) and (31), we can obtain In the m -term approximate solution (7) of the following form: ) ( ) ( 1 1    i m i b u u  + = − = . Thus, due to equations (10) and (31), we can obtain   . 12 ) 2 3 ( ) ( 3 ) ( ) ( )) ( ( ) ( 2 3 2 0 0 0 0 1 b b M ds s Mu s u s u N u       − = − − − = =  (32)   . 12 ) 2 3 ( ) ( 3 ) ( ) ( )) ( ( ) ( 2 3 2 0 0 0 0 1 b b M ds s Mu s u s u N u       − = − − − = =  (32) Using equations (31) and (32), we get ) ( ) ( ) ( 0 1 0 2 u N u u N u − + =  , that is Using equations (31) and (32), we get ) ( ) ( ) ( 0 1 0 2 u N u u N u − + =  , that is ( ) ). ( ) ( ) ( 3 ) ( ) ( ) ( ) ( 1 2 1 0 1 0 0 2 ] [     u ds s u s u M s u s u s u − + − + − − =  (33) ( ) ). 5. Light Deflection in Schwarzschild spacetime via DJM Graphs of deflection angles ) (x DJM  and ) (x HPM  , where M b x / = is a dimensionless impact parameter, are shown in Figure 1. It is known that a photon with an impact parameter of 2 / 3 3 S cr r b b =  , where M rS 2 = is the Schwarzschild radius, may be captured by the central object of mass M [2]. Thus, we have to consider 3 3  x . Fig. 1: Shows The Graphs of Deflection Angles ( ) DJM x  (Continuous Line) and ( ) HPM x  (Dashed Line) Versus the Dimensionless Impact Parameter M b x / = . Fig. 1: Shows The Graphs of Deflection Angles ( ) DJM x  (Continuous Line) and ( ) HPM x  (Dashed Line) Versus the Dimensionless Impact Parameter M b x / = . 5. Light Deflection in Schwarzschild spacetime via DJM k n k n n k C − = are the binomial coefficients.Applying the formulas (36) and (37) to the null trajectory (35) results in the following small deflection angle in DJM approximation (up to the second order): Due to the binomial formula , ) ( 0 k k n n k n k n C     − =  = + where )! (! ! k n k n n k C − = are the binomial coefficients.Applying the formulas (36) and (37) to the null trajectory (35) results in the following small deflection angle in DJM approximation (up to the second order): . 288 28 1 28 4 3 3 2 2 2 7        +      − + = b M b M b M DJM    (38) . 288 28 1 28 4 3 3 2 2 2 7        +      − + = b M b M b M DJM    (38) Note that the angle of light deflection in the Schwarzschild metric obtained earlier (see, for example, [Virbhadra1]) and also derived using the homotopy perturbation method in [Shchigolev4] with the same accuracy is as follows Note that the angle of light deflection in the Schwarzschild metric obtained earlier (see, for example, [Virbhadra1]) and also derived using the homotopy perturbation method in [Shchigolev4] with the same accuracy is as follows . 4 15 4 3 3 2 2        + + = b M b M b M HPM   (39) . 4 15 4 3 3 2 2        + + = b M b M b M HPM   (39) Graphs of deflection angles ) (x DJM  and ) (x HPM  , where M b x / = is a dimensionless impact parameter, are shown in Figure 1. It is known that a photon with an impact parameter of 2 / 3 3 S cr r b b =  , where M rS 2 = is the Schwarzschild radius, may be captured by the central object of mass M [2]. Thus, we have to consider 3 3  x . 6. Conclusion Thus, we have applied the iterative method called Daftardar-Jafari method for studying the deflection of light in General Relativity. First of all, we have represented a brief review of the nonlinear geodesic equations in the spherical symmetry spacetime and the main ideas of DJM. In order to approbate DJM in the problem of deflection of light and present the main steps in solving by this method, we have illus- trated how DJM can be employed to obtain the approximate analytical solution of the null geodesic equation in the simple case of the Schwarzschild metric. For this metric, the approximate solutions to the null-geodesic equation and the deflection angle of light have been obtained. We also compared the obtained result with the similar result presented earlier in the literature. An important advantage of DJM is the simplicity of obtaining approximate solutions by repeated applications of the iterative equations. The analytic and approximate solutions are obtained without any restrictive assumptions for nonlinear terms as required by some existing techniques. Moreover, by solving some examples, it is seems that the DJM appears to be very accurate to employ with reliable results. We used the Maple software for the calculations in this study. [1] S. Weinberg. Gravitation and Cosmology: Principles and Applications of the General Theory of Relativity, John Wiley. Press, New York, 1972. [2] C. W. Misner, K. S. Thorne, and J. A. Wheeler, Gravitation, W. H. Freeman and Co., San Francisco, Calif, USA, 1973. Obviously, solution (35) satisfies the initial condition 0 ) 0 ( = u . Therefore, the deflection angle of light  can be obtained from the equation 0 ) ( = +  u , using the small angle approximation 5. Light Deflection in Schwarzschild spacetime via DJM Indeed, applying the power series of trigonometric functions, ..., !4 !2 1 cos ..., !5 !3 sin 4 2 5 3 − + − = − + − =        ..., !4 !2 1 cos ..., !5 !3 sin 4 2 5 3 − + − = − + − =        To the first and second terms of equation (34), we can represent the approximate solution in the following form: To the first and second terms of equation (34), we can represent the approximate solution in the following form: ( ) . 480 288 28 cos 1 sin ) ( 4 10 3 9 7 3 2 2 2 b M b M b M b u       +       − + − +  (35) ( ) . 480 288 28 cos 1 sin ) ( 4 10 3 9 7 3 2 2 2 b M b M b M b u       +       − + − +  (35) International Journal of Advanced Astronomy 9 Obviously, solution (35) satisfies the initial condition 0 ) 0 ( = u . Therefore, the deflection angle of light  can be obtained from the equation 0 ) ( = +  u , using the small angle approximation Obviously, solution (35) satisfies the initial condition 0 ) 0 ( = u . Therefore, the deflection angle of light  can be obtained from th equation 0 ) ( = +  u , using the small angle approximation ,1 ) cos( , ) sin( −  + −  +      (36) And , ) ( 1     − +  + n n n n (37) ,1 ) cos( , ) sin( −  + −  +      (36) And , ) ( 1     − +  + n n n n (37) Due to the binomial formula , ) ( 0 k k n n k n k n C     − =  = + where )! (! ! References International Journal of Advanced Astronomy 10 [3] R. M. Wald, General Relativity, University of Chicago Press, Chicago, Ill, USA, 1984. [4] M. V. Bebronne, P.G.Tinyakov, "Erratum: Black hole solutions in massive gravity", J. High Energ. Phys., 2011 (2011), 18. https://doi.org/10.1007/JHEP06(2011)018. p g [5] C. Magnan, "Complete calculations of the perihelion precession of Mercury and the deflection of light by the Sun in General Relativity", http://arxiv.org/abs/0712.3709 p g [6] J. Poutanen, "Accurate analytic formula for light bending in Schwarzschild metric", A A, 640 (2020), A24. https 6361/202037471. g "Accurate analytic formula for light bending in Schwarzschild metric", A A, 640 (2020), A24. https://doi.org/10.1051/000 471. [7] R. S. Kuniyal, R. Uniyal, A. Biswas, et al., "Null geodesics and red–blue shifts of photons emitted from geodesic particles around a noncommuta- tive black hole space–time", Int. J. Modern Phys. A, 33(16) (2018), 1850098. https://doi.org/10.1142/S0217751X18500987. Hongsheng Zhang, Jun-Peng Hou, and Liang-Zun Tang, "Perihelion Precession and Deflection of Light in the General Spherica pacetime", Advances in High Energy Physics, Volume 2014, Article ID 604321, 7 pages. https://doi.org/10.1155/2014/604321. [8] Ya-Peng Hu, Hongsheng Zhang, Jun-Peng Hou, and Liang-Zun Tang, "Perihelion Precession and Deflection of Light in the General Spherically Symmetric Spacetime", Advances in High Energy Physics, Volume 2014, Article ID 604321, 7 pages. https://doi.org/10.1155/2014/604321. [9] V Bozza "Comparison of approximate gravitational lens equations and a proposal for an improved new one" Phys Rev D 78 (2008) 103005 [9] V. Bozza, "Comparison of approximate gravitational lens equations and a proposal for an improved new one", Phys. Rev https://doi.org/10.1103/PhysRevD.78.103005. Comparison of approximate gravitational lens equations and a proposal for an improved new one", Phys. Rev. D, 78 (2008), 10300 g/10.1103/PhysRevD.78.103005. [10] K. S. Virbhadra, D. Narasimha, S. M. Chitre, "Role of the scalar field in gravitational lensing", Astron. Astrophys., 337 (1998), 1. https://doi.org/10.48550/arXiv.astro-ph/9801174 [11] K. S. Virbhadra, George F. R. Ellis, "Schwarzschild black hole lensing", Phys. Rev. D, 62 (2000), 084003. https://doi.org/10.1103/PhysRevD.62.084003. [12] M. K. Mak, C. S. Leung, and T. Harko, "Computation of the General Relativistic Perihelion Precession and of Light Deflection via the Laplace- Adomian Decomposition Method", Advances in High Energy Physics, Volume 2018, Article ID 7093592, 15 pages. https://doi.org/10.1155/2018/7093592. p g 13] G. W. Gibbons and M. C. Werner, "Classical and Quantum Gravity Applications of the Gauss–Bonnet theorem to gravitatio Quantum Grav., 25 (2008), 235009. https://doi.org/10.1088/0264-9381/25/23/235009. [14] G. Crisnejo and E. References Gallo, "Weak lensing in a plasma medium and gravitational deflection of massive particles using the Gauss-Bonnet theorem. A unified treatment", Phys. Rev. D, 97 (2018), 124016. https://doi.org/10.1103/PhysRevD.97.124016. [15] J. Y. Kim, "Deflection of light by a Coulomb charge in Born–Infeld electrodynamics", Eur. Phys. J. C, 81 (2021), 508. https://doi.org/10.1140/epjc/s10052-021-09291-6. p g pj 16] W. Javed, M. B. Khadim, A. Ц vgь n, "Weak gravitational lensing by Bocharova–Bronnikov–Melnikov–Bekenstein black Bonnet theorem", Eur. Phys. J. Plus, 135 (2020), 595. https://doi.org/10.1140/epjp/s13360-020-00619-x. [17] J. G. Cramer, R. L. Forward, M. S. Morris, et al., "Natural wormholes as gravitational lenses", Phys. Rev. D, 51 (1995), 3117. https://doi.org/10.1103/PhysRevD.51.3117. [18] K. Jusufi, N. Sarkar, F. Rahaman, et al., "Deflection of light by black holes and massless wormholes in massive gravity", Eur. Phys. J. C, 78 (2018), 349. https://doi.org/10.1140/epjc/s10052-018-5823-z. Homotopy perturbation technique", Computer Methods in Applied Mechanics and Engineering, 178 (1999), 257-26 g/10.1016/S0045-7825(99)00018-3. p g ( ) 20] J. -H. He, "Variational iteration method- a kind of non-linear analytical technique: some examples", International Journal of No p g [20] J. -H. He, "Variational iteration method- a kind of non-linear analytical technique: some examples", International Journal of Non-Linear Mechanics, 34(4) (1999), 699-708. https://doi.org/10.1016/S0020-7462(98)00048-1. p g [20] J. -H. He, "Variational iteration method- a kind of non-linear analytical technique: some 34(4) (1999), 699-708. https://doi.org/10.1016/S0020-7462(98)00048-1. , "Variational iteration method- a kind of non-linear analytical technique: some examples", International Journal of Non-Linear Mec 999), 699-708. https://doi.org/10.1016/S0020-7462(98)00048-1. 0] J , y q p , J 34(4) (1999), 699-708. https://doi.org/10.1016/S0020-7462(98)00048-1. [21] V. Shchigolev, "Analytical Computation of the Perihelion Precession in General Relativity via the Homotopy Perturbation Method", Univ. J Com- put. Math., 3(4) (2015), 45-49. https://doi.org/10.13189/ujcmj.2015.030401. p g j j olev, "Variational iteration method for studying perihelion precession and deflection of light in General Relativity", Int. J. Phys. R 2016), 52. https://doi.org/10.14419/ijpr.v4i2.6314. [22] V. K. Shchigolev, "Variational iteration method for studying perihelion precession and deflection of light in General Rela search, 4(2) (2016), 52. https://doi.org/10.14419/ijpr.v4i2.6314. , ( ) ( ), p g jp [23] V. K. Shchigolev, D. N. Bezbatko, "On HPM approximation for the perihelion precession angle in general relativity", Int. J. Adv. Astronomy, 5(1) (2017), 38. https://doi.org/10.14419/ijaa.v5i1.7279. ( ) p g j [24] V. K. Shchigolev, D. N. Bezbatko, "Studying Gravitational Deflection of Light by Kiselev Black Hole via Homotopy Perturbation Method", Gen. Relativ. Gravit., 51 (2019), 34. https://doi.org/10.1007/s10714-019-2521-6. References [25] V D ft d G jji H J f i "A it ti th d f l i li f ti l ti " J l f M th ti l A l i d A li ti p g [25] V. Daftardar-Gejji, H. Jafari, "An iterative method for solving non-linear functional equations", Journal of Mathematical A 316 (2006) 753 763 https://doi org/10 1016/j jmaa 2005 05 009 316 (2006) 753–763. https://doi.org/10.1016/j.jmaa.2005.05.00 ( ) p g j j [26] M. A. Al-Jawary, S. G. Abd-Al-Razaq, "Analytic and numerical solution for Duffing equations", Int. J. of Basic and Applied Sciences, 5 (2) (2016), 115-119. https://doi.org/10.14419/ijbas.v5i2.5838. [ ] g , p y pp y gy g , p Science and Information Technology, 8(2) (2020), 43-49. https://doi.org/10.13189/csit.2020.080202. [28] S. Bhalekar, V. Daftardar-Gejji, "Convergence of the New Iterative Method", International Journal of Differential Equations, 2011, Article ID 989065, 10 pages. https://doi.org/10.1155/2011/989065.
https://openalex.org/W4396688827	https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0302707&type=printable	English	null	Combining enhanced spectral resolution of EMG and a deep learning approach for knee pathology diagnosis	PloS one	2,024	cc-by	8,775	RESEARCH ARTICLE Combining enhanced spectral resolution of EMG and a deep learning approach for knee pathology diagnosis a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 Ateka KhaderID, Ala’a Zyout, Amjed Al Fahoum* Biomedical Systems and Informatics Engineering Department, Yarmouk University, Irbid, Jordan Ateka KhaderID, Ala’a Zyout, Amjed Al Fahoum* Biomedical Systems and Informatics Engineering Department, Yarmouk University, Irbid, Jordan Biomedical Systems and Informatics Engineering Department, Yarmouk University, Irbid, Jordan * afahoum@yu.edu.jo a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 PLOS ONE RESEARCH ARTICLE OPEN ACCESS Citation: Khader A, Zyout A, Al Fahoum A (2024) Combining enhanced spectral resolution of EMG and a deep learning approach for knee pathology diagnosis. PLoS ONE 19(5): e0302707. https://doi. org/10.1371/journal.pone.0302707 Editor: Fei Yan, Chongqing University Three Gorges Hospital, CHINA Editor: Fei Yan, Chongqing University Three Gorges Hospital, CHINA Gorges Hospital, CHINA Received: September 21, 2023 Accepted: April 9, 2024 Published: May 7, 2024 Copyright: © 2024 Khader et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Received: September 21, 2023 Accepted: April 9, 2024 Published: May 7, 2024 Copyright: © 2024 Khader et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: The data underlying the results presented in the study are available from Roelker et al [33]. Data are stored on Figshare: https://figshare.com/projects/Modular_ Control_of_Walking_in_Unimpaired_Younger_ and_Older_Adults_and_Individuals_with_Knee_ Osteoarthritis/128774. Abstract Knee osteoarthritis (OA) is a prevalent, debilitating joint condition primarily affecting the elderly. This investigation aims to develop an electromyography (EMG)-based method for diagnosing knee pathologies. EMG signals of the muscles surrounding the knee joint were examined and recorded. The principal components of the proposed method were prepro- cessing, high-order spectral analysis (HOSA), and diagnosis/recognition through deep learning. EMG signals from individuals with normal and OA knees while walking were extracted from a publicly available database. This examination focused on the quadriceps femoris, the medial gastrocnemius, the rectus femoris, the semitendinosus, and the vastus medialis. Filtration and rectification were utilized beforehand to eradicate noise and smooth EMG signals. Signals’ higher-order spectra were analyzed with HOSA to obtain information about nonlinear interactions and phase coupling. Initially, the bicoherence representation of EMG signals was devised. The resulting images were fed into a deep-learning system for identification and analysis. A deep learning algorithm using adapted ResNet101 CNN model examined the images to determine whether the EMG signals were conventional or indicative of knee osteoarthritis. The validated test results demonstrated high accuracy and robust metrics, indicating that the proposed method is effective. The medial gastrocnemius (MG) muscle was able to distinguish Knee osteoarthritis (KOA) patients from normal with 96.3 ±1.7% accuracy and 0.994±0.008 AUC. MG has the highest prediction accuracy of KOA and can be used as the muscle of interest in future analysis. Despite the proposed method’s superiority, some limitations still require special consideration and will be addressed in future research. PLOS ONE PLOS ONE Introduction Knee osteoarthritis (OA) is a joint disorder that causes pain and disability, particularly in the elderly [1]. OA results from wear, tear, and deterioration of the articular cartilage [2]. How- ever, OA affects not only the cartilage but is a disease that affects all of the tissue of the joint [3]. There are many reasons for this joint disease, including aging, obesity, injuries, stress, and Funding: The author(s) received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. 1 / 15 PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 PLOS ONE Deep learning approach for detection knee pathology using EMG signal other factors that weaken the muscles around the joint [4]. The symptom that usually leads to discovering OA is severe pain in the knee. The knee joint is complex and comprises bone, ligaments, tendon muscles, cartilage, and fluid. The knee joint plays a role in the movement and stability of the human body, shock absorption, and supporting the body weight [5]. Many methods have been used to detect knee abnormalities, including X-ray, magnetic resonance imaging (MRI), and Computer Tomogra- phy (CT) [6–9]. In the initial investigation of knee pain, X-ray imaging often serves as the pri- mary diagnostic tool due to its accessibility and cost-effectiveness. However, while X-rays can reveal changes in bone structure and joint space narrowing indicative of osteoarthritis, their capacity to detail the condition of soft tissues, such as cartilage, is limited [10]. This limitation is significant given that knee osteoarthritis (KOA) is characterized not only by bone deteriora- tion but also by the degeneration of cartilage and other joint tissues. Advanced imaging modal- ities like magnetic resonance imaging (MRI) and computer tomography (CT) scans offer comprehensive insights into both bone and soft tissue states but come with higher costs and increased examination time, potentially limiting their routine use in clinical practice [10]. Fur- thermore, Inertial Measurement Units (IMUs), including gyroscopes and accelerometers, have been explored for their potential in diagnosing knee pathologies through the analysis of move- ment patterns [11–13]. Electromyography (EMG) is a valuable, economical and non-invasive technique that com- plements existing diagnostic tools by providing insights into muscle activity patterns affected by the disease [14]. EMG signals are widely used in different fields, including rehabilitation medicine, human-machine interface design, prosthesis and robotics control, and clinical diag- nosis [15–17]. Introduction Interestingly [32–34], studies have used the same data sets, which comprise EMG signals for 11 normal subjects and 11 sub- jects with knee OA. The data have been taken from the biceps femoris (BF), vastus medialis (VM), rectus femoris (RF), and semitendinosus (ST) muscles while the subject is walking, sit- ting, and standing. Moreover, in all these studies, the EMG signals from individual muscles were not examined separately. Instead, features were extracted from all four muscles, then the selected features were fed into different machine-learning classifiers. These studies often focused on isolated muscle analysis or conventional machine learning algorithms, which may not fully capture the intricate patterns of neuromuscular activity associated with KOA. Our study extends this body of work by employing a novel integration of high-order spectral analy- sis (HOSA) with deep learning (specifically, the ResNet101 architecture), aimed at enhancing the diagnostic precision of EMG signal analysis for KOA. Unlike previous efforts, our approach considers the complex, nonlinear interactions between muscle activities, leveraging the advanced feature extraction capabilities of HOSA and the sophisticated pattern recognition power of deep learning. HOSA method, such as bispectrum and bicoherence, provides higher resolution than low- order power spectral analysis approaches [35]. These methods can capture nonlinear and non- Gaussian properties to comprehend biological signals’ frequency content and interactions. Nonlinear characteristics such as phase coupling, amplitude modulation, and frequency mod- ulation are present in various biological signals [36]. These nonlinearities are crucial to under- standing the underlying physiological processes and disorders, and they can be detected and quantified using the HOSA technique [35]. Conventional spectrum analysis techniques can occasionally obscure or overlook complex information in biomedical signals. In order to better analyze and classify data, algorithms for HOSA can extract more features and reveal signals’ latent correlations [37]. Methods based on HOSA analysis have shown promise for improving the categorization and recognition of biomedical signals [23, 25, 35–37]. This study aims to apply deep learning and incorporate bicoherence for the first time to EMG data using HOSA to diagnose KOA as a source of significant ramifications. It can facili- tate a more precise diagnosis, allowing knee pathology to be diagnosed and treated sooner. Early diagnosis is the key to effectively controlling OA and halting further joint decline. In addition, it can shed light on how OA develops and the causes of the neuromuscular alter- ations that characterize the condition. Introduction In assessing knee joint function and pathology, various imaging techniques, including fluoroscopy, offer valuable insights, particularly in dynamic conditions [18, 19]. While these methods provide detailed anatomical information during motion, they primarily focus on structural aspects of the knee joint. EMG, on the other hand, complements these imaging techniques by offering unique insights into muscle activity patterns, which are crucial for understanding the functional implications of knee pathologies such as OA [20, 21]. The strength of EMG lies in its ability to capture the neuromuscular dynamics associated with knee movement, providing a direct link to the muscular adaptations or impairments resulting from or contributing to KOA. This capability makes EMG a critical tool in a multidisciplinary approach to diagnosing and understanding KOA, alongside conventional imaging methods. However, the EMG signal is a complex signal affected by intrinsic and extrinsic noise due to motion artifacts, baseline noise, and interference noise [22]. This demands sophisticated pro- cessing analysis techniques to extract a significant insight into muscle-specific changes associ- ated with KOA. Despite it’s complexity, the usefulness of EMG technology remains to be invaluable in understanding the muscle state and function. So using EMG signals for classify- ing the pathological knee by traditional methods could be challenging. Developing an auto- mated system could help diagnose the pathological knee from the EMG data. Machine learning and deep learning have been used in many biomedical applications [23– 26]. They were also utilized to predict and evaluate OA using MRI [27, 28] and X-ray [29, 30]. Furthermore, EMG signals have been investigated using machine learning techniques [31, 32]. Vijayvargiya et al. [33] used EMG for the knee muscles during gait, standing, and sitting posi- tions for normal and abnormal subjects to predict the abnormality. Different machine learning classifiers are used, including the k-nearest neighbor, support vector machine, decision tree, random forest, and extra tree, with the different tree classifiers having the highest accuracy (91%) [33]. Another study used the same data to detect knee abnormalities [32]. Anomaly detection methods have removed abnormal data with the light gradient boosting machine to reach 98% accuracy. 2 / 15 PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 PLOS ONE Deep learning approach for detection knee pathology using EMG signal Furthermore, it has been shown that EMG signals can predict lower limb activities for nor- mal and OA patients using a hybrid deep learning model [34]. PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 Introduction The proposed method will develop more effective treat- ments and recovery protocols with this knowledge. Moreover, by utilizing EMG data, the proposed method may further reduce the cost, hazards due to imaging, and surgical interven- tion of knee pathology diagnosis. The contribution of this work can be validated through the comparison of the performance of the proposed method to that of other credible EMG analysis methods and clinical evaluations. The decision to study the EMG signal from the BF, MG, VM, RF, and ST stems from the objective to capture a comprehensive view of the knee’s neuro- muscular function. By analyzing signals from these four muscles simultaneously, we aim to understand the synergistic muscle activities that characterize knee movement in both healthy individuals and those affected by KOA. The signals were analyzed in parallel, considering the unique contributions of each muscle to knee stability and movement. This parallel analysis allows us to identify patterns that might be indicative of KOA, taking into account the complex interplay between different muscles around the knee. Some specific metrics like accuracy, sen- sitivity, specificity, and area under the receiver operating characteristic curve can be used to quantify the results. Furthermore, investigations with a larger patient cohort can compare classification outputs with clinical assessments and imaging modalities, shedding light on the approach’s clinical 3 / 15 PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 PLOS ONE Deep learning approach for detection knee pathology using EMG signal relevance and practical utility. Long-term follow-up research can be used to evaluate the pro- posed method’s ability to predict disease progression and treatment response. By gathering user feedback and expert opinions, assessing the proposed system’s effectiveness and interpret- ability is also possible. These evaluations will shed light on the impact and efficacy of applying deep learning to EMG data to identify KOA. Methods This study’s methodology focuses on developing and implementing innovative algorithms for achieving the intended outcomes. The first step of the proposed method is to collect Normal and KOA EMG data from various muscles. Utilize a signal processing technique to process the data, aiming to reveal hidden patterns within the signal. Apply a deep learning network to clas- sify the results that emphasize the differences among the categories to facilitate its classifica- tion. Fig 1 depicts the three primary processes of the developed method for EMG type detection: preprocessing, high-order spectrum analysis (HOSA), and diagnosis and recogni- tion using deep learning. EMG preprocessing Motion artifacts are a substantial source of interference for the surface EMG signal in almost all biomedical signals. The EMG signal and the motion artifacts often have a high-amplitude peak next to each other, but the motion artifacts typically occur at low frequencies. Therefore, a high pass filter was used to remove the artifacts effects [39]. The signal was filtered with a Notch Filter of 60 Hz to remove the DC power line interference, a common artifact observed in EMG signals due to electrical noise from the power supply [22]. This preprocessing step is crucial for ensuring the accuracy of EMG signal analysis by eliminating extraneous noise that can obscure the true physiological signals of interest. This approach is consistent with estab- lished practices in EMG signal processing, where mitigating power line interference is recog- nized as essential for reliable data analysis. Then, EMG signal is rectified to determine the strength of the neural drive to the muscle, which is related to the force of the muscular contrac- tion and its output [40]. All EMG processing was computed on MATLAB1 R2022b. Dataset description The EMG signals analyzed in this study were acquired from a publicly available dataset described by Roelker et al [38]. The choice to use a publicly available dataset by Roelker et al. was driven by the dataset’s relevance, quality, and comprehensive documentation of EMG sig- nals under conditions pertinent to the study’s objectives. The dataset encompasses EMG sig- nals collected from 20 subjects during walking at their preferred pace for at least five walking trials. The subjects were ten adults (63.5 ± 3.4 years) without KOA (Normal) and ten adults with KOA (64.0 ± 4.0 years). The healthy subjects did not have any record of any knee injuries and were able to walk without pain or limp. The patients with KOA had suffered from knee KOA, with grade 3 or 4 OA, and were diagnosed by three orthopedic surgeons at The Ohio State Wexner Medical Center. The Telemyo DTS System (Noraxon, Scottsdale, AZ) was employed for signal collection, capturing data from eight key leg muscles at a sampling rate of 1500 Hz. Prior to analysis, signals underwent filtering through a 30–300 Hz sixth-order Butter- worth band-pass filter to ensure the removal of frequencies not pertinent to muscle activity, thereby refining the quality of the data for subsequent deep-learning analysis. Fig 1. Block diagram of the proposed approach. https://doi.org/10.1371/journal.pone.0302707.g001 Fig 1. Block diagram of the proposed approach. Fig 1. Block diagram of the proposed approach. https://doi.org/10.1371/journal.pone.0302707.g001 https://doi.org/10.1371/journal.pone.0302707.g001 4 / 15 PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 PLOS ONE Deep learning approach for detection knee pathology using EMG signal Fig 2. EMG data from five selected muscles before and after preprocessing for the KOA class. https://doi.org/10.1371/journal.pone.0302707.g002 Fig 2. EMG data from five selected muscles before and after preprocessing for the KOA class. https://doi.org/10.1371/journal.pone.0302707.g002 This paper selects Five muscles to build an automated classification system: biceps femoris (BF), medial gastrocnemius (MG), rectus femoris (RF), semitendinosus (ST), vastus medialis (VM), for both EMG types. One sample for each muscle is illustrated in Fig 2. PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 High Order Spectral Analysis (HOSA) HOSA was used for feature extraction and simplifying the complex nature of EMG signals. Higher-order statistics ("cumulants") of a signal are used to generate higher-order spectra (sometimes called polyspectra). Higher-order spectra include the trispectrum (fourth-order spectrum), defined as the Fourier transform of the fourth-order statistics of a stationary signal, and the bispectrum (third-order spectrum), defined as the Fourier transform of the third-order statistics. Fig 3 shows a discrete-time signal’s higher-order spectrum classification map [41, 42]. In contrast to the power spectrum, higher-order spectra are functions of two or more indepen- dent frequencies. Numerical spectrum estimates at higher orders may or may not be statistically significant, but they will always be greater than zero. The larger the degrees of freedom, the more reliable the estimate. The amplitude of the higher-order spectrum is essential when study- ing phase coupling between Fourier components, which holds regardless of the powers of the 5 / 15 PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 PLOS ONE Deep learning approach for detection knee pathology using EMG signal Fig 3. The various higher-order spectra for a deterministic signal. F [.] denotes the k-dimensional Fourier Transform. https //doi org/10 1371/journal pone 0302707 g003 Fig 3. The various higher-order spectra for a deterministic signal. F [.] denotes the k-dimensional Fourier Transform. Fig 3. The various higher-order spectra for a deterministic signal. F [.] denotes the k-dimensional Fourier Transform. htt //d i /10 1371/j l 0302707 003 https://doi.org/10.1371/journal.pone.0302707.g003 component frequencies. Coupling the phases by comparing the strengths at the individual fre- quencies is possible. Normalized spectra can be used to detect and describe nonlinearity in sys- tems because nonlinear interactions produce phase-coupled power at the sum and difference frequencies [43]. The power spectrum (n = 2) and the cumulate spectrum (order n) are com- bined to create the normalized higher-order spectrum or nth-order coherency (bicoherence) index. The definition of the third-order coherence of a signal with discrete time intervals is: BX Bicoherence ð Þ f1; f2 ð Þ ¼ Bx ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi PXðf1ÞPXðf2ÞPXðf1 þ f2Þ p ð1Þ ð1Þ fifififififififififififififififififififififififififififififififififififififififififififi With this function, the power spectrum can be estimated more precisely, and the bispec- trum and bicoherence estimates coincide. Auto-bicoherence and cross-bicoherence with the bicoher and bicoherx procedures can be calculated [44]. High Order Spectral Analysis (HOSA) The HOSA Toolbox in MATLAB1 provides the bicoherence function in two classes of electromyography (EMG) from five mus- cles; the bicoherence representation is formed, and the resulting images are stored with their respective classes, as shown in Fig 4. The deep-learning models will then process the stored images. Fig 4. Samples of a Bicoherence representation for EMG signal from five muscles: (a) The Normal EMG from Biceps Femoris; (b) The Normal EMG from Medial Gastroc; (c) The Normal EMG from Rectus Femoris; (d) The Normal EMG from Semitendinosus; (e) The Normal EMG from Vastus Medialis; (f) The Normal EMG from average muscles; (g) The KOA EMG from Biceps Femoris; (h) The KOA EMG from Medial Gastroc; (i) The KOA EMG from Rectus Femoris; (j) The KOA EMG from Semitendinosus (k) The KOA EMG from Vastus Medialis; (l) The KOA EMG from average muscles. https //doi org/10 1371/jo rnal pone 0302707 g004 Fig 4. Samples of a Bicoherence representation for EMG signal from five muscles: (a) The Normal EMG from Biceps Femoris; (b) The Normal EMG from Medial Gastroc; (c) The Normal EMG from Rectus Femoris; (d) The Normal EMG from Semitendinosus; (e) The Normal EMG from Vastus Medialis; (f) The Normal EMG from average muscles; (g) The KOA EMG from Biceps Femoris; (h) The KOA EMG from Medial Gastroc; (i) The KOA EMG from Rectus Femoris; (j) The KOA EMG from Semitendinosus (k) The KOA EMG from Vastus Medialis; (l) The KOA EMG from average muscles. https://doi.org/10.1371/journal.pone.0302707.g004 PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 6 / 15 PLOS ONE Deep learning approach for detection knee pathology using EMG signal Convolutional neural network CNN is the best option for image-based classification due to its self-feature learning capabili- ties and higher classification results on multi-class classification problems. Many researchers have begun using transfer learning approaches to fine-tune the pre-trained deep learning structures for the desired job [45–48]. To find the optimum model for the suggested technique, an adapted version of the per- tained Resnet101 CNN model is used in this research. The network has been implemented in MATLAB1. The ResNet101 CNN model was chosen due to its deep architecture of 101 layers, facilitating feature extraction and learning from the complex patterns inherent in the bicoher- ence images of EMG signals. The transfer learning techniques were utilized to fine-tune the pre-trained ResNet101 model to our specific task. This approach enables the model to leverage knowledge gained from a related task (image recognition on ImageNet) and apply it to our domain-specific problem of classifying EMG signals for KOA diagnosis. The fully connected layer of ResNet101 was adapted to output two classes, corresponding to normal and KOA con- ditions. This modification ensures that the model’s outputs are directly relevant to our classifi- cation problem. Fig 5 describes the detailed structure of ResNet101 [49]. This preprocessing step ensures that the model can effectively process and learn from our dataset. The data is divided into 70% training, 15% validation, and 15% randomized testing. The study employed a repeated random subsampling validation method, wherein the EMG dataset was randomly partitioned into training and testing sets across five separate iterations. For each iteration, the dataset under- went preprocessing to ensure data cleanliness before being divided. This approach allowed to assess the robustness and consistency of the proposed machine learning classifiers’ predictive performance on different subsets of the data, thereby providing a reliable estimate of their gen- eralization ability to unseen data. The model is built by the following hyperparameters; Adap- tive Moment Estimation (Adam) for efficient gradient descent, mini patch size 32, maximum epochs 60, the initial learning rate 0.001, and the validation frequency 3. All images fed into the model were standardized to a size of 224×224×3, aligning with the input dimensions expected by adapted version of the ResNet101. Results In this study, a total of 20 subjects consisting of normal subjects and subjects with KOA in equal numbers were examined. The dataset was divided into 70/15/15 Train-Validate-Test datasets. The datasets are subsequently trained on a variety of machine-learning classifiers. The data were trained on machine learning classifiers at different five runs. In each run, the datasets were selected arbitrarily and put in train, validation or train set. For each muscle, the confusion matrices were obtained for each run and the features were extracted and further analyzed. In addition to that, the EMG signals from the five muscles together were used to pre- dict the KOA. Fig 6 shows the precision and sensitivity results for the prediction of normal and KOA subjects using machine learning based on EMG signals from different muscles. The results showed that the average precision where more than 90% when using the EMG signals from any of the five muscles and all muscles for both normal and KOA subjects. Medial gas- trocnemius (MG), rectus femoris (RF), semitendinosus (ST), and vastus medialis (VM) have significantly higher precision than using all the muscles to predict KOA. Moreover, the sensi- tivity was more than 92% for prediction KOA for all the muscles, in which the MG muscle has 99.1% followed by ST and VM with 96.7% average sensitivity. Using all muscles together to predict the normal subject has a lower sensitivity of 89.5%. MG and RF had significantly higher sensitivity than using all muscles to predict the normal person (Fig 6B). The specificity and the F-measure for predicting knee abnormality from EMG signals are shown in Fig 7. The average specificity is more than 93% for BF, MG, and RF for prediction KOA. While the average specificity is more than 96% for ST and VM for the prediction of nor- mal knee. Moreover, the specificity for using MG, RF, ST and VM are significantly higher than using all the muscles together to predict KOA. The F- measure is the highest for MG for both prediction KOA and normal subjects with 96.4% and 96.2% respectively. Moreover, the F- measure is significantly higher for MG muscles as compared with all muscles for both predic- tion KOA and normal subjects. VM muscle has a significantly higher F-value as compared with all muscles for prediction KOA. Statistical analysis The results are presented as mean with standard deviations. Statistical analysis was performed using Statistical Package for the Social Science (SPSS, v.21.0, SPSS Inc, Chicago, IL). A one- way analysis of variance (ANOVA) test was used to evaluate whether the type of muscle (e.g., medial gastrocnemius, rectus femoris, etc.) significantly affects the machine learning model’s performance in predicting KOA. The test specifically assesses the effect of muscle type on vari- ous performance metrics of machine learning models, such as accuracy, sensitivity, and speci- ficity. Tukey’s Honest Significant Difference (HSD) post-hoc test was used to investigate the Fig 5. The architecture of ResNet101. https://doi.org/10.1371/journal.pone.0302707.g005 7 / 15 PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 PLOS ONE Deep learning approach for detection knee pathology using EMG signal main effect of the type of muscles on different machine learning performance metrics. Proba- bility (p) values < 0.05 were considered statistically significant. Results Table 1 shows the Area Under the Curve (AUC), Matthews correlation coefficient (MCC) and accuracy for all the fusion matrices to predict knee osteoarthritis (KOA) and normal Fig 6. (a) Precision and (b) sensitivity for prediction the knee osteoarthritis (KOA) and normal subjects using EMG signals from different muscles. Where biceps femoris (BF), medial gastrocnemius (MG), rectus femoris (RF), semitendinosus (ST), vastus medialis (VM), and average of all five muscles (ALL) groups. Data are presented as mean with standard deviations (mean ±STD). P<0.5, significantly higher than ALL group. https://doi.org/10.1371/journal.pone.0302707.g006 Fig 6. (a) Precision and (b) sensitivity for prediction the knee osteoarthritis (KOA) and normal subjects using EMG signals from different muscles. Where biceps femoris (BF), medial gastrocnemius (MG), rectus femoris (RF), semitendinosus (ST), vastus medialis (VM), and average of all five muscles (ALL) groups. Data are presented as mean with standard deviations (mean ±STD). P<0.5, significantly higher than ALL group. https://doi.org/10.1371/journal.pone.0302707.g006 PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 8 / 15 PLOS ONE Deep learning approach for detection knee pathology using EMG signal Fig 7. (a) Specificity and (b) F-measure for prediction the knee osteoarthritis (KOA) and normal subjects using EMG signals from different muscles. Data are presented as mean ±STD. P<0.5, significantly higher than ALL group. https://doi org/10 1371/journal pone 0302707 g007 F-measure for prediction the knee osteoarthritis (KOA) and normal subjects using EMG signals from different muscles. Data a <0.5, significantly higher than ALL group. Fig 7. (a) Specificity and (b) F-measure for prediction the knee osteoarthritis (KOA) and normal subjects using EMG signals f presented as mean ±STD. P<0.5, significantly higher than ALL group. Fig 7. (a) Specificity and (b) F-measure for prediction the knee osteoarthritis (KOA) and normal subjects using EMG signals from different muscles. Data are presented as mean ±STD. P<0.5, significantly higher than ALL group. https://doi.org/10.1371/journal.pone.0302707.g007 subjects using EMG signals. AUC values are more than 0.98 for all muscles when are used indi- vidually to predict the subjects with KOA. While the AUC is 0.964 when the average of all muscles is used for prediction. AUC values are significantly higher for MG, ST and VM as compared to ALL. MCC values are significantly higher for MG and VM as compared to ALL group. The average accuracy is 94.2%, 96.3%, 94.7%, 94.4%, 94.7% and 91% for BF, MG, RF, ST, VM, and ALL muscles respectively. Results Interestingly, MG has the highest accuracy between the average for all the muscles and has significantly higher accuracy than the average of all the muscles together for prediction abnormality. The one-way ANOVA test’s findings, as applied in this study, suggest that certain muscles, such as the medial gastrocnemius, may have a more pronounced role in accurately predicting KOA based on EMG signals. This is evidenced by higher precision, sensitivity, and specificity values compared to other muscles or the aggregate of all studied muscles. https://doi.org/10.1371/journal.pone.0302707.t001 Discussion In this study, EMG signals from five different muscles were used to discriminate KOA subjects from normal controls using a machine learning technique, which could give a quantitative tool for KOA diagnosis. Our results indicate that the EMG signal from the medial gastrocnemius (MG) muscle was able to distinguish KOA patients from normal with 96.3% accuracy and 0.994 AUC. Also, the EMG signals from the biceps femoris (BF), rectus femoris (RF), semiten- dinosus (ST) and vastus medialis (VM) were able to distinguish KOA patients from normal with around 94% accuracy while when the five muscles together yielded only 91% accuracy. While it is acknowledged that EMG signals inherently possess noise due to motion artifacts, baseline fluctuations, and interference [18]. It is essential to highlight the efficacy of advanced Table 1. Area Under the Curve (AUC), Matthews correlation coefficient (MCC) and accuracy using different muscles to predict knee osteoarthritis (KOA) and nor- mal subjects using EMG signals. Muscles BF MG RF ST VM ALL AUC 0.985±0.018 0.994±0.008 0.984±0.015 0.994±0.004* 0.993±0.004* 0.964±0.019 MCC 0.886±0.048 0.927±0.034* 0.893±0.035 0.891±0.018 0.896±0.037* 0.823±0.024 Accuracy(%) 94.2%±2.5 96.3±1.7* 94.7±1.8 94.4±1 94.7±1.9 91±1.1 P<0.5, significantly higher than ALL group. orrelation coefficient (MCC) and accuracy using different muscles to predict knee osteoarthritis (KOA) and nor- ea Under the Curve (AUC), Matthews correlation coefficient (MCC) and accuracy using different muscles to predict knee using EMG signals. P<0.5, significantly higher than ALL group. PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 9 / 15 PLOS ONE Deep learning approach for detection knee pathology using EMG signal signal processing and deep learning techniques employed in this study to mitigate these chal- lenges. These methods significantly enhance the signal-to-noise ratio, allowing for the extrac- tion of meaningful information from the EMG data that is relevant to diagnosing KOA. Furthermore, MRI, despite being the gold standard for visualizing joint structures and assess- ing OA severity, does not provide direct insights into the dynamic muscle activities and neuro- muscular control associated with knee function and pathology [50]. EMG fills this gap by offering real-time, non-invasive measurements of muscle activity, which are crucial for under- standing the biomechanical and functional aspects of knee OA. The complementary nature of EMG and MRI highlights the multidimensional approach required for a comprehensive assessment of KOA. Discussion While MRI excellently depicts structural changes, EMG provides a unique window into functional impairments, allowing for a holistic understanding of the disease’s impact on knee joint stability and movement. Machine learning and deep learning techniques were used to predict the KOA in different studies. Many studies have used the same datasets from UCI machine learning repository by Sanchez et al [51] to predict knee abnormality. EMG signals from RF, BF, VM and ST muscles from 11 healthy subjects and 11 subjects with knee abnormalities were collected during stand- ing, walking, and sitting. In Vijayvargiya et al., Five different classifiers were used in the study including k-nearest neighbor, support vector machine, decision tree, random forest and extra tree [33]. Forty-four features were extracted from four different muscle EMG signals then the backward elimination technique was used to select the relevant features by the p-value test. Machine learning and deep learning techniques were used to predict the KOA in different studies. Many studies have used the same datasets from UCI machine learning repository by Sanchez et al [51] to predict knee abnormality. EMG signals from RF, BF, VM and ST muscles from 11 healthy subjects and 11 subjects with knee abnormalities were collected during stand- ing, walking, and sitting. In Vijayvargiya et al., Five different classifiers were used in the study including k-nearest neighbor, support vector machine, decision tree, random forest and extra tree [33]. Forty-four features were extracted from four different muscle EMG signals then the backward elimination technique was used to select the relevant features by the p-value test. The highest accuracy was 91% in detecting knee abnormality and achieved by Extra Tree Clas- sifier. In another study, the EMG signals were denoised by the Wavelet Denoising method fol- lowed by the extraction of eleven features, then using oversampling to balance the data [52]. After that different classifiers were used such as the extra tree, SVM, MLP, random forest and gradient boosting with extra tree have the highest accuracy (92.5%). In another study with the same dataset, anomaly detection techniques were used to improve the performance of the clas- sifiers [32]. The accuracy was 98.5% when using the iforest anomaly detection method on the light gradient boosting machine model while the accuracy was as low as 85% without any adjustment methods. Zhao et al. Discussion [53] have used XGBoost and cross-validation (CV) to predict the KOA from EMG signals and compare them to SVM and the deep neural network (DNN). XGBoost classifier has higher accuracy than the other two methods (the average accuracy of ten experiments was 96.09%). In our study, the accuracy of prediction the knee OA was higher than 94% when using EMG signals from BF, RF, VM, ST or MG muscles. Previous studies, such as those by Vijayvargiya et al. [33, 52], have indeed explored the potential of machine learning and deep learning techniques for differentiating healthy subjects from those with KOA using EMG signals. Notably, studies have often analyzed EMG signals in isolation or employed conventional machine learning techniques that may not fully capture the complex, non-linear interactions between muscle activities during knee movement. This study advances this research by not only employing a deep learning approach but also by integrating it with the bicoherence analysis of EMG signals. This integration allows for a more advanced under- standing of the non-linear interactions between muscle activities, which is critical for accu- rately diagnosing KOA. The bicoherence method provides a novel way to feature EMG signals for deep learning analysis, enhancing the model’s diagnostic capability beyond what has been reported in previous studies. Chen et al. have used EMG signals from VM, ST, BF, and vastus lateralis (VL) muscles to distinguish between KOA subjects and control based on three entropy measures, i.e., approxi- mate entropy, sample entropy, and fuzzy entropy [54]. They found that using fuzzy entropy features from the EMG signals of VM and BF muscles yielded 92% accuracy, 91.43% sensitivity and 93.33% specificity in distinguishing KOA subjects. Parisi et al. have used a Genetic 10 / 15 PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 PLOS ONE Deep learning approach for detection knee pathology using EMG signal Algorithm-based denoising approach on the EMG signals to maximize mutual information and minimize entropy [55]. The EMG signals of RF, BF, VM, ST, MG, hamstrings medial (MH) and lateral gastrocnemius (LG) muscles were encompassed in the study. The use of Genetic Algorithm-Denoising- Lagrangian Support Vector Machine (GA-D-LSVM) resulted in very high accuracy (99.57%). It should be pointed out that EMG signals from 7 different muscles were used to train the model and the data used in the study were selected from two databases. Discussion In this article, the pre-processing steps were conducted to enhance the quality of the signal and did not eliminate frequency-related features or outliers data or use anomaly detec- tion. The obtained results of this study demonstrated that the MG muscle exhibits distinct pat- terns that are highly predictive of KOA, achieving 96.3% accuracy, while BF, RF, ST and VM muscles were able to predict KOA with around 94% accuracy. A study by Derek et al. has shown that subjects with moderate KOA had higher and prolonged quadriceps and higher lat- eral hamstring activity as compared to a control subject [56]. Interestingly, they found that the activity of both lateral and medial gastrocnemius (MG and LG) muscles increased during the early stance phase which could enhance joint stability during weight acceptance and single-leg stance, especially in severe KOA. This could explain the obtained results that show MG has the highest prediction accuracy of KOA. In a meta-analysis review, they found that patients with KOA have augmented co-contraction, amplitude and duration of lateral knee muscles [57]. RF, VL and BF activation amplitudes were usually increased in moderate KOA patients. Table 2 provides comparative performance results of the proposed method with recently published articles. This comparison demonstrates the advanced accuracy, sensitivity, specific- ity, and AUC metrics achieved by this study, particularly highlighting the utility of combining high-order spectral analysis (HOSA) with deep learning adapted (ResNet101) for diagnosing knee osteoarthritis (KOA) through EMG signals. Furthermore, this juxtaposition against clini- cal evaluations and other EMG methods serves as external validation, reinforcing the reliability and applicability of our findings. By showcasing how the method aligns with or surpasses the diagnostic capabilities of existing approaches, these findings underscore its potential contribu- tion to the field. Notably, the superior performance metrics—especially the prediction accu- racy of the medial gastrocnemius (MG) muscle—underscore the method’s diagnostic precision. Conversely, the study represents a significant step forward in this endeavor, offering new insights into the neuromuscular alterations associated with KOA and paving the way for future research to explore these methods in clinical settings Notably, the classification problem for the four grades of KOA must be addressed, and the deep learning model must be strengthened to withstand the difficulty of distinguishing between healthy knees and lower-grade osteoarthritis. Discussion Future research could therefore be directed in accordance with the subsequent strategies: Expand the sample size to encompass a more extensive spectrum of knee osteoarthritis (KOA) grades, ranging from Grade 1 (severe) to Grade 4 (doubtful). Convert the deep learning model to handle a multi-class classification problem, requiring it to categorize knees into the following five groups: normal, Grade 1, Table 2. Comparison of the proposed method’s performance with other studies. Study Methods Muscles Accuracy Chen et al. [54] fuzzy entropy VM and BF 92% Vijayvargiya et al. [33] Extra Tree Classifier RF, BF, VM and ST 91% Vijayvargiya et al. [52] Extra Tree Classifier RF, BF, VM and ST 92.5% Parisi et al. [55] GA-D-LSVM RF, BF, VM, ST, MG, MH and LG 99.57% Zhao et al. [53] XGBoost RF, BF, VM and ST 96.09% This study HOSA- ResNet101 MG 96.3% h //d i /10 1371/j l 0302707 002 le 2. Comparison of the proposed method’s performance with other studies. PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 11 / 15 PLOS ONE Deep learning approach for detection knee pathology using EMG signal Grade 2, Grade 3, and Grade 4 KOA. Investigate sophisticated feature engineering methodolo- gies in order to extract more informative features from EMG signals that are capable of distin- guishing between various grades of KOA. A number of critical characteristics, including fatigue resistance, co-contraction, and muscle activation patterns, may differ between catego- ries and are therefore essential for precise classification. To mitigate the lack of diversity in the training dataset, employ data augmentation methods that are tailored to EMG signals, includ- ing amplitude scaling and time distortion. It is advisable to integrate deep learning with con- ventional machine learning methods, which have demonstrated potential in comprehending the evolution of KOA. A hybrid approach has the potential to capitalize on the respective mer- its of both methodologies in order to enhance the precision and dependability of diagnostics. Incorporate clinical insights and additional non-EMG data into the model, such as patient demographics, duration of symptoms, and physical examination findings. The utilization of this multimodal approach may facilitate the differentiation of KOA grades by offering a more comprehensive view of the disease. Conduct external validation on the model by employing distinct datasets that were not utilized during the phases of training or testing. Discussion Incorporate the knowledge and skills of radiologists, clinicians, and physiotherapists through collaborative efforts during the model development and validation phases. With their valuable insights, the model’s output can be enhanced to better suit clinical needs and requirements. Through the implementation of these strategies, a more resilient and dependable deep-learning model can be constructed to categorize KOA grades. Such a model would make a substantial contribution to the timely detection, surveillance, and individualized treatment of patients afflicted with KOA. Conclusion In conclusion, higher-order spectral analysis (HOSA) techniques were able to enhance feature extraction of the EMG signals from five knee muscles. Bicoherence representation of EMG sig- nals was fed to a modified deep learning network. The combination of bicoherence and adabted ResNet 101 enabled the achievement of robust metrics for accuracy, precision, and sensitivity. The 5 muscles were able to distinguish Knee OA patients from normal with very good accuracy. Interestingly, the medial gastrocnemius (MG) muscle has the highest predic- tion accuracy. A future research direction could involve expanding the dataset to include a larger number of subjects from diverse demographics, including varying ages, genders, and stages of KOA. This expansion would allow for a more comprehensive analysis of the deep learning model’s performance across a broader spectrum of the population and enhance the model’s generalizability. Additionally, analyzing the dataset for gender-specific patterns in EMG signals related to KOA could provide insights into personalized treatment strategies, potentially leading to more targeted and effective interventions for different patient groups. PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 References 1. Zhang YandJordan JM. Epidemiology of osteoarthritis. Clinics in Geriatric Medicine. 2010; 26(3):355– 69. https://doi.org/10.1016/j.cger.2010.03.001 PMID: 20699159 2. Vijayvargiya A, Singh PL, Verma SM, Kumar RandBansal S. Chapter 12—performance comparison analysis of different classifier for early detection of knee osteoarthritis. In: Dey N, Chaki J, Kumar R, edi- tors. Sensors for health monitoring. 5: Academic Press; 2019. p. 243–57. 3. Primorac D, Molnar V, Rod E, Jeleč Zˇ, Čukelj F, Matisˇić V, et al. Knee osteoarthritis: A review of patho- genesis and state-of-the-art non-operative therapeutic considerations. Genes (Basel). 2020; 11(8). https://doi.org/10.3390/genes11080854 PMID: 32722615 4. Segal NA, Torner JC, Felson D, Niu J, Sharma L, Lewis CE, et al. Effect of thigh strength on incident radiographic and symptomatic knee osteoarthritis in a longitudinal cohort. Arthritis Rheum. 2009; 61 (9):1210–7. https://doi.org/10.1002/art.24541 PMID: 19714608 5. Shamaei K, Cenciarini M, Adams AA, Gregorczyk KN, Schiffman JMandDollar AM. Design and evalua- tion of a quasi-passive knee exoskeleton for investigation of motor adaptation in lower extremity joints. IEEE Transactions on Biomedical Engineering. 2014; 61(6):1809–21. https://doi.org/10.1109/TBME. 2014.2307698 PMID: 24845291 6. Sowers MF, Hayes C, Jamadar D, Capul D, Lachance L, Jannausch M, et al. Magnetic resonance- detected subchondral bone marrow and cartilage defect characteristics associated with pain and x-ray- defined knee osteoarthritis. Osteoarthritis and Cartilage. 2003; 11(6):387–93. https://doi.org/10.1016/ s1063-4584(03)00080-3 PMID: 12801478 7. Bedson JandCroft PR. The discordance between clinical and radiographic knee osteoarthritis: A sys- tematic search and summary of the literature. BMC Musculoskeletal Disorders. 2008; 9(1):116. https:// doi.org/10.1186/1471-2474-9-116 PMID: 18764949 8. Chaisson C, Gale D, Gale E, Kazis L, Skinner KandFelson D. Detecting radiographic knee osteoarthri- tis: What combination of views is optimal? Rheumatology. 2000; 39(11):1218–21. https://doi.org/10. 1093/rheumatology/39.11.1218 PMID: 11085800 9. Langer JE, Meyer SJandDalinka MK. Imaging of the knee. Radiologic Clinics of North America. 1990; 28(5):975–90. PMID: 2202007 10. Lu SJ, Ul Hassan F, Vijayanathan SandGnanasegaran G. Radionuclide bone spect/ct in the evaluation of knee pain: Comparing two-phase bone scintigraphy, spect and spect/ct. Br J Radiol. 2018; 91 (1090):20180168. https://doi.org/10.1259/bjr.20180168 PMID: 29906234 11. Naik GR, Selvan SE, Arjunan SP, Acharyya A, Kumar DK, Ramanujam A, et al. An ica-ebm-based semg classifier for recognizing lower limb movements in individuals with and without knee pathology. IEEE Transactions on Neural Systems and Rehabilitation Engineering. 2018; 26(3):675–86. https://doi. org/10.1109/TNSRE.2018.2796070 PMID: 29522411 12. Hafer JF, Provenzano SG, Kern KL, Agresta CE, Grant JAandZernicke RF. Measuring markers of aging and knee osteoarthritis gait using inertial measurement units. Journal of Biomechanics. Author Contributions Conceptualization: Ateka Khader, Ala’a Zyout, Amjed Al Fahoum. Formal analysis: Ateka Khader, Ala’a Zyout, Amjed Al Fahoum. Methodology: Ateka Khader, Ala’a Zyout, Amjed Al Fahoum. Project administration: Ateka Khader, Amjed Al Fahoum. Software: Ateka Khader, Ala’a Zyout, Amjed Al Fahoum. Supervision: Ateka Khader, Amjed Al Fahoum. Validation: Ateka Khader, Ala’a Zyout, Amjed Al Fahoum. 12 / 15 PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 PLOS ONE Deep learning approach for detection knee pathology using EMG signal Visualization: Ateka Khader, Ala’a Zyout, Amjed Al Fahoum. Writing – original draft: Ateka Khader, Ala’a Zyout, Amjed Al Fahoum. Writing – review & editing: Ateka Khader, Ala’a Zyout, Amjed Al Fahoum. Visualization: Ateka Khader, Ala’a Zyout, Amjed Al Fahoum. Writing – original draft: Ateka Khader, Ala’a Zyout, Amjed Al Fahoum. Writing – original draft: Ateka Khader, Ala’a Zyout, Amjed Al Fahoum. Writing – review & editing: Ateka Khader, Ala’a Zyout, Amjed Al Fahoum. References https://doi.org/10.3390/bioengineering10020249 PMID: 36829743 25. Al Fahoum A, Zyout A, Alquran H and Abu-Qasmieh I. A novel multi-stage bispectral deep learning method for protein family classification. Computers, Materials & Continua. 2023; 76(1). 26. Al Fahoum A and Zyout A. Enhancing early detection of schizophrenia through multi-modal eeg analy- sis: A fusion of wavelet transform, reconstructed phase space, and deep learning neural networks. 5th International Conference on Advances in Signal Processing and Artificial Intelligence (ASPAI’ 2023); Spain2023. 27. Rajamohan HR, Wang T, Leung K, Chang G, Cho K, Kijowski R, et al. Prediction of total knee replace- ment using deep learning analysis of knee mri. Scientific Reports. 2023; 13(1):6922. https://doi.org/10. 1038/s41598-023-33934-1 PMID: 37117260 28. Ashinsky BG, Bouhrara M, Coletta CE, Lehallier B, Urish KL, Lin PC, et al. Predicting early symptomatic osteoarthritis in the human knee using machine learning classification of magnetic resonance images from the osteoarthritis initiative. J Orthop Res. 2017; 35(10):2243–50. https://doi.org/10.1002/jor.23519 PMID: 28084653 29. Ahmed SM and Mstafa RJ. Identifying severity grading of knee osteoarthritis from x-ray images using an efficient mixture of deep learning and machine learning models. Diagnostics. 2022; 12(12):2939. https://doi.org/10.3390/diagnostics12122939 PMID: 36552945 30. Arumugam SR, Balakrishna R, Rajeshram V, Gowr S, Karuppasamy SGandPremnath SP, editors. Pre- diction of severity of knee osteoarthritis on x-ray images using deep learning. 2022 IEEE North Karna- taka Subsection Flagship International Conference (NKCon); 2022 20–21 Nov. 2022. 31. de Dieu Uwisengeyimana JandIbrikci T. Diagnosing knee osteoarthritis using artificial neural networks and deep learning. Biomedical Statistics and Informatics. 2017; 2(3):95. 32. Bansal H, Chinagundi B, Rana PandKumar N. An ensemble machine learning technique for detection of abnormalities in knee movement sustainability. Sustainability. 2022; 14:13464. 33. Vijayvargiya A, Kumar R, Dey NandTavares JMRS, editors. Comparative analysis of machine learning techniques for the classification of knee abnormality. 2020 IEEE 5th International Conference on Com- puting Communication and Automation (ICCCA); 2020 30–31 Oct. 2020. 34. Vijayvargiya A, Singh B, Kumar R, Desai UandHemanth J. Hybrid deep learning approaches for semg signal-based lower limb activity recognition. Mathematical Problems in Engineering. 2022; 2022. 35. Khadra L, Al-Fahoum AandBinajjaj S, editors. A new quantitative analysis technique for cardiac arrhyth- mia using bispectrum and bicoherency. The 26th Annual International Conference of the IEEE Engi- neering in Medicine and Biology Society; 2004: IEEE. 36. Al-Fahoum A and Khadra L, editors. Combined bispectral and bicoherency approach for catastrophic arrhythmia classification. References 2020; 99:109567. https://doi.org/10.1016/j.jbiomech.2019.109567 PMID: 31916999 13. Staab W, Hottowitz R, Sohns C, Sohns JM, Gilbert F, Menke J, et al. Accelerometer and gyroscope based gait analysis using spectral analysis of patients with osteoarthritis of the knee. J Phys Ther Sci. 2014; 26(7):997–1002. https://doi.org/10.1589/jpts.26.997 PMID: 25140082 14. Raez MB, Hussain MSandMohd-Yasin F. Techniques of emg signal analysis: Detection, processing, classification and applications. Biol Proced Online. 2006; 8:11–35. https://doi.org/10.1251/bpo115 PMID: 16799694 15. Al-Ayyad M, Owida HA, De Fazio R, Al-Naami BandVisconti P. Electromyography monitoring systems in rehabilitation: A review of clinical applications, wearable devices and signal acquisition methodolo- gies. Electronics. 2023; 12(7):1520. 16. Al-Fahoum AS and Gharaibeh KH. Prediction of sagittal lower limb joints moments under dynamic con- dition: Feasibility of using emg and arma model identification techniques. International Journal of Exper- imental and Computational Biomechanics. 2014; 2(3):245–64. 13 / 15 PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 PLOS ONE Deep learning approach for detection knee pathology using EMG signal 17. Al-Fahoum A and Gharaibeh KH. Feasibility study for anfis and emg utilization in modeling prosthesis for trans-femoral cut rehabilitation and gait cycle restoration. Biomedical Engineering: Applications, Basis and Communications. 2015; 27(03):1550023. 18. Li G, Kozanek M, Hosseini A, Liu F, Van de Velde SKandRubash HE. New fluoroscopic imaging tech- nique for investigation of 6dof knee kinematics during treadmill gait. J Orthop Surg Res. 2009; 4:6. https://doi.org/10.1186/1749-799X-4-6 PMID: 19284658 19. Amano K, Li QandMa CB. Functional knee assessment with advanced imaging. Curr Rev Musculoske- let Med. 2016; 9(2):123–9. https://doi.org/10.1007/s12178-016-9340-0 PMID: 27052009 20. Ling SM, Conwit RA, Talbot L, Shermack M, Wood JE, Dredge EM, et al. Electromyographic patterns suggest changes in motor unit physiology associated with early osteoarthritis of the knee. Osteoarthritis and cartilage. 2007; 15(10):1134–40. https://doi.org/10.1016/j.joca.2007.03.024 PMID: 17543548 21. Al Amer HS, Sabbahi MA, Alrowayeh HN, Bryan WJandOlson SL. Electromyographic activity of quadri- ceps muscle during sit-to-stand in patients with unilateral knee osteoarthritis. BMC research notes. 2018; 11:1–6. 22. Boyer M, Bouyer L, Roy J-SandCampeau-Lecours A. Reducing noise, artifacts and interference in sin- gle-channel emg signals: A review. Sensors. 2023; 23(6):2927. https://doi.org/10.3390/s23062927 PMID: 36991639 23. Abu-Qasmieh I, Al Fahoum A, Alquran H and Zyout A. An innovative bispectral deep learning method for protein family classification. Computers, Materials & Continua. 2023; 75(2). 24. Al Fahoum AS, Abu Al-Haija AO and Alshraideh HA. Identification of coronary artery diseases using photoplethysmography signals and practical feature selection process. Bioengineering. 2023; 10 (2):249. References 2005 IEEE Engineering in Medicine and Biology 27th Annual Conference; 2006: IEEE. 37. Al-Fahoum A, Al-Fraihat A and Al-Araida A. Detection of cardiac ischaemia using bispectral analysis approach. Journal of medical engineering & technology. 2014; 38(6):311–6. https://doi.org/10.3109/ 03091902.2014.925983 PMID: 25050476 14 / 15 PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 PLOS ONE Deep learning approach for detection knee pathology using EMG signal 38. Roelker SA, Koehn RR, Caruthers EJ, Schmitt LC, Chaudhari AMWandSiston RA. Effects of age and knee osteoarthritis on the modular control of walking: A pilot study. PLoS One. 2021; 16(12):e0261862. https://doi.org/10.1371/journal.pone.0261862 PMID: 34965270 39. Martinek R, Ladrova M, Sidikova M, Jaros R, Behbehani K, Kahankova R, et al. Advanced bioelectrical signal processing methods: Past, present and future approach—part i: Cardiac signals. Sensors. 2021; 21(15):5186. https://doi.org/10.3390/s21155186 PMID: 34372424 40. Merletti R, Rainoldi AandFarina D. Surface electromyography for noninvasive characterization of mus- cle. Exerc Sport Sci Rev. 2001; 29(1):20–5. https://doi.org/10.1097/00003677-200101000-00005 PMID: 11210442 41. Nikias CLandMendel JM. Signal processing with higher-order spectra. IEEE Signal Processing Maga- zine. 1993; 10(3):10–37. 42. Martı´nez M, Antonino-Daviu JA, Ferna´ndez de Co´rdoba Castella´ PandConejero A. Higher-order spec- tral analysis of stray flux signals for faults detection in induction motors. Applied Mathematics and Non- linear Sciences. 2020; 5. 43. Sivagamasundari SAaS. Bispectrum and bicoherence analysis using hosa (higher order spectral analy- sis)–a review. European Journal of Advances in Engineering and Technology. 2018; 5(11):898–902. 44. Swami A, Mendel JMandNikias CL. Higher-order spectral analysis toolbox. The Mathworks Inc. 1998; 3:22–6. 45. Al Fahoum A and Zyout Aa. Early detection of neurological abnormalities using a combined phase space reconstruction and deep learning approach. Intelligence-Based Medicine. 2023 Jan 1; 8:100123. 46. Raghu S, Sriraam N, Temel Y, Rao SVandKubben PL. Eeg based multi-class seizure type classification using convolutional neural network and transfer learning. Neural Networks. 2020; 124:202–12. https:// doi.org/10.1016/j.neunet.2020.01.017 PMID: 32018158 47. He K, Zhang X, Ren SandSun J, editors. Deep residual learning for image recognition. Proceedings of the IEEE conference on computer vision and pattern recognition; 2016. 48. Akbari H, Sadiq MT, Rehman AU, Ghazvini M, Naqvi RA, Payan M, et al. Depression recognition based on the reconstruction of phase space of eeg signals and geometrical features. Applied Acoustics. 2021; 179:108078. 49. Khan A, Khan MA, Javed MY, Alhaisoni M, Tariq U, Kadry S, et al. Human gait recognition using deep learning and improved ant colony optimization. Computers, Materials & Continua. 2022; 70(2). 50. PLOS ONE \| https://doi.org/10.1371/journal.pone.0302707 May 7, 2024 References Bruno F, Barile A, Arrigoni F, Laporta A, Russo A, Carotti M, et al. Weight-bearing mri of the knee: A review of advantages and limits. Acta Biomed. 2018; 89(1-s):78–88. https://doi.org/10.23750/abm. v89i1-S.7011 PMID: 29350638 51. Sanchez JS O. Emg dataset in lower limb. UCI Machine Learning Repository. 2014. 52. Vijayvargiya A, Prakash C, Kumar R, Bansal Sand R.S. Tavares JM. Human knee abnormality detec- tion from imbalanced semg data. Biomedical Signal Processing and Control. 2021; 66:102406. 53. Zhao J, She J, Wang DandWang F. Extreme gradient boosting for surface electromyography classifica- tion on time-domain features. Journal of Advanced Computational Intelligence and Intelligent Informat- ics. 2022; 26(5):722–30. 54. Chen X, Chen J, Liang J, Li Y, Courtney CAandYang Y. Entropy-based surface electromyogram feature extraction for knee osteoarthritis classification. IEEE Access. 2019; 7:164144–51. 55. Parisi LandRaviChandran N. Evolutionary denoising-based machine learning for detecting knee disor- ders. Neural Processing Letters. 2020; 52(3):2565–81. 56. Rutherford DJ, Hubley-Kozey CL, Stanish WDandDunbar MJ. Neuromuscular alterations exist with knee osteoarthritis presence and severity despite walking velocity similarities. Clinical Biomechanics. 2011; 26(4):377–83. https://doi.org/10.1016/j.clinbiomech.2010.11.018 PMID: 21185628 57. Mills K, Hunt MA, Leigh RandFerber R. A systematic review and meta-analysis of lower limb neuromus- cular alterations associated with knee osteoarthritis during level walking. Clin Biomech (Bristol, Avon). 2013; 28(7):713–24. https://doi.org/10.1016/j.clinbiomech.2013.07.008 PMID: 23953330 15 / 15
https://openalex.org/W2087708257	http://bib.irb.hr/datoteka/678240.Plos_Biology_2013.pdf	English	null	NKT Cell-TCR Expression Activates Conventional T Cells in Vivo, but Is Largely Dispensable for Mature NKT Cell Biology	PLoS biology	2,013	cc-by	15,113	Abstract Abbreviations: a-GalCer, a-Galactosylceramide; DN, double-negative; Egr2, early growth response protein 2; GATA-3, GATA binding protein 3; ICOS, inducible T cell co-stimulator; NKT, natural killer T; PLZF, promyelocytic leukemia zinc finger; ROR-yt, Retinoic acid-related Orphan Receptor Gamma; SLAMf, SLAM family; T bet, T-box expressed in T cells; tg, transgenic; Th-POK, T-helper-inducing POZ/Kru¨ppel-like factor; Va14i, Va14-Ja18. * E-mail: supprian@biochem.mpg.de ¤a Current address: Institut fu¨r Klinische Chemie und Pathobiochemie, Klinikum rechts der Isar, Technische Universita¨t Mu¨nchen, Munich, Germany. ¤b Current address: Proteomics and Signal Transduction, Max Planck Institute of Biochemistry, Martinsried, Germany. positive thymocytes. Although several good candidates have been put forward [8–10], the exact nature of the selecting glycolipids remains controversial. Homotypic interactions involving the SLAM family (SLAMf) receptors 1 and 6 are additionally required for NKT cell differentiation [11]. Auto-reactive activation during thymic selection is thought to induce a substantially stronger TCR stimulus in comparison to that during the development of conventional T cells [12,13]. As a consequence, expression of the transcription factors Egr1 and Egr2 is strongly increased [13], which in turn directly induce PLZF, the key transcription factor controlling NKT cell differentiation, migration, and functions [13]. NKT Cell-TCR Expression Activates Conventional T Cells in Vivo, but Is Largely Dispensable for Mature NKT Cell Biology J. Christoph Vahl1, Klaus Heger1, Nathalie Knies1¤a, Marco Y. Hein1¤b, Louis Boon2, Hideo Yagita3, Bojan Polic4, Marc Schmidt-Supprian1* 1 Molecular Immunology and Signaltransduction, Max Planck Institute of Biochemistry, Martinsried, Germany, 2 Bioceros, Yalelaan 46, Utrecht, The Netherlands, 3 Juntendo University School of Medicine, Tokyo, Japan, 4 University of Rijeka School of Medicine, Rijeka, Croatia Abstract Natural killer T (NKT) cell development depends on recognition of self-glycolipids via their semi-invariant Va14i-TCR. However, to what extent TCR-mediated signals determine identity and function of mature NKT cells remains incompletely understood. To address this issue, we developed a mouse strain allowing conditional Va14i-TCR expression from within the endogenous Tcra locus. We demonstrate that naı¨ve T cells are activated upon replacement of their endogenous TCR repertoire with Va14i-restricted TCRs, but they do not differentiate into NKT cells. On the other hand, induced TCR ablation on mature NKT cells did not affect their lineage identity, homeostasis, or innate rapid cytokine secretion abilities. We therefore propose that peripheral NKT cells become unresponsive to and thus are independent of their autoreactive TCR. Citation: Vahl JC, Heger K, Knies N, Hein MY, Boon L, et al. (2013) NKT Cell-TCR Expression Activates Conventional T Cells in Vivo, but Is Largely Dispensable for Mature NKT Cell Biology. PLoS Biol 11(6): e1001589. doi:10.1371/journal.pbio.1001589 Academic Editor: Philippa Marrack, National Jewish Medical and Research Center/Howard Hughes Medical Institute, United States of Americ Received December 14, 2012; Accepted May 7, 2013; Published June 18, 2013 Copyright: 2013 Vahl et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Copyright: 2013 Vahl et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This study was supported by the DFG through an Emmy Noether grant and SFB 1054 to MS-S. The Va14iStopF mice were generated with support from a Sandler Foundation for Asthma Research grant to Klaus Rajewsky. JCV and KH received PhD stipends from the Ernst Schering Foundation and the Boehringer Ingelheim Fonds, respectively. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org Results Correct Timing and Endogenous Control of Va14i-TCR Expression Produces Large Numbers of Bona Fide NKT Cells In order to produce large numbers of NKT cells in a physiological manner and to manipulate the expression of the semi-invariant Va14i-TCR in a conditional fashion, we generated Va14iStopF knock-in mice. To this end we cloned a productive Va14-Ja18 rearrangement, including the Va14 leader exon, intron and 1.8 kb of upstream regulatory sequence, and 0.2 kb intronic sequence downstream of Ja18. These elements were inserted by homologous recombination 39 of Ja1 upstream of the Ca constant region of the Tcra locus (Figure 1A). Expression of putative upstream rearrangements is aborted by four SV40 polyA sites at the 59 end of the construct, and expression of Va14i is rendered conditional through a loxP-flanked STOP cassette. We obtained over 80% (271 of 325) homologous recombinant ES cell clones during gene targeting, indicating an unusually high targeting efficiency of our construct (Figure S1A). The development of conventional T and NKT cells, identified by staining with mouse CD1d-PBS57-tetramers (tetramer+), occurs unperturbed in Va14iStopF/wt heterozygous mice. In homozygous Va14iStopF/F mice, T cell development is abolished due to transcriptional termination of TCRa expression before the Ca exons (Figure 1B). We bred Va14iStopF to CD4-Cre mice, in order to express the inserted Va14i-chain in double-positive thymocytes, mimicking the physiological timing of TCRa-chain rearrangement and expression [26,27]. On average 23 times more thymic and 43 times more splenic NKT cells were generated in these, compared to wild-type mice (Figures 1B and 2A–E). Around 9% of the tetramer+ T cells in CD4-Cre Va14iStopF/wt mice expressed the CD8 co-receptor (over 80% as CD8ab heterodimer; Figures 1C and S1B,C), which is also expressed by some human NKT cells, but normally not in mice [28]. The proportions of CD42 CD82 double negative (DN) and CD4+ cells were comparable between transgenic (tg) and wild-type NKT cells (Figure 1C). Furthermore, the tgNKT cells were largely comparable to wild-type NKT cells with respect to Vb-chain bias (Figure 1D) and surface phenotype (Figure 1E). Finally, we found that NKT cells from CD4-Cre Va14iStopF/wt animals expressed the critical transcription factors promyelocytic leukemia zinc finger (PLZF), GATA binding protein 3 (GATA-3), and T-helper-inducing POZ/Kru¨ppel-like factor (Th-POK) (Figure 1F) [28,29]. Interestingly, we also detected a substantial proportion of the recently described NKT17 subset in the transgenic animals. Results These DN NK1.12 NKT cells express the transcription factor ROR-ct and were shown to produce the cytokine IL-17 upon activation (Figure 1F) [29,30]. ways such as toll-like receptor (TLR) signals. Lipids derived from different bacteria [16–19] were shown to directly activate mouse and human NKT cells in a TLR- and IL-12-independent manner, and NKT cells are required for productive immune responses against these pathogens. NKT cells can also be activated indirectly through cytokines such as IL-12, IL-18, or type I interferons (IFNs) [20]. However, it remains controversial whether, depending on the strength of the cytokine signal, weak responses to self-antigens presented by CD1d are an additional obligate requirement. In one study, CD1d-dependent signals were found to be necessary for full NKT cell activation in response to all tested pathogens [20]. In contrast, others reported that IL-12-dependent NKT cell activa- tion after LPS injection [21] or MCMV infection [22] is independent of either foreign or self-glycolipid antigen presenta- tion by CD1d. Upon activation, the most distinguishing feature of NKT cells is their ability to rapidly produce and secrete large amounts of cytokines (Th1 and Th2 cytokines, among others). Their fast, effector-like response could be based on steady-state expression of cytokine mRNA in mice [23,24] that was suggested to be a consequence of tonic self-reactive activation [2]. Recently, it was reported that human NKT cells do not constitutively express cytokine mRNAs. Instead, rapid cytokine-induced innate IFNc production by NKT cells was suggested to rely on obligate continuous recognition of self-lipids, which retains histone acetylation patterns at the IFNG locus that favor transcription [25]. Another characteristic feature of NKT cells, their surface marker expression reminiscent of memory or recently activated T cells, was also connected to their inherent autoreactivity [2]. Author Summary Author Summary Immune system natural killer T (NKT) cells help to protect against certain strains of bacteria and viruses, and suppress the development of autoimmune diseases and cancer. However, NKT cells are also central mediators of allergic responses. The recognition of one’s own glycolipid antigens (self-glycolipids) in the thymus via the unique Va14i T cell receptor, Va14i-TCR, triggers the NKT cell developmental program, which differs considerably from that of conventional T cells. We generated a mouse model to investigate whether the Va14i-TCR on mature NKT cells constantly recognizes self-glycolipids and to assess wheth- er this TCR is required for survival and continued NKT cell identity. Switching the peptide-recognizing TCR of a mature conventional T cell to a glycolipid-recognizing Va14i-TCR led to activation of the T cells, indicating that this TCR is also autoreactive on peripheral T cells or can signal autonomously. But TCR ablation did not affect the half-life, characteristic gene expression or innate functions of mature NKT cells. Therefore, the inherently autoreactive Va14i-TCR is dispensable for the functions of mature peripheral NKT cells after instructing thymic NKT cell development. Thus the Va14i-TCR serves a similar function to pattern-recognition receptors, in mediating immune recognition of foreign invasion or diseased cells. Introduction Natural Killer T (NKT) cells represent a subset of T cells in mice and humans that express NK cell markers and recognize a small class of glycolipid (auto-) antigens [1,2]. Most mouse NKT cells express an invariant Va14-Ja18 (Va14i) TCRa rearrange- ment (Va24-Ja18 in humans). In principle, all TCRb-chains are able to pair with this Va14i-TCR chain [3]. However, the selection of NKT cells by endogenous glycolipids presented by the monomorphic MHC class I-like CD1d induces a strong bias towards TCRs containing Vb8, Vb7, or Vb2 [1,3], which is abrogated in the absence of selection [3,4]. Recently, crystallo- graphic analysis demonstrated a conserved binding mode of the NKT cell TCR to various glycolipids, where only germline- encoded residues were in direct antigen contact, reminiscent of innate pattern-recognition receptors [5]. Moreover, several observations suggest that this receptor is inherently auto-reactive [1,2] and thereby determines NKT cell identity and influences their function. The expression of several inhibitory NK cell receptors on NKT cells was suggested to control their self- reactivity and avoid autoimmune activation [6,7]. Interestingly, the homeostatic proliferation of NKT cells after adoptive transfer was similar in CD1d-deficient and wild-type mice, indicating that this process is mostly cytokine-driven and does not depend on continued TCR-mediated self-lipid-recogni- tion [14,15]. However, as the transferred cells contained CD1d, a role for antigen could not be completely excluded. In addition, tonic antigen-independent TCR signals might contribute to NKT cell maintenance and phenotype. During immune responses, NKT cell activation depends mostly on two parameters: engagement of the TCR and the presence of proinflammatory cytokines released from antigen-presenting cells activated by innate immune path- During development in the thymus, the few T cells expressing a Va14i-TCR are selected upon recognition of self-lipids on double- June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org 1 Investigating NKT Cell-TCR Signals required for NKT cell homeostasis, lineage identity, and rapid cytokine secretion. June 2013 \| Volume 11 \| Issue 6 \| e1001589 Timing of Transgenic Va14i-TCR Expression Is Critical for Normal NKT Cell Development Premature TCRa expression leads to aberrant T cell development in transgenic mouse models [26,27]. To directly compare the consequence of premature to CD4-Cre-mediated timely Va14i-TCRa-chain expression in our knock-in approach, we bred our mice to a germline Cre-deleter strain (Nestin-Cre) [31]. Compared to CD4-Cre-induced Va14i-TCRa-chain expression, premature expression in Cre-deleter Va14iStopF/wt led to signifi- cantly reduced numbers of NKT cells in thymus and spleen, especially of CD4+ NKT cells (Figure 2A–C). In addition, we To thoroughly address the open questions regarding the nature and importance of TCR signaling for NKT cells, we generated a novel mouse model that allowed us to study the extent of Va14i- TCR-mediated auto-antigen recognition in the periphery and its relevance for NKT cell identity. Furthermore, we monitored the fate of NKT cells after TCR ablation. Our results prove the inherent self-reactivity of the NKT cell TCR and demonstrate that although essential for positive selection, tonic TCR signaling is not June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org 2 Figure 1. The Va14i-TCR knock-in mouse produces large numbers of correctly selected, bona fide NKT cells. (A) Schematic representation of the knock-in transgene. The Va14 promoter, loxP (triangle)-flanked STOP cassette, and pre-rearranged Va14i (Va14-Ja18, red square) sequences were inserted 39 of Ja1 and 59 of the first Ca exon (coding exons are highlighted in blue); 4pA = 4 SV40 polyadenylation sites. AH, arms of homology. Ea, enhancer (black oval). (B) Representative proportions of NKT cells and conventional T cells of total lymphocytes in thymus and spleen. Numbers indicate mean percentages 6 SD of at least seven age-matched mice per genotype. (C) Representative proportions of splenic CD4+, CD8+, and DN (CD42 CD82) NKT cells. Numbers indicate mean percentages 6 SD of seven mice per genotype. (D) The Vb repertoires of splenic NKT cells of the indicated genotypes. Bars indicate means and error bars SD of three independent experiments. (E) Representative flow cytometric analysis of the indicated cell-surface proteins on conventional CD4+ T cells and NKT cells. (F) Intracellular flow cytometric staining of PLZF, GATA-3, ROR-ct, and Th- POK in the depicted NKT cells. Numbers indicate means of the median fluorescence intensities (MFIs), normalized to CD4+ tetramer2 T cells of CTR animals, or percentage of ROR-ct+ cells among DN NKT cells; calculated from three animals per genotype. Histograms are representative of three independent experiments with eight mice in total. Timing of Transgenic Va14i-TCR Expression Is Critical for Normal NKT Cell Development Throughout the figure, NKT cells were gated as tetramer+ TCRb+, conventional (conv) T cells as tetramer2 TCRb+; CTR, CD4-Cre or Va14iStopF/wt. doi:10.1371/journal.pbio.1001589.g001 Investigating NKT Cell-TCR Signals Investigating NKT Cell-TCR Signals Figure 1. The Va14i-TCR knock-in mouse produces large numbers of correctly selected, bona fide NKT cells. (A) Schematic representation of the knock-in transgene. The Va14 promoter, loxP (triangle)-flanked STOP cassette, and pre-rearranged Va14i (Va14-Ja18, red square) sequences were inserted 39 of Ja1 and 59 of the first Ca exon (coding exons are highlighted in blue); 4pA = 4 SV40 polyadenylation sites. AH, arms of homology. Ea, enhancer (black oval). (B) Representative proportions of NKT cells and conventional T cells of total lymphocytes in thymus and spleen. Numbers indicate mean percentages 6 SD of at least seven age-matched mice per genotype. (C) Representative proportions of splenic CD4+, CD8+, and DN (CD42 CD82) NKT cells. Numbers indicate mean percentages 6 SD of seven mice per genotype. (D) The Vb repertoires of splenic NKT cells of the indicated genotypes. Bars indicate means and error bars SD of three independent experiments. (E) Representative flow cytometric analysis of the indicated cell-surface proteins on conventional CD4+ T cells and NKT cells. (F) Intracellular flow cytometric staining of PLZF, GATA-3, ROR-ct, and Th- POK in the depicted NKT cells. Numbers indicate means of the median fluorescence intensities (MFIs), normalized to CD4+ tetramer2 T cells of CTR animals, or percentage of ROR-ct+ cells among DN NKT cells; calculated from three animals per genotype. Histograms are representative of three independent experiments with eight mice in total. Throughout the figure, NKT cells were gated as tetramer+ TCRb+, conventional (conv) T cells as tetramer2 TCRb+; CTR, CD4-Cre or Va14iStopF/wt. doi:10.1371/journal.pbio.1001589.g001 perturbed positive selection (Figure 2F). CD4-Cre Va14iStopF/wt mice produce more NKT cells than any of the previously reported models, including mice with a Va14i allele derived from a NKT cell nuclear transplantation experiment [11,32–35]. A comparison of different Va14i-transgenic models demonstrates that both the correct timing and endogenous control of TCR expression control favor NKT cell development (Table S1). Our analyses therefore showed that physiological timing of Va14i- TCRa-expression at endogenous levels in CD4-Cre Va14iStopF/wt found reduced thymocyte counts and a significant increase of most likely lineage-‘‘confused’’ DN (CD42 CD82) tetramer- negative T cells (Figure 2D,E). In fact Cre-deleter Va14iStopF/wt mice strongly resemble the ‘‘first generation’’ Va11 promoter- driven (Va11p) Va14i transgenic mice in these respects (Table S1) [32]. June 2013 \| Volume 11 \| Issue 6 \| e1001589 Timing of Transgenic Va14i-TCR Expression Is Critical for Normal NKT Cell Development Moreover, in Cre-deleter Va14iStopF/wt mice, we observed increased proportions of Vb9-, Vb10-, and Vb14-containing Va14i-TCRs, which can recognize a-GalCer-loaded tetramers, but most likely not endogenous self-glycolipids [3,4], pointing to June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org PLOS Biology \| www.plosbiology.org 3 Investigating NKT Cell-TCR Signals Figure 2. Premature Va14i-TCR expression impairs NKT and conventional T cell development. (A–E) Absolute cell numbers in thymus and spleen of 7–13 mice of the indicated genotypes: NKT cells (A), CD4+ NKT cells (B), DN NKT cells (C), total cells (D), and DN tetramer2 T cells (E). Bars indicate medians. * p,0.001; p,0.01; * p,0.05; ns, not significant; one-way ANOVA. (A) Mean cell numbers are depicted below the scatter blot. (F) The Vb repertoires of splenic NKT cells of the depicted animals. Data for CD4-Cre Va14iStopF/wt are the same as shown in Figure 1D. Bars indicate means and error bars SD of 3–4 mice per genotype of 3–4 independent experiments. Throughout the figure, NKT cells were gated as tetramer+ TCRb+, conventional (conv) T cells as tetramer2 TCRb+; CTR, CD4-Cre or Va14iStopF/wt. doi:10.1371/journal.pbio.1001589.g002 Figure 2. Premature Va14i-TCR expression impairs NKT and conventional T cell development. (A–E) Absolute cell numbers in thymus and spleen of 7–13 mice of the indicated genotypes: NKT cells (A), CD4+ NKT cells (B), DN NKT cells (C), total cells (D), and DN tetramer2 T cells (E). Bars indicate medians. * p,0.001; p,0.01; * p,0.05; ns, not significant; one-way ANOVA. (A) Mean cell numbers are depicted below the scatter blot. (F) The Vb repertoires of splenic NKT cells of the depicted animals. Data for CD4-Cre Va14iStopF/wt are the same as shown in Figure 1D. Bars indicate means and error bars SD of 3–4 mice per genotype of 3–4 independent experiments. Throughout the figure, NKT cells were gated as tetramer+ TCRb+, conventional (conv) T cells as tetramer2 TCRb+; CTR, CD4-Cre or Va14iStopF/wt. doi:10.1371/journal.pbio.1001589.g002 Figure 2. Premature Va14i-TCR expression impairs NKT and conventional T cell development. (A–E) Absolute cell numbers in thymus and spleen of 7–13 mice of the indicated genotypes: NKT cells (A), CD4+ NKT cells (B), DN NKT cells (C), total cells (D), and DN tetramer2 T cells (E). Bars indicate medians. * p,0.001; p,0.01; * p,0.05; ns, not significant; one-way ANOVA. (A) Mean cell numbers are depicted below the scatter blot. Timing of Transgenic Va14i-TCR Expression Is Critical for Normal NKT Cell Development (F) The Vb repertoires of splenic NKT cells of the depicted animals. Data for CD4-Cre Va14iStopF/wt are the same as shown in Figure 1D. Bars indicate means and error bars SD of 3–4 mice per genotype of 3–4 independent experiments. Throughout the figure, NKT cells were gated as tetramer+ TCRb+, conventional (conv) T cells as tetramer2 TCRb+; CTR, CD4-Cre or Va14iStopF/wt. doi:10.1371/journal.pbio.1001589.g002 mice contributes to the production of large numbers of correctly selected, bona fide NKT cells. and T-bet-expressing NKT cells in CD4-Cre Va14iStopF/wt com- pared to wild-type mice (Figure 3D). The expression of both CD69 and T-bet strongly correlated with NK1.1 surface levels (Figure S1E,F). This also explains the higher intracellular PLZF expression in CD4+ and DN NKT cells of CD4-Cre Va14iStopF/wt animals in comparison to control animals (Figure 1F), as it was shown that PLZF expression is downregulated during NKT cell development [36]. Reduced maturation seems to be a common feature in mice with overabundance of NKT cells (Figure S1G and Table S1) [33]. Indeed, a comparison of different Va14i-tg mice suggests that independently of the total number of NKT cells generated, the size of the homeostatic niche for mature NKT cells appears to be around two million cells (Table S1). NKT Cell Maturation in CD4-Cre Va14iStopF/wt Animals June 2013 \| Volume 11 \| Issue 6 \| e1001589 NKT Cell Maturation in CD4-Cre Va14iStopF/wt Animals To test the functionality of our transgenic NKT cells, we injected CD4-Cre Va14iStopF/wt mice with the NKT cell ligand a- Galactosylceramide (a-GalCer) and determined their cytokine production directly ex vivo. The transgenic NKT cells were able to mount a rapid and robust cytokine response. Although a reduced proportion of transgenic NKT cells responded, in absolute cell numbers there was a 6–10-fold increase compared to wild-type NKT cells (Figure 3A). We did not observe significant steady-state cytokine production by transgenic or control NKT cells, and we detected only minor increases in cytokine levels in the serum of some of these mice (Figure S1D). Since cytokine production also varies with NKT cell maturation, we analyzed NKT cell development in CD4-Cre Va14iStopF/wt mice in more detail. This revealed a strong bias toward immature fractions in the thymus, due to the dramatic increase in NKT cell progenitors. In the periphery, 20% of NKT cells fully matured, as judged by the expression of NK1.1 and other NK cell markers (Figure 3B,C). This view is further supported by the reduced proportion of CD69 IL-15 is critical for the final maturation of NKT cells [37] and together with IL-7 required for their peripheral maintenance [14,38]. NKT cells compete with NK cells for these resources [38]. The halved number of NK cells in CD4-Cre Va14iStopF/wt mice (Figure 3E) suggests that the availability of these and maybe other cytokines might be insufficient due to the dramatically increased NKT cell numbers. The fact that a similar effect was observed in Va11p-Va14itg mice (Figure 3E) underscores this notion. These results let us conclude that while large amounts of NKT cells can June 2013 \| Volume 11 \| Issue 6 \| e1001589 June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org 4 Investigating NKT Cell-TCR Signals Figure 3. NKT cell overproduction affects their maturation and NK cell homeostasis. (A) Intracellular IL-4, IL-13, IFN-c, and TNF expression of splenic CD4+ NKT cells isolated from the depicted animals 90 min after aGalCer injection. Cells were stained directly ex vivo without addition of brefeldin or monensin. Black numbers indicate mean percentages 6 SD, and red numbers indicate mean total NKT cell counts expressing the respective cytokine. Data are from three animals per genotype; FSC, forward scatter. (B) Representative proportions of stage 1 (CD44low NK1.1low), stage 2 (CD44high NK1.1low), and stage 3 (CD44high NK1.1high) thymic and splenic NKT cells. NKT Cell Maturation in CD4-Cre Va14iStopF/wt Animals Numbers indicate mean percentages 6 SD of 10 mice per genotype. (C) Flow cytometric analysis of the depicted markers on thymic and splenic, transgenic, and control NKT cells. Bars indicate means and error bars SD calculated from 4–7 mice. (D) Extracellular and intracellular flow cytometric stainings of CD69 and T-bet in the depicted NKT cell subpopulations. Numbers in representative histogram indicate percentage of CD69high or T-bet+ cells among the indicated NKT cells calculated from eight animals per genotype (CD69) or three animals per genotype (T-bet). Histograms are representative of at least three independent experiments with each at least seven mice in total. (E) Absolute splenic NK cell numbers (NK1.1+ TCRb2 tetramer–) of age-matched 6–12-wk-old animals (7–16 per genotype). Bars indicate medians. * p,0.001; ns, not significant; one-way ANOVA. Throughout the figure, NKT cells were gated as tetramer+ TCRb+, conventional (conv) T cells as tetramer2 TCRb+; CTR, CD4-Cre or Va14iStopF/wt. doi:10.1371/journal.pbio.1001589.g003 Figure 3. NKT cell overproduction affects their maturation and NK cell homeostasis. (A) Intracellular IL-4, IL-13, IFN-c, and TNF expression of splenic CD4+ NKT cells isolated from the depicted animals 90 min after aGalCer injection. Cells were stained directly ex vivo without addition of brefeldin or monensin. Black numbers indicate mean percentages 6 SD, and red numbers indicate mean total NKT cell counts expressing the respective cytokine. Data are from three animals per genotype; FSC, forward scatter. (B) Representative proportions of stage 1 (CD44low NK1.1low), stage 2 (CD44high NK1.1low), and stage 3 (CD44high NK1.1high) thymic and splenic NKT cells. Numbers indicate mean percentages 6 SD of 10 mice per genotype. (C) Flow cytometric analysis of the depicted markers on thymic and splenic, transgenic, and control NKT cells. Bars indicate means and error bars SD calculated from 4–7 mice. (D) Extracellular and intracellular flow cytometric stainings of CD69 and T-bet in the depicted NKT cell subpopulations. Numbers in representative histogram indicate percentage of CD69high or T-bet+ cells among the indicated NKT cells calculated from eight animals per genotype (CD69) or three animals per genotype (T-bet). Histograms are representative of at least three independent experiments with each at least seven mice in total. (E) Absolute splenic NK cell numbers (NK1.1+ TCRb2 tetramer–) of age-matched 6–12-wk-old animals (7–16 per genotype). Bars indicate medians. * p,0.001; ns, not significant; one-way ANOVA. Throughout the figure, NKT cells were gated as tetramer+ TCRb+, conventional (conv) T cells as tetramer2 TCRb+; CTR, CD4-Cre or Va14iStopF/wt. June 2013 \| Volume 11 \| Issue 6 \| e1001589 NKT Cell Maturation in CD4-Cre Va14iStopF/wt Animals doi:10.1371/journal.pbio.1001589.g003 be produced in mice, depending on the mode of Va14i expression, the number of fully mature NKT cells is restricted by homeostatic constraints, some of which are shared with NK cells. this question and to study NKT cell TCR-autoreactivity in the periphery, we investigated the consequences of Va14i-TCR signals for conventional naı¨ve T cells. We wondered whether Va14i-TCR expression on naı¨ve T cells, lacking inhibitory receptors and generally a NKT cell ‘‘identity’’, would lead to activation upon (self-)lipid recognition and what cellular fate(s) are elicited by such activation. The Exchange of the Endogenous TCR Repertoire for a Va14i-Restricted One on Mature Conventional T Cells Leads to a Significant Population of Tetramer+ T Cells To this end, we generated mice enabling us to exchange the endogenous TCR-repertoire present on naı¨ve peripheral T cells for a Va14i-restricted TCR repertoire. The induction of Cre expression in Mx-Cre CaF/Va14iStopF mice inactivates the CaF allele and simultaneously turns on the Va14iStopF allele, leading to substitution of endogenous TCRa-chains with the Va14i TCRa- The strong self-lipid-induced TCR stimulus that early NKT cell progenitors receive in the thymus can be visualized through high GFP expression under the control of the Nur77 gene locus, reporting TCR signal strength [12]. However, the subsequent loss of GFP in mature NKT cells suggests that these cells are either not exposed to or not responsive to self-antigens. In order to answer June 2013 \| Volume 11 \| Issue 6 \| e1001589 June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org 5 Investigating NKT Cell-TCR Signals Investigating NKT Cell-TCR Signals Figure 4. TCR switch on mature conventional T cells. (A) Genetic set-up of the TCR switch experiment. In Mx-Cre CaF/Va14iStopF mice, the endogenous TCRa-chains (Va(x)Ja(y)) are exclusively expressed from the CaF allele. Cre-mediated recombination leads to termination of expression from the CaF allele, and simultaneous start of expression of the Va14i-TCRa-chain from the Va14iStopF allele. (B) T-cell-deficient mice were reconstituted with NKT cell-depleted splenocytes of the indicated genotypes. After 2 wk, the TCR switch was induced by poly(I:C) injection. Eight weeks later, percentages of tetramer+ and tetramer2 T cells (TCRb+) were analyzed in spleen and liver. Black numbers indicate percentages of total lymphocytes, red numbers absolute cell number calculated from 9–17 animals. NKT Cell Maturation in CD4-Cre Va14iStopF/wt Animals (C) Bars indicate means and SD (error bars) of CD4+, CD8+, or DN (CD42 CD82) cells among tetramer2 and tetramer+ T cells, calculated from at least nine mice per genotype. (D, E) The Vb repertoires of the depicted splenic CD4+ (D) or CD8+ (E) T cell subsets isolated from T-cell-deficient animals that received NKT cell-depleted Mx-Cre CaF/Va14iStopF splenocytes. Some of these mice were injected with poly(I:C) 2 wk later to induce the TCR switch. Eight weeks after poly(I:C) injection, the Vb repertoires were analyzed. Data represent means and SD (error bars) of two independent experiments with a total of three mice (tetramer2 without poly(I:C) injection) or eight mice (poly(I:C) injected) per T cell population. Vbs typical for glycolipid selection of NKT cells are highlighted in red. doi:10.1371/journal.pbio.1001589.g004 Figure 4. TCR switch on mature conventional T cells. (A) Genetic set-up of the TCR switch experiment. In Mx-Cre CaF/Va14iStopF mice, the endogenous TCRa-chains (Va(x)Ja(y)) are exclusively expressed from the CaF allele. Cre-mediated recombination leads to termination of expression from the CaF allele, and simultaneous start of expression of the Va14i-TCRa-chain from the Va14iStopF allele. (B) T-cell-deficient mice were reconstituted with NKT cell-depleted splenocytes of the indicated genotypes. After 2 wk, the TCR switch was induced by poly(I:C) injection. Eight weeks later, percentages of tetramer+ and tetramer2 T cells (TCRb+) were analyzed in spleen and liver. Black numbers indicate percentages of total lymphocytes, red numbers absolute cell number calculated from 9–17 animals. (C) Bars indicate means and SD (error bars) of CD4+, CD8+, or DN (CD42 CD82) cells among tetramer2 and tetramer+ T cells, calculated from at least nine mice per genotype. (D, E) The Vb repertoires of the depicted splenic CD4+ (D) or CD8+ (E) T cell subsets isolated from T-cell-deficient animals that received NKT cell-depleted Mx-Cre CaF/Va14iStopF splenocytes. Some of these mice were injected with poly(I:C) 2 wk later to induce the TCR switch. Eight weeks after poly(I:C) injection, the Vb repertoires were analyzed. Data represent means and SD (error bars) of two independent experiments with a total of three mice (tetramer2 without poly(I:C) injection) or eight mice (poly(I:C) injected) per T cell population. Vbs typical for glycolipid selection of NKT cells are highlighted in red. doi:10.1371/journal.pbio.1001589.g004 (unpublished data). To definitely exclude any effect of poly(I:C) injection on our results, we waited 2–4 mo before analyzing the animals after the induced TCR switch. chain (Figure 4A). NKT Cell Maturation in CD4-Cre Va14iStopF/wt Animals As mentioned above, the Va14i-chain can pair with all TCRb-chains [3], although only Vb2-, Vb7-, and Vb8- containing Va14i-TCRs can recognize endogenous lipids such as iGb3 [3,4]. Since TCRs containing one of these Vb-chains constitute approximately 30% of the CD4+ and CD8+ peripheral T cell pool (Figure 1D and unpublished data), we predicted that our genetic switch experiment should generate sufficient numbers of T cells able to recognize self-lipids. We found significant numbers of tetramer+ CD4+ and CD8+ T cells as a result of this switch experiment (Figure 4B–E). ‘‘Unloaded’’ tetramers did not stain these cells, demonstrating that they were not reactive against CD1d itself (Figure S2B). The TCR-switched tetramer+ T cells were predominantly enriched in cells expressing Vb-chains that are associated with high avidity auto-antigen binding: Vb2, Vb8.1/8.2, and Vb7 (Figure 4D,E) [3,4,42]. The exceptions were CD8+ TCR-switched tetramer+ T cells, in which Vb7-expressing cells were not enriched. The bias toward tetramer+ CD8+ T cells (Figure 4C) is most likely due to more efficient Mx-Cre-mediated recombination in these cells [40]. In Mx-Cre transgenic mice, Cre expression can be induced through injection of dsRNA, such as poly(I:C) [39]. However, low- level ‘‘leaky’’ recombination occurs also in absence of an inducer [39,40], leading to increased numbers of tetramer+ T cells in naive Mx-Cre CaF/Va14iStopF mice (Figure S2A). Therefore, splenocytes were depleted of tetramer+ T cells by magnetic cell separation (MACS, Figure S2A), and 206106 purified cells were injected intravenously (i.v.) into recipient animals lacking conventional ab T cells and NKT cells (Ca2/2 or Va14iStopF/F). After cells were allowed to engraft for 2 wk, the TCR switch was induced by poly(I:C) injection. Importantly, except for a short-term activation of the immune system, poly(I:C) injection in Mx-Cre mice per se has no significant long-lasting effect on peripheral conventional T cells [40,41] or on the number and phenotype of NKT cells Sterile Inflammation in Mice Containing TCR-Switched T Cells Sterile Inflammation in Mice Containing TCR-Switched T Cells Investigating NKT Cell-TCR Signals Investigating NKT Cell-TCR Signals Investigating NKT Cell-TCR Signals DCs in some cases, suggested auto-antigen-mediated activation of TCR-switched cells. To verify that the newly assembled Va14i- TCR on conventional T cells is functional, we injected recipients of Mx-Cre CaF/Va14iStopF and control cells with a-GalCer or PBS 2 mo after switch induction. Ninety minutes after a-GalCer, but not PBS, injection, CD4+ and CD8+ tetramer+ T cells produced IFN-c and TNF (Figure 6A), demonstrating the functionality of the newly assembled Va14i-TCR. In comparison to NKT cells from wild-type or CD4-Cre Va14iStopF/wt animals, a smaller proportion of tetramer+ T cells produced cytokines (Figures 6A and S2C). Tetramer+ TCR-switched T cells could also be activated in vitro through a-GalCer-pulsed A20 cells overexpress- ing CD1d (unpublished data) [43]. elevated serum TNF in more than half of these mice (Figure 5F). Elevated levels of other cytokines, such as IL-2, IL-4, IL-5, IL-6, IL-10, IL-17, and IFN-c, were not found in the sera of these mice (unpublished data). Interestingly, we found that 6 (highlighted in red throughout the figure) of 17 spleens containing TCR-switched T cells were almost completely devoid of B cells (Figure 5G) as well as dendritic cells (DCs, Figure 5H), which present lipid antigens to NKT cells via CD1d [1]. Furthermore, tetramer- ‘‘conventional’’ T cells were also strongly reduced in these animals (unpublished data). Together, these results suggest that induced expression of the Va14i-TCR on conventional naı¨ve T cells causes sterile inflammation, possibly due to autoimmune activation. To study the consequences of Va14i-TCR expression on tetramer+ TCR-switched T cells in more detail, we analyzed their surface phenotype and transcription factor expression. Absence of NK cell markers (Figures 6B and S2D) and PLZF expression (Figure 6C) indicated that the Va14i-TCR signals are not sufficient to induce NKT cell differentiation of mature conventional T cells. However, the TCR-switched tetramer+ T cells expressed signifi- Sterile Inflammation in Mice Containing TCR-Switched T Cells Animals containing TCR-switched tetramer+ T cells, but not controls, displayed splenomegaly (Figure 5A,B), characterized by increased numbers of macrophages/monocytes, neutrophils, and Ter119+ erythroid progenitor cells, suggesting an inflammatory state (Figure 5C–E). In line with these findings, we could detect June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org 6 Va14i-TCR Signaling Induces Cellular Activation, But Not NKT Cell Differentiation of TCR-Switched Tetramer+ T Cells The appearance of tetramer+ cells displaying a Vb bias similar to antigen-selected NKT cells, together with signs of inflammation upon TCR switch and the absence of CD1d-expressing B cell and upon TCR switch and the absence of CD1d-expressing B cell and However, the TCR-switched tetramer+ T cells expressed signifi- Figure 5. Signs of sterile inflammation in mice harboring TCR-switched T cells. T-cell-deficient mice were reconstituted with NKT-cell- depleted splenocytes of the indicated genotypes. Spleen weight (A), absolute splenic cell numbers (B–E, G, H), and serum TNF levels (F) of 3–28 mice per genotype were determined 8 wk after poly(I:C) administration where indicated. Bars indicate medians. Red points show six animals with near absence of B cells and dendritic cells. (B) Total splenocytes; (C) Macrophages/monocytes (Mac1+ Gr1int SiglecF2); (D) Neutrophils (Mac1+ Gr1high SiglecF2); (E) Erythroblasts (Ter119+); (G) B cells (B220+ TCRb2); (H) Dendritic cells (CD11c+). * p,0.001; p,0.01; * p,0.05, one-way ANOVA. doi:10.1371/journal.pbio.1001589.g005 Figure 5. Signs of sterile inflammation in mice harboring TCR-switched T cells. T-cell-deficient mice were reconstituted with NKT-cell- depleted splenocytes of the indicated genotypes. Spleen weight (A), absolute splenic cell numbers (B–E, G, H), and serum TNF levels (F) of 3–28 mice per genotype were determined 8 wk after poly(I:C) administration where indicated. Bars indicate medians. Red points show six animals with near absence of B cells and dendritic cells. (B) Total splenocytes; (C) Macrophages/monocytes (Mac1+ Gr1int SiglecF2); (D) Neutrophils (Mac1+ Gr1high SiglecF2); (E) Erythroblasts (Ter119+); (G) B cells (B220+ TCRb2); (H) Dendritic cells (CD11c+). * p,0.001; p,0.01; * p,0.05, one-way ANOVA. doi:10.1371/journal.pbio.1001589.g005 June 2013 \| Volume 11 \| Issue 6 \| e1001589 June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org 7 7 Investigating NKT Cell-TCR Signals Figure 6. TCR-switched tetramer+ T cells display an activated/exhausted phenotype, but no signs of NKT cell differentiation. T-cell- deficient mice were reconstituted with NKT-cell-depleted splenocytes of the indicated genotypes. The TCR switch was induced by poly(I:C) administration. Eight weeks later, the animals were analyzed. (A) Expression of intracellular IFN-c or TNF ex vivo 90 min after aGalCer injection of the indicated mice. Data are representative of two independent experiments with two animals each. (B) Representative histograms of flow cytometric analyses. Surface expression of NKG2D and NK1.1 on T cells (TCRb+) of the indicated surface phenotypes in comparison to NK cells (NKG2D+ TCRb2 CD52 or NK1.1+ TCRb2 CD52) are shown. Va14i-TCR Signaling Induces Cellular Activation, But Not NKT Cell Differentiation of TCR-Switched Tetramer+ T Cells (E–H) Extracellular expression of CD69 (E), PD-1 (F), LAG-3 (G), BTLA (H); * p,0.001; p,0.01; * p,0.05; ns, not significant; one-way ANOVA. doi:10.1371/journal.pbio.1001589.g006 Figure 6. TCR-switched tetramer+ T cells display an activated/exhausted phenotype, but no signs of NKT cell differentiation. T-cell- deficient mice were reconstituted with NKT-cell-depleted splenocytes of the indicated genotypes. The TCR switch was induced by poly(I:C) administration. Eight weeks later, the animals were analyzed. (A) Expression of intracellular IFN-c or TNF ex vivo 90 min after aGalCer injection of the indicated mice. Data are representative of two independent experiments with two animals each. (B) Representative histograms of flow cytometric analyses. Surface expression of NKG2D and NK1.1 on T cells (TCRb+) of the indicated surface phenotypes in comparison to NK cells (NKG2D+ TCRb2 CD52 or NK1.1+ TCRb2 CD52) are shown. Histograms are representative for at least three independent experiments with at least one mouse each. (C–H) Representative histograms of flow cytometric analyses. T cells (TCRb+) of the indicated surface phenotypes, and of wild-type splenic CD4+ NKT cells, are shown. Numbers in representative histograms indicate means of the median fluorescence intensities (MFIs), normalized to CD4+ tetramer2 T cells of animals that received NKT-cell-depleted Mx-Cre CaF/wt splenocytes, 8 wk after poly(I:C) injection. Means were calculated from 6–25 mice. Scatter plots display normalized MFI. Bars indicate medians. Dotted lines indicate medians of the median fluorescence intensities of control CD4+ wild-type NKT cells calculated from 2–6 mice. (C, D) Intracellular PLZF (C) and Egr2 (D) expression. (E–H) Extracellular expression of CD69 (E), PD-1 (F), LAG-3 (G), BTLA (H); * p,0.001; p,0.01; * p,0.05; ns, not significant; one-way ANOVA. doi:10.1371/journal.pbio.1001589.g006 To test whether exhaustion/anergy of tetramer+ TCR-switched T cells prevented a more dramatic form of autoimmune inflammation, we injected mice with PD-L1 and PD-L2 blocking or control antibodies twice a week for 4 consecutive weeks, starting 2 d before switch induction. The administration of these blocking antibodies has previously been shown to efficiently prevent anergy induction of conventional T as well as NKT cells, and to partially reverse the exhaustion of CD8+ T cells [44,45]. However, we did cantly higher levels of Egr2 in comparison to tetramer2 T cells in the same animals (Figure 6D), suggesting that the switched cells receive stronger TCR signals [13]. TCR-switched T cells showed further signs of cellular activation, as they expressed elevated levels of CD69 (Figure 6E). Va14i-TCR Signaling Induces Cellular Activation, But Not NKT Cell Differentiation of TCR-Switched Tetramer+ T Cells Histograms are representative for at least three independent experiments with at least one mouse each. (C–H) Representative histograms of flow cytometric analyses. T cells (TCRb+) of the indicated surface phenotypes, and of wild-type splenic CD4+ NKT cells, are shown. Numbers in representative histograms indicate means of the median fluorescence intensities (MFIs), normalized to CD4+ tetramer2 T cells of animals that received NKT-cell-depleted Mx-Cre CaF/wt splenocytes, 8 wk after poly(I:C) injection. Means were calculated from 6–25 mice. Scatter plots display normalized MFI. Bars indicate medians. Dotted lines indicate medians of the median fluorescence intensities of control CD4+ wild-type NKT cells calculated from 2–6 mice. (C, D) Intracellular PLZF (C) and Egr2 (D) expression. (E–H) Extracellular expression of CD69 (E), PD-1 (F), LAG-3 (G), BTLA (H); * p,0.001; p,0.01; * p,0.05; ns, not significant; one-way ANOVA. doi:10.1371/journal.pbio.1001589.g006 tramer+ T cells display an activated/exhausted phenotype, but no signs of NKT cell differentiation. T-cel ated/exhausted phenotype, but no signs of NKT cell differentiation. T-cell- Figure 6. TCR-switched tetramer+ T cells display an activated/exhausted phenotype, but no signs of NKT cell differentiation. T-cell- deficient mice were reconstituted with NKT-cell-depleted splenocytes of the indicated genotypes. The TCR switch was induced by poly(I:C) administration. Eight weeks later, the animals were analyzed. (A) Expression of intracellular IFN-c or TNF ex vivo 90 min after aGalCer injection of the indicated mice. Data are representative of two independent experiments with two animals each. (B) Representative histograms of flow cytometric analyses. Surface expression of NKG2D and NK1.1 on T cells (TCRb+) of the indicated surface phenotypes in comparison to NK cells (NKG2D+ TCRb2 CD52 or NK1.1+ TCRb2 CD52) are shown. Histograms are representative for at least three independent experiments with at least one mouse each. (C–H) Representative histograms of flow cytometric analyses. T cells (TCRb+) of the indicated surface phenotypes, and of wild-type splenic CD4+ NKT cells, are shown. Numbers in representative histograms indicate means of the median fluorescence intensities (MFIs), normalized to CD4+ tetramer2 T cells of animals that received NKT-cell-depleted Mx-Cre CaF/wt splenocytes, 8 wk after poly(I:C) injection. Means were calculated from 6–25 mice. Scatter plots display normalized MFI. Bars indicate medians. Dotted lines indicate medians of the median fluorescence intensities of control CD4+ wild-type NKT cells calculated from 2–6 mice. (C, D) Intracellular PLZF (C) and Egr2 (D) expression. Maintenance of Mature NKT Cells Is TCR-Independent This does not require acute TCR engagement [21]. However, it has been proposed that the ability of NKT cells to rapidly release IFN-c in this context critically requires continuous weak TCR activation in the steady state [25]. We therefore analyzed IFN-c release of TCR+ and TCR- NKT cells after in vivo injection of LPS, a-GalCer, and PBS (Figure 9A,B). As expected, Egr2 expression could only be detected in NKT cells that were activated through their TCR (Figure 9A). Accordingly, 90 min after a-GalCer injection, the majority of TCR+ NKT cells, but virtually none of the TCR- NKT cells or the CD4+ conventional T cells, produced IFN-c protein (Figure 9B). Interestingly, NKT cell activation through LPS injection in vivo was able to induce similar IFN-c production by TCR- NKT cells in comparison to their TCR+ counterparts (Figure 9B). Our results thus clearly demonstrate that homeostasis and key features defining the nature of NKT cells, namely the unique activated cell-surface phenotype and the innate capacity for instant production of IFN-c, do not require continuous auto-antigen recognition in the mouse. ( g ) Due to complete Cre-mediated recombination in lymphoid progenitors, T cell development is blocked at the double positive stage in Mx-Cre CaF/F mice after induction of Cre [40]. This allowed us to study the T cell decay in the absence of cellular efflux from the thymus. In agreement with previous studies [40,46], we found that loss of the TCR leads to decay of naı¨ve CD4+ CD44low and memory/effector-like CD4+ CD44high T cells with a half-life of 40 d and 297 d, respectively (Figure 7C,D). Interestingly, we observed essentially no decay of receptor-less NKT cells, with a calculated half-life of 322 d (Figure 7E), and could find significant numbers of TCR-deficient NKT cells even 45 wk after TCR deletion (unpublished data). To evaluate the role of TCR signals during in situ homeostatic proliferation, we administered BrdU for 4 wk via the drinking water, starting 2 wk after induced TCR ablation. Naı¨ve CD4+ CD44low as well as CD4+ CD44high memory/effector-like T cells showed significantly decreased BrdU incorporation in TCR-deficient compared to TCR-expressing cells (Figure 7F,G). In contrast, TCR ablation did not affect NKT cell proliferation (Figure 7F,G). Maintenance of Mature NKT Cells Is TCR-Independent Maintenance of Mature NKT Cells Is TCR-Independent The evidence for autoreactivity of the Va14i-TCR on mature peripheral T cells raised the old but still not completely resolved question whether and to what extent interactions with self-lipid- presenting APCs are required for NKT cell maintenance, cellular identity, and function. In order to evaluate the importance of constitutive TCR expression and signaling for NKT cells directly in vivo and for long periods of time, we ablated the TCR on mature T cells using poly(I:C) injection of Mx-Cre CaF/F mice [40]. Two weeks after induced Cre-mediated recombination, around 30% of CD4 and 65% of CD8 T cells had lost functional TCR expression in these mice (Figure 7A and [40]). To unambiguously identify TCR-deficient NKT cells, we developed a robust staining strategy based on CD4, NK1.1, CD5, and CD62L expression (Figure S3A). This limited us to CD4+ NKT cells, but our staining identified over 50% of the total NKT cell populations in thymus and spleen (Figure S3B). Around 65% of the thus identified NKT cells had lost TCR surface expression 2 wk after Cre induction (Figure 7A,B). The evidence for autoreactivity of the Va14i-TCR on mature peripheral T cells raised the old but still not completely resolved question whether and to what extent interactions with self-lipid- presenting APCs are required for NKT cell maintenance, cellular identity, and function. In order to evaluate the importance of constitutive TCR expression and signaling for NKT cells directly in vivo and for long periods of time, we ablated the TCR on mature T cells using poly(I:C) injection of Mx-Cre CaF/F mice [40]. Two weeks after induced Cre-mediated recombination, around 30% of CD4 and 65% of CD8 T cells had lost functional TCR expression in these mice (Figure 7A and [40]). To unambiguously identify TCR-deficient NKT cells, we developed a robust staining strategy based on CD4, NK1.1, CD5, and CD62L expression (Figure S3A). This limited us to CD4+ NKT cells, but our staining identified over 50% of the total NKT cell populations in thymus and spleen (Figure S3B). Around 65% of the thus identified NKT cells had lost TCR surface expression 2 wk after Cre induction (Figure 7A,B). ( g ) Treatment of mice with LPS, a cell wall component of gram- negative bacteria, leads to release of IFN-c by NKT cells via stimulation with IL-12 and IL-18 produced by innate immune cells. Discussion The elucidation of NKT cell function and their intriguing semi-invariant TCR benefited enormously from Va14i-TCR transgenic mouse models [11,32,47,48]. Over the last years, it became increasingly clear that premature expression of trans- genic TCRa chains, including Va14i [11,32], leads to various unwanted side-effects such as impaired b-selection and the generation of large numbers of DN T cells both in the periphery and in the thymus [26,27]. This drawback affects even TCR alleles generated through nuclear transfer of mature NKT cells [33]. For that reason, Baldwin et al. developed a system in which a transgenic CAGGS-promoter-driven TCRa-chain is expressed upon CD4-Cre-mediated excision of a loxP-flanked STOP cassette, mimicking the physiologic expression time point [26]. Likewise, Griewank and colleagues expressed the Va14i-TCR Maintenance of Mature NKT Cells Is TCR-Independent Interestingly, the BrdU incorpo- ration was identical in TCR-deficient CD4+ CD44high T and NKT cells, indicating that in the absence of TCR signals the cytokine-driven expansion of CD4+ CD44high memory/effector- like T and NKT cells is similar (Figure 7F,G). Our results therefore indicate that long-term in situ NKT cell homeostasis is completely independent of TCR-induced signals. Va14i-TCR Signaling Induces Cellular Activation, But Not NKT Cell Differentiation of TCR-Switched Tetramer+ T Cells Interestingly, these T cells displayed also significantly increased surface levels of PD-1, LAG-3, and less frequently, BTLA and TIM-3, which is typical of exhausted/ anergic cells (Figure 6F–H and unpublished data) [44,45]. June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org PLOS Biology \| www.plosbiology.org 8 Investigating NKT Cell-TCR Signals indicates that NKT cells receive stronger TCR signals in the thymus, which is supported by the decreased Egr2 expression of mature thymic TCR-deficient NKT cells (Figure 8A). Surprisingly, in mature NKT cells in thymus and spleen, expression of the TCR-signal-induced key transcription factor PLZF is completely unaffected by TCR ablation (Figure 8B). not observe any dramatic differences in spleen weight or cellularity, or signs of increased inflammation, between animals receiving PD-L blocking or control antibodies (unpublished data). In response to PD-1 blockade, other inhibitory receptors such as LAG-3, BTLA, or TIM-3 might control the TCR-switched T cells. y ( g ) In order to more generally evaluate to what extent NKT cell TCR-expression is required for the maintenance of characteristic lineage-specific gene expression (resembling recently activated T cells), we extensively analyzed the cell-surface phenotype of NKT cells 6 wk after TCR ablation. Of all the analyzed markers, the only significant changes that we observed on splenic NKT cells upon TCR ablation were downregulation of NK1.1, CD4, CD5, and ICOS (Figures 8C,D and S3C–E). NK1.1 expression was also reduced in thymic TCR-deficient NKT cells, in addition to CXCR6 expression (unpublished data). CD5 and ICOS expres- sion were also reduced in TCR-deficient splenic naı¨ve as well as CD62Llow CD4+ T cells (Figure S3C,D). CD4 was upregulated on TCR-deficient CD4+ naı¨ve, but downregulated on NKT and CD4+ CD44high T cells (Figure S3E). Strikingly, all other cell surface markers characteristic for the NKT cell lineage, among them the transcription factors PLZF, GATA-3, T-bet, and Th- POK, as well as many cell surface markers whose expression is also induced upon TCR engagement, remained largely unaffected by loss of the NKT cell TCR (Figure 8D). Taken together, our results showed that expression of the Va14i-TCR on mature conventional T cells is not sufficient to induce a NKT cell differentiation program. Still, it is likely that Va14i-TCR signals induce auto-antigen-mediated activation, possibly to the point of exhaustion. We therefore present strong evidence that the Va14i-TCR can constitutively recognize self- lipids in the naı¨ve steady state situation in vivo. The TCR Is Dispensable for the Identity and Cytokine- Secretion Ability of Mature NKT Cells In absence of de novo T cell generation, we found elevated Egr2 expression in mature thymic, but not splenic, NKT cells compared to DP thymocytes and CD4+ T cells, respectively (Figure 8A). This June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org 9 Investigating NKT Cell-TCR Signals Figure 7. TCR signaling is not required for the steady state homeostasis of mature NKT cells. (A) Percentages of TCRb2 cells of the depicted T cell subsets 2 wk after poly(I:C) injection into Mx-Cre CaF/F mice. Bars show means and SD (error bars) of 3–5 mice. (B) Surface TCRb expression of splenic CD4+ NKT cells (NK1.1+ CD5+ CD62Llow) 2 wk after poly(I:C) injection. Numbers indicate means 6 SD of three independent experiments with altogether five mice per genotype. (C, D) Total cell counts of splenic naı¨ve conventional CD4+ T cells (CD5+ CD44low NK1.12; C) or of memory/effector-like CD4+ T cells (CD5+ CD44high NK1.12; D) from 26 control CaF/F (CTR, TCR+) animals as well as from 24 Mx-Cre CaF/F animals, all after poly(I:C) injection (TCR+, TCR2). (E) Splenic CD4+ NKT cell numbers from in total 32 control CaF/F animals (CTR, TCR+) as well as TCRb+ and TCRb2 CD4+ NKT cell numbers from in total 27 Mx-Cre CaF/F animals, at the indicated time after poly(I:C) injection. (C–E) Half-lives were calculated with GraphPad Prism software using nonlinear regression, one-phase decay analysis. (F) BrdU was administered for 4 wk via the drinking water, starting 2 wk after poly(I:C) injection. Directly afterwards, animals were sacrificed and BrdU incorporation was measured by flow cytometry. Representative blots of 2 CaF/F and 4 Mx-Cre CaF/F mice are shown. (G) Bar chart showing proportion of cells that incorporated BrdU of the indicated T cell subtypes. Bars show means calculated from 2 CaF/F and means and SD (error bars) 4 Mx-Cre CaF/F mice. *** p,0.001; * p,0.05; ns, not significant; one-way ANOVA. doi:10.1371/journal.pbio.1001589.g007 Investigating NKT Cell-TCR Signals Figure 7. TCR signaling is not required for the steady state homeostasis of mature NKT cells. (A) Percentages of TCRb2 cells of the depicted T cell subsets 2 wk after poly(I:C) injection into Mx-Cre CaF/F mice. Bars show means and SD (error bars) of 3–5 mice. (B) Surface TCRb expression of splenic CD4+ NKT cells (NK1.1+ CD5+ CD62Llow) 2 wk after poly(I:C) injection. Numbers indicate means 6 SD of three independent experiments with altogether five mice per genotype. The TCR Is Dispensable for the Identity and Cytokine- Secretion Ability of Mature NKT Cells Here, we present a novel approach, in which the expression of the transgenic Va14i-TCRa-chain, and in the future any other TCRa-chain of interest, can be initiated via CD4-Cre at the DP stage in the thymus, and is under endogenous control June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org PLOS Biology \| www.plosbiology.org 10 Investigating NKT Cell-TCR Signals play a role in the initiation of CD69 expression on NKT cells, as well as in the induction of IL-2Rb, the b-chain of the IL-2 and IL-15 receptors [13]. Moreover, we observed the generation of tetramer+ CD8+ T cells. CD8+ NKT cells are found in the human, but not in wild- type mice. CD8 expression on Va14i NKT cells does not interfere Figure 8. The maintenance of NKT lineage identity does not depend on TCR-signals. CaF/F and Mx-Cre CaF/F mice were injected with poly(I:C) and analyzed 6 wk later. (A, B) Intracellular expression of Egr2 (A) and PLZF (B) in T cells from the depicted mice. Plots are representative for at least three independent experiments. (C) Flow cytometric analysis of NK1.1 expression on splenic naı¨ve (CD62Lhigh CD5+), memory/effector-like (CD62Llow CD5+) CD4+ T cells, and CD4+ NKT cells (NK1.1+ CD5+ CD62Llow), with or without TCR expression. Median fluorescence intensity, normalized to NK1.1 expression of NK cells (NK1.1+ TCRb2 CD52). Bars indicate medians. * p,0.001; p,0.01; * p,0.05; ns, not significant; one- way ANOVA. (D) Flow cytometric expression analysis of extra- and intracellular markers of splenic T cells. Median fluorescence intensities of at least four mice per analyzed protein were normalized to the expression on/in conventional CD4+ T cells (tetramer2 TCRb+) to account for interexperimental variations. Expression of NK1.1, CD122, FasL, and T-bet were normalized to NK cells (NK1.1+ TCRb2 CD52) and then set to 1 for naı¨ve T cells. Data are shown as heatmap, calculated by Perseus software. Blue letters, significantly reduced on splenic TCR2 CD4+ NKT cells in comparison to TCR+ CD4+ NKT cells from CaF/F and Mx-Cre CaF/F mice; analyzed by one-way ANOVA. doi:10.1371/journal.pbio.1001589.g008 g g g Figure 8. The maintenance of NKT lineage identity does not depend on TCR-signals. CaF/F and Mx-Cre CaF/F mice were injected with poly(I:C) and analyzed 6 wk later. (A, B) Intracellular expression of Egr2 (A) and PLZF (B) in T cells from the depicted mice. Plots are representative for at least three independent experiments. The TCR Is Dispensable for the Identity and Cytokine- Secretion Ability of Mature NKT Cells (C, D) Total cell counts of splenic naı¨ve conventional CD4+ T cells (CD5+ CD44low NK1.12; C) or of memory/effector-like CD4+ T cells (CD5+ CD44high NK1.12; D) from 26 control CaF/F (CTR, TCR+) animals as well as from 24 Mx-Cre CaF/F animals, all after poly(I:C) injection (TCR+, TCR2). (E) Splenic CD4+ NKT cell numbers from in total 32 control CaF/F animals (CTR, TCR+) as well as TCRb+ and TCRb2 CD4+ NKT cell numbers from in total 27 Mx-Cre CaF/F animals, at the indicated time after poly(I:C) injection. (C–E) Half-lives were calculated with GraphPad Prism software using nonlinear regression, one-phase decay analysis. (F) BrdU was administered for 4 wk via the drinking water, starting 2 wk after poly(I:C) injection. Directly afterwards, animals were sacrificed and BrdU incorporation was measured by flow cytometry. Representative blots of 2 CaF/F and 4 Mx-Cre CaF/F mice are shown. (G) Bar chart showing proportion of cells that incorporated BrdU of the indicated T cell subtypes. Bars show means calculated from 2 CaF/F and means and SD (error bars) 4 Mx-Cre CaF/F mice. *** p,0.001; * p,0.05; ns, not significant; one-way ANOVA. doi:10.1371/journal.pbio.1001589.g007 of the Tcra locus throughout the lifespan of the cell. In these mice, large numbers of bona fide CD4+ and DN NKT cells were generated. The reduced proportions of fully mature stage 3 NKT cells (NK1.1+, CD69high, T-bet+), as well as the reduced numbers of NK cells, are most likely a consequence of limiting amounts of common differentiation and maintenance factors, such as IL-15 [14,37,50]. In addition, attenuated TCR-signaling due to increased competition for self-antigen/ CD1d-complexes might delay the full maturation of NKT cells in the transgenic animals. TCR signals have been proposed to under direct control of CD4 promoter and enhancer sequences [11]. These are clear improvements, but carry the inbuilt caveats of the respective heterologous expression construct. For example, it has been shown that a large proportion of activated mature T cells loses expression from such transgenic CD4 promoter enhancer constructs [49]. The TCR Is Dispensable for the Identity and Cytokine- Secretion Ability of Mature NKT Cells (B) Intracellular IFN-c expression of the depicted cells stained directly ex vivo without addition of brefeldin or monensin. Splenic cells were isolated 90 min after PBS or aGalCer injection, or 6 h after LPS injection. Plots are representative for at least three independent experiments. doi:10.1371/journal.pbio.1001589.g009 achieved through the generation of mixed bone marrow chimeras with Ja182/2 bone marrow, which cannot give rise to Va14i- NKT cells. with negative selection, avidity for antigen presented by CD1d, or NKT cell function [28]. Instead, it was proposed that the absence of CD8+ NKT cells in the mouse is due to the constitutive expression of the transcription factor Th-Pok in all CD4+ as well as DN NKT cells [28]. Th-Pok has been shown to be crucial for the maturation and function of NKT cells, and directly represses CD8 expression [28]. This scenario fits well with the fact that the CD8+ tetramer+ T cells in the CD4-Cre Va14iStopF/wt (as well as in the Va11p-Va14itg animals) did not express Th-Pok. These cells also lack many other characteristic features of NKT cells, including PLZF expression. Therefore, we refer to them as tetramer+ CD8+ T cells. Our studies were designed to elucidate whether or to what extent the expression of the autoreactive semi-invariant TCR would activate a peripheral mature naı¨ve conventional T cell, convert it into an NKT cell, or induce gene expression typical of NKT cells. We took advantage of the conditional nature of the Va14i-TCR knock-in transgene for a TCR switch experiment on conventional peripheral T cells. Naı¨ve CD4+ T cells inherit a high plasticity [51]. Depending on TCR signaling strength and cytokine environment, they can differentiate in various subsets in periphery. This differentiation includes the induction of specific transcription factors, namely T-bet (Th1), GATA-3 (Th2), ROR-ct (Th17), and FoxP3 (peripherally derived regulatory T cells). For NKT cells, it is believed that strong TCR signaling, together with homotypic interactions involving the SLAM family (SLAMf) receptors 1 and 6, ultimately leads to PLZF induction during thymic development [11,13]. DP thymocytes, presenting auto-antigen via CD1d and also expressing SLAMf members, are crucial for thymic NKT cell selection [11]. The TCR Is Dispensable for the Identity and Cytokine- Secretion Ability of Mature NKT Cells (C) Flow cytometric analysis of NK1.1 expression on splenic naı¨ve (CD62Lhigh CD5+), memory/effector-like (CD62Llow CD5+) CD4+ T cells, and CD4+ NKT cells (NK1.1+ CD5+ CD62Llow), with or without TCR expression. Median fluorescence intensity, normalized to NK1.1 expression of NK cells (NK1.1+ TCRb2 CD52). Bars indicate medians. * p,0.001; p,0.01; * p,0.05; ns, not significant; one- way ANOVA. (D) Flow cytometric expression analysis of extra- and intracellular markers of splenic T cells. Median fluorescence intensities of at least four mice per analyzed protein were normalized to the expression on/in conventional CD4+ T cells (tetramer2 TCRb+) to account for interexperimental variations. Expression of NK1.1, CD122, FasL, and T-bet were normalized to NK cells (NK1.1+ TCRb2 CD52) and then set to 1 for naı¨ve T cells. Data are shown as heatmap, calculated by Perseus software. Blue letters, significantly reduced on splenic TCR2 CD4+ NKT cells in comparison to TCR+ CD4+ NKT cells from CaF/F and Mx-Cre CaF/F mice; analyzed by one-way ANOVA. doi:10.1371/journal.pbio.1001589.g008 Moreover, we observed the generation of tetramer+ CD8+ T cells. CD8+ NKT cells are found in the human, but not in wild- type mice. CD8 expression on Va14i NKT cells does not interfere play a role in the initiation of CD69 expression on NKT cells, as well as in the induction of IL-2Rb, the b-chain of the IL-2 and IL-15 receptors [13]. June 2013 \| Volume 11 \| Issue 6 \| e1001589 June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org 11 Investigating NKT Cell-TCR Signals Figure 9. TCR-signals are not required for the innate activation of NKT cells. (A) Intracellular Egr2 expression of the depicted splenic T cells, 90 min after PBS or aGalCer injection, or 6 h after LPS injection. Plots are representative for at least three independent experiments. (B) Intracellular IFN-c expression of the depicted cells stained directly ex vivo without addition of brefeldin or monensin. Splenic cells were isolated 90 min after PBS or aGalCer injection, or 6 h after LPS injection. Plots are representative for at least three independent experiments. doi:10.1371/journal.pbio.1001589.g009 Figure 9. TCR-signals are not required for the innate activation of NKT cells. (A) Intracellular Egr2 expression of the depicted splenic T cells, 90 min after PBS or aGalCer injection, or 6 h after LPS injection. Plots are representative for at least three independent experiments. The TCR Is Dispensable for the Identity and Cytokine- Secretion Ability of Mature NKT Cells Indeed, the high Egr2 expression of mature NKT cells that matured in the periphery and migrated back to the thymus (Figure 8A) suggests that stronger self-antigens are presented at this location. Interestingly, unlike TCR-switched tetramer+ T cells, Egr2 expression in mature splenic NKT cells was similar to that of conventional mature CD4+ T cells. Our data therefore suggest that in the periphery, the Va14i-TCR can recognize self-lipids, but maturing NKT cells undergo a developmental program that prevents an auto-reactive inflammatory response. At this point, we cannot exclude the possibility that the observed cellular activation was antigen- independent. The fact that the internal control cells, the co- transferred tetramer2 T cells, show no or significantly less signs of activation strongly argues for an involvement of antigen recognition or tonic signaling by the Va14i-TCR. It also remains possible that the transient immune activation caused by the poly(I:C) administration contributes to the observed phenotypes. In all likelihood, this contribution is small, as we never observed any significant immune activation, not to mention loss of CD1d- expressing antigen-presenting B cells and dendritic cells, in Mx-Cre CaF/wt control mice that received poly(I:C). Despite these caveats, our results clearly show that under our experimental conditions, Va14i-TCR expression on conventional naı¨ve T cells leads to their activation and general immune deregulation. Collectively, our data strongly support the view that Va14i- TCR expression on developing NKT cells triggers a program that makes them unresponsive to peripheral self-antigens, which can continuously be recognized by their auto-reactive TCR. NKT cells are extremely potent immune-modulatory cells that upon activation can instantly secrete a large array of cytokines. Although they are selected by high affinity to auto-antigens, similar to regulatory T cells, they are not mainly suppressive cells. Therefore, it seems plausible that NKT cells are rendered ‘‘blind’’ to peripheral auto-antigens, rather than depend on continuous stimulation by self-lipids to maintain their cellular identity and innate functions. By keeping their activated state independent of self-antigen recognition, NKT cells can stay poised to secrete immune-activating cytokines while minimizing the risk of causing damage to self during normal physiology. On the other hand, the presence of the auto-reactive Va14i-TCR serves to detect pathogenic states when a stronger signal is generated by the enhanced presentation of potentially more potent self-antigens or foreign lipids. Genetically Modified Mice S F y To generate Va14iStopF mice, B6 ES cells (Artemis) were transfected, cultured, and selected as previously described for Bruce 4 ES cells [56]. Mx-Cre [39], CaF [40], CD4-Cre [57], Nestin- Cre [31], Va11p-Va14i-tg [32], and Va14iStopF mice were kept on a C57BL/6 genetic background. As we did not observe any differences between CD4-Cre and Va14iStopF/wt mice in NKT cell biology, they were sometimes grouped together as controls. Mice were housed in the specific pathogen-free animal facility of the MPIB. All animal procedures were approved by the Regierung of Oberbayern. These findings seemed to support notions that NKT cell maintenance [52], their activated surface phenotype, and espe- cially their rapid cytokine expression abilities might depend on constant antigen recognition [25]. However, by ablating the TCR on mature NKT cells in situ, we unequivocally demonstrated that long-term mouse NKT cell homeostasis and gene expression are nearly completely independent of TCR signals. In this regard, they are similar to memory T and B cells, which can maintain their numbers, identity, and functional capabilities in the absence of antigen [53,54]. Our results are hard to reconcile with a recent report suggesting that NKT cell maintenance requires lipid presentation by B cells [52]. While there might be some differences between mouse and man, a more likely scenario is that the The TCR Is Dispensable for the Identity and Cytokine- Secretion Ability of Mature NKT Cells Indeed, the high Egr2 expression of mature NKT cells that matured in the periphery and migrated back to the thymus (Figure 8A) suggests that stronger self-antigens are presented at this location. Interestingly, unlike TCR-switched tetramer+ T cells, Egr2 expression in mature splenic NKT cells was similar to that of conventional mature CD4+ T cells. Our data therefore suggest that in the periphery, the Va14i-TCR can recognize self-lipids, but maturing NKT cells undergo a developmental program that prevents an auto-reactive inflammatory response. At this point, we cannot exclude the possibility that the observed cellular activation was antigen- independent. The fact that the internal control cells, the co- transferred tetramer2 T cells, show no or significantly less signs of activation strongly argues for an involvement of antigen recognition or tonic signaling by the Va14i-TCR. It also remains possible that the transient immune activation caused by the poly(I:C) administration contributes to the observed phenotypes. In all likelihood, this contribution is small, as we never observed any significant immune activation, not to mention loss of CD1d- expressing antigen-presenting B cells and dendritic cells, in Mx-Cre CaF/wt control mice that received poly(I:C). Despite these caveats, our results clearly show that under our experimental conditions, Va14i-TCR expression on conventional naı¨ve T cells leads to their activation and general immune deregulation. These findings seemed to support notions that NKT cell maintenance [52], their activated surface phenotype, and espe- cially their rapid cytokine expression abilities might depend on constant antigen recognition [25] However by ablating the TCR observations of Bosma et al. reflect rather acute local activation than true homeostatic requirements. Most of the known functions of NKT cells critically depend on their ability to rapidly secrete large amounts of many different immune-modulatory cytokines shortly after their activation. Still, it is not fully understood how NKT cell activation is triggered in different disease settings, and especially to what extent signaling in response to TCR-mediated recognition of antigens versus activation by proinflammatory cytokines contributes to this. Various studies reported that CD1d- dependent signals were required for full NKT activation in vitro [19,20,55], although most of them contained the caveat of potentially incomplete blockade of CD1d function by blocking antibodies. The TCR Is Dispensable for the Identity and Cytokine- Secretion Ability of Mature NKT Cells These SLAMf receptors are expressed on periph- eral lymphocytes in comparable levels to double positive Given the faithful recapitulation of endogenous TCRa-chain expression timing and strength in our knock-in mice, combined with the extremely high homologous recombination efficiency, we believe that our strategy should prove useful for the generation of further novel TCR-transgenic mouse models. By replacing RAG- mediated Va14 to Ja18 recombination with Cre-mediated activation of Va14i expression in CD4-Cre Va14iStopF/wt mice, we can directly couple conditional gain or loss of gene function with Va14i-TCR expression in NKT cells. NKT cell-specific gene targeting in mice with physiological NKT cell numbers could be June 2013 \| Volume 11 \| Issue 6 \| e1001589 12 PLOS Biology \| www.plosbiology.org Investigating NKT Cell-TCR Signals thymocytes (www.immgen.org). Therefore, lymphocytes, especial- ly marginal zone B cells, which express CD1d to a similar level as DP thymocytes, should be able to present antigen and SLAMf- mediated co-stimulation, to naı¨ve conventional T cells with a newly expressed Va14i-TCR on their surface. The elevated levels of the TCR-induced transcription factor Egr2 in switched tetramer+ T cells suggest that they receive an (auto-)antigenic signal. This finding is in principle in agreement with our finding that tetramer+ TCR-switched T cells are enriched in cells that express Vb2- and Vb8.1-/8.2-containing Va14i-TCRs. These TCRs were shown to have the highest avidity for NKT cell antigens [3]. Furthermore, Vb7-containing Va14i-TCRs were shown to be favored when endogenous ligand concentration are suboptimal in CD1d+/2 mice [42]. In fact, in CD4+ tetramer+ TCR-switched T cells the relative enrichment for Vb7-expressing cells was slightly higher than for Vb2- and Vb8.1-/8.2-expressing cells (unpublished data). However, the interpretation that this advantage is due to antigenic selection is at odds with the fact that Vb7-expressing cells are not enriched in tetramer+ TCR-switched CD8+ T cells. We currently have no satisfactory explanation for this discrepancy. Both CD4+ and CD8+ Va14i-TCR-expressing conventional T cells show features of activation and exhaustion/ anergy, but do not develop into NKT cells, judged by absent PLZF and NK cell marker expression. This indicates that either mature T cells have lost the ability to enter the NKT cell lineage, the peripheral Va14i-TCR signal is not strong enough, or as yet unidentified components of the thymic microenvironment are required to induce an NKT cell fate. The TCR Is Dispensable for the Identity and Cytokine- Secretion Ability of Mature NKT Cells Our experiments, in line with a recent report [21], show that even in the complete absence of TCR signaling for 4 wk, NKT cells can be robustly activated in vivo to produce IFN- c upon LPS injection in similar amounts as their TCR+ counterparts. Thus, we demonstrate that in mouse NKT cells continuous steady-state TCR-signaling is not required to maintain the Ifng locus in a transcriptionally active state, as recently proposed for human NKT cells [25]. Therefore, our results clearly demonstrate that cellular identity and critical functional abilities of mature NKT cells, such as steady-state proliferation and innate cytokine secretion ability, although initially instructed by strong TCR signals, do not require further antigen recognition through their TCR. thymocytes (www.immgen.org). Therefore, lymphocytes, especial- ly marginal zone B cells, which express CD1d to a similar level as DP thymocytes, should be able to present antigen and SLAMf- mediated co-stimulation, to naı¨ve conventional T cells with a newly expressed Va14i-TCR on their surface. The elevated levels of the TCR-induced transcription factor Egr2 in switched tetramer+ T cells suggest that they receive an (auto-)antigenic signal. This finding is in principle in agreement with our finding that tetramer+ TCR-switched T cells are enriched in cells that express Vb2- and Vb8.1-/8.2-containing Va14i-TCRs. These TCRs were shown to have the highest avidity for NKT cell antigens [3]. Furthermore, Vb7-containing Va14i-TCRs were shown to be favored when endogenous ligand concentration are suboptimal in CD1d+/2 mice [42]. In fact, in CD4+ tetramer+ TCR-switched T cells the relative enrichment for Vb7-expressing cells was slightly higher than for Vb2- and Vb8.1-/8.2-expressing cells (unpublished data). However, the interpretation that this advantage is due to antigenic selection is at odds with the fact that Vb7-expressing cells are not enriched in tetramer+ TCR-switched CD8+ T cells. We currently have no satisfactory explanation for this discrepancy. Both CD4+ and CD8+ Va14i-TCR-expressing conventional T cells show features of activation and exhaustion/ anergy, but do not develop into NKT cells, judged by absent PLZF and NK cell marker expression. This indicates that either mature T cells have lost the ability to enter the NKT cell lineage, the peripheral Va14i-TCR signal is not strong enough, or as yet unidentified components of the thymic microenvironment are required to induce an NKT cell fate. Statistics Table S1 Comparison of different Va14i-transgenic mice. (DOCX) Statistical analysis of the results was performed by one-way ANOVA followed by Tukey’s test, or by student t test, in Prism software (GraphPad). The p values are presented in figure legends where a statistically significant difference was found. Table S1 Comparison of different Va14i-transgenic mice. (DOCX) Cre Induction in Mx-Cre Animals (D) Serum cytokine levels, measured by FlowCytomix, of three CTR mice and each five CD4-Cre Va14iStopF/wt and Va11p-Va14itg mice. (E, F) CD69 and intracellular T-bet expression of NK1.1+/NK1.12 NKT cells from CTR and CD4-Cre Va14iStopF/wt mice. Numbers in representative histogram indicate percentage of CD69high or T- bet+ cells among the indicated NKT cells calculated from eight animals per genotype (CD69) or three animals per genotype (T- bet). Histograms are representative of three or more independent experiments with each at least seven mice in total. (G) CD69 expression of CD4+ conventional T cells (filled grey) and NKT cells (black) from Va11p-Va14itg mice. Number in representative histogram indicates percentage of CD69high cells among the NKT cells, calculated from seven animals. Throughout the figure, NKT cells were gated as tetramer+ TCRb+, conventional (conv) T cells as tetramer2 TCRb+; CTR, CD4-Cre or Va14iStopF/wt. (TIF) Figure S2 NKT-cell-depletion before cell transfer and additional analysis of the animals in the TCR-switch experiment. (A) Splenocytes of the indicated genotypes were stained before and after depletion of NKT cells by MACS. Numbers indicate percentages of tetramer+ and tetramer2 T cells (TCRb+). Plots are representative for over 15 independent experiments. (B) Staining with ‘‘unloaded’’ mCD1d- tetramer in comparison to PBS57-loaded mCD1d-tetramer of splenocytes from the same animal. Plots are representative for three independent experiments with five mice in total. (C) Expression of intracellular IFN-c or TNF ex vivo 90 min after aGalCer injection. Data are representative of two independent experiments with two animals each. (D) Representative histograms of flow cytometric analysis of T cells in animals 8 wk after switch induction: CD4+ tetramer–, CD4+ tetramer+, CD8+ tetramer–, and CD8+ tetramer+ T cells (TCRb+). Surface expression of the depicted markers in comparison to NK cells (gated as marker+, TCRb2). Representative plots for at least three independent experiments with at least one mouse each. (TIF) Figure S2 NKT-cell-depletion before cell transfer and additional analysis of the animals in the TCR-switch experiment. (A) Splenocytes of the indicated genotypes were stained before and after depletion of NKT cells by MACS. Numbers indicate percentages of tetramer+ and tetramer2 T cells (TCRb+). Plots are representative for over 15 independent experiments. (B) Staining with ‘‘unloaded’’ mCD1d- tetramer in comparison to PBS57-loaded mCD1d-tetramer of splenocytes from the same animal. Plots are representative for three independent experiments with five mice in total. (C) Expression of intracellular IFN-c or TNF ex vivo 90 min after aGalCer injection. Cre Induction in Mx-Cre Animals At the age of 6–8 wk (or 2 wk after cell transfer for the TCR switch experiment), animals were given a single i.p. injection (400 mg) of poly(I:C) (Amersham). All mice were analyzed 6–8 wk after injection, unless otherwise indicated. June 2013 \| Volume 11 \| Issue 6 \| e1001589 June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org 13 Investigating NKT Cell-TCR Signals Flow Cytometry and Heat Map Generation Single-cell suspensions were prepared and stained with mono- clonal antibodies: B220 (clone RA3-6B2), BTLA (8F4), CD11c (N418), CD122 (TM-b1), CD127 (A7R34), CD160 (eBioCNX46- 3), CD25 (PC61.5), CD28 (37.51), CD38 (90), CD39 (24DMS1), CD4 (RM4-5), CD44 (IM7), CD45RB (C363.16A), CD5 (53-7.3), CD62L (MEL-14), CD69 (H1.2-F3), CD8a (53-6.7), CD8b (H35- 17.2), CD95 (15A7), DX5 (DX5), Egr2 (erongr2), GATA-3 (TWAJ), Gr1 (RB6-8C5), ICOS (7E.17G9), IL-4 (11B11), IL-13 (eBio13A), IL-17A (eBio17B7), IFN-c (XMG1.2), LAG-3 (eBioC9B7W), LFA-1 (M17/4), Ly49A/D (eBio12A8), Ly49C/I (14B11), Ly49G2 (eBio4D11), Mac1 (M1/70), NKG2A (16A11), NKG2D (CX5), NK1.1 (PK136), PD-1 (J43), ROR-ct (AFKJS-9), T-bet (eBio4B10), TCRb (H57-597), Ter119 (TER-119), Th-POK (2POK), and TNF (MP6-XT22) (all from eBioscience). SiglecF (E50-2440) was from BD. TCRb chains were stained with the mouse Vb TCR screening panel (BD). PLZF antibody and the CXCL16-Fc fusion were generous gifts from Derek Sant’Angelo and Mehrdad Matloubian, respectively. mCD1d-tetramers were provided by the NIH tetramer core facility. For intracellular transcription factor stainings, cells were fixed and permeabilized with the FoxP3 staining kit (eBioscience). For intracellular cytokine stainings, mice were injected i.v. in the tail vein with 40 mg of LPS (Sigma) or 2 mg aGalCer (Funakoshi) in a total volume of 200 ml PBS. Afterwards, cells were treated according to manufacturer’s instructions with the Cytofix/Cytoperm kit (BD). For multiplex measurement of cytokines in the serum, we used the mouse Th1/ Th2 10plex Cytomix kit according to manufacturer’s instructions (eBioscience). Samples were acquired on a FACSCanto2 (BD) machine, and analyzed with FlowJo software (Treestar). The heat map was generated using perseus (part of the MaxQuant software [58]). eight showing homologous integration of the knock-in allele. (B) Absolute cell numbers in thymus and spleen of 7–13 mice of the indicated genotypes of CD8+ tetramer+ T cells. Bars indicate medians. *** p,0.001; ns, not significant, one-way ANOVA. (C) CD8a/CD8b expression of splenic CD8a+ NKT cells from CD4- Cre Va14iStopF/wt animals. Numbers indicate mean percentages 6 SD of three mice. Acknowledgments We are grateful to Reinhard Fa¨ssler for support. We wish to thank Julia Knogler and Barbara Habermehl for technical assistance. The PLZF and the CXCL16-Fc fusion protein were generous gifts from Derek Sant’Angelo and Mehrdad Matloubian, respectively. The Va11p-Va14i- tg mice were kindly provided by Stephanie Ganal and Andreas Diefenbach. We are grateful to Xiaojing Yue and Tilman Borggrefe for sharing data on CD4p-Va14i-tg mice with us. We thank Albert Bendelac for input and plasmids for our first attempts to generate TCR-switch mouse models. We wish to thank the NIH tetramer core facility for providing us with mCD1d- PBS57-tetramers. Cre Induction in Mx-Cre Animals Data are representative of two independent experiments with two animals each. (D) Representative histograms of flow cytometric analysis of T cells in animals 8 wk after switch induction: CD4+ tetramer–, CD4+ tetramer+, CD8+ tetramer–, and CD8+ tetramer+ T cells (TCRb+). Surface expression of the depicted markers in comparison to NK cells (gated as marker+, TCRb2). Representative plots for at least three independent experiments with at least one mouse each. (TIF) ELISA Serum TNF levels were determined by ELISA as recommended by the manufacturer (BD). Figure S3 Gating strategy for the TCR-ablation experiments. (A) Gating strategy to identify TCR2 NKT cells. (B) Yield of the applied gating strategy. (C–E) Extracellular expression of the depicted proteins on CD4+ naı¨ve (CD62Lhigh CD5+), CD4+ memory/effector-like (CD62Llow CD5+) T cells, and CD4+ NKT cells (NK1.1+ CD5+ CD62Llow). MFIs were normalized to the expression of CD4+ naı¨ve T cells. (TIF) BrdU Incorporation Mice were fed with 0.5 mg/ml BrdU (Sigma) in the drinking water for 4 consecutive weeks. Directly afterwards, BrdU incorporation was analyzed with a BrdU Flow Kit (BD). Quantitative RT-PCR RNA was isolated (QIAGEN RNeasy Micro Kit) and reverse transcribed (Promega) for quantitative real-time polymerase chain reaction (PCR) using probes and primers from the Universal Probe Library (Roche Diagnostics) according to the manufactur- er’s instructions. References Kuhn R, Schwenk F, Aguet M, Rajewsky K (1995) Inducible gene targeting in mice. Science 269: 1427–1429. 15. McNab FW, Berzins SP, Pellicci DG, Kyparissoudis K, Field K, et al. (2005) The influence of CD1d in postselection NKT cell maturation and homeostasis. J Immunol 175: 3762–3768. 40. Polic B, Kunkel D, Scheffold A, Rajewsky K (2001) How alpha beta T cells deal with induced TCR alpha ablation. Proc Natl Acad Sci USA 98: 8744–8749. 41. Bourgeois C, Hao Z, Rajewsky K, Potocnik AJ, Stockinger B (2008) Ablation of thymic export causes accelerated decay of naive CD4 T cells in the periphery because of activation by environmental antigen. Proc Natl Acad Sci USA 105: 8691–8696. doi:10.1073/pnas.0803732105. 16. Kinjo Y, Tupin E, Wu D, Fujio M, Garcia-Navarro R, et al. (2006) Natural killer T cells recognize diacylglycerol antigens from pathogenic bacteria. Nat Immunol 7: 978–986. doi:10.1038/ni1380. 42. Schu¨mann J, Mycko MP, Dellabona P, Casorati G, Macdonald HR (2006) Cutting edge: influence of the TCR Vbeta domain on the selection of semi- invariant NKT cells by endogenous ligands. J Immunol 176: 2064–2068. 17. Kinjo Y, Wu D, Kim G, Xing G, Poles M, et al. (2005) Recognition of bacterial glycosphingolipids by natural killer T cells. Nature 434: 520–525. doi:nature03407 18. Kinjo Y, Illarionov P, Vela JL, Pei B, Girardi E, et al. (2011) Invariant natural killer T cells recognize glycolipids from pathogenic Gram-positive bacteria. Nat Immunol 12: 966–974. doi:10.1038/ni.2096. 43. Brossay L, Tangri S, Bix M, Cardell S, Locksley R, et al. (1998) Mouse CD1- autoreactive T cells have diverse patterns of reactivity to CD1+ targets. J Immunol 160: 3681–3688. 19. Mattner J, Debord KL, Ismail N, Goff RD, Cantu C, et al. (2005) Exogenous and endogenous glycolipid antigens activate NKT cells during microbial infections. Nature 434: 525–529. doi:10.1038/nature03408. 44. Blackburn SD, Shin H, Haining WN, Zou T, Workman CJ, et al. (2009) Coregulation of CD8+ T cell exhaustion by multiple inhibitory receptors during chronic viral infection. Nat Immunol 10: 29–37. doi:10.1038/ni.1679. 20. Brigl M, Tatituri RVV, Watts GFM, Bhowruth V, Leadbetter EA, et al. (2011) Innate and cytokine-driven signals, rather than microbial antigens, dominate in natural killer T cell activation during microbial infection. J Exp Med 208: 1163– 1177. doi:10.1084/jem.20102555. 45. Parekh VV, Lalani S, Kim S, Halder R, Azuma M, et al. (2009) PD-1/PD-L blockade prevents anergy induction and enhances the anti-tumor activities of glycolipid-activated invariant NKT cells. References 1. Bendelac A, Savage PB, Teyton L (2007) The biology of NKT cells. Annu Rev Immunol 25: 297–336. 1. Bendelac A, Savage PB, Teyton L (2007) The biology of NKT cells. Annu Rev Immunol 25: 297–336. 27. Serwold T, Hochedlinger K, Inlay MA, Jaenisch R, Weissman IL (2007) Early TCR expression and aberrant T cell development in mice with endogenous prerearranged T cell receptor genes. J Immunol 179: 928–938. 2. Gapin L (2010) iNKT cell autoreactivity: what is ‘‘self’’ and how is it recognized? Nat Rev Immunol 10: 272–277. doi:10.1038/nri2743. 28. Engel I, Hammond K, Sullivan BA, He X, Taniuchi I, et al. (2010) Co-receptor choice by V alpha14i NKT cells is driven by Th-POK expression rather than avoidance of CD8-mediated negative selection. J Exp Med 207: 1015–1029. doi:10.1084/jem.20090557. 3. Mallevaey T, Scott-Browne JP, Matsuda JL, Young MH, Pellicci DG, et al. (2009) T cell receptor CDR2 beta and CDR3 beta loops collaborate functionally to shape the iNKT cell repertoire. Immunity 31: 60–71. doi:10.1016/ j.immuni.2009.05.010. j 29. Godfrey DI, Stankovic S, Baxter AG (2010) Raising the NKT cell family. Nat Immunol 11: 197–206. doi:10.1038/ni.1841. 4. Wei D, Curran S, Savage PB, Teyton L, Bendelac A (2006) Mechanisms imposing the Vbeta bias of Valpha14 natural killer T cells and consequences for microbial glycolipid recognition. J Exp Med 203: 1197–1207. 30. Coquet JM, Chakravarti S, Kyparissoudis K, McNab FW, Pitt LA, et al. (2008) Diverse cytokine production by NKT cell subsets and identification of an IL-17- producing CD4-NK1.1- NKT cell population. Proc Natl Acad Sci USA 105: 11287–11292. doi:10.1073/pnas.0801631105. 5. Mallevaey T, Clarke AJ, Scott-Browne JP, Young MH, Roisman LC, et al. (2011) A molecular basis for NKT cell recognition of CD1d-self-antigen. Immunity 34: 315–326. doi:10.1016/j.immuni.2011.01.013. 31. Betz UA, Vosshenrich CA, Rajewsky K, Mu¨ller W (1996) Bypass of lethality with mosaic mice generated by Cre-loxP-mediated recombination. Curr Biol 6: 1307–1316. 6. Kronenberg M, Rudensky AY (2005) Regulation of immunity by self-reactive T cells. Nature 435: 598–604. 32. Bendelac A, Hunziker R, Lantz O (1996) Increased interleukin 4 and immunoglobulin E production in transgenic mice overexpressing NK1 T cells. J Exp Med 184: 1285–1293. 7. Bendelac A, Bonneville M, Kearney J (2001) Autoreactivity by design: innate B and T lymphocytes. Nat Rev Immunol 1: 177–186. 8. Facciotti F, Ramanjaneyulu GS, Lepore M, Sansano S, Cavallari M, et al. References (2012) Peroxisome-derived lipids are self antigens that stimulate invariant natural killer T cells in the thymus. Nat Immunol 13: 474–480. doi:10.1038/ni.2245. 33. Wakao H, Kawamoto H, Sakata S, Inoue K, Ogura A, et al. (2007) A novel mouse model for invariant NKT cell study. J Immunol 179: 3888–3895. 9. Zhou D, Mattner J, Cantu C, Schrantz N, Yin N, et al. (2004) Lysosomal glycosphingolipid recognition by NKT cells. Science 306: 1786–1789. 34. Yue X, Izcue A, Borggrefe T (2011) Essential role of Mediator subunit Med1 in invariant natural killer T-cell development. Proc Natl Acad Sci USA. doi:10.1073/pnas.1109095108. 10. Fox LM, Cox DG, Lockridge JL, Wang X, Chen X, et al. (2009) Recognition of lyso-phospholipids by human natural killer T lymphocytes. PLoS Biol 7: e1000228. doi:10.1371/journal.pbio.1000228. p 35. Thapa P, Das J, McWilliams D, Shapiro M, Sundsbak R, et al. (2013) The transcriptional repressor NKAP is required for the development of iNKT cells. Nat Commun 4: 1582. doi:10.1038/ncomms2580. j p 11. Griewank K, Borowski C, Rietdijk S, Wang N, Julien A, et al. (2007) Homotypic interactions mediated by Slamf1 and Slamf6 receptors control NKT cell lineage development. Immunity 27: 751–762. doi:10.1016/j.immuni.2007.08.020. 36. Kovalovsky D, Uche OU, Eladad S, Hobbs RM, Yi W, et al. (2008) The BTB– zinc finger transcriptional regulator PLZF controls the development of invariant natural killer T cell effector functions. Nat Immunol 9: 1055–1064. doi:10.1038/ ni.1641. 12. Moran AE, Holzapfel KL, Xing Y, Cunningham NR, Maltzman JS, et al. (2011) T cell receptor signal strength in Treg and iNKT cell development demonstrated by a novel fluorescent reporter mouse. J Exp Med 208: 1279–1289. doi:10.1084/jem.20110308. 37. Gordy LE, Bezbradica JS, Flyak AI, Spencer CT, Dunkle A, et al. (2011) IL-15 regulates homeostasis and terminal maturation of NKT cells. J Immunol 187: 6335–6345. doi:10.4049/jimmunol.1003965. 13. Seiler MP, Mathew R, Liszewski MK, Spooner C, Barr K, et al. (2012) Elevated and sustained expression of the transcription factors Egr1 and Egr2 controls NKT lineage differentiation in response to TCR signaling. Nat Immunol 13: 264–271. doi:10.1038/ni.2230. 38. Ranson T, Vosshenrich CAJ, Corcuff E, Richard O, Laloux V, et al. (2003) IL- 15 availability conditions homeostasis of peripheral natural killer T cells. Proc Natl Acad Sci USA 100: 2663–2668. doi:10.1073/pnas.0535482100. 14. Matsuda JL, Gapin L, Sidobre S, Kieper WC, Tan JT, et al. (2002) Homeostasis of V alpha 14i NKT cells. Nat Immunol 3: 966–974. doi:10.1038/ni837. 39. Investigating NKT Cell-TCR Signals Investigating NKT Cell-TCR Signals Performed the experiments: JCV KH NK MSS. Analyzed the data: JCV MYH MSS. Contributed reagents/materials/analysis tools: LB HY BP. Wrote the paper: JCV MSS. Author Contributions The author(s) have made the following declarations about their contributions: Conceived and designed the experiments: JCV MSS. Supporting Information Figure S1 Southern blot screening strategy for the Va14iStopF knock-in allele and NKT cell characterization in Va14i– transgenic mice. (A) DNA of targeted neomycin-resistant embry- onic stem cells was digested with BamHI. The Southern Blot probe contains the untranslated exon 4 of Ca and recognizes a 12.5 kb fragment for the knock-in in comparison to 8.9 kb for the wild-type allele. Representative Southern blot for 325 clones, six of June 2013 \| Volume 11 \| Issue 6 \| e1001589 14 PLOS Biology \| www.plosbiology.org Investigating NKT Cell-TCR Signals 54. Maruyama M, Lam KP, Rajewsky K (2000) Memory B-cell persistence is independent of persisting immunizing antigen. Nature 407: 636–642. doi:10.1038/35036600. 53. Boyman O, Krieg C, Homann D, Sprent J (2012) Homeostatic maintenance of T cells and natural killer cells. Cell Mol Life Sci 69: 1597–1608. doi:10.1007/ s00018-012-0968-7. 51. Zhu J, Yamane H, Paul WE (2010) Differentiation of effector CD4 T cell populations (). Annu Rev Immunol 28: 445–489. doi:10.1146/annurev- immunol-030409-101212. 52. Bosma A, Abdel-Gadir A, Isenberg DA, Jury EC, Mauri C (2012) Lipid-antigen presentation by CD1d(+) B cells is essential for the maintenance of invariant natural killer T cells. Immunity 36: 477–490. doi:10.1016/j.im- muni.2012.02.008. References J Immunol 182: 2816–2826. doi:10.4049/jimmunol.0803648. 21. Nagarajan NA, Kronenberg M (2007) Invariant NKT cells amplify the innate immune response to lipopolysaccharide. J Immunol 178: 2706–2713. 46. Witherden D, van Oers N, Waltzinger C, Weiss A, Benoist C, et al. (2000) Tetracycline-controllable selection of CD4(+) T cells: half-life and survival signals in the absence of major histocompatibility complex class II molecules. J Exp Med 191: 355–364. 22. Tyznik AJ, Tupin E, Nagarajan NA, Her MJ, Benedict CA, et al. (2008) Cutting edge: the mechanism of invariant NKT cell responses to viral danger signals. J Immunol 181: 4452–4456. J 23. Matsuda JL, Gapin L, Baron JL, Sidobre S, Stetson DB, et al. (2003) Mouse V alpha 14i natural killer T cells are resistant to cytokine polarization in vivo. Proc Natl Acad Sci USA 100: 8395–8400. doi:10.1073/pnas.1332805100. 47. Capone M, Cantarella D, Schu¨mann J, Naidenko OV, Garavaglia C, et al. (2003) Human invariant V alpha 24-J alpha Q TCR supports the development of CD1d-dependent NK1.1+ and NK1.1- T cells in transgenic mice. J Immunol 170: 2390–2398. 24. Stetson DB, Mohrs M, Reinhardt RL, Baron JL, Wang Z-E, et al. (2003) Constitutive cytokine mRNAs mark natural killer (NK) and NK T cells poised for rapid effector function. J Exp Med 198: 1069–1076. doi:10.1084/ jem.20030630. 48. Inoue K, Wakao H, Ogonuki N, Miki H, Seino K, et al. (2005) Generation of cloned mice by direct nuclear transfer from natural killer T cells. Curr Biol 15: 1114–1118. 49. Manjunath N, Shankar P, Stockton B, Dubey PD, Lieberman J, et al. (1999) A transgenic mouse model to analyze CD8(+) effector T cell differentiation in vivo. Proc Natl Acad Sci USA 96: 13932–13937. j 25. Wang X, Bishop KA, Hegde S, Rodenkirch LA, Pike JW, et al. (2012) Human invariant natural killer T cells acquire transient innate responsiveness via histone H4 acetylation induced by weak TCR stimulation. J Exp Med 209: 987–1000. doi:10.1084/jem.20111024. 50. Huntington ND, Puthalakath H, Gunn P, Naik E, Michalak EM, et al. (2007) Interleukin 15-mediated survival of natural killer cells is determined by interactions among Bim, Noxa and Mcl-1. Nat Immunol 8: 856–863. doi:10.1038/ni1487. 26. Baldwin TA, Sandau MM, Jameson SC, Hogquist KA (2005) The timing of TCR alpha expression critically influences T cell development and selection. J Exp Med 202: 111–121. doi:10.1084/jem.20050359. June 2013 \| Volume 11 \| Issue 6 \| e1001589 PLOS Biology \| www.plosbiology.org 15 Investigating NKT Cell-TCR Signals Investigating NKT Cell-TCR Signals Investigating NKT Cell-TCR Signals 51. Zhu J, Yamane H, Paul WE (2010) Differentiation of effector CD4 T cell populations (). Annu Rev Immunol 28: 445–489. doi:10.1146/annurev- immunol-030409-101212. 55. Brigl M, Bry L, Kent SC, Gumperz JE, Brenner MB (2003) Mechanism of CD1d-restricted natural killer T cell activation during microbial infection. Nat Immunol 4: 1230–1237. doi:10.1038/ni1002. 52. Bosma A, Abdel-Gadir A, Isenberg DA, Jury EC, Mauri C (2012) Lipid-antigen presentation by CD1d(+) B cells is essential for the maintenance of invariant natural killer T cells. Immunity 36: 477–490. doi:10.1016/j.im- muni.2012.02.008. 56. Schmidt-Supprian M, Bloch W, Courtois G, Addicks K, Israel A, et al. (2000) NEMO/IKK gamma-deficient mice model incontinentia pigmenti. Mol Cell 5: 981–992. doi:S1097-2765(00)80263-4. 57. Lee P, Fitzpatrick D, Beard C, Jessup H, Lehar S, et al. (2001) A critical role for Dnmt1 and DNA methylation in T cell development, function, and survival. Immunity 15: 763–774. 53. Boyman O, Krieg C, Homann D, Sprent J (2012) Homeostatic maintenance of T cells and natural killer cells. Cell Mol Life Sci 69: 1597–1608. doi:10.1007/ s00018-012-0968-7. 58. Cox J, Mann M (2008) MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat Biotechnol 26: 1367–1372. doi:10.1038/nbt.1511. 54. Maruyama M, Lam KP, Rajewsky K (2000) Memory B-cell persistence is independent of persisting immunizing antigen. Nature 407: 636–642. doi:10.1038/35036600. PLOS Biology \| www.plosbiology.org PLOS Biology \| www.plosbiology.org June 2013 \| Volume 11 \| Issue 6 \| e1001589 16
https://openalex.org/W2806036740	https://dergipark.org.tr/tr/download/article-file/480986	Turkish	null	Mermer Blokların AHP Destekli TOPSIS ve GİA Yöntemleri ile Sınıflandırılması	Politeknik dergisi	2,018	cc-by-sa	11,543	JOURNAL of POLYTECHNIC ISSN: 1302-0900 (PRINT), ISSN: 2147-9429 (ONLINE) URL: http://dergipark.gov.tr/politeknik Mermer blokların AHP destekli TOPSIS ve GİA yöntemleri ile sınıflandırılması Classification of marble blocks using AHP assisted TOPSIS and GRA methods Yazar(lar) (Author(s)): Metin ERSOY ORCID: 0000-0001-7997-6847 Bu makaleye şu şekilde atıfta bulunabilirsiniz(To cite to this article): Ersoy M., “Mermer blokların AHP destekli TOPSIS ve GİA yöntemleri ile sınıflandırılması”, Politeknik Dergisi, 22(2): 303-317, (2019). Bu makaleye şu şekilde atıfta bulunabilirsiniz(To cite to this article): Ersoy M., “Mermer blokların AHP destekli TOPSIS ve GİA yöntemleri ile sınıflandırılması”, Politeknik Dergisi, 22(2): 303-317, (2019). Erişim linki (To link to this article): http://dergipark.gov.tr/politeknik/archive Erişim linki (To link to this article): http://dergipark.gov.tr/politeknik/archive DOI: 10.2339/politeknik.428979 Politeknik Dergisi, 2019; 22(2) :303-317 Journal of Polytechnic, 2019; 22 (2) :303-317 ÖZ Bu çalışmada, plaka ve levha olarak kesilen mermer blokların, çok kriterli karar verme yöntemlerinden TOPSIS ve GİA yöntemleri kullanılarak sınıflandırılması yapılmıştır. Bu amaçla ilk aşamada, aynı kökene sahip ancak farklı yapısal özellik, şekil ve boyutta 20 adet blok incelenmiştir. İkinci aşamada mermer bloklar, dairesel testereli blok kesici ile plakalara ayrılmış ve kesim süreci boyunca üretimle ilgili veriler not edilmiştir. Elde edilen veriler, mermer blokların sınıflandırılmasında; blok büyüklüğü, süreksizlik sıklığı, düzgünlük, üretim oranı, paledyen oranı ve pasa oranı olmak üzere toplam 6 kriter olduğunu göstermiştir. Belirlenen kriterlerin önem dereceleri, AHP yöntemin ile belirlenmiştir. Son olarak blokların iyiden kötüye doğru sıralamaları, TOPSIS ve GİA yöntemleri ile gerçekleştirilmiştir. Sonuçlar, TOPSIS ve GİA sonuçları blok büyüklüğü arasında artan doğrusal ilişkiyi gösterirken, süreksizlik sıklığı arasında ise azalan yönde doğrusal ilişkiyi göstermiştir. Öte yandan, bazı bloklar için önem sırasının uygulanan yöntemlere göre farklılık gösterdiği tespit edilmiştir. Bu nedenle, sınıflandırma problemlerinin çözümünde sadece ÇKKV yöntemleri yerine problemin farklı yöntemlerle de çözülerek sonuçların değerlendirilmesi önerilmektedir. Anahtar kelimeler: Mermer, çok kriterli karar verme, AHP, TOPSIS, GİA. Sorumlu Yazar (Corresponding Author) e-posta : metinersoy@aku.edu.tr, metinersoy@yandex.com Araştırma Makalesi / Research Article Metin ERSOY Afyon Meslek Yüksekokulu, Motorlu Arç. ve Ulş. Böl. Raylı Sist. Yol Tekn. Prg., Afyon Kocatepe Üniversitesi, Türkiye (Geliş/Received : 19.12.2017 ; Kabul/Accepted : 20.05.2018) Metin ERSOY* Afyon Meslek Yüksekokulu, Motorlu Arç. ve Ulş. Böl. Raylı Sist. Yol Tekn. Prg., Afyon Kocatepe Üniversitesi, Türkiye (Geliş/Received : 19.12.2017 ; Kabul/Accepted : 20.05.2018) Mermer Blokların AHP Destekli TOPSIS ve GİA Yöntemleri ile Sınıflandırılması Araştırma Makalesi / Research Article ABSTRACT In this paper, marble blocks which are cut as slabs or stripes were classified by multi criteria decision methods, TOPSIS and GRA. For this purpose, at the first stage 20 marble blocks from same basin examined their structural features i.e. cracks, fissures, gaps, sizes etc. At the second stage they were cut by circular saw blade machine as stripes. The cutting operations data showed that there are 6 parameters to classify the blocks such as block sizes, discontinuities, block shapes, production, paledyen and waste ratios. Weighting of the criteria were described by AHP. Finally, the blocks ranking was determined by TOPSIS and GIA methods. Results, when TOPSIS and GRA results showed increasing linear relationship with the block sizes decreasing linear relationship with the discontinuity frequencies. On the other hand, it has been determined that order of importance for some blocks differs according to the methods. For this reason, it was recommended that solving and evaluating the classification problems by different methods instead of using only MCDM methods. Keywords: Marble, multi criteria decision, AHP, TOPSIS, GRA. 1. GİRİŞ (INTRODUCTION)  AHP (Analytic Hierarchy Process),  TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution),  GIA (Grey relation analysis)  GIA (Grey relation analysis)  GIA (Grey relation analysis)  ANP (Analitik Network Process),  SAW (Simple Additive Weighting)  ELECTRE (Elimination Et Choix Traduisant La Realite),  ANP (Analitik Network Process),  SAW (Simple Additive Weighting)  ELECTRE (Elimination Et Choix Traduisant La Realite),  DEMATEL (The Decision Making Trial and Evaluation Laboratory Method),  VIKOR (Visekriterijumska Optimizacija I Kompromisno Resenje) ve  MOORA (Multi-Objective Optimization on the basis of Ratio Analysis), MULTI MOORA şeklinde sıralanabilir. Ayrıca, bunların ve listede yer al- mayan diğer yöntemlerin bulanık mantık (Fuzzy sets) ku- ralları çerçevesinde düzenlenmiş çeşitleri de kullanılmaktadır. Literatürde farklı alanlardaki araştırmalarda farklı yön- temler kullanılmıştır. Örneğin; Yiğit ve Gök’ün [2] GİA ve TOPSIS; Tripathy’nin [3] Taguchi, TOPSIS ve GİA; Arıbaş ve Özcan’ın [4] AHP ve TOPSIS; Karaatlı vd. nin [1] SAW, TOPSIS ve GİA; Güneysu vd. nin [5] AHSve GİA; Bektaş ve Tuna’nın [6] GİA; Muralidhar vd. nin [7] GİA ve TOPSIS; Kou vd. nin [8] TOPSIS, ELECTRE III, GİA, VIKOR, ve PROMETHEE II; Şahin ve Ak- yer’in [9] AHS ve TOPSIS ve Dai vd. nin [10] GİA ve TOPSIS yöntemlerini kullandıkları görülmektedir. Doğaltaşlar konusunda farklı çalışmalarda ise; Ersoy ve Yeşilkaya’ın [11] AHP; Eleren ve Ersoy’un [12] Bulanık TOPSIS; Gökçe ve Sonugür’ün [13] ve Caner ve Akars- lan’ın [14] ANFIS ve YSA; Güvenç vd. in [15] ve Ekin- cioğlu vd. nin [16] YSA; Aras ve Gencer’in [17] ve Oruç vd. nin [18] VZA (Veri Zarflama Analizi); Akkoyun ve Toprak’ın [19] ve Yalçın vd. nin [20] Bulanık mantık yöntemlerini kullandıkları görülmektedir. Bir bloktan üretilen levha ve plaka miktarının tahmin edi- lebilmesi için öncelikle, blok büyüklüğü, bloğun geomet- rik şekli (düzgünlük) ve blok üzerindeki süreksizlikler (kırık, çatlak, tabaka düzlemi vb.) gibi parametrelerin bi- linmesi gerekir. Bu parametreler tür, homojenlik, renk ve desenle ilgili unsurlarla birlikte bloğun satış fiyatını be- lirleyen faktörlerdir. Başka bir ifadeyle; büyük, çatlaksız ve düzgün bloklar, diğer özellikleri sabit olmak koşu- luyla, daha değerlidir. Çeşitli problemin çözümünde tek veya birden fazla ÇKKV yönteminin kullanıldığı çok sayıda yayın bulun- masına rağmen mermer bloklarının sınıflandırılmasına yönelik ilgili bir çalışmaya rastlanmamıştır. Çalışma kap- samında mermer blokları AHP, TOPSIS ve GİA yöntem- leri ile sınıflandırılmıştır. Fabrikada işlenecek blokların satın alınması ya da üretim programının hazırlanmasında, aday blokların sınıflandı- rılması gerekir. 1. GİRİŞ (INTRODUCTION) Doğaltaşların üretimi, büyük kütlelerin kesilmesi gerekliliğinden dolayı tehlikeli ve zor bir süreçtir. Ocakta, formasyonun durumuna göre yaklaşık 10x8x3 m. boyutlarındaki kütleler ana kayadan kesilerek alınır. Bu işleme dağ kesme adı verilir ve delme-çatlatma, elmas tel kesme, kollu kesici ile kesme gibi yöntemler uygulanır. Dağ kesme sırasında kütle zaman zaman parçalanır. Bu parçalardan nispeten büyük ve belli bir geometrik şekle sahip olmayanlar moloz olarak adlandırılır. Daha büyük olan diğer parçalar, gerek taşınabilmesi için gerekse fabrikalardaki blok kesici makinalarda işlenebilir hale getirilebilmesi için boyutlandırılır. Bu boyuttaki parçalara blok adı verilir ve yaklaşık 2.5x2.0x1.5 m. boyutlarında dikdörtgen prizma şekillidirler. Doğaltaşlar, süs eşyalarından inşaatlarda döşeme, kaplama ve mutfak tezgahlarına kadar geniş yelpazede kullanım alanı bulan doğal malzemelerdir. Oluşumlarına göre kayaçlar; magmatik, sedimanter ve metamorfik olmak üzere üç ana grupta değerlendirilir. Ticari olarak, renk ve desen bakımından tercih edilebilir olan ve sahadan yeterli büyüklükte blok elde edilebilen her türlü kayaç, mermer tanımı içinde ele alınmaktadır. Doğaltaşlar, oluşum koşullarından (sıcaklık, basınç, kimyasal etkiler, tektonik hareketler vb.), türlerinin farklılığından dolayı renk, desen, kimyasal ve fiziksel özellikleri bakımından farklıdırlar. Bu farklılıklar kullanım alanı bakımından da belirleyici unsurlardır. Kamyonlarla fabrikaya getirilen bloklar dairesel testereli (ST), lamalı (katrak) veya telli (multitel) blok kesme 303 Metin ERSOY /POLİTEKNİK DERGİSİ, Politeknik Dergisi,2019;22(2):303-317 yüzden bloğun sınıflandırılmasında bilimsel yöntemlerin kullanılması gerekir. Çok kriterli karar verme yöntemleri (ÇKKV), sayısal veya sözel ifadelerin kullanılabildiği, en az üç kritere göre değişken alternatifler arasından karar verici için en uygun olanın belirlenmesinde başarı ile uy- gulanan, matematiksel yöntemlerdir. 1960’lı yıllardan günümüze kadar çok çeşitli ÇKKV yöntemi; istatistiksel analizler, performans analizleri, iş sağlığı ve güvenliği analizleri, proje yönetimi, optimizasyon problemlerinin çözümü gibi önemli konularda karar vermede kullanıl- maktadır [1]. ÇKKV yöntemlerinden en yaygın olanları; makinalarında dilimlenerek levha ve plakalara ayrılır. Levhaların kalınlıkları 2-3 cm arasında değişirken; diğer boyutları blok uzunluğu ve genişliği kadardır. Plakalar ise; 2-4 cm kalınlığında, 30, 40 ve 60 cm eninde ve boyu blok uzunluğu kadar olacak şekilde dilimlenir. Sonra levha ve plakalar kullanılacağı yere göre ebatlama, yüzey işleme (parlatma, çekiçleme, ateşle yakma vs.), kenar iş- leme (pah, perdah vs.), seçme (kalite kontrol) ve amba- lajlama gibi aşamalardan geçerek satışa sunulur. makinalarında dilimlenerek levha ve plakalara ayrılır. Levhaların kalınlıkları 2-3 cm arasında değişirken; diğer boyutları blok uzunluğu ve genişliği kadardır. Plakalar ise; 2-4 cm kalınlığında, 30, 40 ve 60 cm eninde ve boyu blok uzunluğu kadar olacak şekilde dilimlenir. 1. GİRİŞ (INTRODUCTION) Sonra levha ve plakalar kullanılacağı yere göre ebatlama, yüzey işleme (parlatma, çekiçleme, ateşle yakma vs.), kenar iş- leme (pah, perdah vs.), seçme (kalite kontrol) ve amba- lajlama gibi aşamalardan geçerek satışa sunulur. Doğaltaş bloklarının sahadan kesilip piyasaya sunulma- sına kadar olan aşamalarda, üretilen malzemenin büyük bir bölümü toz ve parça taş şeklinde kullanılmaz hale ge- lir ve atığa çıkar. Bir bloktan elde edilebilecek kullanıla- bilir ürün miktarı o bloğun verimi olarak tanımlanır. Üretim kayıplarının çoğu, ocak kayıpları dışında blok kesme aşamasında gerçekleşir. Örneğin sadece kesme yüzeyindeki kayıp, kesici eleman (elmas soket) kalınlığı ile kesilen yüzey alanının çarpımı kadardır. Bu değer 1 cm kalınlık için %50, 2 cm kalınlık için %30, 3 cm kalın- lık için %25 den daha büyük değerlerdir. Bir başka kayıp, bloğun makinaya yerleştirilmeden önce düzgün geomet- rik bir şekle (dikdörtgen prizma) getirilmesi sırasında (sayalama) ortaya çıkar. Blok ne kadar şekilsiz ise (dik- dörtgene benzemez ise) kayıplar da o kadar fazla olur. Blok kesici makinada da kesim sırasında da üç farklı aşa- mada kayıplarla karşılaşılır. Bunlardan birincisi bloğun üst yüzeyinin düzleştirilmesi (tarama) işlemidir. Tarama, istenen ölçülerin sağlanabilmesi için gereklidir. İkinci kayıp ön ve arka yüzeylerden kaynaklanır. Bloğun ilk ve son yüzeylerinin tesviyesi düzgün olmadığı için kullanı- lamaz. Üçüncü önemli kayıp, bloğun en alt kısmıdır. Blok, kesildikçe hafifler ve blok kesme makinasının oluş- turduğu gerilmelere karşı koyamaz hale gelir, oturduğu vagon üzerinde kayar. Bunun sonucunda iş kazaları mey- dana gelebileceğinden dolayı, en alt kısımda 10-15’cm lik bölüm kesilmeden bırakılır (kapak). Blok kesiciden alınan levha ve plakaların kenarları düz olmaz. Ebatlama ve yüzey işleme aşamasına geçilmeden önce bu kenar kı- sımlar, baş kesme, yan kesme ve köprülü kesme adı veri- len makinalarla düzeltilir. Bunun sonucunda, kırık parçalar ve paledyen adı verilen, sınırlı kullanım imkânı olan, genellikle üçgen şekilli parçalar oluşur. Kısaca, ka- yıp oranının bu kadar fazla olduğu bir üretim sürecinde; kapasiteyi artırarak üretim miktarının artırılması yerine kayıpların azaltılarak verimin artırılması daha yararlı ola- caktır. 1. GİRİŞ (INTRODUCTION) Sınıflandırma, elde edilecek levha ve pla- kaların miktarını da etkileyeceği için çok önemlidir. Bu 304 MERMER BLOKLARIN AHP DESTEKLİ TKOPSIS VE GIA YÖNTEMLERİ İLE SINIFLANDI … Politeknik Dergisi, 2019; 22 (2) : 303-317 2. MALZEME VE YÖNTEM (MATERIAL AND METHOD) Kesme makinasının bazı özellikleri (Some specifications of block cutter machine) Blok kesme makinasının bazı özellikleri* Dikey testere çapı (mm) 1000–1750 Yatay testere çapı (mm) 450-650 Dikey-Yatay testere motor güçleri (kw) 110-15 Dikey-Yatay testere devirleri (dev/dk) 1500-3000 Dikey testere soket kalınlığı (mm) 8 Kiriş Hareketi motor gücü (kw) 1,1 Kesici araba motor gücü (kw) 2,2 Köprü (aşağı-yukarı) motor gücü (kw) 2,2 Vagon motor gücü (kw) 1,5 Max. Blok ebatları (mm) 3500x2400x2200 Max. Su ihtiyacı (lt/dk.) 112 Toplam elektrik gücü (kw) 130 Makine ağırlığı (kg) 15000 * Firma verileri Çizelge 1. Kesilen blokların bazı özellikleri (Some specifications of blocks) Kesme tesisine getirilen 20 adet blok, gözlem ve incelemeler için kayıt altına alınmıştır. Seçilen blokların büyüklüğü, taşıyıcı kamyonların kantar ağırlıklarından hesaplanmıştır. Sonra tüm örnekler için kesme parametreleri (kesim yönü, derinliği, hızı, testere devri vs.) belirlenmiş ve kesim sonunda elde edilecek ebat ve şekilleri tasarlanarak, hammaddenin piyasaya uygunluğu belirlenmiştir. Daha sonra hazırlanan kesim planı doğrultusunda dilimlenerek 2x30x∞ cm ebatlı plakalar elde edilmiştir. Bu sırada, her bir örnek için harcanan zaman, elde edilen ürün ve ortaya çıkan pasa miktarı not edilmiştir. g specifications of blocks) Blokların bazı özellikleri* Sertlik (Mohs) ≈4,5 Birim hacim ağırlığı (kg/m³) 2690 Özgül ağırlığı (kg/m³) 2700 Atmosfer basıncına su emme (%) 0,2 – 0,4 Kaynar suda su emme (%) 0,3 – 0,7 Porozite (%) 0,4 Basınç direnci (MPa) 114 Don sonrası basınç direnci (MPa) 98 Darbe direnci (kgf. cm/cm³) 20 Eğilme direnci (MPa) 12,2 Elastisite modülü (kgf/cm²) 78,52x104 Doluluk oranı (%) 99,6 Gözeneklilik derecesi (%) 0,4 Aşınma direnci (cm³/50 cm²) 15,8 Çekme direnci (MPa) 79,54 SiO2-Fe2O3-CaO-MgO (%) 0,8-0,1-54,2-0,6 * Firma verileri Alınan veriler ışığında, üretim (satılabilir), paledyen (değersiz ikinci kalite ürün) ve pasa (atık) oranları değerleri tespit edilmiştir. 2. MALZEME VE YÖNTEM (MATERIAL AND METHOD) kriptokristalin kalsitten oluşmuş pellet, değişik tane boylarına sahip kriptokristalin kalsitten oluşmuş intraklast içermektedir. Kayaçtaki bütün çatlaklar mikromezokristalin kalsit dolguludur. Çalışma; mermer blokların seçimi, tesise getirilmesi ve kesme faaliyetleri sırasında elde edilen verilerin ÇKKV yöntemleri ile değerlendirilerek blokların kalitelerine göre sıralanması olarak özetlenebilir. Çalışmanın aşamaları Şekil 1’de akım şeması halinde verilmiştir. Örneklerin seçiminde, ağırlığı 10 ton altı ve üzeri olmak üzere iki ayrı büyüklük belirlenmiş ve küçük olanlar N.1- N.10 arasında kenarları düzgün ve büyük olanlar ise aşamaları Şekil 1 de akım şeması halinde verilmiştir. 2.1. Analitik Çalışmalar (Analytical Studies) Çalışmada kullanılan örnekler, Sivrihisar bej ticari adıyla bilinen ve Eskişehir ili Babadat köyü kuzeybatı istikametindeki mermer ocaklarında üretilen kireçtaşı N.10 arasında kenarları düzgün ve büyük olanla N.11-N.20 arasında numaralandırılmıştır. K işleminde, dört kolonlu dairesel testereli bir blok k kullanılmıştır. Örneklere ait bazı özellikler Çizelge ve blok kesme makinası özellikleri Çizelge Şekil 1. Çalışmanın aşamaları (The steps of study) Şekil 1. Çalışmanın aşamaları (The steps of study) N.11-N.20 arasında numaralandırılmıştır. Kesim işleminde, dört kolonlu dairesel testereli bir blok kesici kullanılmıştır. Örneklere ait bazı özellikler Çizelge 1’de ve blok kesme makinası özellikleri Çizelge 2’de verilmiştir. 2.1. Analitik Çalışmalar (Analytical Studies) N.11-N.20 arasında numaralandırılmıştır. Kesim işleminde, dört kolonlu dairesel testereli bir blok kesici kullanılmıştır. Örneklere ait bazı özellikler Çizelge 1’de ve blok kesme makinası özellikleri Çizelge 2’de verilmiştir. Çalışmada kullanılan örnekler, Sivrihisar bej ticari adıyla bilinen ve Eskişehir ili Babadat köyü kuzeybatı istikametindeki mermer ocaklarında üretilen kireçtaşı türü mermer bloklardır (Şekil 2). Seçilen örnekler, kriptokristalin kalsit kristalleri, mikrofosil kavkı izleri, 305 Metin ERSOY /POLİTEKNİK DERGİSİ, Politeknik Dergisi,2019;22(2):303-317 Şekil 2. Ocakların konumu ve görünümü (Location of the quarries and a view of the area) Şekil 2. Ocakların konumu ve görünümü (Location of the quarries and a view of the area) Çizelge 1. Kesilen blokların bazı özellikleri (Some specifications of blocks) Blokların bazı özellikleri* Sertlik (Mohs) ≈4,5 Birim hacim ağırlığı (kg/m³) 2690 Özgül ağırlığı (kg/m³) 2700 Atmosfer basıncına su emme (%) 0,2 – 0,4 Kaynar suda su emme (%) 0,3 – 0,7 Porozite (%) 0,4 Basınç direnci (MPa) 114 Don sonrası basınç direnci (MPa) 98 Darbe direnci (kgf. cm/cm³) 20 Eğilme direnci (MPa) 12,2 Elastisite modülü (kgf/cm²) 78,52x104 Doluluk oranı (%) 99,6 Gözeneklilik derecesi (%) 0,4 Aşınma direnci (cm³/50 cm²) 15,8 Çekme direnci (MPa) 79,54 SiO2-Fe2O3-CaO-MgO (%) 0,8-0,1-54,2-0,6 * Firma verileri Çizelge 2. 2.2. Çok Kriterli Karar Verme Yöntemleri (Multi Criteria Decision Making Methods) Çizelge 2. Kesme makinasının bazı özellikleri (Some specifications of block cutter machine) Çalışmada ÇKKV yöntemlerinden AHP, blok veriminde etkin kriterlerin önem derecelerinin belirlenmesinde kullanılırken blokların sınıflandırılmasında TOPSIS ve GİA yöntemleri kullanılmıştır. specifications of block cutter machine) Blok kesme makinasının bazı özellikleri* Dikey testere çapı (mm) 1000–1750 Yatay testere çapı (mm) 450-650 Dikey-Yatay testere motor güçleri (kw) 110-15 Dikey-Yatay testere devirleri (dev/dk) 1500-3000 Dikey testere soket kalınlığı (mm) 8 Kiriş Hareketi motor gücü (kw) 1,1 Kesici araba motor gücü (kw) 2,2 Köprü (aşağı-yukarı) motor gücü (kw) 2,2 Vagon motor gücü (kw) 1,5 Max. Blok ebatları (mm) 3500x2400x2200 Max. Su ihtiyacı (lt/dk.) 112 Toplam elektrik gücü (kw) 130 Makine ağırlığı (kg) 15000 * Firma verileri AHP yöntemi; Thomas L. Saaty [21, 22, 23] tarafından önerilen bir ÇKKV yöntemi olup karar hiyerarşisinin tanımlanabilmesi durumunda kullanılan, kararı etkileyen faktörler açısından karar noktalarının yüzde dağılımlarını dikkate alan bir ÇKKV yöntemidir. Uygulama kısaca;  Hiyerarşik yapının oluşturulması,  Önceliklerin belirlenmesi,  İkili karşılaştırma matrisi ve çözümü,  Öncelik vektörünün oluşturulması,  Hiyerarşik yapının oluşturulması,  Önceliklerin belirlenmesi,  İkili karşılaştırma matrisi ve çözümü,  Öncelik vektörünün oluşturulması,  Hiyerarşik yapının oluşturulması, Ö  Önceliklerin belirlenmesi, İkili karşılaştırma matrisi ve çözümü, 306 MERMER BLOKLARIN AHP DESTEKLİ TKOPSIS VE GIA YÖNTEMLERİ İLE SINIFLANDI … Politeknik Dergisi, 2019; 22 (2) : 303-317 sm1 . . . . smn \| →{𝑆: 𝐴𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑓𝑙𝑒𝑟𝑖𝑛 𝑠ü𝑡𝑢𝑛 𝑣𝑒𝑘𝑡ö𝑟ü; 𝐾: 𝐾𝑎𝑟𝑎𝑟 𝑚𝑎𝑡𝑟𝑖𝑠𝑖} (7) L = \| s11 . . . . s1n . . . . sm1 . . . . smn \| x \| w1 .. wn \| →L = \| L11 .. Lm1 \| → {𝐿: 𝐾𝑎𝑟𝑎𝑟 𝑛𝑜𝑘. % 𝑑𝑎ğ. ; ∑Li = 1, Lmax = En iyi alternatif} (8) 6 Karar noktalarındaki sonuç dağılımının bulunması S = \| s1 .. sm \| →K = \| s11 . . . . s1n . . . . sm1 . . . . smn \| →{𝑆: 𝐴𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑓𝑙𝑒𝑟𝑖𝑛 𝑠ü𝑡𝑢𝑛 𝑣𝑒𝑘𝑡ö𝑟ü; 𝐾: 𝐾𝑎𝑟𝑎𝑟 𝑚𝑎𝑡𝑟𝑖𝑠𝑖} (7) L = \| s11 . . . . s1n . . . . sm1 . . . . smn \| x \| w1 .. wn \| →L = \| L11 .. Lm1 \| → {𝐿: 𝐾𝑎𝑟𝑎𝑟 𝑛𝑜𝑘. % 𝑑𝑎ğ. ; ∑Li = 1, Lmax = En iyi alternatif} (8) TOPSIS yöntemi; Hwang ve Yoon [24] referans alınarak geliştirilmiş, seçilen alternatifin pozitif ideal çözüme olabildiğince yakın ve negatif ideal çözüme de uzak olması mantığını temel alan bir ÇKKV yöntemidir. Uygulama kısaca;  İdeal çözüme göreli yakınlığın hesaplanması ve  İdeal çözüme göreli yakınlığın hesaplanması ve  Alternatiflerin sıralanması aşamalarından oluşur. TOPSIS yönteminde kullanılan bağıntılar Çizelge 4’de verilmiştir. aşamalarından oluşur. TOPSIS yönteminde kullanılan bağıntılar Çizelge 4’de verilmiştir.  Karar matrisinin oluşturulması,  Karar matrisinin oluşturulması, Çizelge 4. TOPSIS Eşitlikleri (TOPSIS Equations) No Hesaplama Eşitlik No 1 Karar matrisinin oluşturulması \| xij . . xin . . . . . . xmj . . xmn \| {i = 1, 2, … . n; j = 1, 2, … . m; n: Kriter sayısı; m: Alternatif sayısı} (9) 2 Karar matrisinin normalize edilmesi rij = xij √∑ Xkj 2 m k=1 (10) 3 Karar matrisinin ağırlıklandırılması vij = wj. rij {∑wj = 1} (11) 4 Pozitif ideal ve negatif ideal kümelerin oluşturulması A∗= {(maxvij\|j ∈J); (minvij\|j ∈J′)} A∗= {v1 ∗, v2 ∗, … . vn ∗} A−= {(minvij\|j ∈J); (maxvij\|j ∈J′)} A−= {v1 −, v2 −, … . vn −} {J: Fayda; J′: Kayıp} (12) 5 Ayrım ölçütlerini hesaplanması si ∗= √∑ (vij −vj ∗)2 n j=1 ↔ si −= √∑ (vij −vj −)2 n j=1 (13) 6 İdeal çözüme göreli yakınlığın hesaplanması ci ∗= si − si −+si ∗ {0 ≤ci ∗≤1; 𝑐𝑚𝑎𝑥 ∗ = 𝐸𝑛 𝑖𝑦𝑖 𝑎𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑓} (14) Çizelge 4. MERMER BLOKLARIN AHP DESTEKLİ TKOPSIS VE GIA YÖNTEMLERİ İLE SINIFLANDI … Politeknik Dergisi, 2019; 22 (2) : 303-317 MERMER BLOKLARIN AHP DESTEKLİ TKOPSIS VE GIA YÖNTEMLERİ İLE SINIFLANDI … Politeknik Dergisi, 2019; 22 (2) : 303-317  Tutarlılık oranının hesaplanması ve  Karar noktalarındaki sonuç dağılımının bulunması aşamalarından oluşur. AHP yönteminde kullanılan bağıntılar Çizelge 3’de verilmiştir  Kriterlerin ağırlık değerlerinin belirlenmesi ve karar matrisinin ağırlıklandırılması,  Pozitif ideal ve negatif ideal kümelerin oluşturulması,  Ayrım ölçütlerini hesaplanması, Çi l 3 AHP i likl i (AHP i )  Tutarlılık oranının hesaplanması ve  Karar noktalarındaki sonuç dağılımının bulunması aşamalarından oluşur. AHP yönteminde kullanılan bağıntılar Çizelge 3’de verilmiştir aşamalarından oluşur. AHP yönteminde kullanılan bağıntılar Çizelge 3’de verilmiştir Ç g ş q No Hesaplama Eşitlik No 1 İkili karşılaştırma matrisinin oluşturulması A = \| 1 . . 1 . an . . . . 1/an . 1 . . 1 \| →{n: Faktör sayısı; aij ∈(1 −9), 1: Eşit önem, 5: Çok önemli,9: Mutlak üstün} (1) 2 Öncelik vektörünün oluşturulması cij = aij ∑aij → Ci = \| c1 ..cn \| (2) 3 Öncelik vektörlerinin birleştirilmesi ve ağırlıkların hesaplanması C = \| c11 . . . . c1n . . . . cn1 . . . . cnn \| →wi = ∑ cij n j=1 n →W = \| w1 .. wn \| (3) 4 Tutarlılığın hesaplanması D = \| 1 . . 1 . an . . . . 1/an . 1 . . 1 \| x \| w1 .. wn \| →D = \| d1 .. dn \| (4) 5 Ei = di wi → = ∑ Ei n i=1 n → CI = −n n−1 →{E: Temel değer; :Max temel değer; CI: Tutarlılık göstergesi} (5) CR = CI RI → {RI: Random gösterge; Tutarlılık oranı CR < %10 olmalı} (6) 6 Karar noktalarındaki sonuç dağılımının bulunması S = \| s1 .. sm \| →K = \| s11 . . . . s1n . . . . sm1 . . . . smn \| →{𝑆: 𝐴𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑓𝑙𝑒𝑟𝑖𝑛 𝑠ü𝑡𝑢𝑛 𝑣𝑒𝑘𝑡ö𝑟ü; 𝐾: 𝐾𝑎𝑟𝑎𝑟 𝑚𝑎𝑡𝑟𝑖𝑠𝑖} (7) L = \| s11 . . . . s1n . . . . sm1 . . . . smn \| x \| w1 .. wn \| →L = \| L11 .. Lm1 \| → {𝐿: 𝐾𝑎𝑟𝑎𝑟 𝑛𝑜𝑘. % 𝑑𝑎ğ. ; ∑Li = 1, Lmax = En iyi alternatif} (8) No Hesaplama 1 İkili karşılaştırma matrisinin oluşturulması A = \| 1 . . 1 . an . . . . 1/an . 1 . . MERMER BLOKLARIN AHP DESTEKLİ TKOPSIS VE GIA YÖNTEMLERİ İLE SINIFLANDI … Politeknik Dergisi, 2019; 22 (2) : 303-317 1 \| →{n: Faktör sayısı; aij ∈(1 −9), 1: Eşit önem, 5: 2 Öncelik vektörünün oluşturulması cij = aij ∑aij → Ci = \| c1 ..cn \| 3 Öncelik vektörlerinin birleştirilmesi ve ağırlıkların hesaplanması C = \| c11 . . . . c1n . . . . cn1 . . . . cnn \| →wi = ∑ cij n j=1 n →W = \| w1 .. wn \| 4 Tutarlılığın hesaplanması D = \| 1 . . 1 . an . . . . 1/an . 1 . . 1 \| x \| w1 .. wn \| →D = \| d1 .. dn \| 5 Ei = di wi → = ∑ Ei n i=1 n → CI = −n n−1 →{E: Temel değer; :Max temel değe CR = CI RI → {RI: Random gösterge; Tutarlılık oranı CR < %10 olmalı} 6 Karar noktalarındaki sonuç dağılımının bulunması S = \| s1 .. sm \| →K = \| s11 . . . . s1n . . . . sm1 . . . . smn \| →{𝑆: 𝐴𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑓𝑙𝑒𝑟𝑖𝑛 𝑠ü𝑡𝑢𝑛 𝑣𝑒𝑘𝑡ö𝑟ü L = \| s11 . . . . s1n . . . . sm1 . . . . smn \| x \| w1 .. wn \| →L = \| L11 .. Lm1 \| → {𝐿: 𝐾𝑎𝑟𝑎𝑟 𝑛𝑜𝑘. % 𝑑𝑎ğ. ; ∑L Eşitlik No No Hesaplama Eşitlik No 1 İkili karşılaştırma matrisinin oluşturulması A = \| 1 . . 1 . an . . . . 1/an . 1 . . 1 \| →{n: Faktör sayısı; aij ∈(1 −9), 1: Eşit önem, 5: Çok önemli,9: Mutlak üstün} (1) 2 Öncelik vektörünün oluşturulması cij = aij ∑a → Ci = \| c1 . \| (2) (1) / n . 1 2 Öncelik vektörünün oluşturulması c cij = aij ∑aij → Ci = \| c1 ..cn \| cn 3 Öncelik vektörlerinin birleştirilmesi ve ağırlıkların hesaplanması n 3 Öncelik vektörlerinin birleştirilmesi ve ağırlıkların hesaplanması wi n n 1 R = CI RI → {RI: Random gösterge; Tutarlılık oranı CR < %10 olmalı} (6) i CR = CI RI → {RI: Random gösterge; Tutarlılık oranı CR < %10 olmalı} CR = CI RI → {RI: Random gösterge; Tutarlılık oranı CR < %10 olmalı} RI 6 Karar noktalarındaki sonuç dağılımının bulunması 6 Karar noktalarındaki sonuç dağılımının bulunması S = \| s1 .. sm \| →K = \| s11 . . . . s1n . . . . MERMER BLOKLARIN AHP DESTEKLİ TKOPSIS VE GIA YÖNTEMLERİ İLE SINIFLANDI … Politeknik Dergisi, 2019; 22 (2) : 303-317 TOPSIS Eşitlikleri (TOPSIS Equations) No Hesaplama Eşitlik No 1 Karar matrisinin oluşturulması \| xij . . xin . . . . . . xmj . . xmn \| {i = 1, 2, … . n; j = 1, 2, … . m; n: Kriter sayısı; m: Alternatif sayısı} (9) 2 Karar matrisinin normalize edilmesi rij = xij √∑ Xkj 2 m k=1 (10) 3 Karar matrisinin ağırlıklandırılması vij = wj. rij {∑wj = 1} (11) 4 Pozitif ideal ve negatif ideal kümelerin oluşturulması A∗= {(maxvij\|j ∈J); (minvij\|j ∈J′)} A∗= {v1 ∗, v2 ∗, … . vn ∗} A−= {(minvij\|j ∈J); (maxvij\|j ∈J′)} A−= {v1 −, v2 −, … . vn −} {J: Fayda; J′: Kayıp} (12) 5 Ayrım ölçütlerini hesaplanması si ∗= √∑ (vij −vj ∗)2 n j=1 ↔ si −= √∑ (vij −vj −)2 n j=1 (13) 6 İdeal çözüme göreli yakınlığın hesaplanması ci ∗= si − si −+si ∗ {0 ≤ci ∗≤1; 𝑐𝑚𝑎𝑥 ∗ = 𝐸𝑛 𝑖𝑦𝑖 𝑎𝑙𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑓} (14) Çizelge 4. TOPSIS Eşitlikleri (TOPSIS Equations) (9) mj mn 2 Karar matrisinin normalize edilmesi vij = wj. rij {∑wj = 1} {J: Fayda; J : Kayıp} 5 Ayrım ölçütlerini hesaplanması (14) 307 Metin ERSOY /POLİTEKNİK DERGİSİ, Politeknik Dergisi,2019;22(2):303-317 GİA yöntemi; ilk defa 1982 yılında Julong Deng [25, 26, 27] tarafından ortaya atılmış olup gri teori, gri ilişkisel analiz, gri modelleme, gri tahmin ve gri karar verme gibi alt başlıklar altında, istatistiksel analizlerden proje yönetimine kadar birçok önemli konuda karar vermede kullanılmaktadır [28]. Uygulama kısaca; (kırık, çatlak, fissür, tabaka düzlemleri vb.) Bloğun kesimi sırasında plaka, süreksizlik düzlemleri boyunca kendini bırakır ve plakanın kırılmasına ya da daha küçük boyutlu ürünler üretilmesine neden olur. Verime etki eden bu unsurun da hesaplamalara katılabilmesi için bloğun kesme doğrultusu boyunca kaç adet süreksizlik içerdiği sayılarak belirlenmiş ve kesme doğrultusuna (blok boyu) bölünerek süreksizlik sıklığı (ortalama 1 metreden geçen süreksizlik) hesaplanmıştır.  Karar matrisinin ve referans serinin oluşturulması,  Verilerin normalize edilmesi,  Verilerin normalize edilmesi,  Mutlak değer tablosu oluşturulması (Uzaklık matrisi), Bloğun kesimi sırasında önce tarama adı verilen tesviye işlemi yapılır. Tesviye işlemi sırasında üretim yapılmadığı için harcanan süre ve enerji boşa geçmiş olur. Blok ne kadar düzgün ve küçükse işlem o kadar kısa sürer. MERMER BLOKLARIN AHP DESTEKLİ TKOPSIS VE GIA YÖNTEMLERİ İLE SINIFLANDI … Politeknik Dergisi, 2019; 22 (2) : 303-317 rin) {Normalize edilmiş referans seri; r0 = 1} (17) rij = xij−minxij max xij−min xij {maksimumdaha iyi durumu} (18) rij = maxxij−xij maxxij−minxij {mininimum daha iyi durumu} (19) rij = 1 − \|xij−xj ∗\| max xij−xj ∗ ,xj ∗−minxij {optimum daha iyi durumu; maxxj < xj ∗< minxj} (20) 3 Uzaklık matrisinin oluşturulması ∆ij= \|r0j −rij\| ∆max= max∆ij ∆min= min∆ij (21) 4 Gri ilişkisel katsayıların hesaplanması ε(r0j,rij) = ∆min+ξ∆max ∆ij+ξ∆max {ξ ∈(0 −1)} (22) 5 Gri ilişkisel derecelerin hesaplanması γ(r0j, rij) = 1 n ∑ ε(r0j, rij) n j=1 {0 ≤γ ≤1} (23) 6 Ağırlıklı gri ilişkisel derecelerin hesaplanması γ∗(r0j, rij) = ∑ wj. ε(r0j, rij) n j=1 {∑wj = 1; 0 ≤γ∗≤1} (24) n 6 Ağırlıklı gri ilişkisel derecelerin hesaplanması γ∗(r0j, rij) = ∑ wj. ε(r0j, rij) n j=1 {∑wj = 1; 0 ≤γ∗≤1} 3. BULGULAR (RESULTS) etkileyen bu süre ve plaka kesimi için harcanan süreler, blokların kesimi sırasında not edilmiştir. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) MERMER BLOKLARIN AHP DESTEKLİ TKOPSIS VE GIA YÖNTEMLERİ İLE SINIFLANDI … Politeknik Dergisi, 2019; 22 (2) : 303-317 Makinanın birim zamandaki üretim miktarını  Gri ilişkisel katsayıların hesaplanması,  Gri ilişkisel derecelerin hesaplanması,  Gri ilişkisel derecelerin hesaplanması,  Kriter ağırlıklarının belirlenmesi ve gri ilişkisel derecelerin güncellenmesi  Kriter ağırlıklarının belirlenmesi ve gri ilişkisel derecelerin güncellenmesi aşamalarından oluşur. GİA yönteminde kullanılan bağıntılar Çizelge 5’de verilmiştir. Çizelge 5. GİA Eşitlikleri (GRA Equations) No Hesaplama Eşitlik No 1 Karar matrisinin ve referans serinin oluşturulması \| xij . . xin . . . . . . xmj . . xmn \| {i = 1, 2, … . n; j = 1, 2, … . m; n: Kriter sayısı; m: Alternatif sayısı} (15) xi = (xi1, xi2, … . xin) {Referans seri; xmax: maksimum daha iyi; xmin: minimum daha iyi; x∗: optimum daha iyi} (16) 2 Karar matrisinin normalize edilmesi (gri ilişkisel oluşum) r0 = (ri1, ri2,… . rin) {Normalize edilmiş referans seri; r0 = 1} (17) rij = xij−minxij max xij−min xij {maksimumdaha iyi durumu} (18) rij = maxxij−xij maxxij−minxij {mininimum daha iyi durumu} (19) rij = 1 − \|xij−xj ∗\| max xij−xj ∗ ,xj ∗−minxij {optimum daha iyi durumu; maxxj < xj ∗< minxj} (20) 3 Uzaklık matrisinin oluşturulması ∆ij= \|r0j −rij\| ∆max= max∆ij ∆min= min∆ij (21) 4 Gri ilişkisel katsayıların hesaplanması ε(r0j,rij) = ∆min+ξ∆max ∆ij+ξ∆max {ξ ∈(0 −1)} (22) 5 Gri ilişkisel derecelerin hesaplanması γ(r0j, rij) = 1 n ∑ ε(r0j, rij) n j=1 {0 ≤γ ≤1} (23) 6 Ağırlıklı gri ilişkisel derecelerin hesaplanması γ∗(r0j, rij) = ∑ wj. ε(r0j, rij) n j=1 {∑wj = 1; 0 ≤γ∗≤1} (24) Çizelge 5. GİA Eşitlikleri (GRA Equations) No Hesaplama Eşitlik No 1 Karar matrisinin ve referans serinin oluşturulması \| xij . . xin . . . . . . xmj . . xmn \| {i = 1, 2, … . n; j = 1, 2, … . m; n: Kriter sayısı; m: Alternatif sayısı} (15) xi = (xi1, xi2, … . xin) {Referans seri; xmax: maksimum daha iyi; xmin: minimum daha iyi; x∗: optimum daha iyi} (16) 2 Karar matrisinin normalize edilmesi (gri ilişkisel oluşum) r0 = (ri1, ri2,… . 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Çalışmada kullanılan düzgün ve düzgün olmayan blok örnekleri (Regular and irregular shaped blocks used in the study) plaka boyutunu olumlu etkilediği için büyük değerlerin “faydalı” olduğu ön görülmüştür. Süreksizlik sıklığı ve düzgünlük ise plaka miktarını ve boyutunu ayrıca pasa miktarını “olumsuz” yönde etkileyecektir. Üretim oranı ile ilgili olarak, temel hedefin kaliteli ürün miktarının fazla olması olduğu için büyük değerlerin “faydalı” ol- Kesim sırasında ortaya çıkan atık miktarının belirlenmesi için, önce blokların birim hacim ağırlığından hacimleri hesaplanmış sonra üretilen plakaların toplam hacimleri, boyutları ölçülerek hesaplanmış ve bu iki değerin farkı alınmıştır. Sonra m3 cinsinden hesaplanan atık miktarı blok ağırlığına oranlanarak bloğun pasa oranı (m3/ton) hesaplanmıştır. Çizelge 6. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Blokların kesimi sonrasında elde edilen ebatlı plakalar (1. Kalite yarı mamul) ölçülerek sonra da blok ağırlığına oranlanarak, her bir blok için m2/ton cinsinden üretim oranları hesaplanmıştır. Ayrıca plakaların ebatlanması ya da kenarlarının düzeltilmesi sırasında, çoğunlukla üçgen şekilli parçalar ortaya çıkar. Bu parça plakalar paledyen olarak adlandırılır ve piyasada düşük fiyata alıcı bulur. Bu yüzden, ortaya çıkan paledyen miktarı da ölçülerek blok ağırlığına oranlanmış ve m2/ton cinsinden paledyen oranları hesaplanmıştır. Mermer blok üretiminde ilk hedef mümkün olduğunca büyük kütlelerin üretilmesidir. Blok kesici makinalarda bir defada daha büyük hacimlerin kesilmesi, makinanın blok bazında çalışma verimini artırır dolayısıyla büyük bloklar daha iyi fiyatlara alıcı bulabilir. Birçok blok kesme makinası, dikdörtgen prizma şekilli malzemelerin kesilmesi için tasarlanmıştır. Düzgün şekle sahip olmayan blokların kesiminde, makina kesme bölgesinde boşluklar olacağı için ancak daha küçük ebatlı plakalar üretilebilir. Dolayısıyla bloğun düzgünlüğü (dikdörtgen prizmaya benzerliği) plaka verimine etken parametrelerdendir. Bu çalışma için 1-5 arasında bir düzgünlük değeri belirlenmiş ve seçilen bloklardan tam anlamıyla geometrik şekle sahip olanlar 5 puan, bir kısmı düzgün olan/olmayanlar 3 puan ve hiç düzgün olmayanlar 1 puan ile derecelendirilmiştir (Şekil 2). Blokların kesimi sırasında kesici elmas soketlerin kalınlığından dolayı kesilen yüzey alanı boyunca toz oluşur. Toz atıklar kısmen endüstride kullanılsa da ciddi bir ekonomik değeri yoktur, genellikle atık sahalarında depolanır. Bunun dışında bloğun ön ve arka yüzeyleri düz olmadığı için ilk ve son yüzey kesimleri de atık olarak depolanır. Ayrıca, kesim sonuna yaklaşıldığında, ağırlık azaldığı için vagon üzerindeki blok, kesme ünitesinin itme gücüne karşı koyamaz ve kayar. Bu da ya kalınlık ölçülerinin tutturulamamasına ya da çeşitli Blokların plaka verimine etken bir başka parametre de tektonik hareketlerden kaynaklanan süreksizliklerdir 308 MERMER BLOKLARIN AHP DESTEKLİ TKOPSIS VE GIA YÖNTEMLERİ İLE SINIFLANDI … Politeknik Dergisi, 2019; 22 (2) : 303-317 Ölçüm ve hesaplama sonuçları Çizelge 6’da verilmiştir. Blokların kesme verimi ile ilgili olarak 6 kriter belirlenmiştir. Bunlar; blok büyüklüğü, süreksizlik sıklığı, düzgünlük, üretim oranı, paledyen oranı ve pasa oranıdır. Karar tabloları hazırlanırken, blok büyüklüğü makinanın bir defada yapabileceği üretim miktarını ve kazalara neden olabilir. Bu yüzden blok tabanından itibaren 10-15 cm lik kısım kesilmeden bırakılır. Bu kalan parça, kapak olarak adlandırılır ve bazı atölye çalışanları tarafından hediyelik eşya yapımı için kullanılabilir. Ancak tüketim çok düşük olduğu için genellikle atık olarak depolanır. Şekil 3. Çalışmada kullanılan düzgün ve düzgün olmayan blok örnekleri (Regular and irregular shaped blocks used in the study) Şekil 3. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Atık oranı ise, hem ekonomik değer ifade etmemesinden hem de depolama zorunluluğundan dolayı “olumsuz” olarak değerlendiril- miştir.  Paledyen oranı (C5); blok büyüklüğüne (C1) göre 3, düzgünlük ve pasa oranına göre 2.  Pasa oranı (C6); blok büyüklüğüne (C1) göre 4, düzgünlüğe (C6) göre 3. Bu kabuller doğrultusunda hazırlanan karar tabloları TOPSIS ve GİA yöntemleri ile analiz edilmeden önce, belirlenen kriterlerin önem dereceleri AHP yöntemi yardımıyla hesaplanmıştır. Eşitlik 2 yardımıyla matris normalize edilmiş ve Eşitlik 3 yardımıyla öncelik vektörü oluşturulmuştur. İkili karşılaştırma matrisi ve hesaplama sonuçları Çizelge 8’de verilmiş olup w değeri her bir kriterin önceliğini yani ağırlığını ifade etmektedir. 3.2. Kriter Ağırlıklarının AHP Yöntemine Göre Hesaplanması (Calculation of Criteria Weighting by AHP Method) Oluşturulan ikili karşılaştırma matrisinin dolayısıyla hesaplanan kriter ağırlıklarının doğruluğunun ölçülmesi için tutarlılık analizi yapılmıştır. Bunun için önce, ilk matris ile öncelik vektörünün matris çarpımı hesaplanmış, sonra bulunan sonuç kriter ağırlığına oranlanarak her bir değerlendirme faktörüne ilişkin temel değerler elde edilmiştir (Eşitlik 4). Maksimum temel AHP yöntemine göre birinci adımda ikili karşılaştırma matrisi oluşturulmuştur (Eşitlik 1). Bu matris 6x6 boyutunda bir kare matris olup kriterlerin üstünlük dereceleri Saaty [22] tarafından ortaya atılan “1-9” ölçeği kullanılarak belirlenmiştir (Çizelge 7). Bu ölçeğe göre Çizelge 7. Karşılaştırma ölçeği (Comparison scale) [22] Önem Tanımlama Açıklama 1 Her iki faktörün eşit öneme sahip olması durumu İki faaliyet amaca eşit düzeyde katkıda bulunur. 3 1. Faktörün 2. faktörden daha önemli olması durumu Tecrübe ve yargı ile bir faaliyet diğerine göre biraz daha fazla derecede tercih edilir. 5 1. Faktörün 2. faktörden çok önemli olması durumu Tecrübe ve yargı ile bir faaliyet diğerine göre kuvvetli derecede tercih edilir. 7 1. Faktörün 2. faktöre nazaran çok güçlü bir öneme sahip olması durumu Bir faaliyet çok kuvvetli bir şekilde tercih edilir ve baskınlığı uygulamada rahatlıkla görülür. 9 1. Faktörün 2. faktöre nazaran mutlak üstün bir öneme sahip olması durumu Bir faaliyetin diğerine tercih edilmesine ilişkin kanıtlar çok büyük bir güvenilirliğe sahiptir. 2,4,6,8 Ara değerler Uzlaşma gerektiğinde kullanmak üzere iki ardışık yargı arasına düşen değerler. Çizelge 7. Karşılaştırma ölçeği (Comparison scale) [22] Çizelge 7. Karşılaştırma ölçeği (Comparison scale) [22] değer (temel değerlerin ortalaması, λ) ve tutarlılık göstergesi (CI), Eşitlik 5 yardımıyla hesaplanmıştır. CR tutarlılık oranı, tutarlılık göstergesinin rassal göstergeye oranıdır ve bu değer %10 dan düşük olduğu durumlarda hesaplamaların doğru olduğu kabul edilir (Eşitlik 6) [29]. Rassal gösterge, matris boyutuna göre değişen bir katsayı olup 6x6 boyutlu bir matris için değeri 1.24’tür. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Blok kesme verileri (Block cutting data) Ölçüm verileri Üretim verileri Verim Büyüklük (ton) Boy (Kesme uzunluğu, m) Süreksizlikler (adet) Süreksizlik sıklığı (1/m) Düzgünlük (1-5) Tarama (saat) Kesme (saat) Üretim miktarı (m2) Paledyen üretimi (≈m2) Pasa üretimi (m3) Üretim oranı (m2/ton) Paledyen oranı (m2/ton) Pasa (Atık) oranı (m3/ton) N.1 4,67 1,40 3 2,14 3 1,78 2,80 21,51 1,89 1,27 4,61 0,40 0,27 N.2 3,77 1,50 3 2,00 3 1,97 2,03 19,80 1,45 0,98 5,25 0,38 0,26 N.3 7,52 2,40 4 1,67 1 2,03 2,88 32,70 4,21 2,06 4,35 0,56 0,27 N.4 7,24 2,30 3 1,30 2 1,17 3,83 33,92 4,93 1,91 4,69 0,68 0,26 N.5 7,29 2,15 3 1,40 1 2,05 4,78 29,22 5,08 2,02 4,01 0,70 0,28 N.6 7,75 1,20 2 1,67 1 2,08 2,92 28,72 9,42 2,12 3,71 1,22 0,27 N.7 6,29 1,50 4 2,67 3 1,67 4,33 28,36 2,13 1,73 4,51 0,34 0,27 N.8 6,63 2,20 4 1,82 1 2,42 4,58 25,14 3,62 1,89 3,79 0,55 0,28 N.9 2,28 2,20 3 1,36 2 0,58 3,08 12,00 1,16 0,58 5,27 0,51 0,26 N.10 7,55 1,50 2 1,33 2 4,08 5,92 34,94 10,17 1,90 4,63 1,35 0,25 N.11 19,78 2,90 3 1,03 4 3,67 12,33 147,60 5,16 4,30 7,46 0,26 0,22 N.12 16,05 2,70 2 0,74 5 3,83 11,17 137,60 4,44 3,13 8,57 0,28 0,19 N.13 17,89 2,50 2 0,80 5 4,25 13,75 172,24 3,27 3,14 9,63 0,18 0,18 N.14 23,87 2,90 1 0,34 3 4,58 15,42 186,39 17,46 4,80 7,81 0,73 0,20 N.15 13,56 2,10 1 0,48 5 5,33 4,50 108,00 6,75 2,75 7,97 0,50 0,20 N.16 16,79 2,60 4 1,54 4 3,33 10,67 116,10 3,27 3,85 6,92 0,19 0,23 N.17 18,88 2,60 1 0,38 3 2,83 13,17 134,64 14,17 4,04 7,13 0,75 0,21 N.18 15,84 2,30 1 0,43 3 3,33 10,00 119,40 9,60 3,31 7,54 0,61 0,21 N.19 25,18 2,60 1 0,38 5 5,50 12,50 230,40 5,78 4,64 9,15 0,23 0,18 N.20 17,04 2,20 1 0,45 5 3,58 12,42 144,00 7,98 8,87 8,45 0,47 0,52 Çizelge 6. Blok kesme verileri (Block cutting data) Çizelge 6. Blok kesme verileri (Block cutting data) 309 Metin ERSOY /POLİTEKNİK DERGİSİ, Politeknik Dergisi,2019;22(2):303-317 paledyen oranına (C5) göre çok önemli 5 ve süreksizlik sıklığına (C2) göre 1. paledyen oranına (C5) göre çok önemli 5 ve süreksizlik sıklığına (C2) göre 1. olduğu, paledyenin ikincil ürün olmasından ve ekonomik değerinin düşük olmasından dolayı büyük değerlerin “olumsuz” olduğu ön görülmüştür. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Normalleştirme sürecinde (gri ilişkisel oluşum), en sık kullanılan yöntemlerden birisi olan lineer veri ön işleme yönteminden yararlanılmış ve kriterlerin maksimum daha iyi (C1, C3, C4) durumunda Eşitlik 18, minimum daha iyi (C2, C5, C6) durumunda da Eşitlik 19 kullanılmıştır. Referans serinin normalleştirilmiş değeri daima “1” olarak hesaplanır (Eşitlik 16). GİA yönteminde sonraki aşama, referans seri ile karşılaştırılacak m tane serinin tanımlanması ve mutlak değer tablosunun oluşturulmasıdır. Mutlak değer tablosu, Eşitlik 21 de verildiği gibi, normalleştirilmiş karar matrisindeki kriter elemanlarının referans seri ile farkları hesaplanarak oluşturulmuştur (Çizelge 13). TOPSIS yönteminde son aşama, ayrım ölçütlerinin (Eşitlik 13) ve ideal çözüme göre yakınlığın (Eşitlik 14) hesaplanarak öncelik sıralamasının yapılmasıdır (Çizelge Son olarak, gri ilişkisel katsayılar ve gri ilişkisel dereceler hesaplanmıştır. Gri ilişkisel katsayıların Çizelge 10. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Tutarlılığın ölçülmesiyle ilgili hesaplamalar Çizelge 9’da verilmiştir Çizelge 11’e göre, en iyi blokların sırasıyla N.19, N.13 ve N.15 olduğu, en kötü blokların da N.7, N.1 ve N.6 olduğu söylenebilir. Çizelge 11’e göre, en iyi blokların sırasıyla N.19, N.13 ve N.15 olduğu, en kötü blokların da N.7, N.1 ve N.6 olduğu söylenebilir. Sonuç olarak, 0.37 ile C4 (üretim oranı) en önemli kriter olurken, 0.05 ile C1 (büyüklük) ve C3 (düzgünlük) kriterlerinin diğerlerine göre önemlerinin daha düşük olduğu görülmektedir. Sonuç olarak, 0.37 ile C4 (üretim oranı) en önemli kriter olurken, 0.05 ile C1 (büyüklük) ve C3 (düzgünlük) kriterlerinin diğerlerine göre önemlerinin daha düşük olduğu görülmektedir. 3.4. Blokların GİA Yöntemine Göre Sınıflandırıl- ması (Classification of The Blocks by GRA Method) 3.3. Blokların TOPSIS Yöntemine Göre Sınıflan- dırılması (Classification of The Blocks by TOPSIS Method GİA yönteminde birinci adım karar matrisinin oluşturulması ve referans serinin belirlenmesidir. Karar matrisi olarak TOPSIS yönteminde oluşturulan matris kullanılmıştır. Referans seri ise ideal kriterleri içeren sanal ve/veya gerçek bir seridir. Dolayısıyla referans seri, mevcut alternatiflerin maksimum, minimum veya optimum değerlerinin faydalı olduğu kriterlerinden oluşturulabildiği gibi karar verici tarafından belirlenen hayali bir alternatifin kriterlerinden de oluşturulabilir [30]. Eşitlik 15 ve Eşitlik 16 ya göre oluşturulan karar matrisi ve referans seri Çizelge 12’de verilmiştir. TOPSIS (İdeal Çözüm ile Benzerlik Tercihi Tekniği) yöntemine göre önce karar matrisi oluşturulmuştur. Karar matrisinin satırları alternatifleri (incelenen bloklar) sütunları ise blokların sınıflandırılmasında etken kriteri (Ci; {i=1, 2, …. 6}) ifade etmektedir (Eşitlik 9). Sonra Eşitlik 10 kullanılarak matris elemanları normalize edilmiş ve standart karar matrisi oluşturulmuştur (Çizelge 10). Standart karar matrisinin elemanları, önceki bölümde AHP yöntemi ile hesaplanan kriter ağırlıkları ile çarpılarak ağırlıklı karar matrisi elde edilmiştir (Eşitlik 11). Ağırlıklı karar matrisinde her bir kriter için fayda ya da kayıp getirisine göre pozitif ideal ve negatif ideal çözümler oluşturulmuştur. Burada C1 (büyüklük), C3 (düzgünlük) ve C4 (üretim oranı) fayda sağladığı için bu kriterlerin en büyük değerleri, C2 (süreksizlik sıklığı), C5 (paledyen üretim oranı) ve C6 (pasa oranı) kayıplarla ilgili kriterler olduğu için bu kriterlerin en küçük değerleri pozitif ideal çözüm (A) olarak belirlenmiştir. Aynı şekilde, C1, C3 ve C4 kriterlerinin en küçük değerleri ile C2, C5 ve C6 kriterlerinin en büyük değerleri de negatif ideal çözüm kümesinin elemanlarıdır. Kriterleri tanımlayan veriler farklı birimlerle ölçüldüğü için, karar matrisi normalize edilerek aynı birime dönüştürülmüştür. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Yapılan hesaplamalar sonucunda CR tutarlılık oranının 0.0764  Blok büyüklüğü (C1); blok düzgünlüğüne (C3) göre 2.  Süreksizlik sıklığı (C2); blok büyüklüğüne (C1) ve pasa oranına (C6) göre 6, düzgünlüğe (C3) ve paledyen oranına (C5) göre 4.  Üretim oranı (C4); blok büyüklüğüne (C1) göre 7, düzgünlük (C3) ve pasa oranına (C6) göre 6,  Üretim oranı (C4); blok büyüklüğüne (C1) göre 7, düzgünlük (C3) ve pasa oranına (C6) göre 6, Çizelge 8. Kriter ağırlıklarının belirlenmesi (Definition of the criteria weights) Kriterler İkili karşılaştırma matrisi Kriterlerin % dağılımları ve öncelik vektörü C1 C2 C3 C4 C5 C6 C1 C2 C3 C4 C5 C6 w Büyüklük (ton) C1 1 1/6 2 1/7 1/3 1/4 C1 0,05 0,06 0,11 0,05 0,03 0,02 0,05 Süreksizlik sıklığı (1/m) C2 6 1 4 1 4 6 C2 0,28 0,35 0,22 0,37 0,35 0,39 0,33 Düzgünlük (1-5) C3 1/2 1/4 1 1/6 1/2 1/3 → C3 0,02 0,09 0,06 0,06 0,04 0,02 → 0,05 Üretim oranı (m2/ton) C4 7 1 6 1 5 6 C4 0,33 0,35 0,33 0,37 0,44 0,39 0,37 Paledyen oranı (m2/ton) C5 3 1/4 2 1/5 1 2 C5 0,14 0,09 0,11 0,07 0,09 0,13 0,11 Pasa oranı (m3/ton) C6 4 1/6 3 1/6 1/2 1 C6 0,19 0,06 0,17 0,06 0,04 0,06 0,10 Toplam 1 1 1 1 1 1 1,00 Çizelge 9. Tutarlılığın ölçülmesi (Calculation of consistency) 1 1/6 2 1/7 1/3 1/4 0,05 0,32 6,0395 λ= 6,4740 6 1 4 1 4 6 0,33 2,21 6,7461 CI= 0,0948 1/2 1/4 1 1/6 1/2 1/3 X 0,05 = 0,30 → 6,1787 CR= 0,0764 7 1 6 1 5 6 0,37 2,47 6,6903 3 1/4 2 1/5 1 2 0,11 0,71 6,7658 CR<%10 4 1/6 3 1/6 1/2 1 0,10 0,62 6,4234 Tutarlıdır Çizelge 8. Kriter ağırlıklarının belirlenmesi (Definition of the criteria weights) riterler İkili karşılaştırma matrisi Kriterlerin % dağılımları ve önceli 310 MERMER BLOKLARIN AHP DESTEKLİ TKOPSIS VE GIA YÖNTEMLERİ İLE SINIFLANDI … Politeknik Dergisi, 2019; 22 (2) : 303-317 11). İdeal çözüme yakınlık (C), 0 ile 1 arasında değişir ve 1’e en yakın değer en ideal alternatifi gösterir. olduğu görülmüştür. Bu durumda ikili karşılaştırmanın tutarlı olduğu söylenebilir. Tutarlılığın ölçülmesiyle ilgili hesaplamalar Çizelge 9’da verilmiştir Sonuç olarak, 0.37 ile C4 (üretim oranı) en önemli kriter olurken, 0.05 ile C1 (büyüklük) ve C3 (düzgünlük) kriterlerinin diğerlerine göre önemlerinin daha düşük olduğu görülmektedir. olduğu görülmüştür. Bu durumda ikili karşılaştırmanın tutarlı olduğu söylenebilir. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Gri ilişkisel derecelerin hesaplanması, kriter ağırlıklarının eşit olması durumunda gri ilişkisel katsayıların ortalaması hesaplanarak (Eşitlik 23), kriter ağırlıklarının verilmiş olması durumunda da her bir gri ilişkisel katsayı ile ilgili kriter ağırlığının çarpılması ve satır boyunca toplanması şeklinde yapılır (Eşitlik 24). Gri ilişkisel dereceler de 0 ile 1 arasında değişir ve değeri en Çizelge 14’e göre, en iyi blokların sırasıyla N.19, N.13 ve N.12 olduğu, en kötü blokların da N.6, N.8 ve N.7 olduğu söylenebilir. Çizelge 11. Standart karar matrisinin çözümü (Solving the standard decision matrix) Ağırlıklı karar matrisi Ayrım ölçütleri İdeal çözüme yakınlık Öncelik sıralaması C1 C2 C3 C4 C5 C6 w 0,05 0,33 0,05 0,37 0,11 0,10 S* S- C* N.1 0,004 0,115 0,010 0,058 0,015 0,023 0,117 0,052 0,31 19 N.2 0,003 0,107 0,010 0,066 0,015 0,021 0,106 0,059 0,36 16 N.3 0,006 0,089 0,003 0,055 0,021 0,023 0,100 0,065 0,39 15 N.4 0,006 0,070 0,007 0,059 0,026 0,022 0,085 0,081 0,49 12 N.5 0,006 0,075 0,003 0,050 0,026 0,023 0,095 0,075 0,44 14 N.6 0,006 0,089 0,003 0,047 0,046 0,023 0,112 0,058 0,34 18 N.7 0,005 0,143 0,010 0,057 0,013 0,023 0,141 0,045 0,24 20 N.8 0,006 0,097 0,003 0,048 0,021 0,024 0,111 0,058 0,34 17 N.9 0,002 0,073 0,007 0,066 0,019 0,021 → 0,081 0,082 0,50 11 N.10 0,006 0,071 0,007 0,058 0,051 0,021 0,095 0,076 0,44 13 N.11 0,017 0,055 0,013 0,094 0,010 0,018 0,046 0,112 0,71 9 N.12 0,013 0,040 0,016 0,108 0,010 0,016 0,026 0,130 0,83 4 N.13 0,015 0,043 0,016 0,121 0,007 0,015 0,025 0,136 0,84 2 N.14 0,020 0,018 0,010 0,098 0,028 0,017 0,032 0,140 0,82 5 N.15 0,011 0,025 0,016 0,100 0,019 0,017 0,027 0,136 0,83 3 N.16 0,014 0,082 0,013 0,087 0,007 0,019 0,073 0,089 0,55 10 N.17 0,016 0,021 0,010 0,090 0,028 0,018 0,039 0,135 0,77 8 N.18 0,013 0,023 0,010 0,095 0,023 0,017 0,033 0,135 0,80 6 N.19 0,021 0,021 0,016 0,115 0,009 0,015 0,007 0,151 0,96 1 N.20 0,014 0,024 0,016 0,106 0,018 0,043 0,035 0,138 0,80 7 Pozitif ideal (A) ve negatif ideal (A-) çözümler A 0,021 0,018 0,016 0,121 0,007 0,015 A- 0,002 0,143 0,003 0,047 0,051 0,043 Çizelge 12. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Karar matrisi ve standart karar matrisinin oluşturulması (Creation of decision and standard decision matrixes) Karar matrisi Standart karar matrisi C1 C2 C3 C4 C5 C6 C1 C2 C3 C4 C5 C6 N.1 4,67 2,14 3,00 4,61 0,40 0,27 0,074 0,350 0,199 0,157 0,145 0,232 N.2 3,77 2,00 3,00 5,25 0,38 0,26 0,060 0,327 0,199 0,179 0,138 0,222 N.3 7,52 1,67 1,00 4,35 0,56 0,27 0,120 0,272 0,066 0,148 0,201 0,234 N.4 7,24 1,30 2,00 4,69 0,68 0,26 0,115 0,213 0,133 0,160 0,245 0,226 N.5 7,29 1,40 1,00 4,01 0,70 0,28 0,116 0,228 0,066 0,137 0,251 0,237 N.6 7,75 1,67 1,00 3,71 1,22 0,27 0,123 0,272 0,066 0,126 0,437 0,234 N.7 6,29 2,67 3,00 4,51 0,34 0,27 0,100 0,435 0,199 0,154 0,122 0,235 N.8 6,63 1,82 1,00 3,79 0,55 0,28 0,106 0,297 0,066 0,129 0,196 0,244 N.9 2,28 1,36 2,00 5,27 0,51 0,26 → 0,036 0,223 0,133 0,180 0,183 0,219 N.10 7,55 1,33 2,00 4,63 1,35 0,25 0,120 0,218 0,133 0,158 0,484 0,216 N.11 19,78 1,03 4,00 7,46 0,26 0,22 0,315 0,169 0,265 0,254 0,094 0,186 N.12 16,05 0,74 5,00 8,57 0,28 0,19 0,256 0,121 0,332 0,292 0,099 0,167 N.13 17,89 0,80 5,00 9,63 0,18 0,18 0,285 0,131 0,332 0,328 0,066 0,150 N.14 23,87 0,34 3,00 7,81 0,73 0,20 0,380 0,056 0,199 0,266 0,263 0,172 N.15 13,56 0,48 5,00 7,97 0,50 0,20 0,216 0,078 0,332 0,272 0,179 0,173 N.16 16,79 1,54 4,00 6,92 0,19 0,23 0,267 0,251 0,265 0,236 0,070 0,196 N.17 18,88 0,38 3,00 7,13 0,75 0,21 0,301 0,063 0,199 0,243 0,270 0,183 N.18 15,84 0,43 3,00 7,54 0,61 0,21 0,252 0,071 0,199 0,257 0,218 0,179 N.19 25,18 0,38 5,00 9,15 0,23 0,18 0,401 0,063 0,332 0,312 0,083 0,157 N.20 17,04 0,45 5,00 8,45 0,47 0,52 0,272 0,074 0,332 0,288 0,168 0,445 311 Metin ERSOY /POLİTEKNİK DERGİSİ, Politeknik Dergisi,2019;22(2):303-317 büyük olan seçenek en iyi alternatif olarak değerlendirilir. hesaplanmasında kullanılan Eşitlik 22 deki ξ katsayısı, ∆ij ile ∆max arasındaki farkın en uç değer olma ihtimalini ortadan kaldırmak için kullanılır ve 0 ile 1 arasında değişir [31]. Bu çalışmada ξ=0.5 alınmıştır. Bu çalışmada kriter ağırlıkları AHP yöntemi ile belirlendiği için Eşitlik 24 kullanılmış ve ağırlıklı gri ilişkisel dereceler hesaplanmıştır. Gri ilişkisel katsayılar, gri ilişkisel dereceler ve öncelik sıralaması Çizelge 14’de verilmiştir. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Gri ilişkisel oluşum ve mutlak değer tablosu (The tables of grey relation generating and distance matrix) Normalize edilmiş karar matrisi Mutlak değer (uzaklık) tablosu C1 C2 C3 C4 C5 C6 C1 C2 C3 C4 C5 C6 Referans seri 1 1 1 1 1 1 0,00 0,00 0,00 0,00 0,00 0,00 N.1 0,10 0,23 0,50 0,15 0,81 0,72 0,90 0,77 0,50 0,85 0,19 0,28 N.2 0,07 0,29 0,50 0,26 0,83 0,76 0,93 0,71 0,50 0,74 0,17 0,24 N.3 0,23 0,43 0,00 0,11 0,68 0,72 0,77 0,57 1,00 0,89 0,32 0,28 N.4 0,22 0,59 0,25 0,17 0,57 0,74 0,78 0,41 0,75 0,83 0,43 0,26 N.5 0,22 0,55 0,00 0,05 0,56 0,70 0,78 0,45 1,00 0,95 0,44 0,30 N.6 0,24 0,43 0,00 0,00 0,11 0,72 0,76 0,57 1,00 1,00 0,89 0,28 N.7 0,18 0,00 0,50 0,13 0,87 0,71 0,82 1,00 0,50 0,87 0,13 0,29 N.8 0,19 0,37 0,00 0,01 0,69 0,68 0,81 0,63 1,00 0,99 0,31 0,32 N.9 0,00 0,56 0,25 0,26 0,72 0,77 → 1,00 0,44 0,75 0,74 0,28 0,23 N.10 0,23 0,57 0,25 0,16 0,00 0,78 0,77 0,43 0,75 0,84 1,00 0,22 N.11 0,76 0,70 0,75 0,63 0,93 0,88 0,24 0,30 0,25 0,37 0,07 0,12 N.12 0,60 0,83 1,00 0,82 0,92 0,94 0,40 0,17 0,00 0,18 0,08 0,06 N.13 0,68 0,80 1,00 1,00 1,00 1,00 0,32 0,20 0,00 0,00 0,00 0,00 N.14 0,94 1,00 0,50 0,69 0,53 0,93 0,06 0,00 0,50 0,31 0,47 0,07 N.15 0,49 0,94 1,00 0,72 0,73 0,92 0,51 0,06 0,00 0,28 0,27 0,08 N.16 0,63 0,49 0,75 0,54 0,99 0,84 0,37 0,51 0,25 0,46 0,01 0,16 N.17 0,73 0,98 0,50 0,58 0,51 0,89 0,27 0,02 0,50 0,42 0,49 0,11 N.18 0,59 0,96 0,50 0,65 0,64 0,90 0,41 0,04 0,50 0,35 0,36 0,10 N.19 1,00 0,98 1,00 0,92 0,96 0,97 0,00 0,02 0,00 0,08 0,04 0,03 N.20 0,64 0,95 1,00 0,80 0,76 0,00 0,36 0,05 0,00 0,20 0,24 1,00 Çizelge 14. Gri ilişkisel katsayılar, gri ilişkisel dereceler ve öncelik sıralaması (Grey relation coefficients, grey relation ranks and priority orders) Çizelge 13. Gri ilişkisel oluşum ve mutlak değer tablosu (The tables of grey relation generating and distance matrix) 4. TARTIŞMA VE SONUÇ (DISCUSSION AND CONCLUSIONS) TOPSIS ve GİA sonuçları karşılaştırıldığında, aynı veriler ve kriter ağırlıkları kullanılmış olmasına rağmen N.4, N.9, N.11, N.13, N.16, N.17, N.19 numaralı bloklar dışındaki 13 adet bloğun sıralamadaki yerlerinin farklı olduğu görülmektedir (Çizelge 15). 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Karar matrisi ve referans seri (Decision matrix and reference set) C1 C2 C3 C4 C5 C6 Maksimum iyi Minimum iyi Maksimum iyi Maksimum iyi Minimum iyi Minimum iyi Referans seri 25,18 0,34 5,00 9,63 0,18 0,18 N.1 4,67 2,14 3,00 4,61 0,40 0,27 N.2 3,77 2,00 3,00 5,25 0,38 0,26 N.3 7,52 1,67 1,00 4,35 0,56 0,27 N.4 7,24 1,30 2,00 4,69 0,68 0,26 N.5 7,29 1,40 1,00 4,01 0,70 0,28 N.6 7,75 1,67 1,00 3,71 1,22 0,27 N.7 6,29 2,67 3,00 4,51 0,34 0,27 N.8 6,63 1,82 1,00 3,79 0,55 0,28 N.9 2,28 1,36 2,00 5,27 0,51 0,26 N.10 7,55 1,33 2,00 4,63 1,35 0,25 N.11 19,78 1,03 4,00 7,46 0,26 0,22 N.12 16,05 0,74 5,00 8,57 0,28 0,19 N.13 17,89 0,80 5,00 9,63 0,18 0,18 N.14 23,87 0,34 3,00 7,81 0,73 0,20 N.15 13,56 0,48 5,00 7,97 0,50 0,20 N.16 16,79 1,54 4,00 6,92 0,19 0,23 N.17 18,88 0,38 3,00 7,13 0,75 0,21 N.18 15,84 0,43 3,00 7,54 0,61 0,21 N.19 25,18 0,38 5,00 9,15 0,23 0,18 N.20 17,04 0,45 5,00 8,45 0,47 0,52 Çizelge 11. Standart karar matrisinin çözümü (Solving the standard decision matrix) 312 MERMER BLOKLARIN AHP DESTEKLİ TKOPSIS VE GIA YÖNTEMLERİ İLE SINIFLANDI … Politeknik Dergisi, 2019; 22 (2) : 303-317 4 TARTIŞMA VE SONUÇ (DISCUSSION AND 3 N 1 N 7 l bl kl 2 N 12 N 14 N 20 Çizelge 13. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Gri ilişkisel oluşum ve mutlak değer tablosu (The tables of grey relation generating and distance matrix) Normalize edilmiş karar matrisi Mutlak değer (uzaklık) tablosu C1 C2 C3 C4 C5 C6 C1 C2 C3 C4 C5 C6 Referans seri 1 1 1 1 1 1 0,00 0,00 0,00 0,00 0,00 0,00 N.1 0,10 0,23 0,50 0,15 0,81 0,72 0,90 0,77 0,50 0,85 0,19 0,28 N.2 0,07 0,29 0,50 0,26 0,83 0,76 0,93 0,71 0,50 0,74 0,17 0,24 N.3 0,23 0,43 0,00 0,11 0,68 0,72 0,77 0,57 1,00 0,89 0,32 0,28 N.4 0,22 0,59 0,25 0,17 0,57 0,74 0,78 0,41 0,75 0,83 0,43 0,26 N.5 0,22 0,55 0,00 0,05 0,56 0,70 0,78 0,45 1,00 0,95 0,44 0,30 N.6 0,24 0,43 0,00 0,00 0,11 0,72 0,76 0,57 1,00 1,00 0,89 0,28 N.7 0,18 0,00 0,50 0,13 0,87 0,71 0,82 1,00 0,50 0,87 0,13 0,29 N.8 0,19 0,37 0,00 0,01 0,69 0,68 0,81 0,63 1,00 0,99 0,31 0,32 N.9 0,00 0,56 0,25 0,26 0,72 0,77 → 1,00 0,44 0,75 0,74 0,28 0,23 N.10 0,23 0,57 0,25 0,16 0,00 0,78 0,77 0,43 0,75 0,84 1,00 0,22 N.11 0,76 0,70 0,75 0,63 0,93 0,88 0,24 0,30 0,25 0,37 0,07 0,12 N.12 0,60 0,83 1,00 0,82 0,92 0,94 0,40 0,17 0,00 0,18 0,08 0,06 N.13 0,68 0,80 1,00 1,00 1,00 1,00 0,32 0,20 0,00 0,00 0,00 0,00 N.14 0,94 1,00 0,50 0,69 0,53 0,93 0,06 0,00 0,50 0,31 0,47 0,07 N.15 0,49 0,94 1,00 0,72 0,73 0,92 0,51 0,06 0,00 0,28 0,27 0,08 N.16 0,63 0,49 0,75 0,54 0,99 0,84 0,37 0,51 0,25 0,46 0,01 0,16 N.17 0,73 0,98 0,50 0,58 0,51 0,89 0,27 0,02 0,50 0,42 0,49 0,11 N.18 0,59 0,96 0,50 0,65 0,64 0,90 0,41 0,04 0,50 0,35 0,36 0,10 N.19 1,00 0,98 1,00 0,92 0,96 0,97 0,00 0,02 0,00 0,08 0,04 0,03 N.20 0,64 0,95 1,00 0,80 0,76 0,00 0,36 0,05 0,00 0,20 0,24 1,00 Çizelge 14. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Gri ilişkisel katsayılar, gri ilişkisel dereceler ve öncelik sıralaması (Grey relation coefficients, grey relation ranks and priority orders) Gri ilişkisel katsayılar Gri ilişkisel dereceler Öncelik sıralaması C1 C2 C3 C4 C5 C6 w 0,05 0,33 0,05 0,37 0,11 0,10 ϒ* N.1 0,36 0,39 0,50 0,37 0,72 0,64 0,45 17 N.2 0,35 0,41 0,50 0,40 0,74 0,67 0,47 13 N.3 0,39 0,47 0,33 0,36 0,61 0,64 0,45 16 N.4 0,39 0,55 0,40 0,37 0,54 0,66 0,48 12 N.5 0,39 0,52 0,33 0,35 0,53 0,63 0,45 15 N.6 0,40 0,47 0,33 0,33 0,36 0,64 0,41 20 N.7 0,38 0,33 0,50 0,37 0,79 0,63 0,43 18 N.8 0,38 0,44 0,33 0,34 0,62 0,61 0,43 19 N.9 0,33 0,53 0,40 0,40 0,64 0,68 → 0,49 11 N.10 0,39 0,54 0,40 0,37 0,33 0,69 0,46 14 N.11 0,68 0,63 0,67 0,58 0,88 0,81 0,66 9 N.12 0,56 0,75 1,00 0,74 0,86 0,90 0,77 3 N.13 0,61 0,72 1,00 1,00 1,00 1,00 0,89 2 N.14 0,90 1,00 0,50 0,62 0,51 0,87 0,77 4 N.15 0,50 0,90 1,00 0,64 0,65 0,86 0,76 5 N.16 0,58 0,49 0,67 0,52 0,98 0,76 0,59 10 N.17 0,65 0,97 0,50 0,54 0,51 0,82 0,71 8 N.18 0,55 0,93 0,50 0,59 0,58 0,84 0,72 7 N.19 1,00 0,97 1,00 0,86 0,93 0,95 0,93 1 N.20 0,58 0,91 1,00 0,72 0,67 0,33 0,75 6 Çizelge 13. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) TOPSIS yöntemine göre, N.6, N.8 ve N.15 no lu bloklar 2 sıra, N.3, N.5, N.10 ve N.18 no lu bloklar da 1 sıra daha öncelikli hesaplanmıştır. GİA yöntemine göre ise, N.2 no lu blok 3 sıra, N.1 ve N.7 no lu bloklar 2 sıra, N.12, N.14 ve N.20 no lu bloklar da 1 sıra daha önceliklidirler. Çalışmanın başlangıcında, takip kolaylığı sağlaması bakımından, küçük boyutlu bloklar N.1-N.10 arasında, daha büyük olanlar ise N.11-N.20 arasında numaralandırılmıştır. Çizelge 15 incelendiğinde N.11-N.20 arası numaralandırılmış olan blokların ilk 10 sırayı aldığı görülmektedir. Bu durum, mermer ocaklarında büyük boyutlu blokların üretilmek istenmesinin nedenini açıklamaktadır. Çizelge 14. Gri ilişkisel katsayılar, gri ilişkisel dereceler ve öncelik sıralaması (Grey relation coefficients, grey relation ranks and priority orders) Gri ilişkisel katsayılar Gri ilişkisel dereceler Öncelik sıralaması C1 C2 C3 C4 C5 C6 w 0,05 0,33 0,05 0,37 0,11 0,10 ϒ* N.1 0,36 0,39 0,50 0,37 0,72 0,64 0,45 17 N.2 0,35 0,41 0,50 0,40 0,74 0,67 0,47 13 N.3 0,39 0,47 0,33 0,36 0,61 0,64 0,45 16 N.4 0,39 0,55 0,40 0,37 0,54 0,66 0,48 12 N.5 0,39 0,52 0,33 0,35 0,53 0,63 0,45 15 N.6 0,40 0,47 0,33 0,33 0,36 0,64 0,41 20 N.7 0,38 0,33 0,50 0,37 0,79 0,63 0,43 18 N.8 0,38 0,44 0,33 0,34 0,62 0,61 0,43 19 N.9 0,33 0,53 0,40 0,40 0,64 0,68 → 0,49 11 N.10 0,39 0,54 0,40 0,37 0,33 0,69 0,46 14 N.11 0,68 0,63 0,67 0,58 0,88 0,81 0,66 9 N.12 0,56 0,75 1,00 0,74 0,86 0,90 0,77 3 N.13 0,61 0,72 1,00 1,00 1,00 1,00 0,89 2 N.14 0,90 1,00 0,50 0,62 0,51 0,87 0,77 4 N.15 0,50 0,90 1,00 0,64 0,65 0,86 0,76 5 N.16 0,58 0,49 0,67 0,52 0,98 0,76 0,59 10 N.17 0,65 0,97 0,50 0,54 0,51 0,82 0,71 8 N.18 0,55 0,93 0,50 0,59 0,58 0,84 0,72 7 N.19 1,00 0,97 1,00 0,86 0,93 0,95 0,93 1 N.20 0,58 0,91 1,00 0,72 0,67 0,33 0,75 6 14. Gri ilişkisel katsayılar, gri ilişkisel dereceler ve öncelik sıralaması (Grey relation coefficients, grey relation ranks and priority orders) 3 sıra, N.1 ve N.7 no lu bloklar 2 sıra, N.12, N.14 ve N.20 no lu bloklar da 1 sıra daha önceliklidirler. Çalışmanın başlangıcında, takip kolaylığı sağlaması bakımından, küçük boyutlu bloklar N.1-N.10 arasında, daha büyük olanlar ise N.11-N.20 arasında numaralandırılmıştır. Çizelge 15 incelendiğinde N.11-N.20 arası numaralandırılmış olan blokların ilk 10 sırayı aldığı görülmektedir. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Bu durum, mermer ocaklarında büyük boyutlu blokların üretilmek istenmesinin nedenini açıklamaktadır. 3 sıra, N.1 ve N.7 no lu bloklar 2 sıra, N.12, N.14 ve N.20 no lu bloklar da 1 sıra daha önceliklidirler. Çalışmanın başlangıcında, takip kolaylığı sağlaması bakımından, küçük boyutlu bloklar N.1-N.10 arasında, daha büyük olanlar ise N.11-N.20 arasında numaralandırılmıştır. Çizelge 15 incelendiğinde N.11-N.20 arası numaralandırılmış olan blokların ilk 10 sırayı aldığı görülmektedir. Bu durum, mermer ocaklarında büyük boyutlu blokların üretilmek istenmesinin nedenini açıklamaktadır. TOPSIS ve GİA sonuçları karşılaştırıldığında, aynı veriler ve kriter ağırlıkları kullanılmış olmasına rağmen N.4, N.9, N.11, N.13, N.16, N.17, N.19 numaralı bloklar dışındaki 13 adet bloğun sıralamadaki yerlerinin farklı olduğu görülmektedir (Çizelge 15). TOPSIS yöntemine göre, N.6, N.8 ve N.15 no lu bloklar 2 sıra, N.3, N.5, N.10 ve N.18 no lu bloklar da 1 sıra daha öncelikli hesaplanmıştır. GİA yöntemine göre ise, N.2 no lu blok 313 Metin ERSOY /POLİTEKNİK DERGİSİ, Politeknik Dergisi,2019;22(2):303-317 Çizelge 15. TOPSIS ve GİA sıralamalarının karşılaştırılması (Comparison of TOPSIS and GRA orders) Blok No TOPSIS GİA DEĞERLENDİRME C* Öncelik sırası Ύ* Öncelik sırası N.1 0,31 19 0,45 17 GİA yöntemine göre daha öncelikli N.2 0,36 16 0,47 13 GİA yöntemine göre daha öncelikli N.3 0,39 15 0,45 16 TOPSIS yöntemine göre daha öncelikli N.4 0,49 12 0,48 12 TOPSIS ve GİA sonuçları eşit N.5 0,44 14 0,45 15 TOPSIS yöntemine göre daha öncelikli N.6 0,34 18 0,41 20 TOPSIS yöntemine göre daha öncelikli N.7 0,24 20 0,43 18 GİA yöntemine göre daha öncelikli N.8 0,34 17 0,43 19 TOPSIS yöntemine göre daha öncelikli N.9 0,50 11 0,49 11 TOPSIS ve GİA sonuçları eşit N.10 0,44 13 0,46 14 TOPSIS yöntemine göre daha öncelikli N.11 0,71 9 0,66 9 TOPSIS ve GİA sonuçları eşit N.12 0,83 4 0,77 3 GİA yöntemine göre daha öncelikli N.13 0,84 2 0,89 2 TOPSIS ve GİA sonuçları eşit N.14 0,82 5 0,77 4 GİA yöntemine göre daha öncelikli N.15 0,83 3 0,76 5 TOPSIS yöntemine göre daha öncelikli N.16 0,55 10 0,59 10 TOPSIS ve GİA sonuçları eşit N.17 0,77 8 0,71 8 TOPSIS ve GİA sonuçları eşit N.18 0,80 6 0,72 7 TOPSIS yöntemine göre daha öncelikli N.19 0,96 1 0,93 1 TOPSIS ve GİA sonuçları eşit N.20 0,80 7 0,75 6 GİA yöntemine göre daha öncelikli Çizelge 15. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) TOPSIS ve GİA sıralamalarının karşılaştırılması (Comparison of TOPSIS and GRA orders) Plaka veriminin yüksek olması üreticinin karını da artıracağı için önemli olan kesilecek bloktan elde edilecek m3 bazında plaka miktarıdır. Dolayısıyla blokların satın alınması ya da fabrikada kesim için seçimi sırasında bloğun büyüklüğü ve bünyesindeki süreksizlikler dikkatle incelenir ve elde edilebilecek plaka miktarı tahmin edilmeye çalışılır. Bu sebeple, TOPSIS ve GİA yöntemleri ile sınıflandırılan bloklar için hesaplanan ideal çözüme yakınlık ve gri ilişkisel katsayılar ile blok büyüklüğü, üretim oranı ve süreksizlik sıklığı arasındaki ilişkiler araştırılmıştır. doğrusal ilişki olduğu anlaşılmaktadır. TOPSIS e göre 0.76 ve GİA e göre 0.78 gibi ciddiye alınacak R2 değerlerine göre büyük boyutlu blokların daha kaliteli olduğu söylenebilir. Şekil 5’de ÇKKV sonuçları ve süreksizlik sıklığı dağılım grafikleri verilmiştir. Şekil 5a ve Şekil 5b’ye göre, süreksizlik sıklığı değerlerinin homojen değiştiği ve ÇKKV sonuçları ile arasında azalan doğrusal ilişki olduğu görülmektedir (R2TOPSIS=0.89, R2GİA=0.71). İki Şekil 5’de ÇKKV sonuçları ve süreksizlik sıklığı dağılım grafikleri verilmiştir. Şekil 5a ve Şekil 5b’ye göre, süreksizlik sıklığı değerlerinin homojen değiştiği ve ÇKKV sonuçları ile arasında azalan doğrusal ilişki olduğu görülmektedir (R2TOPSIS=0.89, R2GİA=0.71). İki Şekil 4. TOPSIS ve GİA sonuçları ile blok büyüklüğü arasındaki ilişki (Relationship between TOPSIS and GRA Results and Block Sizes) Şekil 4. TOPSIS ve GİA sonuçları ile blok büyüklüğü arasındaki ilişki (Relationship between TOPSIS and GRA Results and Block Sizes) Şekil 4’de TOPSIS (ideal çözüme yakınlık) ve GİA (gri ilişkisel derece) sonuçları ve blok büyüklüğü (ton) dağılım grafikleri verilmiştir. Şekil 4a ve Şekil 4b incelendiğinde ÇKKV sonuçlarının 10 ton altı ve üstü olmak üzere iki ayrı bölümde gruplandığı dolayısıyla blok büyüklüğü ile ÇKKV sonuçlarının arasında yöntemin R2 değerlerindeki 0.18 puanlık farka göre, süreksizliklerin analizinde TOPSIS yaklaşımının GİA yaklaşımına göre daha etkin olduğu söylenebilir. Ayrıca, süreksizlik sıklığının blok kalitesinin belirlenmesinde önemli bir parametre olduğu da anlaşılmıştır 314 MERMER BLOKLARIN AHP DESTEKLİ TKOPSIS VE GIA YÖNTEMLERİ İLE SINIFLANDI … Politeknik Dergisi, 2019; 22 (2) : 303-317 Şekil 5. TOPSIS ve GİA sonuçları ile süreksizlik sıklığı arasındaki ilişki (Relationship between discontinuity frequences and TOPSIS and GRA Results) Şekil 5. TOPSIS ve GİA sonuçları ile süreksizlik sıklığı arasındaki ilişki (Relationship between discontinuity frequences and TOPSIS and GRA Results) Şekil 6’da ÇKKV sonuçları ve üretim oranı dağılım grafikleri verilmiştir. Şekil 6a ve Şekil 6b’ye göre, ÇKKV sonuçları ile üretim oranı arasında R2=0.88 ve R2=0.96 gibi yüksek derecede artan doğrusal ilişki olduğu söylenebilir. KAYNAKLAR (REFERENCES) [1] Karaatlı M., Ömürbek N., Budak İ. ve Dağ O., “Çok kriterli karar verme yöntemleri ile yaşanabilir illerin sıralanması”, Selçuk University Journal of Institute of Social Sciences 33: 215-228, (2015). [2] Yiğit A. M. and Gök M., “Tire Selection with TOPSIS and GRA Methods in Multi Criteria Decision Making”, Ordu University Journal of Social Science Research, 7(3): 423-431, (2017). [3] Tripathy S. and Tripathy D.K., “Multi-attribute optimization of machining process parameters in powder mixed electro-discharge machining using TOPSIS and grey relational analysis”, Engineering Science and Technology, an International Journal 19: 62–70. DOI: 10.1016/j.jestch.2015.07.010, (2016) [4] Arıbaş M. ve Özcan U., “Akademik araştırma projelerinin AHP ve TOPSIS yöntemleri kullanılarak değerlendirilmesi”, Journal of Polytechnic, 19(2): 163- 173, (2016). Çalışmada elde edilen sonuçlar aşağıdaki şekilde özetlenebilir. Çalışmada elde edilen sonuçlar aşağıdaki şekilde özetlenebilir. Çalışmada elde edilen sonuçlar aşağıdaki şekilde özetlenebilir.  Blokların kalitesine etken parametreler önem sırasına göre üretim oranı, süreksizlik sıklığı, paledyen oranı, atık oranı, blok büyüklüğü ve düzgünlüktür. Renk, desen ve homojenlik gibi faktörler de müşterinin kişisel tercihine göre değişmektedir. [5] Güneysu Y., Er B. ve Ar İ.M.. “Türkiye’deki ticari bankaların performanslarının AHS ve GİA yöntemleri ile incelenmesi”, KTU SBE Sos. Bil. Derg., 9: 71-93, (2015). [6] Bektaş H. ve Tuna K., “Borsa İstanbul Gelişen İşletmeler Piyasası’nda İşlem Gören Firmaların Gri İlişkisel Analiz ile Performans Ölçümü”, Çankırı Karatekin University Journal of the Faculty of Economics and Administrative Sciences, 3(2): 185-198, (2013).  TOPSIS ve GİA sonuçları ile süreksizlik sıklığı arasında azalan doğrusal ilişki görülürken blok büyüklüğü ve üretim oranı arasında artan doğrusal ilişki olduğu görülmüştür. İlişki düzeyleri göz önüne alındığında, blok seçme işleminde bu üç kriterden yararlanılabilir. [7] Muralidhar P., Ravindranath K. and Srihari V., “The Influence of GRA and TOPSIS for Assortment of Green Supply Chain Management Strategies in Cement Industry”, International Journal of Supply Chain Management, 2(1): 49-54, (2013).  Bazı blokların önem sıralarının her iki yöntemde de farklı hesaplanmasının iki önemli nedeni olduğu görülmüştür. Bunlardan birincisi, TOPSIS ve GİA yöntemlerinde kriterlerin normalizasyonunda yaklaşımlarının farklı olması, ikincisi ise bu bloklara ait kriterlerin verilerinin ortalamadan sapmalarının (standart sapma) yüksek olmasıdır. Buna göre alternatifler arasında kriterlerin değişim aralığının yüksek olduğu durumlarda problemin birden fazla ÇKKV yöntemi uygulanarak çözüme ulaşılmasının daha sağlıklı olacağı düşünülmektedir. [8] Kou G., Lu Y., Peng Y. and Shi Y., “Evaluation of classification algorithms using MCDM and RANK correlaion,” International Journal of Information Technology & Decision Making, 11(1): 197-225. DOI: 10.1142/S0219622012500095, (2012). [9] Şahin Y. 3.1. Blok Kesme Gözlem Sonuçları (Block Cutting Observation Results) Blok kesme operasyonunu bir sonucu olarak karşımıza çıkan üretim oranı kuşkusuz blok kalitesinin belirlenmesinde en önemli kriterdir. öncelikli hesaplananlara göre daha yüksek olduğu görülmektedir. C1 kriteri, 0.05 ile en düşük öneme sahip kriterlerden biridir. Buna göre, her iki yöntemde de eşit öneme sahip olan örnekler için baz alınan verilerin ortalamadan sapma oranı daha düşüktür. Dolayısıyla kriter verilerinin değişim aralığının ve ortalamadan sapma derecesinin, TOPSIS ve GİA yöntemlerinin Şekil 6. TOPSIS ve GİA sonuçları ile üretim oranı arasındaki ilişki (Relationship between Production ratio and TOPSIS and GRA) ilişki (Relationship between Production ratio and TOPSI Şekil 6. TOPSIS ve GİA sonuçları ile üretim oranı arasındaki ilişki (Relationship between Production ratio and TOPSIS an GRA) problemin çözümüne yaklaşım tekniğini etkilediği, sonuçları (sıralamanın) değiştirdiği söylenebilir. TOPSIS yaklaşımında kriterlerin normalize edilmesi işleminde, kriterlerin ideal çözümlere vektörel uzaklıkları hesaplanmaktadır. Ayrıca yöntem pozitif ideal çözüme en yakın alternatifi ararken negatif ideal çözüme de en uzak olan alternatifi ön plana çıkarmaktadır. GİA yaklaşımında ise normalizasyon işleminde, doğrusal veri ön işleme yöntemi uygulanmaktadır. Bu yönteme göre tüm kriterlerin kendi aralarındaki değişimi lineerdir ve yöntem referans seriye en çok benzeyen alternatifi ön plana çıkarmaktadır. Yaklaşımlardaki bu farklılık, 13 adet bloğun sıralamadaki yerlerinin de farklı olmasına neden olmuştur. Şekil 7 de TOPSIS ve GİA sonuçları grafiksel olarak karşılaştırılmıştır. problemin çözümüne yaklaşım tekniğini etkilediği, sonuçları (sıralamanın) değiştirdiği söylenebilir. Şekil 7. TOPSIS ve GİA sonuçlarının karşılaştırılması (Comparison of TOPSIS and GRA results) Şekil 7. TOPSIS ve GİA sonuçlarının karşılaştırılması (Comparison of TOPSIS and GRA results) Bazı blokların TOPSIS ve GİA önceliklerinin farklı hesaplanmasının nedenini araştırmak üzere, bu bloklara ait kriterlerin verilerinin ortalamaları ve standart sapmaları ayrı ayrı hesaplanmış ve Çizelge 16’da verilmiştir. Çizelgeye göre önceliği farklı hesaplanan kriterlerin standart sapmalarının, C1 kriteri dışında, aynı 315 Metin ERSOY /POLİTEKNİK DERGİSİ, Politeknik Dergisi,2019;22(2):303-317 Çizelge 16. TOPSIS ve GİA önem sıralamalarının farklı olduğu örnekler (The samples whose TOPSIS and GRA orders are different) Kriterler w C* ≠ Ύ* C* = Ύ* Ort. Std. Ort. Std. C1 0,05 10,6 6,04 15,43 7,89 C2 0,33 1,32 0,76 0,97 0,47 C3 0,05 2,77 1,54 3,57 1,27 C4 0,37 5,78 1,94 7,18 1,82 C5 0,11 0,62 0,32 0,40 0,24 C6 0,10 0,27 0,08 0,22 0,03 w: Kriter ağırlığı; ≠: Önem sırası farklı; =: Önem sırası eşit Çizelge 16. TOPSIS ve GİA önem sıralamalarının farklı olduğu örnekler (The samples whose TOPSIS and GRA orders are different) KAYNAKLAR (REFERENCES) KAYNAKLAR (REFERENCES) ve Akyer H., “Ülke kaynaklarının verimli kullanımı: 4x4 arama ve kurtarma aracı seçiminde AHS ve TOPSIS yöntemlerinin uygulaması”, Süleyman Demirel University The Journal of Visionary, 3(5): 72- 87, (2011). [10] Dai J., Qi J., Chi J., Chen S., Yang J., Ju L., Chen B., “Integrated water resource security evaluation of Beijing based on GRA and TOPSIS”, Front. Earth Sci. China, 4(3), 357–362. DOI 10.1007/s11707-010-0120-7, (2010). Gerçekleştirilen çalışma TOPSIS ve GİA yöntemlerinin mermer blokların sınıflandırılması işleminde kullanılabilir olduğunu göstermiştir. Bu yöntemlerin traverten, oniks, hakiki mermer gibi diğer mermer türlerinin de sınıflandırılmasında kullanılabileceği düşünülmektedir. Ayrıca, problemin diğer yöntemlerle de çözümünün ve elde edilen sonuçlardaki muhtemel farklılıkların araştırılması, hem ÇKKV yöntemlerinin birbirlerine göre üstün ya da zayıf olduğu durumları ortaya çıkarması hem de çözümün daha sağlıklı olması bakımından önemlidir. [11] Ersoy, M. and Yeşilkaya, L., “Choice of marble block cutting machine by using Analytic Hierarchy Process (AHP) method”, International Journal of Information Technology and Business Management, 19(1): 67-80, (2013). [12] Eleren, A. ve Ersoy, M., “Mermer blok kesim yöntemlerinin bulanık TOPSIS yöntemiyle değerlendirilmesi”, Madencilik, 46(3): 9-22, (2007). [13] Gökçe, B. ve Sonugür, G., “ANFIS ve YSA Yöntemleri ile İşlenmiş Doğal Taş Üretim Sürecinde Verimlilik Analizi”, Afyon Kocatepe Üniversitesi Fen ve Mühendislik Bilimleri Dergisi, 16: 174-185, DOI: 10.5578/fmbd.13951, (2016). [14] Caner M. ve Akarslan E., “Mermer Kesme İşleminde Spesifik Enerji Faktörünün ANFIS ve YSA Yöntemleri 316 MERMER BLOKLARIN AHP DESTEKLİ TKOPSIS VE GIA YÖNTEMLERİ İLE SINIFLANDI … Politeknik Dergisi, 2019; 22 (2) : 303-317 ile Tahmini”, Pamukkale Mühendislik Bilimleri Dergisi., 15(2): 221-226, (2009). [22] Saaty T.L., “How to Make a Decision: The Analytic Hierarchy Process”, Interfaces, 24(6): 19-43, (1994). [23] Saaty T.L., “Decision making with the analytic hierarchy process”, Int. J. Services Sciences, 1(1): 83-98, (2008). [15] Güvenç, U., Dursun, M. ve Çimen H., “Mermer Kesme İşleminde Kesim Süresinin Yapay Sinir Ağı Tabanlı Modellenmesi”, SDU International Technologic Science, 3(2): 9-16, (2011). [24] Hwang, C. L., & Yoon, K. P., “Multiple Attribute Decision Making: Methods And Applications”, New York: Springer- Verlag, (1981). [16] Ekincioğlu, G., Güney, A., Akbay, D. ve Altındağ, R., “Dairesel Testereli Kesme Makinelerinin Saatlik Üretim Miktarının Mermer Yüzey Sertliğine Bağlı Olarak Yapay Sinir Ağı (YSA) Ve Regresyon Analizleri (RA) İle Tahmin Edilmesi”, Türkiye 9. Uluslararası Mermer ve Doğaltaş Kongresi ve Sergisi, 13-25 Aralık 2017, Antalya, Türkiye, 87-96, (2017). [25] Deng J., “Control Problems of Grey System”, System and Control Letters, 1(5): 288-294, (1982). KAYNAKLAR (REFERENCES) [26] Deng J., “Introduction to grey system theory”, The Journal of Grey System, 1(1): 1-24, (1989). [27] Deng J., “Basis of Grey Theory”, Huazhong University of Science and Technology press Wuhan, (2002). [17] Aras, G. ve Gencer C., “Muğla İlindeki Mermer İşletmelerine Yönelik Veri Zarflama Analizi Örnek Olayı”, İstanbul Üniversitesi İktisat Fakültesi Ekonometri ve İstatistik Dergisi, 13: 139-153, (2011). [28] Üstünışık N. Z., “Türkiye'deki iller ve bölgeler bazında sosyo-ekonomik gelişmişlik sıralaması araştırması: Gri ilişkisel analiz yöntemi ve uygulaması [Yük. Lis Tezi, Dn: Güngör Z.]”, Gazi Üniversitesi Fen Bilimleri Enstitüsü, Ankara, 132 s., (2007). [18] Oruç, K. O., Çuhadar, M., Kılınç, M. ve Osmancık, S., “Veri Zarflama Analizi İle Mermer İşletmelerinin Etkinlik Ölçümü”, SDU Department of Econometrics, 15th International Symposium on Econometrics, Operations Research and Statistics, 977-994, (2014). [29] Forman, E. H ve Selly, M. A. “Decision By Objectives (How To Convince Others That You Are Right)”, World Scientific Pub. Co., USA, Petersburg, (2001). [30] Ersoy, M., Çelik, M. Y., Yeşilkaya, L., Mermer Blok Kesme Makinası Seçiminde, GİA (Gri İlişkisel Analiz) Yönteminin Uygulanması, Türkiye 9. Uluslararası Mermer ve Doğaltaş Kongresi ve Sergisi, 13-15 Aralık, Antalya, Türkiye, 73-86, (2017). [19] Akkoyun Ö. ve Toprak, Z. F., “Mermer Blok Kalitelerinin Bulanık Mantık Yöntemi ile Belirlenmesi”, Bilimde Modern Yöntemler Sempozyumu, BMYS'2008, 15–17 Ekim 2008, Eskişehir Osmangazi Üniversitesi, (2008). [31] Wen, K.L., “The grey system analysis and its application in gas breakdown and var compensator finding”, International Journal of Computational Computing, 2(1): 21-44, (2004). [20] Yalçın, B., Ucun, İ. ve Koru, M., “Mermer Kesme Testerelerinde Oluşan Kesme Kuvvetinin Bulanık Mantık (BM) Yöntemiyle Modellenmesi”, Gazi Üniv. Müh. Mim. Fak. Dergisi, 22(2): 329-336, (2007). [21] Saaty T.L., “How to Make a Decision: The Analytic Hierarchy Process”, European Journal of Operational Research, 48: 9-26, (1990). 317
https://openalex.org/W2123583205	https://europepmc.org/articles/pmc3935900?pdf=render	English	null	Evaluation of the biocompatibility of experimentally manufactured Portland cement: an animal study	Journal of clinical and experimental dentistry	2,014	cc-by	3,609	J Clin Exp Dent. 2014;6(1):e17-21. J Clin Exp Dent. 2014;6(1):e17-21. Biocompability of portland cement doi:10.4317/jced.51210 http://dx.doi.org/10.4317/jced.51210 Journal section: Biomaterials and Bioengineering in Dentistry Publication Types: Research Journal section: Biomaterials and Bioengineering in Dentistry Publication Types: Research Sibel Koçak 1, Hülya Erten 2, Emre Baris 3, Serkan Türk 4, Tayfun Alaçam 5 Sibel Koçak 1, Hülya Erten 2, Emre Baris 3, Serkan Türk 4, Tayfun Alaçam 5 1 Department of Endodontics, Faculty of Dentistry, Bülent Ecevit University, Zonguldak, Turkey 2 Department of Restoratif Dentistry, Faculty of Dentistry, Gazi University, Ankara, Turkey 3 Department of Oral Pathology, Faculty of Dentistry, Gazi University, Ankara, Turkey 4 Department of Endodontics, Faculty of Dentistry, Gazi University, Ankara, Turkey 5 Turkish Cement Manufacturers’ Association, Ankara, Turkey Correspondence: Department of Endodontics Faculty of Dentistry Bülent Ecevit University Kozlu 67600, Zonguldak, Turkey sibel_tazegul@yahoo.com Correspondence: Department of Endodontics Faculty of Dentistry Bülent Ecevit University Kozlu 67600, Zonguldak, Turkey sibel_tazegul@yahoo.com Koçak S, Erten H, Baris E, Türk S, Alaçam T. Evaluation of the biocom patibility of experimentally manufactured portland cement: an animal study. J Clin Exp Dent. 2014;6(1):e17-21. Received: 17/10/2013 Accepted: 25/10/2013 Received: 17/10/2013 Accepted: 25/10/2013 Received: 17/10/2013 Accepted: 25/10/2013 Article Number: 51210 http://www.medicinaoral.com/odo/indice.htm © Medicina Oral S. L. C.I.F. B 96689336 - eISSN: 1989-5488 eMail: jced@jced.es Indexed in: Scopus DOI® System Article Number: 51210 http://www.medicinaoral.com/odo/indice.htm © Medicina Oral S. L. C.I.F. B 96689336 - eISSN: 1989-5488 eMail: jced@jced.es Indexed in: Scopus DOI® System Abstract Objectives: The purpose of this study was to evaluate the biocompatibility of MTA and the experimentally manu factured portland cement (EMPC). Objectives: The purpose of this study was to evaluate the biocompatibility of MTA and the experimentally manu factured portland cement (EMPC). Study design: Twenty one Sprague Dawley (SD) rats were allocated to testing of three groups. Group I and Group II included ProRoot MTA and the EMPC. The materials were mixed with distilled water and placed in polyethylene tubes. The tubes were implanted subcutaneously in the dorsal region of the animals. Group III served as control; the implanted polyethylene tubes remained empty. At 7, 14, and 28 days after the implantation, the animals were sacri ficed and the implants were removed with the surrounding tissues. The specimens were prepared for histological examination to evaluate the inflammatory response.i Results: No significant difference was found between tissue reactions against the tested materials (p>0.05). Also, control group showed similar results (p>0.05). Conclusions: Results suggest that the EMPC has the potential to be used in clinical conditions in which ProRoot MTA is indicated. MTA and the EMPC show comparable biocompatibility when evaluated in vivo. Although the results are supportive for the EMPC, more studies are required before the safe clinical use of the EMPC. Key words: Mineral trioxide aggregate, portland cement, subcutanous implantation. e17 J Clin Exp Dent. 2014;6(1):e17-21. Biocompability of portland cement Results Macroscopic examination at the implant sites revealed that wound healing was satisfactory and without infec tion at all evaluation periods. The ratio of tissue reaction to the implanted materials is shown in table 1. An experimentally manufactured Portland cement (EMPC) was developed as an alternative to MTA by the Turkish Cement Manufacturers’ Association. Com ponents such as clay or chalk are taken directly from nature -including the arsenic- which appeared in PC. EMPC comprises pure components. Parameter Material N Mean count SD 7 days Inflammation MTA 7 2,429 0,535 EMPC 7 2,286 0,488 Control 7 2,000 0,577 Mast cell MTA 7 2,000 0,577 EMPC 7 2,143 0,690 Control 7 1,286 0,488 14 days Inflammation MTA 7 1,429 0,535 EMPC 7 1,429 0,535 Control 7 1,571 0,976 Mast cell MTA 7 1,857 0,900 EMPC 7 1,714 0,951 Control 7 1,000 0,000 28 days Inflammation MTA 7 1,286 0,951 EMPC 7 1,143 0,690 Control 7 1,000 0,000 Mast cell MTA 7 1,714 0,756 EMPC 7 1,286 0,488 Control 7 0,857 0,378 Table 1. Number of implants and intensity of inflammatory response at different periods of the study Table 1. Number of implants and intensity of inflammatory response at different periods of the study The aim of this study was to evaluate the reaction of rat subcutaneous connective tissue against the implantation of polyethylene tubes filled with MTA and EMPC. Introduction ics. The dorsal skin was shaved, and the implants were removed with their surrounding tissues. The samples were kept in a formalin solution. Biocompatibility of a material is the ability of a material to perform with an appropriate host response in a spe cific situation. This means that the tissue of the patient that comes into contact with the materials does not suffer from any toxic, irritating, inflammatory, allergic, geno toxic, or carcinogenic reaction (1). After histologic processing, the tissue was serially sec tioned longitudinally with microtom (Leica SM-2000R, Leica Corp, Germany) set at 5–6 µm. The samples were stained with hematoxylin-eosin for the histological eval uation using Unna’s method for the evaluation of mast cells. Histological evaluations were made under a light microscope (Nikon Eclypse E-600, Nikon Corp, Japan) at 40 x, 100 x, 200 x, and 400 x magnification. The ob server was blinded to the procedure. Evaluation of in flammatory cell and mast cell infiltration was performed according to Salman et al (12). Statistical analysis was performed using the Friedman and Wilcoxon sign tests for intragroup comparison and Kruskal-Wallis and Mann-Whitney U tests for intergroup comparison. Mineral trioxide aggregate (MTA) was developed as a retrofilling material in the 1990s. A number of biocom patibility studies have been conducted either in vitro or in vivo, and the results showed that MTA presents good sealing ability and tissue healing (2-9). The chemical, physical, and biological properties of Portland cement (PC) were analyzed. Estrela et al. (10) reported that PC contains the same principal chemical elements as MTA, except for the bismuth oxide in MTA that increases the radiopacity of the material. Saidon et al. �� (11) �� reported that MTA and PC have similar proper ties but that MTA is an expensive material whereas PC is an economic cement. Material and Methods Approval for the animal use protocol presented below was sought and given by the Animal Ethic Committees at Hacettepe University (No: B.30.2.HAC.0.01.00.05/42). 21 male Sprague-Dawley rats, each weighing 225 to 250 g, were used in this experiment. Each animal was anesthetized by an intraperitoneal injection of ketamine hydrochloride and xylazine. Afterward, the dorsal skin was shaved and disinfected. Three incisions were made in the skin using a No. 15 scalped blade, and 2-cm pockets were created by the blunt dissection of the incisions. MTA was prepared according to the manufacturer’s instructions. EMPC powder was mixed with a sterile saline solution. Fresh ly mixed test materials were applied with an amalgam carrier into clean, sterile polyethylene tubes (Estern Medikit; Haryana, India) with a 1.3-mm inner diameter and 5-mm length. Each implant was carefully placed in a pocket, and the third incision received an empty steril ized tube to serve as a control. To prevent interactions of materials, the tubes were placed at least 2 cm apart. The skin was closed with 4/0 silk sutures. The evaluations were made 7, 14, and 28 days after surgical implanta tion. During each examination period, 7 animals were sacrificed by administration of a high dose of anesthet - 7 Days y The means of inflammation grades for the MTA and EMPC groups were 2.43 ± 0.54 and 2.49 ± 0.49, respec tively (severe to very severe infiltration of lymphocyte e18 J Clin Exp Dent. 2014;6(1):e17-21. Biocompability of portland cement and 1.00 ± 0.00, respectively. There was no statistically significant difference between the groups (p<0.05). and plasma cells). In the control group, the mean infla mmation was 2.00 ± 0.58, which consisted of moderate infiltration of chronic inflammatory cells. There was no statistically significant difference between the groups (p<0.05), (Fig. 1). The means of inflammation grades for the MTA and EMPC groups were 1.29 ± 0.95 and 1.14 ± 0.69, in cluding mild infiltration of inflammatory cells. In the control group, the mean inflammation was 1.00 ± 0.00. There was no statistically significant difference between the groups (p<0.05), (Fig. 3). Perivascular mast cells were observed around the end of the implants in all groups. The means of the number of mast cells for the MTA, EMPC, and control groups were 2.00 ± 0.58 2.14 ± 0.69, and 1.29 ± 0.49, respectively. Material and Methods There was a statistically significant difference between the groups (p<0.05); however, no difference was found between MTA and PCRA (p>0.05). The means of the number of mast cells for the MTA, EMPC, and control groups were 1.71 ± 0.76, 1.29 ± 0.49, and 0.86 ± 0.38, respectively. There was no statistically significant difference between the groups (p<0.05).li Discussion Tissue reactions associated with ProRoot and EMPC im plants were comparable. After 28-day intervals, inflam matory processes associated with most of the implants decreased significantly, suggesting that both materials are equally biocompatible. Our results were in accor dance with previously published studies in which the tis sue reaction to PC was compared with the tissue reaction to ProRoot MTA (11,25,26,27). Materials must not have a deleterious effect when in contact with tissues before they are marketed and used in dental practice (13). Following ISO/6876 and 10993-5 regulations, in vitro cytotoxicity tests, such as tissue and cell culture assays, are important to provide initial evi dence in the study of dental materials, and are critical to identify those components exercising cytotoxic effects (14). However, these tests lack the interaction of the ma terial with cells in the tissue, and those attracted to the site reaction (15). Mast cells are key elements in the innate immune sys tem and have been termed the antennae of the immune response for their ability to detect changes in their en vironment and communicate these to other cells in the vicinity. Mast cells are located throughout the body in close proximity to epithelial surfaces, near blood ves sels, nerves, and glands, placing them at strategic loca tions to detect invading pathogens. Mast cells express a number of receptors that allow them to recognize di verse stimuli (28). In our study, we observed that EMPC stimulated mast cells much like MTA. Both materials demonstrated a similar effect on inflammatory response and wound healing. As a second step, in vivo implantation of materials in laboratory animals was proposed, which provides much more information about the inflammatory and immune responses developed by the test material (16,17). The present study evaluated the inflammatory reaction of the cellular subcutaneous tissue of rats to Teflon tubes filled with MTA or EMPC. The subcutaneous implantation was considered a suitable secondary test for evaluation of biocompatibility proper ties of endodontic materials. Many studies have evaluated material biocompatibility by using different implantation vehicles, such as polyethylene tubes, silicon tubes, den tin tubes, and Teflon tubes (18-21). In the present study, Teflon tubes were used because of their inert nature and ability to bring a test material into contact with living tis sue in a controlled and effective manner (22). References 1. Williams DF. On the mechanisms of biocompatibility. Biomaterials. 2008;29:2941-53. 2. Abdullah D, Ford TR, Papaioannou S, Nicholson J, McDonald F. An evaluation of accelerated Portland cement as a restorative material. Biomaterials. 2002;23:4001-10. 3. Holland R, de Souza V, Nery MJ, Otoboni Filho JA, Bernabé PF, Dezan Júnior E. Reaction of rat connective tissue to implanted dentin tubes filled with mineral trioxide aggregate or calcium hydroxide. J Endod. 1999;25:161-6. This study has demonstrated that all of groups that are implanted into the dorsal connective tissue of rats pro mote a moderate to severe inflammatory reaction over a 7-day period. The inflammatory reaction decreased with time. Similarly, Gomes-Filho et al. (24) showed that both MTA and Portland cement cause moderate reactions at 7 days, which decreased with time. During the 7-day con trol, the moderate to severe inflammatory reaction in all groups could be the result of the surgical trauma. 4. Mitchell PJ, Pitt Ford TR, Torabinejad M, McDonald F. Osteo blast biocompatibility of mineral trioxide aggregate. Biomaterials. 1999;20:167-73. 5. Koh ET, McDonald F, Pitt Ford TR, Torabinejad M. Cellular respon se to Mineral Trioxide Aggregate. J Endod. 1998;24:543-7. 6. Koh ET, Torabinejad M, Pitt Ford TR, Brady K, McDonald F. Mine ral trioxide aggregate stimulates a biological response in human osteo blasts. J Biomed Mater Res. 1997;37:432-9. 5. Koh ET, McDonald F, Pitt Ford TR, Torabinejad M. Cellular respon se to Mineral Trioxide Aggregate. J Endod. 1998;24:543-7. 6. Koh ET, Torabinejad M, Pitt Ford TR, Brady K, McDonald F. Mine ral trioxide aggregate stimulates a biological response in human osteo blasts. J Biomed Mater Res. 1997;37:432-9. Connective tissue healing was remarkable for both ex perimental materials at the 28-day observation period. Inflammation grades at 14- and 28-day intervals for both MTA and EMPC significantly decreased when compared to 7-day controls. Similarly, Shahi et al. �� (25) �� demon strated that Portland cement shows a statistically signifi cant decrease in inflammation grades at 7-, 15-, 30-, and 60-day intervals. Menezes et al. (8) indicated that MTA and Portland cement show a decrease in inflammation severity in the subcutaneous connective tissue in rats at 7-, 30-, and 60-day intervals. 7. Torabinejad M, Hong CU, Lee SJ, Monsef M, Pitt Ford TR. Inves tigation of mineral trioxide aggregate for root-end filling in dogs. J Endod. 1995;21:603-8. 8. Menezes R, Bramante CM, Letra A, Carvalho VG, Garcia RB. - 14 Days Mild inflammatory cell infiltration around the implants was observed in all groups after 7 and 14 days. This in flammatory infiltration comprised mostly plasma cells and lymphocytes. Numerous neutrophils were seen in the 7-day groups while the 14-day groups showed a few neutrophils. At day 28, all groups had fewer inflamma tory cells and presented mature fibrous tissue. The means of inflammation grades for both MTA and EMPC groups were 1.43 ± 0.54, which consisted of mo derate infiltration of predominantly chronic inflamma tory cells. In the control group, the mean inflammation was 1.58 ± 0.98, which consisted of moderate infiltration of chronic inflammatory cells. There was no statistically significant difference between the groups (p<0.05), (Fig. 2). The severity of the mast cells decreased over time; however, there was no statistically significant difference between the groups. The means of the number of mast cells for the MTA, EMPC, and control groups were 1.86 ± 0.90, 1.71 ± 0.95, day (H&E, x100). e19 Fig. 1. A) EMPC, 7th day (H&E, x100); B) MTA, 7th day (H&E, x100); C) Control, 7th day (H&E, x100). Fig. 2. A) EMPC, 14th day (H&E, x200), B) MTA, 14th day (H&E, x100), 3c: Control, 14th day (H&E, x100). Fig. 3. A)EMPC, 28th day (H&E, x100), B) MTA, 28th day (H&E, x100), C) Control, 28th day (H&E, x200). Fig. 1. A) EMPC, 7th day (H&E, x100); B) MTA 7th day (H&E x100); C) Control 7th d MTA, 7th day (H&E, x100); C) Control, 7th day (H&E, x100). ) MTA, 7th day (H&E, x100); C) Control, 7th day Fig. 1. A) EMPC, 7th day (H&E, x100); B) 4th day (H&E, x100). Fig. 2. A) EMPC, 14th day (H&E, x200), B) MTA, 14th day (H&E, x100), 3c: Control, 14th day (H&E, x100). 14th day (H&E, x100). e19 Fig. 3. A)EMPC, 28th day (H&E, x100), B) MTA, 28th day (H&E, x100), C) Control, 28th day (H&E, x200). h e19 Fig. 3. A)EMPC, 28th day (H&E, x100), B) MTA, 28th day (H&E, x100), C) Control, 28th day (H&E, x200). J Clin Exp Dent. 2014;6(1):e17-21. Biocompability of portland cement Discussion A previous study evaluated the reaction of the subcutaneous tissue of rats after 7 and 30 days, implanted with dentine tubes filled with MTA, calcium hydroxide, or PC �� (23)�� . The au thors found similar results for the 3 materials in speci mens stained with hematoxylin and eosin. After 7 days, a mild to moderate inflammatory reaction was observed in all groups; after 30 days, fibrous connective tissue was found in contact with the materials. The results of the present study demonstrate that all the implanted materials are well tolerated by tissues and have acceptable biocompatibility. However, before ex trapolation of these results to an applicable human clini cal situation, further studies are necessary to evaluate the suitability of the experimentally-manufactured Port land cement. In conclusion, EMPC is as biocompatible as MTA, with no significant differences. References Healing process of dog dental pulp after pul potomy and pulp covering with mineral trioxide aggregate or Portland cement. Braz Dent J. 2001;12:109-13. 28. Suzuki T, Chow CW, Downey GP. Role of innate immune cells and their products in lung immunopathology. Int J Biochem Cell Biol. 2008;40:1348-61. 28. Suzuki T, Chow CW, Downey GP. Role of innate immune cells and their products in lung immunopathology. Int J Biochem Cell Biol. 2008;40:1348-61. l The authors declare that they have no conflict of interest. l The authors declare that they have no conflict of interes References Calcium salts deposition in rat connective tissue after 20. Holland R, de Souza V, Nery MJ, Bernabé oF, Filho JA, Junior ED, Murata SS. Calcium salts deposition in rat connective tissue after the implantation of calcium hydroxide-containing sealers. J Endod. 2002;28:173-6. the implantation of calcium hydroxide-containing sealers. J Endod. 2002;28:173-6. 21. Kim JS, Baek SH, Bae KS. In vivo study on the biocompatibility of newly developed calcium phosphate-based root canal sealers. J Endod. 2004;30:708-11. 22. Tronstad L, Barnett F, Flax M. Solubility and biocompatibility of calcium hydroxide-containing root canal sealers. Endod Dent Trauma tol. 1988;4:152-9. 23. Holland R, de Souza V, Nery MJ, Faraco Júnior IM, Bernabé PF, Otoboni Filho JA, Dezan Júnior E. Reaction of rat connective tissue to implanted dentin tube filled with mineral trioxide aggregate, Portland cement or calcium hydroxide. Braz Dent J. 2001;12:3-8. cement or calcium hydroxide. Braz Dent J. 2001;12:3-8. y 24. Gomes-Filho JE, Rodrigues G, Watanabe S, Estrada Bernabé PF, Lodi CS, Gomes AC, Faria MD, Domingos Dos Santos A, Silos Mo raes JC. Evaluation of the tissue reaction to fast endodontic cement (CER) and Angelus MTA. J Endod. 2009;35:1377-80. raes JC. Evaluation of the tissue reaction to fast endodontic cement (CER) and Angelus MTA. J Endod. 2009;35:1377-80. (CER) and Angelus MTA. J Endod. 2009;35:1377-80. 25. Shahi S, Rahimi S, Yavari HR, Mokhtari H, Roshangar L, Abasi MM, Sattari S, Abdolrahimi M. Effect of mineral trioxide aggregates and Portland cements on inflammatory cells. J Endod. 2010;36:899- 903. 26. Torabinejad M, Hong CU, McDonald F, Pitt Ford TR. Physical and chemical properties of a new root-end filling material. J Endod. 1995;21:349-53. 27. Holland R, de Souza V, Murata SS, Nery MJ, Bernabé PF, Otoboni Filho JA, Dezan Júnior E. Healing process of dog dental pulp after pul potomy and pulp covering with mineral trioxide aggregate or Portland cement. Braz Dent J. 2001;12:109-13. 27. Holland R, de Souza V, Murata SS, Nery MJ, Bernabé PF, Otoboni Filho JA, Dezan Júnior E. Healing process of dog dental pulp after pul potomy and pulp covering with mineral trioxide aggregate or Portland cement. Braz Dent J. 2001;12:109-13. 28. Suzuki T, Chow CW, Downey GP. Role of innate immune cells and their products in lung immunopathology. Int J Biochem Cell Biol. 2008;40:1348-61. 27. Holland R, de Souza V, Murata SS, Nery MJ, Bernabé PF, Otoboni Filho JA, Dezan Júnior E. References Histologic evaluation of pulpotomies in dog using two types of mi neral trioxide aggregate and regular and white Portland cements as wound dressings. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2004;98:376-9. 9. Masuda YM, Wang X, Hossain M, Unno A, Jayawardena JA, Saito K, Nakamura Y, Matsumoto K. Evaluation of biocompatibility of mi neral trioxide aggregate with an improved rabbit ear chamber. J Oral Rehabil. 2005;32:145-50. e20 Biocompability of portland cement J Clin Exp Dent. 2014;6(1):e17-21. 10. Estrela C, Bammann LL, Estrela CR, Silva RS, Pécora JD. An timicrobial and chemical study of MTA, Portland cement, calcium hydroxide paste, Sealapex and Dycal. Braz Dent J. 2000;11:3-9. ydroxide paste, Sealapex and Dycal. Braz Dent J. 2000;11:3-9 11. Saidon J, He J, Zhu Q, Safavi K, Spångberg LS. Cell and tissue , , Q, , p g g reactions to mineral trioxide aggregate and Portland cement. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 2003;95:483-9. 12. Salman MA, Quinn F, Dermody J, Hussey D, Claffey N. Histolo gical evaluation of repair using a bioresorbable membrane beneath a resin-modified glass ionomer after mechanical furcation perforation in dogs’ teeth. J Endod. 1999;25:181-6. dogs’ teeth. J Endod. 1999;25:181-6. g 13. Lawrence WH, Malik M, Autian J. Development of a toxicity pro gram for dental materials and products. J Biomed Mat Res 1974;8:11- 34. 14. Camps J, About I. Cytotoxicity testing of endodontic sealers: a new method. J Endod. 2003;29:583-6. 15. Martínez Lalis R, Esaín ML, Kokubu GA, Willis J, Chaves C, Gra na DR. Rat subcutaneous tissue response to modified Portland cement, na DR. Rat subcutaneous tissue response to modified Portland cement, new mineral trioxide aggregate. Braz Dent J. 2009;20:112-7. 16. Recommended standard practices for biological evaluation of den tal materials. Fédération Dentaire International, Commission of Den tal Materials, Instruments, Equipment and Therapeutics. Int Dent J. 1980;30:140-88. 6. Recommended standard practices for biological evaluation o 17. Watts A, Paterson RC. Initial biological testing of root canal sea ling materials--a critical review. J Dent. 1992;20:259-65. 18. Costa CA, Teixeira HM, do Nascimento AB, Hebling J. Biocompa tibility of two current adhesive resins. J Endod. 2000;26:512-6. 19. Zmener O. Tissue response to a new methacrylate-based root canal sealer: preliminary observations in the subcutaneous connective tissue of rats. J Endod. 2004;30:348-51. 20. Holland R, de Souza V, Nery MJ, Bernabé oF, Filho JA, Junior ED, Murata SS. Acknowledgement 2 This research has been supported by Gazi University Scientific Re search Projects Coordination Department. Project Number: 11/2004- 03. e21
https://openalex.org/W3187104866	https://backend.orbit.dtu.dk/ws/files/266486798/futureinternet_13_00198_v2.pdf	English	null	A Survey on Botnets: Incentives, Evolution, Detection and Current Trends	Future internet	2,021	cc-by	30,840	Citation (APA): Thanh, S. N., Stege, M., El-Habr, P. I., Bang, J., & Dragoni, N. (2021). Survey on botnets: Incentives, evolution, detection and current trends. Future Internet, 13(8), Article 198. https://doi.org/10.3390/fi13080198 General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.  Users may download and print one copy of any publication from the public portal for the purpose of private study or research.  You may not further distribute the material or use it for any profit-making activity or commercial gain  You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Survey on botnets: Incentives, evolution, detection and current trends Survey on botnets: Incentives, evolution, detection and current trends Thanh, Simon Nam; Stege, Mads; El-Habr, Peter Issam; Bang, Jesper; Dragoni, Nicola Thanh, Simon Nam; Stege, Mads; El-Habr, Peter Issam; Bang, Jesper; Dragoni, Nicola Thanh, Simon Nam; Stege, Mads; El-Habr, Peter Issam; Bang, Jesper; Dragoni, Nicola Published in: Future Internet Link to article, DOI: 10.3390/fi13080198 Publication date: 2021 Document Version Publisher's PDF, also known as Version of record General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. Downloaded from orbit.dtu.dk on: Oct 24, 2024 Downloaded from orbit.dtu.dk on: Oct 24, 2024 Downloaded from orbit.dtu.dk on: Oct 24, 2024 Citation (APA): Thanh, S. N., Stege, M., El-Habr, P. I., Bang, J., & Dragoni, N. (2021). Survey on botnets: Incentives, evolution, detection and current trends. Future Internet, 13(8), Article 198. https://doi.org/10.3390/fi13080198 General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. Users may download and print one copy of any publication from the public portal for the purpose of private study or research Keywords: botnet; malware; security; IoT Link back to DTU Orbit Link back to DTU Orbit Citation (APA): Thanh, S. N., Stege, M., El-Habr, P. I., Bang, J., & Dragoni, N. (2021). Survey on botnets: Incentives, evolution, detection and current trends. Future Internet, 13(8), Article 198. https://doi.org/10.3390/fi13080198 this document breaches copyright please contact us providing details, and we will remove access to the work immediate ur claim. future internet Article A Survey on Botnets: Incentives, Evolution, Detection and Current Trends Thanh Vu †, Mads Stege † , Peter Issam El-Habr † , Jesper Bang † and Nicola Dragoni ,‡ Simon Nam Thanh Vu †, Mads Stege † , Peter Issam El-Habr † , Jesper Bang † and Nicola DTU Compute, Technical University of Denmark, 2800 Kgs. Lyngby, Denmark; s200361@student.dtu.dk (S.N.T.V.); s165243@student.dtu.dk (M.S.); s165202@student.dtu.dk (P.I.E.-H.); s144211@student.dtu.dk (J.B.) Correspondence: ndra@dtu.dk; Tel.: +45-45-25-37-31 † These authors contributed equally to this work. ‡ Current address: Richard Petersens Plads, 2800 Kgs. Lyngby, Denmark. DTU Compute, Technical University of Denmark, 2800 Kgs. Lyngby, Denmark; s200361@student.dtu.dk (S.N.T.V.); s165243@student.dtu.dk (M.S.); s165202@student.dtu.dk (P.I.E.-H.); 144211@ t d t dt dk (J B ) q y ‡ Current address: Richard Petersens Plads, 2800 Kgs. Lyngby, Denmark. Abstract: Botnets, groups of malware-infected hosts controlled by malicious actors, have gained prominence in an era of pervasive computing and the Internet of Things. Botnets have shown a capacity to perform substantial damage through distributed denial-of-service attacks, information theft, spam and malware propagation. In this paper, a systematic literature review on botnets is presented to the reader in order to obtain an understanding of the incentives, evolution, detection, mitigation and current trends within the ﬁeld of botnet research in pervasive computing. The literature review focuses particularly on the topic of botnet detection and the proposed solutions to mitigate the threat of botnets in system security. Botnet detection and mitigation mechanisms are categorised and brieﬂy described to allow for an easy overview of the many proposed solutions. The paper also summarises the ﬁndings to identify current challenges and trends within research to help identify improvements for further botnet mitigation research. Citation: Thanh Vu, S.N.; Stege, M.; El-Habr, P.I.; Bang, J.; Dragoni, N. A Survey on Botnets: Incentives, Evolution, Detection and Current Trends. Future Internet 2021, 13, 198. https://doi.org/10.3390/ﬁ13080198 Academic Editors: Ammar Alazab, Mamoun Alazab, Ansam Khraisat and Savitri Bevinakoppa Received: 17 June 2021 Accepted: 29 July 2021 Published: 31 July 2021 1. Introduction Botnets are one of the most prominent threats to system and IoT security in the recent age of cloud-enabled pervasive computing. New pervasive computing architectures, such as always-on mobile devices and Internet-of-Things, provide additional infection vectors for botnet attacks. Due to the large increase in interconnected devices and system platforms, the types and attack patterns of botnets are constantly changing [1–3]. As an example, the IoT botnet Mirai has seen growth from approx. 143,000 occurrences to 225,000 occurrences from 2018 to 2019 alone [4]. For these reasons, it is important to ﬁrst get an understanding of the anatomy of botnets, their evolution up until now and what mitigation mechanisms and tools are available to combat botnet-based attacks. Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- iations. Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- iations. A botnet is a network of malware-infected hosts, which are typically controlled by a Command and Control (C&C) server. The C&C server architecture allows for distributed malicious attacks on either the infected hosts or other interconnected hosts over LAN or the internet [5,6]. C&C servers are commonly known as the botmasters, while infected hosts are simply referred to as bots [1]. Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). p y Botnets are commonly divided into two general architectural structures, centralised and Peer-to-Peer (P2P). These structures are deﬁned by how commands are transmitted throughout the C&C channel. In centralised botnets, as seen in Figure 1, a central C&C server is responsible for sending commands to bots. Meanwhile, in a P2P network, the botnet commands are propagated throughout the P2P overlay network, as seen in Figure 2. Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ Centralised botnets are usually more efﬁcient but are less resilient to countermeasures, as the centralised C&C server acts as a single point of failure for the entire botnet [6,7]. creativecommons.org/licenses/by/ https://www.mdpi.com/journal/futureinternet Future Internet 2021, 13, 198. https://doi.org/10.3390/ﬁ13080198 2 of 43 Future Internet 2021, 13, 198 Figure 1. Example of a centralised C&C botnet structure. 1.2. Outline The paper is laid out as a systematic literature review with particular focus on botnet detection and corresponding mitigation mechanisms to identify current trends in botnet attacks. Section 1 gives a general introduction to botnets, as well as the research questions of this paper. Section 2 describes the previous surveys and literature reviews made by other researchers to describe the potential contribution of this paper. Section 3 describes the methodology used for the paper. Sections 4 and 5 cover the incentives and evolution of botnet attacks respectively, giving an overview of the development and reasoning behind this kind of attack (research questions 1 and 2). Section 6 details the different mitigation and detection mechanisms proposed in research to combat botnets (research question 3). Section 7 provides an analysis on the development and trends in botnets and how to potentially mitigate current attacks (research question 4). Lastly, Section 8 concludes the paper. 1.1. Contribution and Research Questions This systematic literature review presents a survey on the incentives and evolution of botnets as well as detection and mitigation mechanisms developed to combat botnets. The main contribution of this paper is a diverse overview of these topics according to mostly peer-reviewed literature during the period 2005–2021, with a particular focus on botnet detection and mitigation. The second contribution is an analysis of the evolution of botnets and mitigation strategies in order to develop an idea of the current trends and challenges within the ﬁeld of botnets. The speciﬁc research questions asked by this paper are: The speciﬁc research questions asked by this paper are: 1. What incentives are behind the development of botnet attacks? 2. How have botnet attacks evolved over time? 3. What has the research industry proposed to mitigate the threat of botnets? 4. What current trends and challenges related to botnets have been identiﬁed by con- temporary research? 4. What current trends and challenges related to botnets have been identiﬁed by con- temporary research? 1. Introduction Figure 2. Example of a decentralised (P2P) C&C botnet structure. Botnets can be used for numerous kinds of distributed attacks such as Distribu Denial of Service (DDoS) attacks, malicious software distribution, piracy, extortion many others. Initially, botnets spread by the use of Internet Relay Chat (IRC), but prese the attack vectors of botnets are much more varied. These attack vectors include ﬁle-sha networks, infected email attachments, infected websites and vulnerability attacks [1,8]. rise of internet-connected pervasive devices provides botnets with a larger attack sur d l bl h t t i f t P i t b t t tt k h Mi i d Z h Figure 1. Example of a centralised C&C botnet structure. Figure 1. Example of a centralised C&C botnet structure. Figure 1. Example of a centralised C&C botnet structure. Figure 2. Example of a decentralised (P2P) C&C botnet structure. Figure 2. Example of a decentralised (P2P) C&C botnet structure. Figure 2. Example of a decentralised (P2P) C&C botnet structure. Botnets can be used for numerous kinds of distributed attacks such as Distributed Denial of Service (DDoS) attacks, malicious software distribution, piracy, extortion and many others. Initially, botnets spread by the use of Internet Relay Chat (IRC), but presently, the attack vectors of botnets are much more varied. These attack vectors include ﬁle-sharing networks, infected email attachments, infected websites and vulnerability attacks [1,8]. The rise of internet-connected pervasive devices provides botnets with a larger attack surface and more vulnerable hosts to infect. Prominent botnet attacks such as Mirai and Zeus show how the pervasive era of computing and the interconnected internet has caused the rise Future Internet 2021, 13, 198 3 of 43 and evolution of increasingly complex botnets, making continued research within the ﬁeld pertinent [2,9,10]. and evolution of increasingly complex botnets, making continued research within the ﬁeld pertinent [2,9,10]. 1.1. Contribution and Research Questions 2. Related Work ([7]—25 references from 2015–2015) ([7]—36, 2005–2013) Covers more kinds of botnet types such as IoT botnets, mobile botnets, VANET-based botnets and their related challenges and trends. [10] Describes botnet evolution, attack threats and actors, but not go into detection and mitigation techniques against botnets. Thirty-one references from 1998–2009. Covers the same points and also describes different detection categories and speciﬁc mitigation mechanisms. [22–27] Limited scope of botnet detection techniques ([22]—34 detection/mitigation references from 1997–2008), ([23]—38, 2004–2011), ([24]—9, 2008–2019), ([25]—7, 2019–2017), ([26]—11, 2008–2015), ([27]—20, 2004–2013). Includes a larger breadth of more recent detection papers, such as [28,29] and more than 100 more papers compared to [22]. [30] Focuses speciﬁcally on DDoS botnet attacks without covering detection strategies. 145 references from 1993–2015. Describes potential attack threats of botnets while also covering detection mechanisms. [31,32] Mentions only IoT-based botnets. ([31]—36 references from 2010 to 2019), ([32]—122, 2004–2021). Covers IoT-based botnets and also includes other types of botnets, such as mobile botnets, social botnets and VANET-based botnets. [33,34] Discusses various botnet detection categories in general, but does not highlight the speciﬁcs of each technique. ([33]—34 references from 2005 to 2010). Highlights and describes each detection technique individually including the strengths and novelty of each botnet detection approach. [35,36] Only compares machine-learning based botnet detection techniques. ([35]—38 references from 1995–2020), ([36]—25, 2001–2017) Compares machine-learning based detection techniques as well as many more types (IoT, social botnets and more). Reference [11] from 2012 gives a short overview of botnets characteristics, their activi- ties, detection mechanisms and challenges. The survey is, with 39 references, quite limited in scope. Likewise, papers such as [13,14] provide pertinent introductions to the topics covered by the research questions of this paper. Like [11], however, the papers do not quite cover the breadth and depth of available botnet research however. Reference [1] is an excellent literature review on botnets and goes into more depth on the general topic of botnets with a detailed timeline of botnets from 1993 and beyond. The literature review also goes into defence mechanisms, the then-current scope of detection techniques and future challenges. The paper is a bit older (2013) and therefore lacks some of the newer developments in botnet detection and mitigation. Likewise, Reference [16] also touches upon the topics of detection, mitigation, future challenges and evolution, but is also a bit on the older side (2014). 2. Related Work Many surveys and systematic literature reviews on botnets can be found in the lit- erature, although their scope and focus vary signiﬁcantly. Table 1 gives an overview of such related works, with emphasis on their main contribution and on how this paper can enhance the state of the art on botnets research. Table 1. Novelty of this paper with respect to related surveys. Number of references within each research question and year range is compared with the contribution of this paper to quantitatively show the novelty of this paper (years: 2006–2021, incentives: 13 references, evolution: 33, detection/mitigation: 134, trends/challenges: 41). For rows with multiple references, a shorthand format (ref—numOfPapers, yearSpanOfReferences) is used. Paper Main Contribution and Reference Metrics This Paper [11] Offers only generalised information about botnets and botnet detection/mitigation strategies. Thirty-nine references from 2007 to 2012. Describes speciﬁc botnet detection mechanisms, advantages, disadvantages, for instance, the Shieldnet framework to detect botnets in vehicular networks [12]. [13,14] Focuses on describing different kinds of botnet attacks ([13]) and on the threats represented by botnets ([14]), without going into speciﬁc mitigation strategies. ([13]—27 references from 2006 to 2016) ([14]—60 references from 2003 to 2018) Describes both botnet evolution and threats, and offers insight into different detection and mitigation mechanisms. Describes speciﬁc botnet detection mechanisms, advantages, disadvantages, for instance, the Shieldnet framework to detect botnets in vehicular networks [12]. Describes both botnet evolution and threats, and offers insight into different detection and mitigation mechanisms. 4 of 43 Future Internet 2021, 13, 198 Table 1. Cont. Paper Main Contribution and Reference Metrics This Paper [1,15–17] Offer great insight on botnet research, types of botnets as well as detection and mitigation mechanisms. However, both papers do not include more recent studies and publications (all are pre-2014). ([1]—205 references), ([7]—36), ([15]—49), ([16]—217), ([17]—28). Covers the same points as aforementioned papers, but also includes more recent research from 2014–2021 such as [18,19] and more. [20] Presents potential challenges of mobile botnets, but does not include any more recent research (paper from 2012). 40 references from 2002–2012. Presents more recent papers on mobile botnets such as [18,19] and more. [7,21]. Covers only generalised botnet types and few speciﬁc recent types, such as cloud botnets and social botnets ([7]) or covers P2P botnets only ([21]). 2. Related Work Reference [7] also discusses the current challenges, defence mechanisms and suggested mitigation techniques. The paper, however, limits the scope of these discussions to purely P2P-based botnets. Reference [20] investigates botnets on mobile devices and their potential damage, but is limited by its age and the relative newness of smartphone technology at the time (2012). Other surveys and detection comparisons, such as [17,21,24–27,33–36], also focus on detection and mitigation mechanisms. Common among them is that they primarily focus on detection techniques and comparing the effectiveness of the techniques in limited Future Internet 2021, 13, 198 5 of 43 scenarios. This is a factor which this paper attempts to remedy, by also including mitigation mechanisms as well as adding a more broad perspective on botnets in general. scenarios. This is a factor which this paper attempts to remedy, by also including mitigation mechanisms as well as adding a more broad perspective on botnets in general. Some earlier papers such as [10] discuss the threats botnets pose to the general infor- mation security landscape. The paper looks into how law enforcement can act upon the criminals behind botnets and focuses mostly on botnets from the perspective of informa- tion security. Reference [10] does not, however, go into speciﬁc detection or mitigitation mechanisms. Reference [22] touches upon and analyses the use of honeynets, honeypots, signature-based detection with IDS, anomaly-based detection with network analysers such as Botsniffer, mining-based detection and DNS-based detection. Furthermore, the survey explores the use of abnormally recurring NXDOMAIN reply rates as a method of detection. The survey is quite limited in scope; for instance, the paper only presents 13 different pa- pers within botnet detection approaches while this paper has more than 100. Reference [23] proposes detection, prevention, investigation and mitigation techniques by classifying the evolved strategies into ﬁve categories: anomaly, signature, DNS, data mining and hybrid technique. Again, the paper is limited in scope with only 39 different detection papers mentioned. Reference [25] addresses four different major botnet detection approaches: signature-based, anomaly-based, DNS-based and mining based detection but does so with four pages and only seen papers mentioned. Reference [30] focuses primarily on botnets used in DDoS attacks. The paper goes into depth about the life cycle, communication mechanisms and attack types within DDoS-enabled botnets. The paper does not discuss any mitigation mechanisms, however. 2. Related Work Reference [31] is another survey with a speciﬁc focus, namely IoT botnets, which gives a very good introduction to the speciﬁc topic of IoT botnets, but otherwise does not cover any other kinds of botnets. Reference [24] endorses convolutional neural network (CNN) as being one of the best-performing techniques for detecting botnets in IoT devices. While a newer systematic review [32] answers the ques- tions of how IoT botnets are formed, what kind of communication and scenarios involve IoT botnets, and which methods currently exist to detect IoT botnets. y A detailed survey [15] touches on problems with other botnet detection papers such as the lack of public dataset, lack of comparison with other papers, very few botnets in datasets, inaccurate outcomes of experiments and more. According to [15], the general best practice of botnet detection is using the most general behavioural features to generate a hybrid detection method where multiple detection algorithms work together as botnets evolve faster than ever. Furthermore, the paper appeals for dataset improvements and studies to compare methods used in detection. Like [1], the paper is a bit on the older side (2013). While many surveys have gone into great depth on speciﬁc areas of botnets, such as detection, it is the opinion of the authors of this paper that a comprehensive systematic literature review with updated literature is needed. Like [1,15], the paper should focus on the current state of botnet evolution, detection, mitigation and current trends and challenges, as well as provide new insights and ideas through more recent (2013+) research. This will allow the research community a more holistic source of reference for the current state of botnets in 2021. 3.1. Search Strategy The PICO (Population, Intervention, Comparison and Outcomes) criteria to identify relevant search queries from the paper’s research questions [39]. The criteria for this speciﬁc paper are deﬁned as follows: • Population: The paper is interested in all research focused on botnet incentives, evo- lution, detection and mitigation, including other surveys. Malware in general is considered too broad, and only papers focused speciﬁcally on botnets are included. • Intervention: Does not apply as all papers within the research space of botnets are • Population: The paper is interested in all research focused on botnet incentives, evo- lution, detection and mitigation, including other surveys. Malware in general is d d b d d l f d ﬁ ll b l d d • Population: The paper is interested in all research focused on botnet incentives, evo- lution, detection and mitigation, including other surveys. Malware in general is considered too broad, and only papers focused speciﬁcally on botnets are included. • Intervention: Does not apply as all papers within the research space of botnets are interesting for the purpose of the survey. y p p p y • Intervention: Does not apply as all papers within the research space of botnets are interesting for the purpose of the survey. g p p y • Comparison: Different approaches to the detection of botnets in particular are compared to identify advantages/disadvantages. The frequency distribution of detection and mitigation mechanisms described in papers are also compared. g p p p • Outcomes: Expected results are an overview of botnet progression and mitigation mechanisms as well as an identiﬁcation of current trends based on the aforemen- tioned overview. Two important keywords were identiﬁed from these criteria, Botnet and Security. As the main goal of the paper is to provide a mitigation-oriented analysis of botnet papers and current trends in botnet security, these two keywords were deemed as the most important. Initially, ﬁve sources—Google Scholar, DTU FindIt, ACM Digital Library, Scopus and IEEE Explore—were used for database query of botnet papers. The ﬁrst query of ‘Botnet’ in title produced more than 18,200 papers, too much to realistically process. Additionally, IEEE Explore, DTU FindIt, ACM Digital Library and Scopus found 1455, 5912, 1379 and 3411 papers respectively. Instead, a second query: ‘botnet’ in the title and ‘security’ in abstract was used to both exclude some potentially unrelated papers and to include both the identiﬁed keywords. 3.1. Search Strategy y For the second query, Google Scholar was removed as it did not provide the option of searching within abstracts. In total, 306, 224, 85 and 399 papers were found on DTU FindIt, IEEE Explore, ACM Digital Library and Scopus, respectively, with the new query. This query did ﬁnd multiple duplicates between the sources that were removed in the ﬁrst exclusion step. In total, some ~630 papers in total (unique, not including duplicates) were found. 3. Methodology This section describes the search and paper selection methodology used to select literature for this paper. The methodology contains elements from both [37,38], which provide guidelines on how to write a systematic literature review and how to use snowball sampling for paper inclusion respectively. An overview of each step of the paper selection process can be found in Figure 3, with more detailed description of each step being described later. 6 of 43 Future Internet 2021, 13, 198 Figure 3. Methodology steps for paper selection and how many papers were left at each exclu- sion/inclusion step. Figure 3. Methodology steps for paper selection and how many papers were left at each exclu- sion/inclusion step. Figure 3. Methodology steps for paper selection and how many papers were left at each exclu- sion/inclusion step. Figure 3. Methodology steps for paper selection and how many papers were left at each exclu sion/inclusion step. 3.2.2. Title and Abstract Review After the initial exclusion each paper was assigned to one reviewer for a quick title and abstract review. The purpose of this exclusion step was twofold: ﬁrst to exclude any irrelevant papers and second to identify which research questions could be answered by the paper (e.g., other survey, detection paper, mitigation). If the abstract of a paper did not give any indication of being useful for the research questions, the paper was excluded. A total of 304 papers were included in the next step. 3.2. Exclusion/Inclusion Process Several exclusion steps and one inclusion step were executed to identify which papers to include in this paper. 7 of 43 Future Internet 2021, 13, 198 3.2.4. Full Text Review A ﬁnal full text review was performed for the remaining papers. Reviewers were reassigned the papers they reviewed for the previous step to exclude any redundant papers. A short summary for each included paper was written in order to allow all reviewers to understand the contribution of each paper, without reading it themselves. At this point 204 papers remained with a certain guarantee of being useful for the purpose of this literature review. • Peer Reviewed. • Peer Reviewed. Excluding papers older than 2005 might mean some of the initial papers on botnets might be missed. However, because backwards snowball sampling of references is used later, those papers should be included during that step. Only two papers were excluded because they were not available through DTU FindIt due to a paywall. The remaining number of papers was 462 at this phase. Some non-peer-reviewed internet sources were included in the paper for deﬁnitions or additional perspectives. 3.2.3. Introduction/Conclusions Review The penultimate exclusion step involved a review of title, abstract, introduction and conclusion of each paper, with two reviewers being assigned to each paper. Reviewers were assigned to papers that they did not review in the previous exclusion step, allowing for a total of three different reviewer opinions on all papers. Each paper was excluded if one reviewer found the paper either lacking or otherwise irrelevant for this paper. This step was also used to classify the contents of each paper in subcategories, e.g., detection papers focusing on machine learning approaches or detection papers focusing on API call logs. The writing of each paper was also considered. A paper was excluded if both reviewers had issues understanding the main purpose of the paper. A total of 221 papers were left after this review. 3.2.1. Initial Exclusion 3.2.1. Initial Exclusion The initial exclusion step excluded papers based on the following exclusion criteria, any papers not meeting the all criteria were excluded • Papers from 2005 or newer • English language papers only • Botnet-related papers only • Open Access or free for DTU students to read through DTU FindIt • Has a Digital Object Identiﬁer (DOI) [40] • Peer Reviewed. • Papers from 2005 or newer • English language papers only • Botnet-related papers only • Open Access or free for DTU students to read through DTU FindIt p g • Has a Digital Object Identiﬁer (DOI) [40] • Has a Digital Object Identiﬁer (DOI) [40] 3.2.5. Backwards Snowball Sampling Finally, a backwards snowball sampling method was used to include any papers that were missed during the initial query. The process involved going through the references of each included paper and see if any reference might be relevant for the purpose of this paper. After snowballing, the ﬁnal number of peer-reviewed references included in this paper was 224. Future Internet 2021, 13, 198 8 of 43 4. Incentives For the purpose of clariﬁcation, Table 2 below details a number of papers discussed in this section. As to the purpose and incentive of botnets, a great many differing desires may be present. This is in no small part due to the multitude of different targets and aspirations for the various botnets. To further complicate matters, not all botnets are necessarily entirely malicious. There exist both malevolent and benevolent botnets, seeking out potential targets to further their respective inherent agendas. The latter of these will be touched on in Section 4.2. For now, the malevolent type of botnets will be the focus of attention. Table 2. Table of motivations behind botnet-based attacks. The columns describe the motivation, type of attack, known affected targets, attack vector(s) and the case study/paper describing the attack. Motivation Type of Attack Target Vector of Attack Papers Disruption Denial of service. DDoS. Hosting Service Provider. IoT devices. [41] Political afﬁliation Censorship. DDoS. Military complex computers. C&C-based botnet. [42] Disruption and Destabilisation of national power grids. DDoS. Metering infrastructure. Internet-connected computers and IoT devices. [43] Sensitive data Password cracking. Cracking/Brute-forcing. Common people. Various. [44–46] Cyber espionage. (Spear-)phishing and malware Multiple victims listed. Various. [31,47] Security patching Vulnerability scanning (benevolent). Security patching Unsecure IoT devices. Other IoT devices. [48] Botnet spooﬁng (ﬁghting other botnets). Mitigation. Malicious botnets. Existing botnets. [49] Miscellaneous pieces of work detailing motivations. Various types discussed. Various targets discussed. Multiple types discussed. [10,50] Table 2. Table of motivations behind botnet-based attacks. The columns describe the motivation, type o affected targets, attack vector(s) and the case study/paper describing the attack. 4.1.1. Designated Targets 4.1.1. Designated Targets To understand the incentives behind the development of botnets, one must ﬁrst understand the ubiquitous nature of botnets as a whole. Botnets may target a great many different objectives, sectors or groups in modern society, a natural conclusion given botnets’ capacity to mobilise great numbers. The following unordered list of targets are but a handful of the potential victims and sought out results of botnets: • Groups of political disparity or political critics, as discussed in Nazario’s paper [42]; p p p y p , p p [ ]; • National power grids and critical service providers, necessary infrastructure of mod- ern day’s increasingly technologically dependent societies, as described by Dabrowski et al. [54] and Sgouras et al. [43]; g • Civilian peoples’ information and passwords [44,45]; • Espionage and intelligence gathering of foreign nations [47]; • Cracking encrypted or hashed data [46]. The difference in targets of botnets is a great incentive in the development of botnets. They can target a broad range of victims, allowing the botnet master to either target whole groups of victims, or a single institute or individual. The versatile nature of botnets caters to a extensive list of use cases, leading to an ever growing demand for powerful, subtle and specialised breed of malware for botnet-based attacks. 4.1.2. Reasons for Attack As touched on brieﬂy in the prior sections, botnets are developed and utilised for a number of use cases. Having gone over how diverse the targets of botnets may be, it is evident that the reasons must be just as diverse [10]. The same range of importance of targets is seen in the reasons for botnets, varying from the single user credentials for petty thieving to nation-spanning acts of terrorism. Another major reason for the usage of IoT devices as the speciﬁc source of infection and attack of botnets is found in the very foundation of modern-day state of IoT. The devices are often mass produced using cheap, potentially outdated, components. While the capabilities of the devices are limited, they all have the ability to connect to the internet and perform some level of basic processing [31]. 4.2. Benevolent Botnets While exceedingly rare, not all botnets are malicious. A scant few, such as the Hajime botnet, is an example of a neutral if not beneﬁcial botnet [48]. Built on a similar method of infection as Mirai, Hajime distinguishes itself from its cousin in a number of ways, such as: • A decentralised P2P distributed hash table, rather than Mirai’s C&C approach. A decentralised P2P distributed hash table, rather than Mirai’s C&C approach. • A far greater number of ways to infect new hosts. g y • The usage of a custom made protocol for disseminating ﬁles. g y • The usage of a custom made protocol for disseminating ﬁles. Another interesting differene, is that is has never been used in a documented hostile attack on a service or platform. The only instances of potentially questionable actions performed by Hajime have been acts of broadening its sphere of inﬂuence to new IoT devices. In a remarkable act of selﬂessness, the botnet actively patches discovered security holes on infected devices, rendering many attack vectors used by other botnets mute. Other botnets are created by researchers to intentionally overtake and disable malicious botnets, propagating the harmless version instead [49]. 4.1. Malevolent Botnets In the world of malevolent botnets, there exist two main types of incentives for the development of a botnet. These two incentives are: • A desire to harm a designated target or group of targets. • A desire to harm a designated target or group of targets. • A desire to harm a designated target or group of targets. • A desire to better one’s (often the C&C master) monetary situation. • A desire to better one’s (often the C&C master) monetary situation. Concerning the ﬁrst driving force, harming a designated target, a great many tools can be utilised to cause harm. One such method, as described in Kolias et al.’s paper [41], is through a Distributed Denial of Service attack (DDoS). This is showcased in the Mirai botnet back in 2016. Mirai, Japanese for “uture”, was not the ﬁrst botnet to emerge. As touched on in Osagie et al.’s paper [50], several botnets had already emerged, dating all the way from the late 1980s and early 1990s. It was, however, capable of performing an excessively powerful attack against the French webservice provider, OVH, with a peak throughput of 1.1 Tbps [51]. The reasoning for this attack, as it turned out, was based on the fact that OVH hosted a popular tool for Minecraft Server hosts [52]. Ironically, this tool helps to mitigate DDoS attacks against servers. p g g References [44–46] present some of the possibilities within the scope of monetary gain from botnets, either via actively cracking user credentials through various means or by cracking entire pieces of encrypted data. Another example would be barring the user from accessing a service or device they own or rely on, as documented by [53]. 9 of 43 Future Internet 2021, 13, 198 5. Evolution of Botnets For the purpose of clariﬁcation, Table 3 below details a number of papers discussed in this section. Botnets, as a deﬁned type of software, ﬁrst saw the light of day in the late 1980s, with botnet toolkits going back to December 1993, with the release of the IRC-based Eggdrop [50]. Its original intention was for the C-based Eggdrop to be able to share data in between instances and act in a coordinated manner. While the original botnet was benevolent and Future Internet 2021, 13, 198 10 of 43 10 of 43 served a honourable purpose, the derivatives have since been used for mostly malicious purposes, however. This section will go over papers and sources detailing the various differences and iterations a number of different botnets have gone through. Table 3. Evolution of botnets and their associated papers. The ﬁrst column describes the novelty and executive summary of the botnet evolution in question. The second and third columns explain the botnet attack vectors and the year of ﬁrst mention. Lastly, the table lists the associated papers. Note: The papers listed is ordered by the year of the earliest documented occurrence of the described topic related to botnets. Table 3. Evolution of botnets and their associated papers. The ﬁrst column describes the novelty and executive summary of the botnet evolution in question. The second and third columns explain the botnet attack vectors and the year of ﬁrst mention. Lastly, the table lists the associated papers. Note: The papers listed is ordered by the year of the earliest documented occurrence of the described topic related to botnets. of the botnet evolution in question. The second and third columns explain the botnet attack vectors and the year of ﬁrst mention. Lastly, the table lists the associated papers. Note: The papers listed is ordered by the year of the earliest documented occurrence of the described topic related to botnets. Associated Area of Interest Vector of Attack Year Papers First recorded appearance. IRC forums. Late 1980s [50] Honeypots is an often employed tool to detect botnets. New botnets have shown a capacity to identify and avoid detection from such measure. N/A. 2004 [55,56] An analysis and discussion of botnets based off of the Darknet. Darknet. 2006 [57] ZeuS botnet and its role as one of the most inﬂuential botnets in the world. Various systems. 5. Evolution of Botnets 2007 [9] Botnets have begun showcasing active methods and tools to circumvent detection. Various means discussed. 2007 [7,58–60] HTTP-based botnets are explored and discussed along with a multitude of different other botnets. Browsers and extensions. 2007 [61] Description of various botnet characteristics, the latest research and insight into botnets. N/A. 2009 [62,63] New type of botnet capable of impersonating human reaction patterns, a factor otherwise used to identify botnets typically. Various systems. 2009 [64] Smartphones have grown powerful enough to be a potential vector of attack, for a botnet. This is explored in detail. Smartphones. 2010 [65–70] Botnets as a service is a newly founded concept, and is explored in details. Typical SaaS centers. 2011 [71] New type of botnet structure, based around a P2P-oriented basis is investigated, discussed and analysed for potential vectors of attack. None formally disclosed, architecture discussed instead. 2011 [72–77] Other botnets use obfuscation tactics to hide the true identity/position of the C&C’s location, showcasing a trend of botnets growing more versatile and elusive to researchers. N/A. 2013 [78] More kinds of botnet susceptible hosts become more common, leading to new potential vectors of attack. Browsers, extensions, smartphones and online clipboards. 2013 [79,80] Vehicles can also be a potential vector for botnets, such as GHOST. GHOST seeks out VANETs in cars to utilise the VANET control channel for communication. Automobiles and other vehicles. 2016 [81] IoT devices have become equipped with enough processing power to pose a sizeable threat. The generally poor safety implementations and the scale of IoT networks, makes them a good candidate for attack vectors. IoT devices. 2016 [82–84] Proposals for self-evolving botnets. Unknown vulnerabilities in hosts. 2016 [85] Cryptocurrencies have lead to explorations into new areas of potential botnets. Discussion and debate on the architecture. Blockchain structures. 2019 [86] Future Internet 2021, 13, 198 11 of 43 11 of 43 Botnets spanning hundreds of thousands of individual systems was a common sight in the early 2000’s, with a few outliers in the millions of devices. The typical infection vector of insecure networking or lack of security updates have long passed, for new, more modern, more intricate and more obfuscated angles of attack [62]. In order to get a solid foundation on the state of modern day botnets and the threats they pose, Ogu et al.’s paper [63] from 2019 showcases some of the latest research and insight into the world of botnets. 5.1. Disguises and Subterfuge In the early days, botnets would often attempt to avoid attention from authorities and government(s) by purposely avoiding targeting or utilising their systems. However, botnets have grown more and more clever and even capable of detecting a variety of detection mechanisms. Honeypots, devices purposely designed to be easy targets of botnets, can now be identiﬁed and avoided to help prevent detection [55,56]. Honeypot avoidance is not the only measure to avoid detection. Obfuscation of the C&C’s location, as described by Wang et al.’s paper [78], highlights just one method of evasive action botnets may utilise. Botnets may also use dynamic IP ranges to quickly and easily circumvent IP blockages [58], or even fortify and defend its C&C center against Sybil and other routing table pollution attacks [7,59,60]. 5. Evolution of Botnets This consolidation of information is a great starting point for researchers looking into furthering their research on botnets and the issues the world faces in that regard. An interesting case of a recent wide spanning botnet is the ZeuS botnet. Etaher el al’s paper [9] on ZeuS offers up an important explanation on one of the most inﬂuential botnets of today, with victims’ losses in the region of hundreds of millions of dollars. ZeuS is an example of a botnet, which, with a staggering 3.6 million infected devices, proved extremely damaging to the American banking sector. As botnets become more commonplace, the availability of botnet-based attacks also increases for non-malicious actors. Botnets-as-a-service is a phenomenon that has also become common, allowing individuals to perform attacks such as DDoS without ﬁrst developing and propagating their own botnet [71]. Finally, Sood et al. presents a recount of HTTP-based botnets in their paper [61], going over various botnets from ZeuS, SpyEye, ICE 1X, Citadel, Carberp, etc. The paper looks into the design and operation of these, summarising their ﬁndings in a list of various mitigation strategies. 5.2. P2P-Based Botnets and Their Intricacies As brieﬂy mentioned previously in Section 4, some botnets utilise a P2P-based chain of command, over the usual C&C-based approach typical of botnets [72]. This decentralisation of the command structure helps to obfuscate the position of the commanding bot, as well as help defend against typical counter attacks against the botnet, such as key pollution from seized bots. Overall, this increases the resilience of P2P botnets manyfold, as no single-point-of-failure exists within the C&C structure [73]. This is explained in detail in Yan et al.’s paper [74], which also proposes a novel botnet called AntBot. AntBot is one of many new examples of more resilient botnets, showcasing the developments of this worrying trend. This type of hardened P2P-based botnet is also explored and explained in detail in Andriesse et al. [75]. In order to counteract this phenomenon in botnet evolution, entirely new approaches much be made, such as [76], which proposes a different take on detection of P2P botnets, based on its behaviour. Some papers, such as [77], have attempted to model the resilience of P2P botnets to help researchers identify weaknesses and potential mitigation against P2P botnets. These papers all try and tackle the developing threat. 5.3. Extension and Browser Based Botnets 5.5. Vehicular Botnets and Its Effect on Modern Trafﬁc Having touched on smartphone-based botnets, it is no surprise that vehicles are becoming increasingly vulnerable to botnet takeovers. Vehicular ad hoc networks (VANETs) are expected to play an increasing role in trafﬁc safety as well as the driving experience. The ability for cars to communicate with one another may very well revolutionise the way people drive. VANETs are, however, under threat of new types of botnets, as touched on in [81]. 5.4. Smartphone-Based Botnets As smartphones have grown more and more powerful and full of personal information, botnet creators increasingly look towards these pocket sized computers for new possibilities. Mobile botnets show disturbing results as a botnet vector of attack [65–67]. Interestingly, something as simple as an SMS sent from one smartphone to another can also prove to be highly potent, as some botnets have taken to this method to relay messages from the C&C to the bots [88,89]. Of further note within the ﬁeld of mobile botnets, Malatras et al.’s taxonomy [68], and [69] by Rodriguez-Gomez et al. are both of great use to model and formalise botnets. There is also Pieterse and Olivier’s paper [70] on this type of botnets, in which they present a valuable take on the evolution of this niche of botnets. All three papers provide excellent introduction and supplementary understanding of the various characteristics and interesting highlights of mobile botnets. Smartphone services such as Googles Push Notiﬁcation Service (PNS) is also con- sidered to be exploited by botnet devs as C&C channels [90]. Android is not the only targeted OS, as seen in [91], where Apple’s iPhone was the target of the iKee.B botnet, which collected system information such as SMS, network conﬁguration, os name and os version. 5.3. Extension and Browser Based Botnets Simple browser extensions for Google’s Chrome or Mozilla’s Firefox have in recent years seen a growing surge of interest from users. The ability to add additional functionality and capability to a browser, such sa blocking ads, easily downloading high-resolution Future Internet 2021, 13, 198 12 of 43 12 of 43 images, etc. have made these small pieces of software an attractive tool. While the user’s browser may be open about what permissions each individual extensions requires to function, the actual implementation and usage of these requirements are often uncharted territory to most users. This makes malicious browser extensions an excellent point of attack, as browsers often have permission to add, edit and delete ﬁles on the host system. This is showcased and documented in Perrotta and Hao’s paper [79] from 2018. The paper’s proposed extension-based botnet is but one take on a new variety of botnets, offering a number of different capabilities. p In a similar tone, massive online social media that connect people with one another have also grown vulnerable to modern botnets. This new breed of botnets, typically nicknamed Social Network Botnets (SNB), are capable of inﬁltrating deep into social networks such as Facebook without being caught or stopped by defence measures. Boshmaf et al.’s paper [80] details how such an SNB can be conceived and details how it performs on Facebook over a period of eight weeks. Likewise, not only have social medias fallen prey to this new type of botnets. Online clipboards and publicly available cloud storage services have turned out to be effective measures to act as C&C centres for botnets, as described in Yin et al. [87]. Other examples include the proposed social botnet DR-SNBot by Yin et al., which argue that bots hiding within social networks are more resistant to to destruction compared to other types of botnets [60]. 5.6. Blockchain-Based Botnets While blockchain has, for a large part, often been associated with cryptocurrencies, new methods and developments showcase a new type of botnets emerging based around blockchain. Bock et al. touches on this, in their assessment [86], providing a broad overview of the associated risks and relates the problems with this new type of botnets to existing C&C-based botnets. 13 of 43 13 of 43 Future Internet 2021, 13, 198 5.8. Atypical New Botnet Variants Every once in a while, entirely new botnets pop up, bringing either new features, capabilities or counteractions to known botnet mitigation tools. Chen et al. [64] discusses a new type of botnet, a so called ’Delay-Tolerant Botnet’, pieces of botnet-enabling malware capable of impersonating human reaction times. This helps it avoid detection for longer, as reaction times are often a measure when identifying botnets and their attacks. Abu Rajab et al. presents an analysis of a botnet within the Darknet [57], showcasing how botnets make up a substantial amount of internet trafﬁc. As a general model, Kudo et al. proposes the concept of self-evolving botnets [85] which models their behaviour through a stochastic epidemic model of botnet features. The behaviour of infection shows quick propagation and the model indicates that self-evolving botnets should be prevented from spreading early. 5.7. IoT-Based Botnets Back in Section 4, a brief recount was made as to the reasons why IoT devices were especially popular botnet slaves. This is further explained, discussed and evaluated in a number of papers, including [82,83]. Situations such as poorly conﬁgured devices, the role of IoT in botnets as well as real life scenarios involving IoT devices capabilities for usage in attacks. Likewise, Mendes, Aloi and Pimenta’s paper [84] on IoT based botnets offers great insight into various architectures employed by botnets. 6. Detection and Mitigation This section describes the botnet mitigation and detection strategies proposed within research. On the topic of detection and mitigation of botnets, the two components are often conjoined in research, as the mechanisms for detecting a botnet often correlates to its behaviour and infection vector. Through this, a mitigation strategy can be built to counteract the identiﬁed vector or behaviour, which either partially or completely nulliﬁes the botnet. For that purpose, it was decided to follow the example from prior peers, and conjoin the two elements in this section as well. The distribution of papers and subcategories of detection papers can be seen in Figure 4. Figure 4. Distribution of botnet detection and mitigation mechanisms for this paper. Figure 4. Distribution of botnet detection and mitigation mechanisms for this paper. Future Internet 2021, 13, 198 14 of 43 14 of 43 6.1. Detection Mechanisms—Techniques This section covers all papers, which are related to detection approaches and compares several techniques for each categories. 6.1.1. Neural Network Detection Mechanisms For the purpose of clariﬁcation, Table 4 above gives an overview of a number of neural network based detection techniques and their related papers: Table 4. Papers describing detection of botnets using neural network based techniques. Each row describes the overall technique, known advantages, disadvantages, detection rate and related papers. Technique Advantage(s) Disadvantage(s) Detection Rate Papers Back Propagation (BP). Can detect botnets with no false positives as well as low expected error rate in higher error environments. Only tested on a certain types of botnet trafﬁc and characteristics of other botnets. 99% detection rate, 95.7% accuracy and FP rates of 0.00952 or lower. [92] PSI-Graph Much faster than FCGs, better FNR, FPR and accuracy. N/A. Accuracy of 98.7%, FNR 1.83% and FPR 0.78%. [93] Convolutional Neural Network (CNN) Can automatically extract features of botnets and has higher accuracy than traditional NN. Reference [94]: Training process requires GPU power. Reference [95] does not yet support transfer learning. Best accuracy of ResNet is 99.32%. Reference [95] 99.98% accuracy for DenseNet and 83.15% for SVM. Reference [96] achieves up to 98.6% botnet detection accuracy on the self-tests and about 90% on the cross-evaluation test. [94–97] Artiﬁcial Neural Network (ANN) & MLP-ANN. Reference [98] has low computational overhead. Reference [99] uses supervised learning approach to obtain high TPR and can further be used in Transfer Learning projects. Reference [99] does not have hybrid models. 6. Detection and Mitigation Reference [98] managed to get minimum of 87.56% TPR during testing. Reference [99] got 100% accuracy and TPR. [98,99] ML & DL. A combination of multiple ML and DL models including comparisons. N/A. N/A. [97] NN & AIS. Reference [29] can provide endpoint protection. Reference [100] does not need prior knowledge of botnets. Reference [29] requires command and control server. N/A. [29,100] Machine Learning, Deep Learning, t-distributed stochastic neighbor embedding & Deep Neural Network. Considers a broad amount of ML and DL techniques. N/A. DNN takes a long time to train and also around 1.3 s to execute detection logic where other methods found in this paper is faster or requires less training time. [101–103] Nonnegative Tucker decomposition. Memory-efﬁcient. Normally requires too high a computational cost to run in real time. N/A. [104] NN with blockchains. Lightweight: small memory and low-power processors needed for devices. N/A. N/A. [29] 6.1. Detection Mechanisms—Techniques 15 of 43 Future Internet 2021, 13, 198 15 of 43 Neural network-based detection of botnets is just one of many proposed methods of botnet detection. X.G. Li and J.F. Wang [92] proposes using back propagation (BP) neural network to detect botnets based on trafﬁc characteristics. Other detection methods, such as the one proposed by [93], also use similar neural network methods for detecting IoT-based botnets using PSI-Graph generation with potentially fewer resources. Reference [99] uses a model based neural network approach to classify IoT botnets; the paper compares the MLP-ANN mode with the N-BaIoT model. MLP-ANN requires a supervised learning approach, meaning it can become even more effective by training with more data and can run on very limited computing resources. N-BaIoT on the other hand works unsupervised (USML) but requires a larger resource overhead. Reference [100] uses a biology-inspired artiﬁcial immune system approach to model botnets as infections within a network body. The microorganisms within the artiﬁcial immune system are trained to act upon spam and scanning related botnet activity. Other papers focus more on applying neural networks to detect irregularities within network trafﬁc. Dhalka et al. compares several contemporary botnet detection techniques, k-means clustering, neural network and recurrent neural network. Their paper [105] compares the algorithms in terms of several factors, including positive/negative rates, sensitivity, speciﬁcity and more. The paper identiﬁes the neural network method as the best solution based on the chosen measures, with a caveat that the neural network method may not be practical. 6. Detection and Mitigation There has been a growth in papers related to mitigating botnets found in IoT devices, as this industry is growing exponentially without regards to security. Alexander and Allison Nixon propose an Industry Security Association committee to be created and publish security standards which manufactures are required to follow [106]. p y q For non-IoT botnets, other papers such as [25] address four different major botnet detection approaches: signature-based, anomaly-based, DNS-based and mining based detection. The paper evaluates previous surveys and illustrates botnets architectures, topologies, communication protocols, attacking method and, their destinations, impedi- ment approaches, and detection techniques. Similar neural network identiﬁcation systems such as [107] work by analysing botnet trafﬁc, using a more adaptive and ﬂexible stream mining algorithm to classify botnets. Reference [94] also proposes a similar network analysis approach with a neural network-based P2P model to monitor botnet trafﬁc and recognise patterns using the ResNet architecture. In another similar approach, the neural network-based detection and mitigation system called BoNeSSy also analyses network trafﬁc to detect and mitigate botnet behaviour [98]. If an application identiﬁes a threat, BoNeSSy will notify the administrator and take appropriate security actions to isolate the potential threat. Chu et al. proposes a combination of machine learning and classiﬁcation mining [108] for botnet detection. g Jithu et al. [102] propose a deep learning method that detects botnets in IoT devices using anomaly detection. The technique employed in the paper reaches an accuracy of 94% and recognises the need for IoT security with a predicted number of 24.1 billion IoT devices by 2030. Abdullah et al. [103] propose using a Local Global Best Bat Algorithm with neural networks (LGBA-NN), which achieves a 99.89% accuracy in their study using the N-BaIoT dataset. Their study includes comparing LGBA-NN with less effective implementations of PSO-NN and BA-NN. Deep learning, which employs neural networks, has been used by Taheri et al. [95], who proposes a deep learning-based botnet detection engine that takes raw network trafﬁc data as input and transforms them into images. These images are then input into a deep convolutional neural network (CNN), DenseNet, for classiﬁcation of normal and botnet trafﬁc data. CNN approaches are endorsed by [24] as being one of the best performing techniques for detecting botnets in IoT devices along with Recurrent Neural Network (RNN) and Artiﬁcial Neural Network (ANN). 6. Detection and Mitigation A similar approach to [24] is using deep learning to construct algorithms to detect IoT-based botnets and botnet attacks. Sriram et al. [101] propose an algorithm that analyses the network ﬂow and can be used to secure “smart city applications”. This includes health care, power grid infrastructure, water Future Internet 2021, 13, 198 16 of 43 treatment facilities, trafﬁc controlling, etc. Additionally, the ﬂow of networks can be utilised for further analysis and learning, to enhance the performance of the algorithm. The authors of "Real-time botnet detection using non-negative tucker decomposition" [104], propose a method for detecting group activities from extracted features in darknet trafﬁc using tensor factorisation. While this method requires too high computational costs to run in real time, they propose implementing a two-step algorithm in order to achieve fast, memory-efﬁcient factorisation. More nontraditional methods like [96] seek to identify botnets through the usage of power consumption as the parameter for their CNN model. In [29] a similar lightweight solution is also mentioned for use in small memory capacity devices with low-power processors, since these are not able to have reliable anti-malware systems. It is based on the use of NeuroMesh, which is a combination of neural and mesh detection networks used to secure the devices. It can detect and delete malware and implements IP-based blacklist and whitelist access control to provide secure channel for IoT devices via the Bitcoin communication protocol. 6.1.2. Machine Learning and Network-Based Detection Mechanisms For the purpose of clariﬁcation, Table 5 below details a number of papers that goes over machine learning-based detection: Table 5. Papers describing detection of botnets using machine learning based techniques. Each column describes the overall technique, known advantages, disadvantages, detection rate and related papers respectively. Technique Advantage(s) Disadvantage(s) Detection Rate Papers Network ﬂow analysis Real-time detection. High detection rate (up to 98%) of known signatures. Can be implemented in current SDN solutions. Supports a wide variety of detection approaches. Requires training data for unknown attacks. Too many features in selection results in unnecessary overhead. Varies between 85.34%. Reference [109] to >99% [26]. [26,28,97, 109–129] Honeypots Ability to train model on unknown variants. Training needs to be supplemented by simulated network trafﬁc. 99%. [130,131] DNS-based proﬁling of Mirai botnets Uses live datasets based on honeypot infected botnets, low computational time (<0.2 s). Limited to mirai(-like) botnet variants. >99%. [132] Cloud-based detection ofﬂoad Linear scaling with number of computational hosts. 6. Detection and Mitigation [139] Multi-phase trafﬁc ranking mechanism Reduces the false classiﬁcation rate of normal IP trafﬁc. Needs further work to detect different kind of HTTP botnets. The experiment data is limited. N/A. [140] Multi clusters for classiﬁcation Achieve high accuracy and reliability. Outperforms individual clustering algorithms in training time. Increased algorithm runtime complexity. N/A. [141,142] ML on DNS query data Better than IDS based detection on newer botnets variants. Takes time to train. Most ML algorithms score over 85% accuracy on DGA botnets, among which the random forest algorithm gives the best results with an overall classiﬁcation accuracy of 90.80%. [143] Table 5. Cont. Better than IDS based detection on newer botnets variants. Takes time to train. Neural network is not the only method to use the N-BaIoT dataset, as seen in [97], where Bashlite and Mirai found their way into various IoT devices. These included door- bells, baby monitors, security cameras and a webcam. Detection models were developed for each device using numerous machine learning modes, including deep learning models. Similar machine learning methods have been used by Long Mai and Dong Kun Noh [141] using cluster ensembles to increase detection reliability compared to other clustering mechanisms. Instead of classifying ﬂow clusters in either a botnet ﬂow or normal ﬂow, the algorithm uses multiple clusters for the same trafﬁc and a link algorithm to do the ﬁnal classiﬁcation. Self-adapting systems for detecting, clustering and classiﬁcation of botnets is proposed by Lysenko et al. [136], who use a semi-supervised fuzzy c-means clustering technique. The system is also able to double as mitigation as it can reconﬁgure corporate networks and execute more speciﬁc actions such as reducing request timeouts, decreasing allowed HTTP request size and blocking source hostname and IP addresses. Reference [142] also applies a clustering machine learning algorithm to detect Internet Relay Chat (IRC) trafﬁc containing botnet behaviour. The approach however is based off a fuzzy cross association clustering algorithm to study the relationship between known trafﬁc and unknown trafﬁc. Unknown trafﬁc can then be checked to verify or disprove the appearance of a botnet within the IRC trafﬁc. Machine learning can be very helpful when it comes to detecting different kinds of botnets, but recently, bot herders [144] have begun to use well-crafted concept drifts based on known machine learning techniques to defend against ML assisted detection. 6. Detection and Mitigation Instead of classifying ﬂow clusters in either a botnet ﬂow or normal ﬂow, the algorithm uses multiple clusters for the same trafﬁc and a link algorithm to do the ﬁnal classiﬁcation. Self-adapting systems for detecting, clustering and classiﬁcation of botnets is proposed by Lysenko et al. [136], who use a semi-supervised fuzzy c-means clustering technique. The system is also able to double as mitigation as it can reconﬁgure corporate networks and execute more speciﬁc actions such as reducing request timeouts, decreasing allowed HTTP request size and blocking source hostname and IP addresses. Reference [142] also applies a clustering machine learning algorithm to detect Internet Relay Chat (IRC) trafﬁc containing botnet behaviour. The approach however is based off a fuzzy cross association clustering algorithm to study the relationship between known trafﬁc and unknown trafﬁc. Unknown trafﬁc can then be checked to verify or disprove the appearance of a botnet within the IRC trafﬁc. Machine learning can be very helpful when it comes to detecting different kinds of botnets, but recently, bot herders [144] have begun to use well-crafted concept drifts based on known machine learning techniques to defend against ML assisted detection. Through a new ML algorithm consisting of a combination of ANN and DT, Rezaei [145], has obtained a detection accuracy of 100%. The technique has a noticeable 11.36 s duration detection time using 20 features to detect botnets in IoT. Seungjin et al. [146] refers to what they call smart factory (SF), which is a combination of AI and ML. They tested two differ- ent ML techniques, Weka and R-studio, achieving 95.3% and 96% accuracy, respectively. Pandey et al. [126] use RF to classify the data into multiple units and then SVM to reclassify every sub-entity to improve accuracy. Their RF-SVM hybrid ML model achieved 85.3% accuracy while RF-Naive Bayers reached 83.36% and lastly RF-KNN-LR 79.56% accuracy. Hidayah et al. [147] obtained up to 92% accuracy using ML algorithms that ﬁlter and classify data to detect the botnets C&C server. Siqlang et al. [148] studied the use of Table 5. Cont. Technique Advantage(s) Disadvantage(s) Detection Rate Papers Network trafﬁc data mining Allows for detection of botnets inside complex trafﬁc before the attack. Does not require network changes. Method needs to be deployed by ISPs. Difﬁculties when using the NAT technology. Detection: 98% NaiveBayes: 89% BayesNet: 87%. 6. Detection and Mitigation Real-time detection not possible. N/A. [133] Minimization of ML feature selection Lower computational cost while keeping high accuracy rate. Limited to IoT-based botnets ([134]), does not improve weaknesses to unknown signatures. 98.97% ([134]) and 75–99% ([135]). [134,135] VM Hypervisor detection agent Allows OS-level passive detection and monitoring. Only applicable for VM-based hosts. N/A. [5] Self-adaptive system with fuzzy c-means clustering Can choose security scenarios and adapt mitigation procedures depending on the attack. High resilience. Dependent on the speciﬁc system network used during testing. High percentage of known and unknown. Multi-vector cyberattacks: 70%. [136] Network botnet ﬁngerprinting and signature Large data throughput. High accuracy. Reference [137]: better real-time performance compared to CPU-based approaches (800–1300% speedup) Speciﬁc parameters in the used dataset, which need more features. Reference [137]: requires dedicated GPU hardware. Very few false positives. Accuracy close to 100%. [137,138] ng detection of botnets using machine learning based techniques. Each column describes the overall ntages, disadvantages, detection rate and related papers respectively. 17 of 43 17 of 43 Future Internet 2021, 13, 198 Table 5. Cont. Technique Advantage(s) Disadvantage(s) Detection Rate Papers Network trafﬁc data mining Allows for detection of botnets inside complex trafﬁc before the attack. Does not require network changes. Method needs to be deployed by ISPs. Difﬁculties when using the NAT technology. Detection: 98% NaiveBayes: 89% BayesNet: 87%. [139] Multi-phase trafﬁc ranking mechanism Reduces the false classiﬁcation rate of normal IP trafﬁc. Needs further work to detect different kind of HTTP botnets. The experiment data is limited. N/A. [140] Multi clusters for classiﬁcation Achieve high accuracy and reliability. Outperforms individual clustering algorithms in training time. Increased algorithm runtime complexity. N/A. [141,142] ML on DNS query data Better than IDS based detection on newer botnets variants. Takes time to train. Most ML algorithms score over 85% accuracy on DGA botnets, among which the random forest algorithm gives the best results with an overall classiﬁcation accuracy of 90.80%. [143] Neural network is not the only method to use the N-BaIoT dataset, as seen in [97], where Bashlite and Mirai found their way into various IoT devices. These included door- bells, baby monitors, security cameras and a webcam. Detection models were developed for each device using numerous machine learning modes, including deep learning models. Similar machine learning methods have been used by Long Mai and Dong Kun Noh [141] using cluster ensembles to increase detection reliability compared to other clustering mechanisms. 6. Detection and Mitigation Through a new ML algorithm consisting of a combination of ANN and DT, Rezaei [145], has obtained a detection accuracy of 100%. The technique has a noticeable 11.36 s duration detection time using 20 features to detect botnets in IoT. Seungjin et al. [146] refers to what they call smart factory (SF), which is a combination of AI and ML. They tested two differ- ent ML techniques, Weka and R-studio, achieving 95.3% and 96% accuracy, respectively. Pandey et al. [126] use RF to classify the data into multiple units and then SVM to reclassify every sub-entity to improve accuracy. Their RF-SVM hybrid ML model achieved 85.3% accuracy while RF-Naive Bayers reached 83.36% and lastly RF-KNN-LR 79.56% accuracy. Hidayah et al [147] obtained up to 92% accuracy using ML algorithms that ﬁlter y y y y Hidayah et al. [147] obtained up to 92% accuracy using ML algorithms that ﬁlter and classify data to detect the botnets C&C server. Siqlang et al. [148] studied the use of Future Internet 2021, 13, 198 18 of 43 18 of 43 unsupervised detection of botnet activities and used the Frequent pattern tree algorithm provided by Weka. They achieved up to 100% accuracy varying with the thresholds chosen and up to 100% precision. Mehdi [149] found that using both ML and DL techniques based on a somewhat hybrid combination of cooperative game theory, accuracy and learning times could greatly be improved. For SVM, he obtained 11.62% improved accuracy and 154.41 s better learning time and for LSTM, 0.24% better accuracy and 222.72 s better learning time. Mehdi also found that these methods achieved an accuracy of 99.98% and higher using 10 or more features for detection. Using KNN, Bjatt et al. [150] achieved in scenarios up to 98% accuracy and provided comparisons to other methods such as Spark-ELM, CCD and Bclus. The proposed method detects botnets based on a forecastive anomaly detection approach, where the ﬁrst progres- sion is the instance creation and the second is Cataloging. After the progressions, they use Graph Structure Based Detection of Anomaly (GSBDA) to detect hazardous anomalies and lastly use a KNN to identify the botnet accurately. Ali and Fatemeh [151] uses DNS queries to extract features from network trafﬁc and then apply ML to generate a botnet detection report. 6. Detection and Mitigation Their studies included testing DT, SVM, RF and Logical regression as their ML algo- rithms and obtained accuracies of 98%, 96%, 99% and 93% respectively. Panda et al. [152] claim 100% accuracy using two different approaches, the ﬁrst approach is scatter search (ScS) combined with CNN and the other method is ScS combined with Deep Multilayer perceptron (DMLP). They tested their implementation on the UNSW-NB15 dataset. where 66% of the data were used for training and the remaining 34% for testing. g g g Another general category within machine learning algorithms is the use of network anomaly [26] focused algorithms. This kind of mechanism of clustering with machine learning can be found in [138], where a new method called BotFingerPrint (BotFP) is presented. BotFP is supposed to be a more lightweight method that can handle a large number of data easily. BotFP is also designed to detect malicious network activities such as port scans and DDoS attacks. Kozik and Chora´s introduce techniques [124] used in big data and machine learning to identify botnet trafﬁc in networks. The multi-scale analysis model is used to extract botnet features from network trafﬁc, which are then classiﬁed using a random forest machine learning algorithm. Poisson sampling is further used to train the random forest model by under-sampling benign trafﬁc. Chen et al. [125] propose a method similar to Kozik and Chora´s [124] with a conversation-based detection mechanism by using a random forest algorithm to classify botnet conversations in network ﬂows. Conversations are classiﬁed depending on their duration, size and distribution of topics. The random forest algorithm is used for selection of probable botnet ﬂows for detection using a separate machine learning algorithm trained with random forest. Besides using random forest, Reference [126] found Support Vector Machine (SVM), Naive Bayes (NB), K- Nearest Neighbour and Linear Regression algorithms to be possible detection mechanisms. Furthermore, Reference [109] conducted an analysis of various machine learning algorithms for botnet DDoS attack detection, including SVM, ANN, NB, Decision Tree (DT) and USML. According to [109], when considering only DDoS attacks, Unsupervised Learning (USML) stands out as the better option to differentiate between botnet trafﬁc and legitimate network trafﬁc. Kirubavathi and Anitha also present an approach for detecting botnets through net- work trafﬁc ﬂow behaviour analysis and machine learning. The proposed method [127] extracts network features such as small packets, packet ratio, initial packet length and bot-response packets. 6. Detection and Mitigation Reference [115] proposes the use of SoftFlow to capture packages and generate NetFlow for machine learning. The paper applied this method to two botnet datasets to test if the method was able to differentiate between legitimate Alexa trafﬁc, Citadel and Zeus botnet trafﬁc. More methods based on existing industry frameworks have also been tested. References [116,117] use Cisco’s Netﬂow for analysis along with a custom-made detection framework to detect botnets. The botnet propagation model uses a modiﬁed Susceptible, Infectious or Recovered SIRS epidemiological model to estimate if there will be an epidemic of the given botnet and then uses the developed framework to mitigate the infection. p g Moving into machine learning combined with the use of honeypots to detect botnet- enabling malware. Ruchi and Kumar [130] proposes using ThingPot which is a virtual IoT honeypot capable of catching various botnet binaries by emulating different IoT commu- nication protocols along with entire IoT platform behaviours. However, with honeypots becoming more normal in the line of defence against botnets, bot herders also become better at bypassing them. Therefore, Reference [131] seeks to make honeypots more efﬁcient and more effective. Owen et al. seeks to use DNS trafﬁc analysis models with a proﬁling scheme of Mirai-like botnet activity captured globally in distributed honeypots [132]. It discusses features useful in proﬁling botnets in the past and suggests a number of improvements. The suggested solution can bring down botnet detection time signiﬁcantly while maintaining high levels of accuracy under random forest formulation. A great amount of botnet detection mechanisms, most of which are based on network analysis, will not use real time detection, as the high number of data overwhelm most CPU detection-based systems. Because of this, Che-Lun and Hsiao-Hsi propose the use of GPU based detection over CPU-based detection to gain a speedup in real time detection [137]. By using GPU based detection, packet loss would occur less frequently as the throughput capacity of the detection system increases. This allows for a very noticeable speedup. Using an approach designed to reach near real-time detection, but without the speedup beneﬁt proposed by Che-Lun and Hsiao-Hsi, Reference [118] seeks to detect Command & Control servers using autonomous methods. This method eliminates the need to manually detect C&C signatures from an intrusion detection system (IDS). GNU Anubis, which is an SMTP message submission daemon, feeds all the IDS data and extracts all frequent strings. 6. Detection and Mitigation The data are then classiﬁed using three machine learning algo- rithms, Boosted Decision Tree (DT), Naive Bayesian (NB) Classiﬁer and Support Vector Machine (SVM) to classify benign and botnet trafﬁc. In common with Kirubavathi and Anitha, Lin et al. [139] propose a method to identify P2P botnet trafﬁc using data mining on network trafﬁc with NB algorithm. Furthermore, Reference [23] proposes detection, prevention, investigation and mitigation using anomaly, signature, DNS, data mining and hybrid techniques. Lin et al. also proposes the use of J48 and Bayesian networks to be applied to the monitored trafﬁc data, while Lee et al. addresses the use of a ranking algo- Future Internet 2021, 13, 198 19 of 43 19 of 43 rithm to clustering-based botnet detection algorithms [140]. The ranking algorithm gives a higher ranking for source/destination IP pairs with identiﬁed suspicious behaviour. The paper argues that only using k-means clustering results in a large degree of false positives, and that the problem can be solved by ranking the resulting clusters by suspicious TCP and ICMP trafﬁc per source/destination IP pair. Further endorsing the use of k-means, Li et al., propose a botnet detection mechanism using the particle swarm optimisation and K-means algorithms to identify botnet network behaviour [128]. Su et al. proposes a machine learning approach to detect P2P botnets in software-deﬁned networks (SDN) [129]. Detection results are provided to an OpenFlow controller in the SDN, which creates rules to control how botnet source packets are handled at the network switching level. p g Along with network analysis, ﬁlters can be applied to help extract relevant features from network trafﬁc such as connection duration, service type, connection state and more. In [110] by Indre and Lemnaru, the features are provided to a static ﬁlter, binary classiﬁcation ﬁlter and a malware detection ﬁlter. These ﬁlters can reject the connection based on static header rules, general behaviour logic and speciﬁc cyber-attack detection, respectively. Also acting on network behaviour and feature set extraction are multiple papers [111–114], which propose detecting HTTP-based Command & Control servers using behavioural analysis. The feature set found by the papers can be used to further train machine learning algorithms to become even better. Other papers make use of similar methods. Reference [28] uses a supervised machine learning algorithm using a random forest classiﬁer to identify anomalies in IoT networks. 6. Detection and Mitigation Then, a ranking function will assign high scores to trafﬁc-class-distinguished strings, as these are more likely to be good C&C signatures. The authors conclude that the method is a meaningful way to extract C&C signatures in real-world applications. Future Internet 2021, 13, 198 20 of 43 In other near real-time detection mechanisms, Reference [119] proposes an open-source network-based botnet detection and mitigation tool called BotFlex. The tool functions as an intrusion detection system (IDS), passively listening to network trafﬁc and determining botnet trafﬁc from various parameters such as blacklists, C&C detection, outbound spam and more. Toby J. Richer [120] introduces an entropy-based detection mechanism to better detect botnet trafﬁc with variance in beacons to C&C servers. The introduction of an entropy-based measure of delay variance allows for the detection of both ﬁxed-delay and variable-delay beacons. As Sadhan and Moura experimented with tinyP2P and SLINGbot to detect periodic botnet behaviour in botnet trafﬁc by analysing control plane trafﬁc [121]. A somewhat similar approach is BotGM [122], which identiﬁes network trafﬁc behaviour using graph-based mining techniques to detect botnet behaviour. The approach also models the dependencies among network ﬂows to trace back to the root botnet propagators. A study done by Rui et al. [123] shows the behaviour of the Grum, Cutwail and Bobax botnet. The study shows that once a host is infected, a number of Unknown TCP packets are sent on port 80 (in fact HTTP trafﬁc). After multiple SIP invite packets and NBNS queries, the bots usually change a bit in behaviour. The bots behaved like expected with unknown UDP trafﬁc as well as a high amount of HTTP trafﬁc, DNS trafﬁc and SMTP packets for DoS attacks. Their study also shows that these bots mostly infect countries in Europe and America. Supporting this is [5], which further proves the effectiveness of behaviour- based detection systems. On a virtual machine (VM), a detection agent is installed, which monitors the processes and their spawned processes to build a behaviour proﬁle and bot process activity log(s). Calculating the Jaccard similarity coefﬁcient between the behaviour proﬁle and process activity logs is used to indicate if the host is infected or not. In their experiment, they show that their bot behaviour proﬁles and passive detection agent can distinguish bot hosts with no false positives and no false positives. 6. Detection and Mitigation Other research, Reference [133], has also focused on increasing the throughput of real- time DNS-based botnet detection mechanisms. The paper in question proposes ofﬂoading fuzzy pattern recognition of suspected botnet trafﬁc to the cloud, executing the detection in parallel and allowing for near real-time detection. Hoang and Ngyuen have tested several machine learning approaches for domain name systems (DNS) botnet detection, ﬁnding random forest to be the best choice [143] when it comes to use DNS query data. Refer- ences [134,135] further propose the reduction in the network features used for detecting botnet trafﬁc in order to speed up the detection process. A feature minimisation exercise shows the possibility to reduce the selected feature set while still providing a high degree of precision. 6.1.3. Domain Name System (DNS) Based Detection For the purpose of clariﬁcation, Table 6 below details a number of papers that cover DNS-based detection. 6.1.3. Domain Name System (DNS) Based Detection 3. Domain Name System (DNS) Based Detection 6.1.3. Domain Name System (DNS) Based Detection For the purpose of clariﬁcation, Table 6 below details a number of papers that cover DNS-based detection. For the purpose of clariﬁcation, Table 6 below details a number of papers that cover DNS-based detection. Table 6. Papers describing detection of botnets using DNS-based techniques. Each column describes the overall technique, known advantages, disadvantages, detection rate and related papers, respectively. Technique Advantage(s) Disadvantage(s) Detection Rate Papers DNS trafﬁc monitoring Can detect known and unknown botnets by monitoring DNS trafﬁc anomalies. Can detect C&C server migration. The huge size of trafﬁc occurring in network environments is computationally intensive, and due to this, most DNS trafﬁc monitoring methods are also not real-time. Positive rate 90% and negative positive rate 5%. Reference [153] managed accuracy of 95% with 0.1% false positives. Reference [154] archived 98.52% True positive as minimum and 0.39% False positive as highest. [22,23,25,153– 158] Table 6. Papers describing detection of botnets using DNS-based techniques. Each column describes the overall technique, known advantages, disadvantages, detection rate and related papers, respectively. describing detection of botnets using DNS-based techniques. Each column describes the overall technique, ges, disadvantages, detection rate and related papers, respectively. The huge size of trafﬁc occurring in network environments is computationally intensive, and due to this, most DNS trafﬁc monitoring methods are also not real-time. 21 of 43 Future Internet 2021, 13, 198 Table 6. Cont. 6. Detection and Mitigation Technique Advantage(s) Disadvantage(s) Detection Rate Papers DNS trafﬁc monitoring assisted by mining Can detect known and unknown bots. Can successfully locate C&C trafﬁc. Most methods have low false positive rates and can detect encrypted communication. Not real-time. Random forest can archive up to 99% accuracy. [22,23,25,132] Network anomalies Can detect known and unknown bots. Not useful for detecting C&C trafﬁc. Not real-time viable. Positive rate of close to 95% and False positive rate lower than 3.5%. [22,23,25,133, 159] Signature-based detection Can detect most known simple botnets. Cannot detect unknown or more advanced known botnets. N/A. [22,23,25,159] Feature-based detection (CAFE/CSTA) Can reduce the number of non-active C&C suspected Domain Names by 79.96% with false positive rate of 0.69%. Can be affected by malformed DNS answers or DNS cache poisoning attacks. N/A. [160] Support vector machine SVM Can detect existence of botnets in small to medium sized networks. Anti-malware software can increase the false positive rate. Detection rate of 0.935 and false positive rate of 0.02. [161] Table 6. Cont. Domain Name System (DNS)-based detection algorithms are another frequently used approach to combat botnet threats. Most DNS approaches use an allow– deny-list concept to distinguish Domain generation algorithm (DGA) botnets from legitimate trafﬁc. This method is used in [158], where it is seen that most of the domains and their trafﬁc will be allowed by the list. Meanwhile, the rest of the trafﬁc will be clustered using the density-based spatial clustering of applications with the DBSCAN algorithm. The clusters are further analysed to identify botnet domains. This method is similar to [153], which tries to detect botnet-based DNS trafﬁc by the use of Power Spectral Density analysis, a signal processing technique. The method used in [158] shares many similarities with [162], which further adds the use of botnet-generated domain names identiﬁcation using entropy measurements and n-gram scores. The domain names are then measured using a k-means clustering algorithm to identify domains which are likely to be generated by the same botnet DGA. References [163,164] instead use the lexical properties and semantic patterns of real domain names to train their proposed detection schemes. Wang et al. [154] exploit the behaviour of DGA botnets to identify potential botnet trafﬁc. Botnets have a high number of failed DNS lookups stemming from the use of DGA algorithms to generate domain names. The algorithm ﬁlters botnet-generated domain names and clusters them using the Chinese Whispers algorithm. 6. Detection and Mitigation The clusters are then classiﬁed using a supervised machine learning algorithm, based on DNS query times and query amount. Truong and Cheng take several of these algorithms and compares their ability to detect DGA based botnets. Their paper [165] includes a comparison of Naive Bayes, K-nearest neighbour, random forest, support vector machine and decision tree. A more speciﬁc usage of Naive Bayes along with AdaBoost, C4.5 and SVM for Flickr proﬁling as proposed by Natarajan et al. [166]. The multilevel social network proﬁle analysis method is used to detect the Stegobot on social networking websites along with identifying a range of image malware, botcargo and stego images used to identify Stegobot. Reference [167] proposes a new method of combined detection, mitigation and clean-up for next-generation botnet combating. The system consists of ﬁve modules with a task each. This system should be able to communicate, report, detect and heal itself when botnet-enabling malware enters the system. Detection is based on DNS Future Internet 2021, 13, 198 22 of 43 22 of 43 host ﬁles and network inbound ports, which are analysed by the administrator along with a MD5 checksum of the tcp.sys ﬁle. p y Monitoring activity from DNS-queries during C&C communication or updates and applying semi-supervised fuzzy c-means clustering to produce security scenarios is the basis of the self-adaptive system called BotGRABBER [161]. Not much different is the method proposed by Sharalfaldin et al. in [168], where a novel botnet detection framework, BotViz, is presented. BotViz uses a combination of DNS-based analysis of host PC DNS records and API hook forensics on memory dumps to detect potentially vulnerable systems. Forensics are done through an analysis module that uses a k-cluster machine learning algorithm to decide whether or not a host might be compromised by a botnet. Other papers seek to develop methods for botnet detection based on botnet behaviour called C&C Tracer. The C&C Tracer [160] works by using C&C active behaviour feature extracting (CAFE), domain name status querying (DNSQ) and C&C status tracing analyser (CSTA) along with allow lists from multiple external sources such as the Honeypot project and Shadowserver Foundation. An analysis done by Ichise et al. [156] to test the feasibility of botnet detection through domain name system (DNS) records. The analysis shows that in the 5.5 million DNS TXT record queries obtained from their campus network, around 2293 queries where classiﬁed as “unconﬁrmed”. 6.1.4. Detection Mechanisms—Pervasive Computing Paradigms The segment highlights different detection techniques employed in various types of pervasive computing paradigms. These paradigms show different ways in which hosts can establish communication channels and networks, which also affect how botnets can be detected within those networks. 6. Detection and Mitigation In their further investigation, ~22% of these queries were targeting suspicious URLs identiﬁed by virustotal [169]. A similar approach is used by Jin et al. in [157], which proposes a novel DNS-based detection approach for detecting botnet activity. The paper focuses on direct outbound DNS queries on non-standard authoritative name servers to identify botnets, which use TXT records to send commands. The paper ﬁnds that a similar 19% of identiﬁed potentially malicious DNS queries have been ﬂagged by online websites, such as [169], for being used for botnet activity. Reference [159] also proposes a similar idea, but with a focus on UDP network trafﬁc, focusing on DNS MX queries, the DNS packet request and various behaviour that might be botnet attacks based on UDP trafﬁc. Reference [170] talks about a proﬁling dataset. “UMUDGA: a dataset for proﬁling DGA-based botnet” aims to enable researchers to move the data collection, organisation and pre-processing phases forward. Ensuring the availability of good datasets also help the general research community in providing novel detection mechanisms. 6.1.4. Detection Mechanisms—Pervasive Computing Paradigms 6.1.5. IoT and P2P Botnets With the Mirai attack in 2016, some focus have shifted towards IoT networks as potential vulnerable hosts for botnet infection. Therefore, multiple mechanisms for IoT botnet detection have been proposed (see Table 7), both speciﬁcally against Mirai and also some more general mechanisms [171]. Reference [172] speciﬁcally targets Mirai and other known types of attacks with a quantum-inspired detection algorithm. The algorithm matches network trafﬁc headers with a predeﬁned table of IoT botnet attack signatures to detect malicious packets. The authors acknowledge that while not all kinds of botnet attacks have been considered in their approach, the method shows very high true positive rate for detecting known types of IoT botnet behaviour. Reference [173] proposes a sparse- representation framework for botnet detection on the IoT edge. The sparse-representation factor is determined from the network trafﬁc of each individual IoT device, which is then compared against a threshold to determine potential malicious trafﬁc. This allows the network controller to cut off any potentially infected IoT devices. Reference [174] argues for the use of logistic regression of IoT trafﬁc to calculate the probability of an infected device. The regression is based on multiple network parameters including ports, number Future Internet 2021, 13, 198 23 of 43 23 of 43 of requests, mean packet size and more. Finally, Reference [175] proposes using a local agent on IoT devices in an installation to collaboratively compute security events to detect botnet attacks. Botnet attacks are determined on the basis of the difference in DDoS trafﬁc and benign network trafﬁc, which is collectively decided upon by the agents. Table 7. Papers describing detection of botnets in IoT and P2P-based environments. Each column describes the overall technique, known advantages, disadvantages, detection rate and related papers, respectively. Technique Advantage(s) Disadvantage(s) Detection Rate Papers Quantum computing to combat Mirai Fast results. High accuracy. Only targets botnet with known signatures. N/A. [172] Sparse-representation framework on IoT Faster than compared approaches. Tested on limited IoT botnet dataset. 90%. [173] Logistic regression of trafﬁc High accuracy. High precision. Can only detect during propagation phase. >99%. [174] Collaborative multi-agent Can detect large scale DDoS attacks. Lightweight: can be installed on hardware with limited resources. Needs a minimum level of collaboration across organisations. Training environment does not reﬂect real-world environments: A percentage of agents may not acting as excepted. Depends on framework implementation. [175–177] Bitcoin Miners detection Experiments are showing excellent accuracy. 6.1.5. IoT and P2P Botnets Since all agents are able to communicate Future Internet 2021, 13, 198 24 of 43 with each other, they can exchange relevant information such as collection of trafﬁc metrics to identify ongoing DDoS attacks and victims. This exchange procedure is implemented via a blockchain smart contract, which is co-maintained by all nodes in the system. The involved blockchain technology ensures integrity among all the nodes and allows for the collaboration of the distributed nodes without a need of a third party. Another purpose of this technology is in [177], where blockchain is implemented as a framework using HyperLedger to give traceability of the hardware. By the use of a physically unclonable function (PUF), all the IoT-connected devices are sure to be unique. In this way, blockchains is used for veriﬁcation in order to compare and identify these devices with their unique ﬁngerprint ID. Among all these papers, the blockchain-based structure is often used for integrity, decentralisation and transparency among the participants of the chain. It allows these agents to communicate in a much more secure way and therefore make the detection mechanism more reliable and efﬁcient. However, according to a different perspective, A. Zareh and H. R. Shahriari detail another type of target in [178], namely called “botcoins”, which are Bitcoin miner botnets. They propose the use of dynamic analysis of instruction traces in suspicious executable binary ﬁles. A constant parameter value in the assembly exist in all botcoin implementations, which can be detected at the assembly language level. Compared to the other blockchain approaches, the detection strategy in [178] does not iuse blockchain as a communication channel, but analyses how a speciﬁc type of botnet functions. The approach described in [180] uses two factors to determine if a P2P host is part of a botnet: host living-time and command search frequency. The paper argues that P2P botnets exhibit longer-term peer connections and high search request frequency compared to benign P2P trafﬁc. Because legitimate P2P peer connection time is usually short and pull-style communication is uncommon, botnet-behaviour can be detected by those two factors. Likewise, Reference [181] also approaches the detection of P2P botnets by the P2P search frequency. The detection mechanism speciﬁed in the paper also considers the number of P2P peers, the argument being that P2P botnets have a larger number of peer connections compared to normal P2P trafﬁc. 6.1.5. IoT and P2P Botnets Only applicable on speciﬁc botnet types. Depending on the variety of techniques that are utilised. [178] Classiﬁcation from trafﬁc frequency and behaviour. No statistical trafﬁc patterns need to be known in advance. Does not rely on payload data. Does not require monitoring of individual host. Effective (very high detection rate). Scalable low false positive rate. Requires installation at all network boundaries. Unable to detect botnets with low amount of requests. Up to 100%. [76,179–187] Graph-based approach Computationally efﬁcient. Needs a certain amount of training data. Accuracy depends on datasets inspired of the evolving Internet state. N/A. [188] PageRank algorithm using clustered cloud computing Efﬁcient performance (clustering). Scalable and Effective. Easily usable at low costs. High computational needs. 99%. [189] SMTP analysis Can catch both text and image-based botnet spam mails. Limited to mail-based botnets. Detects 96.23% of botnets spam mails with no false positive. [190,191] Botnet application sandboxing Computationally cheaper compared to contemporary intrusion detection systems. Legitimate emails can be ﬂagged wrongly as spam. Not tested [192] Evidential reasoning Can improve botnet detection rates. Lacks an uncertainty evaluation model. Up to 90%. [193] Blockchains are another useful paradigm, which can be included in botnet detection techniques. In the paper [176], the use of lightweight agents included in many IoT installa- tions is discussed. The main goal is to provide a secure communication channel, such as i t t k b t h d ( t) Si ll t bl t i t describing detection of botnets in IoT and P2P-based environments. Each column describes the overall n advantages, disadvantages, detection rate and related papers, respectively. Blockchains are another useful paradigm, which can be included in botnet detection techniques. In the paper [176], the use of lightweight agents included in many IoT installa- tions is discussed. The main goal is to provide a secure communication channel, such as a private network, between each node (agent). Since all agents are able to communicate Blockchains are another useful paradigm, which can be included in botnet detection techniques. In the paper [176], the use of lightweight agents included in many IoT installa- tions is discussed. The main goal is to provide a secure communication channel, such as a private network, between each node (agent). 6.1.5. IoT and P2P Botnets The paper also looks at the periodicity of messages sent, with the argument being that bots periodically request commands from the botmaster. In [189], detecting peer-to-peer botnets using a high-level abstraction of parallel computing called MapReduce is discussed. MapReduce is aiming to divide input data into multiple inputs to make applying functions easier to them. MapReduce also helps running the tasks in parallel over multiple servers. Reference [182] uses both periodicity and active peer connections to determine if a host is part of a P2P botnet. The mechanism described also looks at the ratio of small packets vs. large packets to indicate C&C queries by bot hosts. Other papers such as [183] propose the use of of traditional network trafﬁc analysis based on packet feature selection to detect P2P-based botnets. PeerHunter looks at the number of mutual peers between hosts to detect P2P botnet participants. The number of mutually connected nodes indicates the number of potential botnet communities and is used to identify candidates for botnet detection [179]. Reference [76] further builds on top of PeerHunter to identify whether the previously identiﬁed communities are part of a botnet or not. The detection mechanism for the communities use a network ﬂow analysis method to detect botnets, with the primary factors being the ratio of egress/ingress packets, mutual contacts ratio and destination diversity ratio. Reference [188] uses a graph- based approach to detect P2P trafﬁc, instead opting to exploit the structural properties of the botnet P2P overlay network. The approach checks the number and size of weakly- connected components, average node degree and InO ratio of the P2P overlay network graph to determine if the P2P network is a botnet. Reference [185] uses ﬁrewall logs and the number of outbound connections to detect botnet behaviour. If the number of outbound connections suddenly increases above a threshold, the user is informed. Reference [184] detects HTTP botnet trafﬁc in streaming logs by the use of Lanczos method. The log entries and time slots are put into a ma- trix to check for correlation with botnet behavioural trafﬁc. The paper primarily focuses Future Internet 2021, 13, 198 25 of 43 25 of 43 on comparison to principal component analysis (PCA) and shows that Lanczos method achieves similar results with a 25% reduction in runtime compared to similar approaches. Reference [186] proposes a multi-faceted detection mechanism based on both host and network analysis. 6.1.5. IoT and P2P Botnets The network analysis is based on known botnet behaviour while host analysis is based on the expected host processes and behaviour. If behaviour exceeds or goes beyond expected thresholds, the approach assumes that botnet activity is happening. Reference [187] proposes a method for detecting HTTP based botnets and C&C commu- nication in the cloud using trafﬁc analysis. The paper looks at ﬁve instances of packet capture and analyses the HTTP (TCP packets) trafﬁc, calculating entropy of the captures with TCP payload, length of payload and frequency of each character in the payload. Their test shows that C&C communication is relative similar and can be used for detecting C&C communication of botnets in the cloud. For spam-based botnets, Reference [190] proposes a method for the detection from spam mails received by those botnets. By looking at the mail header, the detection mecha- nism determines if the mail came from a botnet by looking at the sender’s IP, the country of the domain name and the MX host of the sender. If the countries do not match, the sender is assumed to be part of a spam botnet. Reference [191] tackles botnet detection from a cyber-security standpoint, using a multiple detection mechanisms and aggregating the detection results in a central detection log for consideration. The used methods include honeypots, spam collection and recognition as well as high-level analysis based on known botnets. The techniques are based on a case study of the techniques applied at ACDC (Advanced Cyber Defense Centre) in Europe. Reference [192] attempts to detect botnets by blocking botnet-infected hosts from sending mails. The proposed framework uses a whitelisting approach for running soft- ware within hosts, only allowing mails to be sent by authorised applications with a per- application encryption key. A process that sends mail without the authorisation key is ﬂagged as a potential malware. Reference [193] uses a unique approach based on evidential reasoning detection botnets. In this approach, the actions of hosts are mined and reasoned to determine if the actions performed are within the expectations of the host. If not, the host may be detected as being part of a botnet. 6.1.5. IoT and P2P Botnets For the purpose of clariﬁcation, Table 8 below details a number of papers that goes over mobile botnet based detection: For the purpose of clariﬁcation, Table 8 below details a number of papers that goes over mobile botnet based detection: over mobile botnet based detection: Table 8. Papers describing detection of botnets in mobile devices. Each column describes the overall technique, known advantages, disadvantages, detection rate and related papers, respectively. Technique Advantage(s) Disadvantage(s) Detection Rate Papers Risk factor based on multi-category features High accuracy in botnet apps. Generates a pattern for Android botnet detection. Only for static analysis. No response mechanism. Reference [194]: 93.1%. [18,194] Dynamic real-time analysis N/A. Limited analysis for risk factor. N/A. [19] Machine learning General high detection performance. Bad detection performance when the bot coexists with other applications that communicate with many hosts. Achieves 0.93 of the F-measure score by using graphlets of TCP and UDP with 10% of total trafﬁc in 3-minute duration [195]. 99.49% accuracy is achieved [196]. [195,196] Application monitoring Mitigation: user warnings if something is suspicious. SMS and social network applications are not monitored. N/A. [197] Table 8. Papers describing detection of botnets in mobile devices. Each column describes the overall tec advantages, disadvantages, detection rate and related papers, respectively. Future Internet 2021, 13, 198 26 of 43 26 of 43 Another point of interest in botnet detection is the detection of smartphone-based botnets. Some papers, such as Abdullah and Saudi [18], propose assessing the potential risk of malicious apps by evaluating the API calls used by given apps. Apps shown to behave more like botnets are categorised as higher-risk and might potentially be blocked. Reference [18] also attempts to evaluate apps based on risk factors, weighing botnet- behaving apps as higher-risk compared to more benign apps. Reference [19] compares an app’s permissions with a list of known harmful permissions and creates a threat level hierarchy on the basis of said permissions. Reference [194] extends the approach used in [19] by detecting botnets on the basis of both permissions and also the used API calls of each app. Reference [195] proposes the use of a graphlet-based machine learning algorithm on smartphone communication and then executing principal components analysis to identify P2P botnets on smartphones. Other approaches such as [196] run as an active agent on the smartphone OS to capture run-time data. 6.1.5. IoT and P2P Botnets The data is then labelled using a machine learning algorithm to determine whether or not an app acts like a botnet. Reference [197] instead asks the user to specify trusted apps and what permissions a given app should have according to the user. Any apps performing unauthorised or suspicious actions are ﬂagged, and the user is informed. Periodic scans are performed to identify new threats and inform the user of unused apps. 6.1.7. Vehicle Networks With the development of autonomous vehicles, vehicular ad hoc networks (VANETs) have been designed to provide trafﬁc safety by allowing the ad hoc transmission of safety information between vehicles. This additional communication makes VANETs a likely target for malicious attackers. Reference [198] introduces novel attack VANETs and propose a honeypot approach to notify nearby vehicles to ignore messages stemming from vehicles infected by botnets. The paper also proposes the use of localisation mechanisms to limit the exposure to far-away botnets. Reference [12] introduces Shieldnet, which employs a set of machine learning algorithms to detect the use of the GHOST [81] vehicular botnet. The algorithm detects suspicious activity by searching for outlier data within the Basic Safety Messages (BSM) ﬁelds of VANET broadcasts, also isolating known infected hosts using a reputation-based identiﬁcation system. 6.2. Mitigation Mechanisms After detecting a botnet and the threat they can represent, mitigation and countermea- sures have to be deployed to limit the propagation of the botnet-enabling malware and protect the devices from being compromised. Mitigation mechanisms for botnets can be either reactive or proactive and can occur at different levels. The following section lists some of the mitigation strategies that can be employed when dealing with botnet and botnet-based attacks. These countermeasures can be found in Table 9 below. Table 9. This table details the speciﬁc mitigation methods described in the following section, the advantages and disadvan- tages, as well as associated papers. Mitigation Mechanism Advantages Disadvantages Papers Best practices for end-users and organisations Increases overall organisational security, considered best-practice, many well-known standards (i.e., ISO 27001). Does not speciﬁcally target botnets, high user inconvenience cost. [53,200] Network-level blocking and packet analysis Very high protection rate, many solutions and detection frameworks. Very botnet-speciﬁc, can introduce additional latency at network edge. [5,84,98,136,176,201–205] Honeypots and botnet isolation No effect on internal networks. Low cost. Lower protection rate compared to network-level blocking, requires additional logically separated network. [13,22,33,206–208] Attacking P2P botnets Helps mitigate botnet threat for others. Can target speciﬁc botnets. Hinders P2P advantages compared to centralised C&C models. Only targets speciﬁc botnets (P2P-based). Low efﬁciency for organisations. [209,210] IoT-speciﬁc mitigation strategies Low or ofﬂoaded compute resource cost. Some solutions provide general integrity for IoT-based networks. Speciﬁc for IoT-based threats. Still few and untested options compared to network-level blocking. [2,106,175,211,212] Community-driven approaches Potentially quicker adaption to newer botnets. Free and Open Source for organisations to use. Dependent on community development. No de-facto standard decided yet. [119,213] Botnet mitigation with ethical issues (spreading anti-botnets, attacking suspected hosts) Mitigates botnets for others. Slows botnet propagation. Ethically questionable or illegal. Very speciﬁc per-botnet. [49,82,214–217] In general, following IT best practices is a good way to avoid the propagation and infection of botnets. An article of Justice news [53] coming from the Department of Justice (DOJ) of the United States announced “a multi-national effort to disrupt the Gameover Zeus Botnet”. The GameOver Zeus (GOZ) botnet is described as being capable of infecting victim computers to harvest credentials and banking information in order to gather millions of dollars from companies and customers. Therefore, a cybersecurity alert [200] at the National Cyber Awareness System has been released to explain how the botnet works and how attacks can be avoided. 6.1.8. Social Network Botnets Social network-based botnets (SnB) have become a major security issue in the past few years. Their incentives are based on sensitive information stealing, and perform complex communication procedures. Publicly available resources are highly vulnerable and provide an obfuscation layer in the C&C communication for botnets. The study [87] of this new malware method is essential to understand the actual challenges in detecting and mitigating social botnets. Social networks contain information such as sensitive and personal data of both registered and unregistered users. Moreover, it acts as a human- driven communication channel to share, talk and learn. This tool can be helpful in some respects, but it can also be destructive, e.g., propagating the inﬂuence of botnets [199]. To understand the behaviour of social network botnets, T. Yin and Y. Zhang and S. Li detail [60] the design and implementation of a Social Network-based botnet called DR-SNBot. The paper presents the necessary framework to deploy a C&C channel on the Sina blog website with a nickname generation algorithm and divide-and-conquer strategy. Compared to [60,166], it contains strong analysis on how to detect covert SnB in the real world. It is focused on the Stegobot and how to monitor host proﬁle activity from a social network, and by extension, differentiate a normal proﬁle from a Stegobot’s one. Proﬁles are analysed by looking at their number of friends, likes and shares. A Stegobot has predictable patterns and communicates secret messages via carrier images, called “stego images”, through the content sharing system from social networks. Their strategy is to study statistical correlation and build a classiﬁcation algorithm using Machine Learning to identify malicious activities and suspicious accounts. 27 of 43 Future Internet 2021, 13, 198 27 of 43 6.2. Mitigation Mechanisms 6.2. Mitigation Mechanisms This assessment and mitigation document is written in collaboration with Department of Homeland Security (DHS), the DOJ and the Federal Bureau of Investigation (FBI). From this source, it is possible to get a grasp on the default methods of countermeasures used against every common botnet or malware: • Updating/changing passwords: typically, botnets will try to access credentials from all connected devices and web accounts. The best way of protection is to follow • Updating/changing passwords: typically, botnets will try to access credentials from all connected devices and web accounts. The best way of protection is to follow Future Internet 2021, 13, 198 28 of 43 the rules of ensuring high entropy of random password generation and execute frequent updates. the rules of ensuring high entropy of random password generation and execute frequent updates. • Updating devices: infections are coming from unwanted vulnerabilities. Updating the operating system and the integrated software can help prevent devices from being compromised. g p • Updating/using anti-malware and anti-virus tools: remediation tools and anti-viruses can erase malware infection and protect the device against new ones. • Being aware: the hardest part to protect from is human behaviour. There ar multiple incentives, but botnets such as GOZ are mostly coming from spam and phishing messages, which can be avoided if the potential victim is aware of this potential threat source. However, individual techniques are often not efﬁcient enough to eradicate such threats. The Justice news article [53] explains the authorisation and capacity of redirecting requests made by the infected computers away from the malicious operators. With the evolution of the botnets detailed in this report, cyber defence needs to evolve and new mitigation techniques need to handle more complex attacks. Moreover, some of the newly described strategies only target speciﬁc types of botnets. 6.2.2. Network-Level Blocking and Packet Analysis 6.2.2. Network-Level Blocking and Packet Analysis Within technical mitigation for botnet propagation, the use of network-level blocking is one of the most cited strategies. Many detection papers focus on network-level detection, which can be used by intrusion detection systems to block and contain botnets. In [202], an autonomous system (AS) is used to mitigate botnet threats. The AS stores a list of hosts’ IP addresses and a threshold per host based on classiﬁcation. Categories can be “Blacklist”, “Whitelist”, “Suspected Attacker” and “Possible Victim”. AS are connected synchronously via the Ethereum blockchain. The threshold is monitored by every AS and refreshed after 20 s. Another way to ensure packets blocking is the software-deﬁned networking approach [84,203]. The main purpose is to analyse the incoming packets rate at deﬁned IoT switches to separate legitimate from malicious communication. Legitimate trafﬁc is accepted, while malicious ones are blocked. Many mitigation strategies use a locally installed agent on host machines to block detected botnet trafﬁc, informing the user of the infected nature of their machine [201]. Blocking can also be performed at the edge of the service provider but would face high implementation costs and requires some coordination across ISPs [204]. At the network level, removal of malware can be performed by agents installed locally or by the use of a continuous communication protocol with a master device. This validates of the integrity of local hosts and allows administrators to perform removal of botnet-enabling malware from hosts, either automatically or manually [5,176,205]. g y y In [136], the authors propose a self-adaptive system for mitigation. In corporate area networks for instance, resilience can be ensured by using scenario-driven adaptive reconﬁguration of networks. Scenarios are assessed and based on cluster analysis coming from previous botnet attacks. Moreover, the described system can apply more advanced actions such as reducing requests timeouts, decreasing allowed HTTP request size and blocking source hostname and IP addresses. In [98], a neural network-based system called BoNeSSy will notify the administrator if a threat is found and apply appropriate security actions such as blocking IP addresses or putting the system or suspicious network segment under surveillance. Many papers describe the detection of botnets using Machine Learning clustering via statistical behaviour correlation, but some of them are lacking of speciﬁc countermeasures description. Some characteristics can be countered by packets or IP addresses blocking. 6.2.4. Attacking P2P Botnets Reference [209] proposes the use of poisoning of the routing table of P2P botnets as a potential mitigation method. By disrupting the majority of entries in the shared routing table of P2P botnets, it becomes possible to hinder some of the advantages that these types of botnets enjoy over the centralised model, such as resource efﬁciency and fault tolerance. Reference [210] also proposes disrupting P2P botnets but uses an optimised and tailored Sybil attack to inﬁltrate botnets and therefore mitigate them by disrupting or even taking them down from the inside. Placing Sybil nodes in the botnet shows that random placement is just as effective as informed placement due to the nature of P2P botnets. These nodes are able to disrupt communication between other nodes within the P2P botnet. 6.2.3. Honeypots and Botnet Isolation One of the most frequently described strategies [13] for mitigation is to isolate the bot- net in order to perform information gathering and analysis of its behaviour and interaction via, for instance, honeypots [33,206] and honeynets [22]. From this information collection Future Internet 2021, 13, 198 29 of 43 and assessment, it is possible to categorise the botnet based on behavioural characteristics and botnet structure. Organisations and researchers are producing and studying many methods of mitigation with various qualities and limitations. Honeypot behaviour has been shown to be detectable by intelligent botnets however. Although this is the case, the research and deployment of honeypots still has value for the scientiﬁc and industrial com- munities. The continued research in covert honeypots is therefore paramount to continue reaping the insights gained by using honeypots [207,208]. 6.2.6. Community Driven Tools against Botnets Reference [119] proposes the use of a community driven framework, BotFlex, to contin- ually improve mitigation of botnets across the entire IT community. The approach attempts to standardise network-based intrusion detection systems with an extensible module sys- tem. Other researchers and corporations can contribute to the system with modules to improve upon BotFlex. In other community-driven approaches, Reference [213] proposes the use of a botnet defence description language to describe the tasks and information shar- ing primitives between devices handling botnet defence. Some community-driven efforts attempt to detect and prevent botnets by providing databases with known spam bots such as the The Spamhaus Project [218] and IBM X-Force exchange [219], where IT researchers can report suspected IP addresses and see a list of IP addresses along with a % indicator of how likely the IP is used for C&C. Furthermore Structured Threat Information eXpression (STIX) is used for exchanging cyber threat intelligence (CTI) as described in [220]. Dog et al. [221] examined the value of sharing IDS logs between enterprises and not just sharing IP addresses, domains and speciﬁc attacks. The study shows that intelligence sharing can provide good strategic threat information for enterprises. 6.2.7. Botnet Mitigation with Potential Ethical Issues 6.2.5. Mitigation against IoT Attacks and Botnets Learning from the Mirai botnet attack illustrates multiple general best practices, which can be used as a mitigation against IoT botnets. These methods include changing default credentials, closing unused service ports like telnet, detecting disabled watchdogs (Mirai speciﬁc) and the use of automated scripts to validate the implementation of the proposed mitigation [2,106]. Other proposed mitigation methods include switching from telnet to SSH (if possible) or changing the default service ports of services. Ensuring proper isolation of users and service account permissions and disabling any unencrypted communications (like HTTP) might mitigate some IoT botnet attacks [211]. Known ports vulnerable to attacks should also be continually monitored to quickly react to suspicious trafﬁc [212]. Some local IoT agents have also been proposed to collectively mitigate the potential damage of DDoS attacks targeting local IoT installations [175]. 6.2.6. Community Driven Tools against Botnets 6.2.7. Botnet Mitigation with Potential Ethical Issues Reference [214] discusses the ethical implications of ﬁghting botnets with sinkholes. The information gathered by these sinkholes can be sold to government agencies, politi- cians, contractors and many more. This information includes geographical location of compromised hosts, operation system including version and the ability to target these Future Internet 2021, 13, 198 30 of 43 already compromised hosts for future botnet or malware attacks. On the bright side, it could also help ISPs to provide their customers the service of malware protection. Another popular approach to mitigate botnets is to use their propagation mechanisms to propagate harmless versions of the given botnet. Actively ttacking the Mirai botnet and other IoT botnets to mitigate their threat has also been proposed [82]. Some researchers have tried to attack spam botnets to send unknowing users to more safe sites [215]. Another example, Reference [49], attempts to mitigate the Conﬁcker botnet by spreading an anti-botnet, which blocks Conﬁcker from executing and overtakes the propagation mechanism of Con- ﬁcker to spread the anti-botnet instead. These approaches can be considered ethically problematic, as they intentionally spread (harmless as they might be) self-propagating malware [216,217]. 7. Current Trends and Challenges For the purpose of clariﬁcation, Table 10 below details a number of papers discussed in this section. The table denotes the overall topic, the overall trends within aforementioned topic, the relative interest for this speciﬁc trend, and a listing of all associated papers. Table 10. Overview of papers discussing current trends and topics concerning botnets. The columns describes the trends of the overall associated area of interest, the detailed topics discussed in each paper, the relative interest amongst the associated trend and ﬁnally a listing of all the associated papers. Trend Topics Within Trend Relative Interest Papers Pervasive Computing Spread of botnets in home appliances 2 out 18 papers listed. [31,201] Spread of botnets in mobile phones 2 out 18 papers listed. [222,223] Spread of botnets in (non)-autonomous vehicles. 2 out 18 papers listed. [81,198] Remotely disrupting the controls of an autonomous vehicle. 1 out of 18 papers listed. [81] Smartphones exploited via insufﬁcient app certiﬁcation process. 2 out of 18 papers listed. [37,197] Lack of restrictions hinders the process of avoiding botnet apps on mobile devices. 2 out of 18 papers listed. [207,211] Various proposals for IoT malware protection, both generalised and specialised. 4 out of 18 papers listed. [2,205,207, 211] Usage of honeypots helps make more real-life like data for mitigation strategies. 1 out of 18 papers listed. [224] No standardised way to protect pervasive computing device hurts development of mitigation strategies. 1 out of 18 papers listed. [225] Best-practices in IT security yearns for standardising security in IoT and mobile devices. 1 out of 18 papers listed. [2] Increasing complexity of botnets Most ﬁrewalls and intrusion detection systems are not able to ﬁlter IPv6 trafﬁc. 2 out of 3 papers listed [226,227] Modern botnets can circumvent traditional detection methods using encrypted channels for traditionally unencrypted trafﬁc. 1 out of 3 papers listed. [228] Social Botnets Social botnets can be used for multiple purposes, including spam, C&C and falsifying/impersonating user behaviour. 2 out of 4 papers listed. [87,199] It is growing increasingly harder for users to discern between true and false information, beneﬁting botnets. 1 out of 4 papers listed. [166] New and more advanced counter measures are necessary to combat this new development of social botnets. 1 out of 4 papers listed. [1] Trend 31 of 43 Future Internet 2021, 13, 198 Table 10. Cont. 7. Current Trends and Challenges The Continued Spread of Botnets within Pervasive Computing (VANETs, IoT and Mobile) The most common trend and challenge within botnet research between 2013 and 2021 is the continued spread of botnets anchored in pervasive computing devices. With the increase in computational power within normally benign devices, such as home appliances [31,201], mobile phones [222,223] and (non)-autonomous vehicles [81,198], a higher potential of malicious activity within these devices becomes viable. As devices are allowed more computational headroom, botnets’ ability to perform increasingly effec- tive obfuscation techniques to mask their existence within pervasive devices grows ever more concerning. Section 5 explains that the damage of botnets has mostly been within information channels, with attacks on the availability of computing system and acquisition of user credentials and national intelligence data being some of the primary targets of botnets. With pervasive computing, however, that threat of disruption transitions into the physical realm. Remotely disrupting the controls of an autonomous vehicle [81] can have potentially fatal results for the people within. Smart devices such as pacemakers and other computer- enabled medical devices may also provide a potentially fatal target for malicious actors or terrorists [234]. The mitigation of these attacks may vary greatly, depending on the speciﬁc scenario and device in question. Some devices, such as smartphones, are shown to be ripe for exploitation. An example of this is the app certiﬁcation process, which has been shown to be insufﬁcient [37,197] to prevent malicious apps from getting into various app stores. Furthermore, some device operating systems, like Android, do not place strict limitations on installing apps through unauthorised sources or package repositories. This considerably complicates the process of avoiding botnet apps on mobile. For IoT devices, attacks such as the Mirai botnet [207,211] shows the lack of basic security conﬁguration and security investment within IoT development. Some solutions for IoT malware protection, both generalised and specialised, have been proposed within research though [2,205,207,211]. Attempts have also been made to make more real-life data based on honeypots available for researchers, in order to propose more IoT mitigation strategies [224]. However, so Table 10. Cont. The current state of botnets and botnet research is consistently changing. As described in Section 5, on the topic of botnet evolution, botnets are constantly evolving. This section lists some of the general trends and challenges that have been identiﬁed during the reading for this paper. 7.1. 7. Current Trends and Challenges Trend Topics Within Trend Relative Interest Papers Machine learning and neural networks for botnet detection New research on machine learning based detection schemes shows high rates of true positive detection of botnet behaviour. 8 out of 9 papers listed. [24,92,99,109, 124,125,127, 136] Discussion on the ability to train for zero-day vulnerabilities with custom created datasets. 1 out of 9 papers listed. [170] Proactive botnet mitigation Proactive botnet mitigation techniques show promising results. 3 out of 6 papers listed. [130,207,229] Proactive mitigation strategies and tools need to be developed for both the local and international stage. 2 out of 6 papers listed. [230,231] Users willing to pay for botnet prevention, but lack awareness. 1 out of 6 papers listed. [232] Cloud-based botnets Cloud services can be used for C&C communications between bots and bot masters, masquerading as benign user trafﬁc. 1 out of 3 papers listed. [187] Cloud services offer options for researchers to create botnets without hassle. 2 out of 3 papers listed. [87,233] Table 10. Cont. Trend Topics Within Trend Relative Interest Papers Machine learning and neural networks for botnet detection New research on machine learning based detection schemes shows high rates of true positive detection of botnet behaviour. 8 out of 9 papers listed. [24,92,99,109, 124,125,127, 136] Discussion on the ability to train for zero-day vulnerabilities with custom created datasets. 1 out of 9 papers listed. [170] Proactive botnet mitigation Proactive botnet mitigation techniques show promising results. 3 out of 6 papers listed. [130,207,229] Proactive mitigation strategies and tools need to be developed for both the local and international stage. 2 out of 6 papers listed. [230,231] Users willing to pay for botnet prevention, but lack awareness. 1 out of 6 papers listed. [232] Cloud-based botnets Cloud services can be used for C&C communications between bots and bot masters, masquerading as benign user trafﬁc. 1 out of 3 papers listed. [187] Cloud services offer options for researchers to create botnets without hassle. 2 out of 3 papers listed. [87,233] The current state of botnets and botnet research is consistently changing. As described in Section 5, on the topic of botnet evolution, botnets are constantly evolving. This section lists some of the general trends and challenges that have been identiﬁed during the reading for this paper. 7.1. 7. Current Trends and Challenges With the differences in architecture and use-scenario of vehicles, IoT and mobile, a completely standardised approach across platforms might be a stretch. Some general best-practises within IT security, like avoiding default credentials, closing unused services and continuous validation of the platform, still apply for all pervasive computing devices [2]. 7.3. Social Botnets Social botnets are another challenge that have been under development for a while and are predicted to threaten online security and the integrity of information online. Social botnets speciﬁcally use social platforms for multiple purposes, including spam, C&C and falsifying or impersonating user behaviour [87] at an increasing rate [199]. This gives botnet developers more tools and ways to avoid typical detection mechanisms by communicating through seemingly benign social user accounts. Furthermore, as the amount of information processing of individuals increases and social botnets for the purpose of spam grow more advanced, it becomes increasingly unlikely for casual social network users to distinguish between true and false information [166]. Online social networks have recently stepped up by increasingly removing false social accounts. However, additional research is necessary to provide more generalised as well as specialised anti-measures for social botnets [1]. 7.2. Increasing Complexity of Botnets Ravishankar expects future botnet threats to include encrypted communication, where a bot herder would encrypt the bot binary with a strong public/private key pair. Self- destruction mechanisms where the bot deletes registry ﬁles to try and enforce the user to reinstall the operation system and thereby get rid of most logged evidence, makes it hard for antivirus companies to analyse the botnets. Decentralised botnets such as P2P do not suffer from the vulnerability of a single point of failure, making the take down more complex. Tor-based onion routing can obfuscate communication to make eavesdropping and trafﬁc analysis almost impossible. Tor can also be used for the bot herder to stay anonymous while setting up a new botnet. IPv6 can be misused to carry edited binary ﬁles and instructions to bots, malware tunnelling would also be possible in some situations, and lastly, most ﬁrewalls and intrusion detection systems are not yet able to ﬁlter IPv6 trafﬁc [226,227]. Traditional detection methods by performing trafﬁc analysis of DNS queries can be prevented by modern botnets utilising encrypted channels for traditionally unencrypted trafﬁc [228]. 7. Current Trends and Challenges The Continued Spread of Botnets within Pervasive Computing (VANETs, IoT and Mobile) The most common trend and challenge within botnet research between 2013 and 2021 is the continued spread of botnets anchored in pervasive computing devices. With the increase in computational power within normally benign devices, such as home appliances [31,201], mobile phones [222,223] and (non)-autonomous vehicles [81,198], a higher potential of malicious activity within these devices becomes viable. As devices are allowed more computational headroom, botnets’ ability to perform increasingly effec- tive obfuscation techniques to mask their existence within pervasive devices grows ever more concerning. Section 5 explains that the damage of botnets has mostly been within information channels, with attacks on the availability of computing system and acquisition of user credentials and national intelligence data being some of the primary targets of botnets. With pervasive computing, however, that threat of disruption transitions into the physical realm. Remotely disrupting the controls of an autonomous vehicle [81] can have potentially fatal results for the people within. Smart devices such as pacemakers and other computer- enabled medical devices may also provide a potentially fatal target for malicious actors or terrorists [234]. The mitigation of these attacks may vary greatly, depending on the speciﬁc scenario and device in question. Some devices, such as smartphones, are shown to be ripe for exploitation. An example of this is the app certiﬁcation process, which has been shown to be insufﬁcient [37,197] to prevent malicious apps from getting into various app stores. Furthermore, some device operating systems, like Android, do not place strict limitations on installing apps through unauthorised sources or package repositories. This considerably complicates the process of avoiding botnet apps on mobile. For IoT devices, attacks such as the Mirai botnet [207,211] shows the lack of basic security conﬁguration and security investment within IoT development. Some solutions for IoT malware protection, both generalised and specialised, have been proposed within research though [2,205,207,211]. Attempts have also been made to make more real-life data based on honeypots available for researchers, in order to propose more IoT mitigation strategies [224]. However, so Future Internet 2021, 13, 198 32 of 43 far no standardised ways to protect pervasive computing devices from botnets have been implemented. The lack of data sets for large IoT botnets in the wild is also seen as a challenge for the further development of mitigation against botnets targeting IoT installations [225]. 7.4. Machine Learning and Neural Networks for Botnet Detection Detecting botnets using machine learning and neural networks has gained prominence amongst researchers and developers. Section 6 shows a clear majority of recent papers focusing on these techniques in order to achieve a high true positive rate of detection of botnet behaviour [24,92,99,109,124,125,127,136]. The additional advantages of potential real-time detection and mitigation and cloud-ofﬂoading for training/learning has allowed these techniques to establish themselves as the solution to botnet detection for the foresee- able future. Some parameters, such as the ability to train for zero-day botnet behaviour, are still a topic for discussion. Some papers, such as Zago et al.’s [170], have already tried to create data sets suitable to train machine learning algorithms for detecting certain patterns occurring within botnets. It is therefore expected that research will continue to attempt to improve upon the detection techniques based on machine learning and neural networks. 7.5. Proactive Botnet Mitigation Most botnet detection mechanisms speciﬁed in Section 6 are reactive by nature. This allows botnets to ﬂourish outside of properly protected and monitored networks. Because Future Internet 2021, 13, 198 33 of 43 33 of 43 of the aforementioned reactive nature of both the detection and mitigation mechanisms, the proposed techniques only allow for protection at the local network level. This leaves many unprotected users vulnerable to botnet attacks, which may lead to technical and ﬁnancial headaches for entities such as ISPs and the corporations supporting said users. Some proactive botnet mitigation techniques, such as honeypots and botnets overtaking other botnets show promising results, as documented in [130,207,229]. These new methods help raise user awareness while rendering other botnets harmless. An increased focus on proactive mitigation and detection strategies is necessary, not just to mitigate botnets at the local network level, but the international stage as well [230,231]. Research shows that users are willing to pay for services from their ISPs to prevent botnet attacks, but the lack of awareness nevertheless hurts the potential reach of such offerings [232]. 7.6. Cloud-Based Botnets Finally, the normalisation of cloud-computing has allowed for greater computing power to both aid and combat botnets. This increase in computing resources and publicly available communication services, such as Google Cloud, has also been exploited by hackers to create botnets in the cloud. As cloud deployments are virtually instantaneous and on-demand, the cloud allows bot masters to dynamically scale the size of their botnets to match the needed computing power for attacks. Additionally, cloud services have been shown to be used for C&C communications between bots and bot masters, masquerading as benign user trafﬁc [187]. Mitigation techniques based on IPs and locations are shown to be ineffective due to the relative in-deterministic nature of cloud deployment locality. The cloud can also be seen as a potentially attractive option for botnet research due to its low price, allowing researchers to instantly create botnets without having to make ethically questionable decisions such as infecting user computers in the wild [87,233]. References 1. Silva, S.S.; Silva, R.M.; Pinto, R.C.; Salles, R.M. Botnets: A survey. Comput. Netw. 2013, 57, 378–403. Botnet Activity: Analysis, Detection and Shutdown. [CrossRef] 2. Margolis, J.; Oh, T.T.; Jadhav, S.; Kim, Y.H.; Kim, J.N. An In-Depth Analysis of the Mirai Botnet. In Proceedings of the 2017 International Conference on Software Security and Assurance (ICSSA), Altoona, PA, USA, 24–25 July 2017; pp. 6–12. [CrossRef] 3. Haria, S. The growth of the hide and seek botnet. Netw. Secur. 2019, 2019, 14–17. [CrossRef] , g , , [ ] 4. ENISA Threat Landscape 2020—Botnet. Available online: https://www.enisa.europa.eu/publications/enisa-threat-landscape- 2020-botnet (accessed on 28 May 2021). 5. Hsiao, S.; Chen, Y.-N.; Sun, Y.S.; Chen, M.C. A cooperative botnet proﬁling and detection in virtualized environment. In Proceedings of the 2013 IEEE Conference on Communications and Network Security (CNS), National Harbor, MD, USA, 14–16 October 2013; pp. 154–162. [CrossRef] gency Cybersecurity. Available online: https://www.enisa.europa.eu/topics/csirts-in-europe/glossary/ n 30 July 2020). 6. European Union Agency Cybersecurity. Available online: https://www.enisa.europa.eu/topics botnets (accessed on 30 July 2020). 7. Zhang, W.; Wang, Y.J.; Wang, X.L. A Survey of Defense against P2P Botnets. In Proceedings of the 2014 IEEE 12th International Conference on Dependable, Autonomic and Secure Computing, Dalian, China, 24–27 August 2014; pp. 97–102. [CrossRef] 8 I lli N H k th A B t t V hi l f O li C i CERT C d C t 2005 28 19 39 [C R f] 8. Ianelli, N.; Hackworth, A. Botnets as a Vehicle for Online Crime. CERT Coord. Cent. 2005, 28, 19–39. [C 8. Ianelli, N.; Hackworth, A. Botnets as a Vehicle for Online Crime. CERT Coord. Cent. 2005, 28, 19–39. [CrossRef] 8. Ianelli, N.; Hackworth, A. Botnets as a Vehicle for Online Crime. CERT Coord. Cent. 2005, 28, 19–39. [CrossRef] 9. Etaher, N.; Weir, G.R.S.; Alazab, M. From ZeuS to Zitmo: Trends in Banking Malware. In Proceedings of the 2015 IEEE 9. Etaher, N.; Weir, G.R.S.; Alazab, M. From ZeuS to Zitmo: Trends in Banking Malware. In Proce Trustcom/BigDataSE/ISPA, Helsinki, Finland, 20–22 August 2015; Volume 1, pp. 1386–1391. [CrossRe r, N.; Weir, G.R.S.; Alazab, M. From ZeuS to Zitmo: Trends in Banking Malware. In Proceedings of com/BigDataSE/ISPA Helsinki Finland 20 22 August 2015; Volume 1 pp 1386 1391 [CrossRef] 10. Elliott, C. Botnets: To what extent are they a threat to information security? Inf. Secur. Tech. Rep. 2010, 15, 79–103. Computer Crime—A 2011 Update. [CrossRef] 11. References Eslahi, M.; Salleh, R.; Anuar, N.B. Bots and botnets: An overview of characteristics, detection and challenges. In Proceedings of the 2012 IEEE International Conference on Control System, Computing and Engineering, Penang, Malaysia, 23–25 November 2012; pp. 349–354. [CrossRef] 12. Garip, M.T.; Lin, J.; Reiher, P.; Gerla, M. SHIELDNET: An Adaptive Detection Mechanism against Vehicular Botnets in VANETs. In Proceedings of the 2019 IEEE Vehicular Networking Conference (VNC), Los Angeles, CA, USA, 4–6 December 2019; p. 9062790. [CrossRef] 13. Garg, S.; Sharma, R.M. Anatomy of botnet on application layer: Mechanism and mitigation. In Proceedings of the 2017 2nd International Conference for Convergence in Technology, I2CT 2017 Mumbai, India, 7–9 April 2017; pp. 1024–1029. [CrossRef] 14 L T K i H O S i Th f B C b S I P di f h 2019 6 h I i l C f 13. Garg, S.; Sharma, R.M. Anatomy of botnet on application layer: Mechanism and mitigation. In Proceedings of the 2017 2nd International Conference for Convergence in Technology, I2CT 2017 Mumbai, India, 7–9 April 2017; pp. 1024–1029. [CrossRef] 14. Lange, T.; Kettani, H. On Security Threats of Botnets to Cyber Systems. In Proceedings of the 2019 6th International Conference on Signal Processing and Integrated Networks (SPIN) Noida India 7–8 March 2019; pp 176–183 [CrossRef] 14. Lange, T.; Kettani, H. On Security Threats of Botnets to Cyber Systems. In Proceedings of the 2019 6th on Signal Processing and Integrated Networks (SPIN), Noida, India, 7–8 March 2019; pp. 176–183. [Cr 15. García, S.; Zunino, A.; Campo, M. Survey on network-based botnet detection methods. Secur. Commun. Netw. 2014, 7, 878–903. [CrossRef] 16. Karim, A.; Salleh, R.; Shiraz, M.; Shah, S.; AWAN, I.; Anuar, N. Botnet detection techniques: Review, future trends and issues. J. Zhejian Univ. Comput. Electron. 2014, 15, 943–983. [CrossRef] 17. Khehra, G.; Sofat, S. Botnet Detection Techniques: A Review. In Proceedings of the 2018 Second Inte Intelligent Computing and Control Systems (ICICCS), Madurai, India, 14–15 June 2018; pp. 1319–1326 17. Khehra, G.; Sofat, S. Botnet Detection Techniques: A Review. In Proceedings of the 2018 Second International Conference on Intelligent Computing and Control Systems (ICICCS), Madurai, India, 14–15 June 2018; pp. 1319–1326. [CrossRef] 18. Abdullah, Z.; Saudi, M. RAPID-Risk Assessment of Android Permission and Application Programming Interface (API) Call for 18. Abdullah, Z.; Saudi, M. RAPID-Risk Assessment of Android Permission and Application Programming Interface (API) Call for Android Botnet. Int. J. Emerg. Technol. 8. Conclusions This paper sought out to produce a novel systematic literature review detailing different subjects related to botnets, a growing subgroup of malware-enabled attacks. Botnets are widely used by malicious actors with various motivations and intentions, from simple denial-of-service attacks to advanced cyber espionage. The actors behind botnets therefore span a large range from security researchers spreading anti-botnets to foreign nations attempting to destabilise infrastructure. The relatively simple structure and potential payoff of a successful attack has been a driving force in the evolution of botnets for decades. The adaptability of botnets are seen in their evolution towards modern platforms such as vehicles, smartphones and IoT devices. Modern botnets are still evolving rapidly, and more advanced counter-detection mechanisms and command-and-control channels are being introduced. It has been shown that recent detection mechanisms based on machine learning and artiﬁcial neural networks provide very high rates of detecting botnet threats. Both these approaches provide additional accuracy to common network behaviour-based approaches. These detection techniques support traditional mitigation strategies such as security best practices and network-level blocking to reduce the risk and impact of botnet attacks. In particular, the spread of pervasive computing paradigms such as the Internet of Things and vehicular networks provide a fertile ground for botnets to spread. Current trends point towards the increase of computing power within pervasive computing as an enabler for botnets to enable malware to spread. Other now-commonplace computing paradigms, such as cloud computing and interconnected social networks, have also seen an increase in interest as potential enablers of botnets. Author Contributions: Conceptualization and methodology, all authors contributed equally; inves- tigation and writing—original draft preparation, S.N.T.V., M.S., P.I.E.-H. and J.B.; writing—review and editing, all authors contributed equally; supervision, N.D.; project administration, N.D.; funding acquisition, N.D. All authors have read and agreed to the published version of the manuscript. 34 of 43 34 of 43 Future Internet 2021, 13, 198 Funding: This research was partially funded by Industriens Fond (Danish Industry Foundation), project “CIDI–Cybersecure IoT in Danish Industry”, grant number 2018-0197. Data Availability Statement: Not applicable, the study does not report any data. Conﬂicts of Interest: The authors declare no conﬂict of interest. Conﬂicts of Interest: The authors declare no conﬂict of interest. References Learn. 2018, 7, 49–54. [CrossRef] , ; , pp g Android Botnet. Int. J. Emerg. Technol. Learn. 2018, 7, 49–54. [CrossRef] me Analysis of Android Applications by Calculating Risk Factor to Identify Botnet Attack. Lect. Notes Electr 2. [CrossRef] 19. Kothari, S. Real Time Analysis of Android Applications by Calculating Risk Factor to Identify Bot Eng. 2020, 570, 55–62. [CrossRef] 20. Eslahi, M.; Salleh, R.; Anuar, N.B. MoBots: A new generation of botnets on mobile devices and networks. In Proceedings of the ISCAIE 2012—2012 IEEE Symposium on Computer Applications and Industrial Electronics, Kota Kinabalu, Malaysia, 3–4 December 2012; p. 6482109. [CrossRef] 21. Kaur, N.; Singh, M. Botnet and botnet detection techniques in cyber realm. In Proceedings of the 2016 on Inventive Computation Technologies (ICICT), Tamilnadu, India, 26–27 August 2016; Volume 3, pp. 21. Kaur, N.; Singh, M. Botnet and botnet detection techniques in cyber realm. In Proceedings of the 2016 International Conference on Inventive Computation Technologies (ICICT), Tamilnadu, India, 26–27 August 2016; Volume 3, pp. 1–7. [CrossRef] 22 F il M Sh h i A R d S A S f B d B D i I P di f h 2009 Thi d I i l 22. Feily, M.; Shahrestani, A.; Ramadass, S. A Survey of Botnet and Botnet Detection. In Proceedings of the 2009 Third International Conference on Emerging Security Information, Systems and Technologies, Athens, Greece, 18–23 June 2009; pp. 268–273. [CrossRef] 23. Abdullah, R.; Abu, N.; Abdollah, M.; Muhamad Noh, Z.A. Understanding the Threats of Botnets Detection: A Wide Scale Survey. Res. J. Inf. Technol. 2014, 6, 135–153. [CrossRef] 24. Gaonkar, S.; Dessai, N.F.; Costa, J.; Borkar, A.; Aswale, S.; Shetgaonkar, P. A Survey on Botnet Detection Techniques. In Proceedings of the 2020 International Conference on Emerging Trends in Information Technology and Engineering (IC-ETITE), Vellore, India, 24–25 February 2020; pp. 1–6. [CrossRef] Future Internet 2021, 13, 198 35 of 43 25. Shetu, S.F.; Saifuzzaman, M.; Moon, N.N.; Nur, F.N. A Survey of Botnet in Cyber Security. In Proceedings of the 2019 2nd International Conference on Intelligent Communication and Computational Techniques (ICCT), Jaipur, India, 28–29 September 2019; pp. 174–177. [CrossRef] pp 26. Stevanovic, M.; Pedersen, J.M. An analysis of network trafﬁc classiﬁcation for botnet detection. In Proceedings of the 2015 International Conference on Cyber Situational Awareness, Data Analytics and Assessment (Cybersa), London, UK, 8–9 June 2015; p. 8. [CrossRef] p 27. References Haddadi, F.; Le Cong, D.; Porter, L.; Zincir-Heywood, A.N. On the Effectiveness of Different Botnet Detection Approaches. In Information Security Practice and Experience; Lopez, J., Wu, Y., Eds.; Springer International Publishing: Cham, Switzerland, 2015; pp. 121–135. 28. Alazzam, H.; Alsmady, A.; Shorman, A.A. Supervised Detection of IoT Botnet Attacks. In Proceedings of the Second International Conference on Data Science, E-Learning and Information Systems, DATA ’19, Dubai, United Arab Emirates, 2–5 December 2019; Association for Computing Machinery: New York, NY, USA, 2019; [CrossRef] p g y 29. Falco, G.; Li, C.; Fedorov, P.; Caldera, C.; Arora, R.; Jackson, K. NeuroMesh: IoT security enabled by a blockchain powered botnet vaccine. ACM Int. Conf. Proc. Ser. 2019, 148162, 1–6. [CrossRef] 30. Hoque, N.; Bhattacharyya, D.K.; Kalita, J.K. Botnet in DDoS Attacks: Trends and Challenges. IEEE Commun. Surv. Tutor. 2015, 17, 2242–2270. [CrossRef] 31. Dange, S.; Chatterjee, M. IoT Botnet: The Largest Threat to the IoT Network. Adv. Intell. Syst. Comput. 2020, 1049, 137–157. [CrossRef] 32. Wazzan, M.; Algazzawi, D.; Bamasaq, O.; Albeshri, A.; Cheng, L. Internet of Things Botnet Detection Approaches: Analysis and Recommendations for Future Research. Appl. Sci. 2021, 11, 5713. [CrossRef] 33. Raghava, N.S.; Sahgal, D.; Chandna, S. Classiﬁcation of Botnet Detection Based on Botnet Architechture. In Proceedings of the 2012 International Conference on Communication Systems and Network Technologies, Bangalore, India, 3–7 January 2012; pp. 569–572. [CrossRef] n, C. The Research on Approaches for Botnet Detection. Energy Procedia 2011, 13, 9726–9732. [CrossRef] pp 34. Zhang, W.; Jin, C. The Research on Approaches for Botnet Detection. Energy Procedia 2011, 13, 9726 35. Soe, Y.N.; Feng, Y.; Santosa, P.I.; Hartanto, R.; Sakurai, K. Machine learning-based IoT-botnet attack detection with sequential architecture. Sensors 2020, 20, 4372. [CrossRef] [PubMed] 36. Abraham, B.; Mandya, A.; Bapat, R.; Alali, F.; Brown, D.E.; Veeraraghavan, M. A Comparison of Machine Learning Approaches to Detect Botnet Trafﬁc. In Proceedings of the 2018 International Joint Conference on Neural Networks (IJCNN), Rio de janeiro, Brazil, 8–13 July 2018; pp. 1–8. [CrossRef] y pp 37. Petersen, K.; Vakkalanka, S.; Kuzniarz, L. Guidelines for conducting systematic mapping studies in software engineering: An update. Inf. Softw. Technol. 2015, 64, 1–18. [CrossRef] p f f 38. Wohlin, C. Guidelines for Snowballing in Systematic Literature Studies and a Replication in Software Engineering; EASE ’14.; Association for Computing Machinery: New York, NY, USA, 2014. [CrossRef] 39. Petticrew, M.; Roberts, H. References Systematic Reviews in the Social Sciences: A Practical Guide; John Wiley & Sons: Hoboken, NJ, USA, 2008. [CrossRef] 41. Kolias, C.; Kambourakis, G.; Stavrou, A.; Voas, J. DDoS in the IoT: Mirai and other botnets. Computer 2017, 50, 80–84. [CrossRef] 42 N i J P liti ll ti t d d i l f i tt k C t l I f S S 2009 3 163 181 [C R f] 41. Kolias, C.; Kambourakis, G.; Stavrou, A.; Voas, J. DDoS in the IoT: Mirai and other botnets. Computer 2017, 50, 80–84. [CrossRef] 42. Nazario, J. Politically motivated denial of service attacks. Cryptol. Inf. Secur. Ser. 2009, 3, 163–181. [CrossRef] 41. Kolias, C.; Kambourakis, G.; Stavrou, A.; Voas, J. DDoS in the IoT: Mirai and other botnets. Comput 42. Nazario, J. Politically motivated denial of service attacks. Cryptol. Inf. Secur. Ser. 2009, 3, 163–181. 43. Sgouras, K.I.; Kyriakidis, A.N.; Labridis, D.P. Short-term risk assessment of botnet attacks on advanced metering infrastructure. IET Cyber-Phys. Syst. Theory Appl. 2017, 2, 143–151. [CrossRef] 44. Li, Z.; Liao, Q.; Blaich, A.; Striegel, A. Fighting botnets with economic uncertainty. Secur. Commun. Netw. 2011, 4, 1104–1113. [CrossRef] 45. Salamatian, S.; Huleihel, W.; Beirami, A.; Cohen, A.; Médard, M. Why Botnets Work: Distributed Brut Synchronization. IEEE Trans. Inf. Forensics Secur. 2019, 14, 2288–2299. [CrossRef] y f [ ] 46. Dev, J.A. Usage of botnets for high speed MD5 hash cracking. In Proceedings of the 2013 3rd International Conference on Innovative Computing Technology, Intech 2013, London, UK, 29–31 August 2013; p. 6653658. [CrossRef] p g gy g p 47. Bederna, Z.; Szadeczky, T. Cyber espionage through Botnets. Secur. J. 2020, 33, 43–62. [CrossRef] 48. Herwig, S.; Harvey, K.; Hughey, G.; Roberts, R.; Levin, D. Measurement and Analysis of Hajime, a Peer-to-peer IoT Botnet. In Proceedings of the Network and Distributed Systems Security (NDSS) Symposium, San Diego, CA, USA, 24–27 February 2019; p. 15. [CrossRef] 49. Xiang, C.; Lihua, Y.; Shuyuan, J.; Zhiyu, H.; Shuhao, L. Botnet spooﬁng: Fighting botnet with itself. Secur. Commun. Netw. 2015, 8, 80–89. [CrossRef] 0. Osagie, M.S.U.; Enagbonma, O.; Inyang, A.I. The H 50. Osagie, M.S.U.; Enagbonma, O.; Inyang, A.I. The Historical Perspective of Botnet Tools. arXiv 2019, arXiv:1904.00948. [CrossRef]. 51. Goodin, D. Record-Breaking DDoS Reportedly Delivered by >145 k hacked Cameras. Available online: arstechnica.com/ information-technology/2016/09/botnet-of-145k-cameras-reportedly-deliver-internets-biggest-ddos-ever/ (accessed on 30 July 2021). y 52. Fruhlinger, J. References The Mirai Botnet Explained: How Teen Scammers and CCTV Cameras almost Brought down the Internet. Available online: csoonline.com/article/3258748/the-mirai-botnet-explained-how-teen-scammers-and-cctv-cameras-almost-brought- down-the-internet.html (accessed on 30 July 2021). 36 of 43 Future Internet 2021, 13, 198 53. Ofﬁce of Public Affairs (USA Department of Justice)—“U.S. Leads Multi-National Action against Gameover Zeus Botnet and Cryptolocker Ransomware, Charges Botnet Administrator”. Available online: https://www.justice.gov/opa/pr/us-leads-multi- national-action-against-gameover-zeus-botnet-and-cryptolocker-ransomware (accessed on 30 July 2021). g g yp y 54. Dabrowski, A.; Ullrich, J.; Weippl, E.R. Botnets causing blackouts: How coordinated load attacks can d Elektrotechnik Und Informationstechnik 2018, 135, 250–255. [CrossRef] 54. Dabrowski, A.; Ullrich, J.; Weippl, E.R. Botnets causing blackouts: How coordinated load attacks can destabilize the power grid. Elektrotechnik Und Informationstechnik 2018, 135, 250–255. [CrossRef] 55. Zou, C.C.; Cunningham, R. Honeypot-aware advanced botnet construction and maintenance. Proc. Int. Conf. Dependable Syst. Netw 2006 2006 1633509 [CrossRef] f 55. Zou, C.C.; Cunningham, R. Honeypot-aware advanced botnet construction and maintenance. Proc. I Netw. 2006, 2006, 1633509. [CrossRef] 56. Zeng, J.; Tang, W.; Liu, C.; Hu, J.; Peng, L. Efﬁcient detect scheme of botnet command and control communication. Commun. Comput. Inf. Sci. 2012, 307, 576–581. [CrossRef] 57. Abu Rajab, M.; Zarfoss, J.; Monrose, F.; Terzis, A. A multifaceted approach to understanding the botnet phenomenon. In Proceed- ings of the 6th ACM SIGCOMM Conference on Internet Measurement, Rio de Janeriro, Brazil, 25–27 October 2006; pp. 41–52. [CrossRef] 58. Heron, S. Working the botnet: How dynamic DNS is revitalising the zombie army. Netw. Secur. 2007, 2007, 9–11. [CrossRef] 59. Liu, C.; Lu, W.; Zhang, Z.; Liao, P.; Cui, X. A recoverable hybrid C C botnet. In Proceedings of the 2011 6th International 58. Heron, S. Working the botnet: How dynamic DNS is revitalising the zombie army. Netw. Secur. 2007, 2007, 9–11. [CrossRef] 59. Liu, C.; Lu, W.; Zhang, Z.; Liao, P.; Cui, X. A recoverable hybrid C C botnet. In Proceedings of the 2011 6th International Conference on Malicious and Unwanted Software, Fajardo, PR, USA, 18–19 October 2011; pp. 110–118. [CrossRef] g y g Conference on Malicious and Unwanted Software, Fajardo, PR, USA, 18–19 October 2011; pp. 110–118. [CrossRef] 60. Yin, T.; Zhang, Y.; Li, S. DR-SNBot: A Social Network-Based Botnet with Strong Destroy-Resistance. In Proceedings of the 2014 9th IEEE International Conference on Networking, Architecture, and Storage, Tianjin, China, 6–8 August 2014; pp. 191–199. [CrossRef] 61. Sood, A.K.; Zeadally, S.; Enbody, R.J. An Empirical Study of HTTP-based Financial Botnets. IEEE Trans. Dependable Secur. References Comput. 2016, 13, 6991594. [CrossRef] 62. Wang, H.; Gong, Z. Collaboration-based botnet detection architecture. In Proceedings of the 2009 2nd International Conference on Intelligent Computing Technology and Automation, ICICTA 2009, Changsha, China, 10–11 October 2009; Volume 2, p. 5287910. [CrossRef] 63. Ogu, E.C.; Ojesanmi, O.A.; Awodele, O.; Kuyoro, S. A botnets circumspection: The current threat landscape, and what we know so far. Information 2019, 10, 337. [CrossRef] 64. Chen, Z.; Chen, C.; Wang, Q. Delay-Tolerant botnets. In Proceedings of the International Conference on Computer Communica- tions and Networks, ICCCN, San Francisco, CA, USA, 3–6 August 2009; p. 5235321. [CrossRef] 65. Anagnostopoulos, M.; Kambourakis, G.; Gritzalis, S. New facets of mobile botnet: Architecture and ev 2016, 15, 455–473. [CrossRef] oulos, M.; Kambourakis, G.; Gritzalis, S. New facets of mobile botnet: Architecture and evaluation. Int. J. In –473. [CrossRef] 66. Hamon, V. Android botnets for multi-targeted attacks. J. Comput. Virol. Hacking Tech. 2015, 11, 193–202. [CrossRef] 6. Hamon, V. Android botnets for multi-targeted att 67. Mulliner, C.; Seifert, J.P. Rise of the iBots: Owning a telco network. In Proceedings of the 5th IEEE International Conference on Malicious and Unwanted Software, Malware 2010, Nancy, France, 19–20 October 2010; p. 5665790. [CrossRef] net, E.; Beslay, L. Mobile Botnets Taxonomy and Challenges. In Proceedings of the 2015 European Intelligence atics Conference, EISIC 2015, Manchester, UK, 7–9 September 2015; p. 7379739. [CrossRef] 68. Malatras, A.; Freyssinet, E.; Beslay, L. Mobile Botnets Taxonomy and Challenges. In Proceedings of the 20 and Security Informatics Conference, EISIC 2015, Manchester, UK, 7–9 September 2015; p. 7379739. [C 69. Rodriguez-Gomez, R.A.; Macia-Fernandez, G.; Garcia-Teodoro, P. Survey and taxonomy of botnet research through life-cycle. ACM Comput. Surv. 2013, 45, 2501659. [CrossRef] 70. Pieterse, H.; Olivier, M.S. Android botnets on the rise: Trends and characteristics. In Proceedings of the 2012 Information Security for South Africa—Proceedings of the ISSA 2012 Conference, Johannesburg, South Africa, 15–17 August 2012; p. 6320432. [CrossRef] [ ] 71. Chang, W.; Wang, A.; Mohaisen, A.; Chen, S. Characterizing botnets-as-a-service. In Proceedings of the Sigcomm 2014 ACM Conference on Special Interest Group on Data Communication, Chicago, IL, USA, 17–22 August 2014; Volume 44, pp. 585–586. [CrossRef] 72. Li, H.; Hu, G.; Yang, Y. Research on P2P botnet network behaviors and modeling. Commun. Comput. [CrossRef] 73. Aanjankumar, S.; Poonkuntran, S. An efﬁcient soft computing approach for securing information over GAMEOVER Zeus Botnets with modiﬁed CPA algorithm. Soft Comput. 2020, 24, 16499–16507. References [CrossRef] g f p 74. Yan, G.; Ha, D.T.; Eidenbenz, S. AntBot: Anti-pollution peer-to-peer botnets. Comput. Netw. 2011, 55, 1941–1956. [CrossRef] 74. Yan, G.; Ha, D.T.; Eidenbenz, S. AntBot: Anti-pollution peer-to-peer botnets. Comput. Netw. 2011, 55, 1941–1956. [CrossRef] 75. Andriesse, D.; Rossow, C.; Stone-Gross, B.; Plohmann, D.; Bos, H. Highly resilient peer-to-peer botnets are here: An analysis of Gameover Zeus. In Proceedings of the 2013 8th International Conference on Malicious and Unwanted Software: “The Americas”, Malware 2013, Fajardo, PR, USA, 22–24 October 2013; p. 6703693. [CrossRef] 76. Zhuang, D.; Morris Chang, J. Enhanced PeerHunter: Detecting Peer-To-Peer Botnets Through Network-Flow Level Community Behavior Analysis. IEEE Trans. Inf. Forensics Secur. 2019, 14, 8536452. [CrossRef] y f 77. Rossow, C.; Andriesse, D.; Werner, T.; Stone-Gross, B.; Plohmann, D.; Dietrich, C.J.; Bos, H. SoK: P2PWNED—Modeling and evaluating the resilience of peer-to-peer botnets. In Proceedings of the IEEE Symposium on Security and Privacy, Berkeley, CA, USA, 19–22 May 2013; p. 6547104. [CrossRef] y p 78. Wang, T.; Wang, H.; Liu, B.; Shi, P. What is the pattern of a botnet? In Proceedings of the 12th IEEE International Conference on Trust, Security and Privacy in Computing and Communications, Trustcom 2013, Melbourne, Australia, 16–18 July 2013; p. 6680849. [CrossRef] 37 of 43 37 of 43 Future Internet 2021, 13, 198 79. Perrotta, R.; Hao, F. Botnet in the browser: Understanding threats caused by malicious browser extensions. IEEE Secur. Priv. 2018, 16, 8425617. [CrossRef] 80. Boshmaf, Y.; Muslukhov, I.; Beznosov, K.; Ripeanu, M. Design and analysis of a social botnet. Comput. Netw. 2013, 57, 556–578. [CrossRef] 81. Garip, M.T.; Reiher, P.; Gerla, M. Ghost: Concealing vehicular botnet communication in the VANET control channel. In Proceedings of the 2016 International Wireless Communications and Mobile Computing Conference, IWCMC 2016, Paphos, Cyprus, 5–9 September 2016; p. 7577024. [CrossRef] yp p p 82. Yamaguchi, S. Botnet defense system: Concept, design, and basic strategy. Information 2020, 11, 516. [CrossRef] 83 B i E I l N B d I f Thi S i C 2017 50 7842850 [C R f] 82. Yamaguchi, S. Botnet defense system: Concept, design, and basic strategy. Information 2020, 11, 516. [C 83. Bertino, E.; Islam, N. Botnets and Internet of Things Security. Computer 2017, 50, 7842850. [CrossRef] Islam, N. Botnets and Internet of Things Security. Com 84. Mendes, L.D.; Aloi, J.; Pimenta, T.C. Analysis of IoT botnet architectures and recent defense proposals. Proc. Int. Conf. Microelectron. References A novel graph-based approach for IoT botnet detection. Int. J. Inf. Secur. 2020, 19, 567–577. [CrossRef] 94. Pei, Z.; Gan, G. Research on p2p botnet trafﬁc identiﬁcation technology based on neural network. IOP Conf. Ser. Earth Environ. Sci. 2020, 428, 012011. [CrossRef] 95. Taheri, S.; Salem, M.; Yuan, J.S. Leveraging image representation of network trafﬁc data and transfer learning in botnet detection. Big Data Cogn. Comput. 2018, 2, 37. [CrossRef] [ ] Sun, M.; Zhou, G. IoT botnet detection via power consumption modeling. Smart Health 2020, 15, 100103 96. Jung, W.; Zhao, H.; Sun, M.; Zhou, G. IoT botnet detection via power consumption modeling. Smar [CrossRef] 97. Kim, J.; Shim, M.; Hong, S.; Shin, Y.; Choi, E. Intelligent detection of iot botnets using machine learning and deep learning. Appl. Sci. 2020, 10, 7009. [CrossRef] [ ] 98. Nogueira, A.; Salvador, P.; Blessa, F. A botnet detection system based on neural networks. In Proceedings of the 5th International Conference on Digital Telecommunications, ICDT 2010, Athens, Greece, 13–19 June 2010; p. 5532380. [CrossRef] 98. Nogueira, A.; Salvador, P.; Blessa, F. A botnet detection system based on neural networks. In Proceedings of the 5th International Conference on Digital Telecommunications, ICDT 2010, Athens, Greece, 13–19 June 2010; p. 5532380. [CrossRef] 99. Javed, Y.; Rajabi, N. Multi-Layer Perceptron Artiﬁcial Neural Network Based IoT Botnet Trafﬁc Classiﬁcation. Adv. Intell. Syst. g , ; , ; , y g Conference on Digital Telecommunications, ICDT 2010, Athens, Greece, 13–19 June 2010; p. 5532380. [CrossRef] 99. Javed, Y.; Rajabi, N. Multi-Layer Perceptron Artiﬁcial Neural Network Based IoT Botnet Trafﬁc Classiﬁcation. Adv. Intell. Syst. Comput. 2020, 1069, 973–984. [CrossRef] 99. Javed, Y.; Rajabi, N. Multi-Layer Perceptron Artiﬁcial Neural Network Based IoT Botnet Trafﬁc Classiﬁcation. Adv. Intell. Syst. Comput. 2020, 1069, 973–984. [CrossRef] p 100. Zeidanloo, H.R.; Hosseinpour, F.; Borazjani, P.N. Botnet detection based on common network behaviors by utilizing Artiﬁcial Immune System(AIS). In Proceedings of the ICSTE 2010—2010 2nd International Conference on Software Technology and Engineering, San Juan, PR, USA, 3–5 October 2010; Volume 1, p. 5608967. [CrossRef] 101. Sriram, S.; Vinayakumar, R.; Alazab, M.; Soman, K.P. Network ﬂow based IoT botnet attack detection using deep learning. In Proceedings of the IEEE Infocom 2020—IEEE Conference on Computer Communications Workshops, Infocom WKSHPS 2020, Toronto, ON, Canada, 6–9 July 2020; p. 9162668. [CrossRef] y p 102. Jithu, P.; Shareena, J.; Ramdas, A.; Haripriya, A.P. References ICM 2019, 2019, 9021715. [CrossRef] 85. Kudo, T.; Kimura, T.; Inoue, Y.; Aman, H.; Hirata, K. Behavior analysis of self-evolving botnets. In Proceedings of the IEEE CITS 2016—2016 International Conference on Computer, Information and Telecommunication Systems, Kunming, China, 6–8 July 2016; p. 7546428. [CrossRef] 86. Bock, L.; Alexopoulos, N.; Saracoglu, E.; Muhlhauser, M.; Vasilomanolakis, E. Assessing the Threat of Blockchain-based Botnets. Ecrime Res. Summit Ecrime 2019, 2019, 9037600. [CrossRef] 87. Yin, J.; Lv, H.; Zhang, F.; Tian, Z.; Cui, X. Study on advanced botnet based on publicly available resources. In Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artiﬁcial Intelligence and Lecture Notes in Bioinformatics); Springer: Cham, Switzerland, 2018; Volume 11149, pp. 57–74. [CrossRef] pp 88. Hua, J.; Sakurai, K. A SMS-Based Mobile Botnet Using Flooding Algorithm. In Information Security Theory and Practice. Security and Privacy of Mobile Devices in Wireless Communication; Ardagna, C.A., Zhou, J., Eds.; Springer: Berlin/Heidelberg, Germany, 2011; pp. 264–279. pp 89. Geng, G.; Xu, G.; Zhang, M.; Yang, Y.; Yang, G. An improved SMS based heterogeneous mobile botnet model. In Proceedings of the 2011 IEEE International Conference on Information and Automation, ICIA 2011, Shenzhen, China, 6–8 June 2011; p. 5948987. [CrossRef] 90. Lee, H.; Kang, T.; Lee, S.; Kim, J.; Kim, Y. Punobot: Mobile Botnet Using Push Notiﬁcation Service in Android. In Information Security Applications; Kim, Y., Lee, H., Perrig, A., Eds.; Springer International Publishing: Cham, Switzerland, 2014; pp. 124–137. 91. Porras, P.; Saidi, H.; Yegneswaran, V. An Analysis of the iKee.B iPhone Botnet. In Security and Privacy in Mobile Information and Communication; Springer: Berlin/Heidelberg, Germany, 2010; pp. 141–152. 90. Lee, H.; Kang, T.; Lee, S.; Kim, J.; Kim, Y. Punobot: Mobile Botnet Using Push Notiﬁcation Service in Android. In Information Security Applications; Kim, Y., Lee, H., Perrig, A., Eds.; Springer International Publishing: Cham, Switzerland, 2014; pp. 124–137. 91 P P S idi H Y V A A l i f th iK B iPh B t t I S it d P i i M bil I f ti d 91. Porras, P.; Saidi, H.; Yegneswaran, V. An Analysis of the iKee.B iPhone Botnet. In Security and Priv Communication; Springer: Berlin/Heidelberg, Germany, 2010; pp. 141–152. p g g y pp 92. Li, X.G.; Wang, J.F. Trafﬁc detection of transmission of botnet threat using BP neural network. Neural Netw. World 2018, 28, 511–522. [CrossRef] 93. Nguyen, H.T.; Ngo, Q.D.; Le, V.H. References Intrusion Detection System for IOT Botnet Attacks Using Deep Learning. SN Comput. Sci. 2021, 2, 1–8. [CrossRef] p 103. Alharbi, A.; Alosaimi, W.; Alyami, H.; Rauf, H.T.; Damasevicius, R. Botnet Attack Detection Using Local Global Best Bat Algorithm for Industrial Internet of Things. Electronics 2021, 10, 1341. [CrossRef] 104. Kanehara, H.; Takahashi, T.; Murakami, Y.; Inoue, D.; Shimamura, J.; Murata, N. Real-time botnet detection using nonnegative tucker decomposition. Proc. ACM Symp. Appl. Comput. 2019, 147772, 1337–1344. [CrossRef] 105. Bansal, A.; Mahapatra, S. A Comparative Analysis of Machine Learning Techniques for Botnet Detection. In Proceedings of the 10th International Conference on Security of Information and Networks, Jaipur, India, 13–15 October 2017; pp. 91–100. [CrossRef] Future Internet 2021, 13, 198 38 of 43 38 of 43 106. Eustis, A.G. The Mirai Botnet and the Importance of IoT Device Security. In Proceedings of the 16th International Conference on Information Technology-New Generations (ITNG 2019), Las Vegas, NV, USA, 1–3 April 2019; Latiﬁ, S., Ed.; Springer International Publishing: Cham, Switzerland, 2019; pp. 85–89. g pp 107. Ribeiro, G.H.; De Faria Paiva, E.R.; Miani, R.S. A comparison of stream mining algorithms on botnet detection. In Proceedings of the 15th International Conference on Availability, Reliability and Security, Dublin, Ireland, 25–28 August 2020; pp. 1–10. [CrossRef] [ ] 108. Chu, Z.; Han, Y.; Zhao, K. Botnet Vulnerability Intelligence Clustering Classiﬁcation Mining and Countermeasure Algorithm Based on Machine Learning. IEEE Access 2019, 7, 8935236. [CrossRef] 109. Tuan, T.A.; Long, H.V.; Son, L.H.; Kumar, R.; Priyadarshini, I.; Son, N.T.K. Performance evaluation of Botnet DDoS attack detection using machine learning. Evol. Intell. 2020, 13, 283–294. [CrossRef] 110. Indre, I.; Lemnaru, C. Detection and prevention system against cyber attacks and botnet malware for information systems and Internet of Things. In Proceedings of the 2016 IEEE 12th International Conference on Intelligent Computer Communication and Processing, ICCP 2016, Cluj-Napoca, Romania, 8–10 September 2016; p. 7737142. [CrossRef] g j p p p 111. Park, Y.; Kengalahalli, N.V.; Chang, S.Y. Distributed Security Network Functions against Botnet Attacks in Software-deﬁned Networks. In Proceedings of the 2018 IEEE Conference on Network Function Virtualization and Software Deﬁned Networks, NFV-SDN 2018, Verona, Italy, 27–29 November 2018; p. 8725657. [CrossRef] 112. Lu, W.; Tavallaee, M.; Ghorbani, A.A. Automatic Discovery of Botnet Communities on Large-Scale Communication Networks. References AsSadhan, B.; Bashaiwth, A.; Al-Muhtadi, J.; Alshebeili, S. Analysis of P2P, IRC and HTTP trafﬁc for botnets detection. Peer-to-Peer Netw. Appl. 2018, 11, 848–861. [CrossRef] g p g y p g p 117. AsSadhan, B.; Bashaiwth, A.; Al-Muhtadi, J.; Alshebeili, S. Analysis of P2P, IRC and HTTP trafﬁc for botnets detection. Peer-to-Peer Netw. Appl. 2018, 11, 848–861. [CrossRef] 118. Zand, A.; Vigna, G.; Yan, X.; Kruegel, C. Extracting probable command and control signatures for detecting botnets. In Proceedings of the 29th Annual ACM Symposium on Applied Computing, Gyeongju, Korea, 24–28 March; pp. 1657–1662. [CrossRef] 119. Khattak, S.; Ahmed, Z.; Syed, A.A.; Khayam, S.A. BotFlex: A community-driven tool for botnet detection. J. Netw. Comput. Appl. 2015, 58, 144–154. [CrossRef] 120. Richer, T.J. Entropy-based detection of botnet command and control. In Proceedings of the Australasian Computer Science Week Multiconference, Geelong, Australia, 31 January–3 February 2017; p. a75. [CrossRef] 121. AsSadhan, B.; Moura, J.M. An efﬁcient method to detect periodic behavior in botnet trafﬁc by analyzing control plane trafﬁc. J. Adv. Res. 2014, 5, 435–448. [CrossRef] 122. Lagraa, S.; François, J.; Lahmadi, A.; Miner, M.; Hammerschmidt, C.; State, R. BotGM: Unsupervised graph mining to detect botnets in trafﬁc ﬂows. In Proceedings of the 2017 1st Cyber Security in Networking Conference, CSNET 2017, Rio de Janeiro, Brazil, 18–20 October 2017; pp. 1–8. [CrossRef] pp 123. Sousa, R.; Rodrigues, N.; Salvador, P.; Nogueira, A. Analyzing the Behavior of Top Spam Botnets. In Proceedings of the 2012 IEEE International Conference on Communications (ICC), Ottawa, ON, Canada, 10–15 June 2012; pp. 6540–6544. [CrossRef] 123. Sousa, R.; Rodrigues, N.; Salvador, P.; Nogueira, A. Analyzing the Behavior of Top Spam Botnets. In Proceedings of the 2012 IEEE International Conference on Communications (ICC), Ottawa, ON, Canada, 10–15 June 2012; pp. 6540–6544. [CrossRef] 124. Kozik, R.; Chora´s, M. Pattern Extraction Algorithm for NetFlow-Based Botnet Activities Detection. Secur. Commun. Netw. 2017, 2017, 6047053. [CrossRef] 124. Kozik, R.; Chora´s, M. Pattern Extraction Algorithm for NetFlow-Based Botnet Activities Detection. Secur. Commun. Netw. 2017, 2017, 6047053. [CrossRef] 125. Chen, R.; Niu, W.; Zhang, X.; Zhuo, Z.; Lv, F. An Effective Conversation-Based Botnet Detection Metho 2017, 4934082. [CrossRef] 126. Pandey, A.; Thaseen, S.; Aswani Kumar, C.; Li, G. Identiﬁcation of botnet attacks using hybrid machine learning models. Adv. Intell. Syst. Comput. 2021, 1179, 249–257. [CrossRef] y p [ ] 127. Kirubavathi, G.; Anitha, R. Botnet detection via mining of trafﬁc ﬂow characteristics. Comput. Electr. Eng. References In Proceedings of the 4th International Symposium on Information, Computer, and Communications Security, ASIACCS ’09, Auckland, New Zealand, 9–12 July 2009; Association for Computing Machinery: New York, NY, USA, 2009; pp. 1–10. [CrossRef] In Proceedings of the 4th International Symposium on Information, Computer, and Communications Security, ASIACCS 09, Auckland, New Zealand, 9–12 July 2009; Association for Computing Machinery: New York, NY, USA, 2009; pp. 1–10. [CrossRef] 113. Goyal, M.; Sahoo, I.; Geethakumari, G. HTTP Botnet Detection in IOT Devices using Network Trafﬁc Analysis. In Proceedings of the 2019 International Conference on Recent Advances in Energy-Efﬁcient Computing and Communication, ICRAECC 2019, Nagercoil, India, 7–20 March 2019; p. 8995160. [CrossRef] y p g y pp 113. Goyal, M.; Sahoo, I.; Geethakumari, G. HTTP Botnet Detection in IOT Devices using Network Trafﬁc Analysis. In Proceedings of the 2019 International Conference on Recent Advances in Energy-Efﬁcient Computing and Communication, ICRAECC 2019, Nagercoil, India, 7–20 March 2019; p. 8995160. [CrossRef] 114. Heydari, B.; Yajam, H.; Akhaee, M.A.; Salehkalaibar, S. Utilizing Features of Aggregated Flows to Identify Botnet Network Trafﬁc. In Proceedings of the 2017 14th International ISC (Iranian Society of Cryptology) Conference on Information Security and Cryptology, ISCISC 2017, Shiraz, Iran, 6–7 September 2017; p. 8488370. [CrossRef] 115. Haddadi, F.; Morgan, J.; Filho, E.G.; Zincir-Heywood, A.N. Botnet behaviour analysis using IP ﬂows: With http ﬁlters using classiﬁers. In Proceedings of the 2014 IEEE 28th International Conference on Advanced Information Networking and Applications Workshops, IEEE Waina 2014, Victoria, BC, Canada, 13–16 May 2014; p. 6844605. [CrossRef] 116. Yong, W.; Tefera, S.H.; Beshah, Y.K. Understanding botnet: From mathematical modelling to integrated detection and mitiga- tion framework. In Proceedings of the 13th ACIS International Conference on Software Engineering, Artiﬁcial Intelligence, Networking, and Parallel/distributed Computing, SNPD 2012, Kyoto, Japan, 8–10 August 2012; p. 6299259. [CrossRef] 6 o g, ; e e , S ; es , U e s g o e o e c o e g o eg e e ec o g tion framework. In Proceedings of the 13th ACIS International Conference on Software Engineering, Artiﬁcial Intelligence, Networking, and Parallel/distributed Computing, SNPD 2012, Kyoto, Japan, 8–10 August 2012; p. 6299259. [CrossRef] 117. AsSadhan, B.; Bashaiwth, A.; Al-Muhtadi, J.; Alshebeili, S. Analysis of P2P, IRC and HTTP trafﬁc for botnets detection. Peer-to-Peer Netw. Appl. 2018, 11, 848–861. [CrossRef] Networking, and Parallel/distributed Computing, SNPD 2012, Kyoto, Japan, 8–10 August 2012; p. 6299259. [CrossRef] 117. References 2016, 50, 91–101. [CrossRef] 128. Li, S.H.; Kao, Y.C.; Zhang, Z.C.; Chuang, Y.P.; Yen, D.C. A network behavior-based botnet detection mechanism using PSO and K-means. ACM Trans. Manag. Inf. Syst. 2015, 6, 3. [CrossRef] g f y 129. Su, S.C.; Chen, Y.R.; Tsai, S.C.; Lin, Y.B. Detecting P2P Botnet in Software Deﬁned Networks. Secur. Commun. Netw. 2018, 2018, 4723862. [CrossRef] Future Internet 2021, 13, 198 39 of 43 39 of 43 130. Vishwakarma, R.; Jain, A.K. A Honeypot with Machine Learning based Detection Framework for defending IoT based Botnet DDoS Attacks. In Proceedings of the 2019 3rd International Conference on Trends in Electronics and Informatics (ICOEI), Tirunelveli, India, 23–25 April 2019; pp. 1019–1024. [CrossRef] p pp 131. Al-Hakbani, M.M.; Dahshan, M.H. Avoiding honeypot detection in peer-to-peer botnets. In Proceedings of the ICETECH 2015—2015 IEEE International Conference on Engineering and Technology, Coimbatore, India, 20 March 2015; p. 7275017. [CrossRef] [ ] 132. Dwyer, O.P.; Marnerides, A.K.; Giotsas, V.; Mursch, T. Proﬁling iot-based botnet trafﬁc using DNS. In Proceedings of the 2019 IEEE Global Communications Conference, Globecom 2019, Waikoloa, HI, USA, 9–13 December 2019; p. 9014300. [CrossRef] 133. Wang, K.; Huang, C.Y.; Tsai, L.Y.; Lin, Y.D. Behavior-based botnet detection in parallel. Secur. Commun. Netw. 2014, 7, 1849–1859. [CrossRef] [ ] 132. Dwyer, O.P.; Marnerides, A.K.; Giotsas, V.; Mursch, T. Proﬁling iot-based botnet trafﬁc using DNS. In Proceedings of the 2019 IEEE Global Communications Conference, Globecom 2019, Waikoloa, HI, USA, 9–13 December 2019; p. 9014300. [CrossRef] , K.; Huang, C.Y.; Tsai, L.Y.; Lin, Y.D. Behavior-based botnet detection in parallel. Secur. Commun. Netw. 201 sRef] 133. Wang, K.; Huang, C.Y.; Tsai, L.Y.; Lin, Y.D. Behavior-based botnet detection in parallel. Secur. Commun. Netw. 2014, 7, 1849–1859. [CrossRef] 134. Bahsi, H.; Nomm, S.; La Torre, F.B. Dimensionality Reduction for Machine Learning Based IoT Botnet Detection. In Proceedings of the 2018 15th International Conference on Control, Automation, Robotics and Vision (ICARCV), Singapore, 18–21 November 2018; pp. 1857–1862. [CrossRef] pp 135. Beigi, E.B.; Jazi, H.H.; Stakhanova, N.; Ghorbani, A.A. Towards effective feature selection in machine learning-based botnet detection approaches. In Proceedings of the 2014 IEEE Conference on Communications and Network Security, CNS 2014, San Francisco, CA, USA, 29–31 October 2014; p. 6997492. [CrossRef] p 136. Lysenko, S.; Savenko, O.; Bobrovnikova, K.; Kryshchuk, A. Self-adaptive system for the corporate area network resilience in the presence of botnet cyberattacks. Commun. Comput. Inf. Sci. 2018, 860, 385–401. [CrossRef] 137. References Hung, C.L.; Wang, H.H. Parallel botnet detection system by using GPU. In Proceedings of the 2014 IEEE/ACIS 13th International Conference on Computer and Information Science, ICIS 2014, Taiyuan, China, 4–6 June 2014; p. 6912109. [CrossRef] 138. Blaise, A.; Bouet, M.; Conan, V.; Secci, S. Botnet Fingerprinting: A Frequency Distributions Scheme for L IEEE Trans. Netw. Serv. Manag. 2020, 17, 9097931. [CrossRef] g 139. Lin, S.C.; Chen, P.S.; Chang, C.C. A novel method of mining network ﬂow to detect P2P botnets. Peer-to-Peer Netw. Appl. 2014, 7, 645–654. [CrossRef] 140. Lee, Y.C.; Tseng, C.M.; Liu, T.J. A HTTP botnet detection system based on ranking mechanism. In Proceedings of the 2017 12th International Conference on Digital Information Management, ICDIM 2017, Fukuoka, Japan, 12–14 September2017; pp. 115–120. [CrossRef] [ ] 141. Mai, L.; Noh, D.K. Cluster Ensemble with Link-Based Approach for Botnet Detection. J. Netw. Syst. Manag. 2018, 26, 616–639. [CrossRef] 142. Lu, W.; Ghorbani, A.A. Botnets Detection Based on IRC-Community. In Proceedings of the IEEE GLOBECOM 2008—2008 IEEE Global Telecommunications Conference, New Orleans, LA, USA, 30 November–4 December 2008; pp. 1–5. [CrossRef] 143. Hoang, X.D.; Nguyen, Q.C. Botnet detection based on machine learning techniques using DNS query data. Future Internet 2018, 10, 43. [CrossRef] 144. Wang, Z.; Qin, M.; Chen, M.; Jia, C.; Ma, Y. A learning evasive email-based P2P-Like botnet. China Commun. 2018, 15, 15–24. [CrossRef] 146. Lee, S.; Abdullah, A.; Jhanjhi, N.Z.; Kok, S.H. Honeypot Coupled Machine Learning Model for Botnet Detection and Classiﬁcation in IoT Smart Factory—An Investigation. MATEC Web Conf. 2021, 335, 04003. [CrossRef] 147. Ibrahim, W.N.H.; Anuar, S.; Selamat, A.; Krejcar, O.; Gonzalez Crespo, R.; Herrera-Viedma, E.; Fujita, H. Multilayer Framework for Botnet Detection Using Machine Learning Algorithms. IEEE Access 2021, 9, 9359784. [CrossRef] 148. Hao, S.; Liu, D.; Baldi, S.; Yu, W. Unsupervised detection of botnet activities using frequent pattern tree mining. Complex Intell. Syst. 2021, 1–9. [CrossRef] 149. Asadi, M. Detecting IoT botnets based on the combination of cooperative game theory with dee approaches. J. Ambient. Intell. Humaniz. Comput. 2021, 1–15. [CrossRef] approaches. J. Ambient. Intell. Humaniz. Comput. 2021, 1–15. [CrossRef] 150. Bhatt, P.; Thakker, B. A Novel Forecastive Anomaly Based Botnet Revelation Framework for Competing Concerns in Internet of Things J Appl Secur Res 2021 16 258 278 [CrossRef] . A Novel Forecastive Anomaly Based Botnet Revelation Framework for Competing Concerns in Internet of ur. Res. 2021, 16, 258–278. [CrossRef] 150. Bhatt, P.; Thakker, B. References A Novel Forecastive Anomaly Based Botnet Revelation Framework for Competin Things. J. Appl. Secur. Res. 2021, 16, 258–278. [CrossRef] 151. Soleymani, A.; Arabgol, F. A Novel Approach for Detecting DGA-Based Botnets in DNS Queries Using Machine Learning Techniques. J. Comput. Netw. Commun. 2021, 2021, 4767388. [CrossRef] q p 152. Panda, M.; Allah A. Mousa, A.; Ella Hassanien, A. Developing an Efﬁcient Feature Engineering and Machine Learning Model for Detecting IoT-Botnet Cyber Attacks. IEEE Access 2021, 9, 91038–91052. [CrossRef] 153. Kwon, J.; Lee, J.; Lee, H.; Perrig, A. PsyBoG: A scalable botnet detection method for large-scale DNS trafﬁc. Comput. Netw. 2016, 97, 48–73. [CrossRef] 154. Wang, T.S.; Lin, H.T.; Cheng, W.T.; Chen, C.Y. DBod: Clustering and detecting DGA-based botnets using DNS trafﬁc analysis. Comput. Secur. 2017, 64, 1–15. [CrossRef] p 155. Zhao, D.; Traore, I.; Sayed, B.; Lu, W.; Saad, S.; Ghorbani, A.; Garant, D. Botnet detection based on traf ﬂow intervals. Comput. Secur. 2013, 39, 2–16. [CrossRef] Sayed, B.; Lu, W.; Saad, S.; Ghorbani, A.; Garant, D. Botnet detection based on trafﬁc behavior analysis and put. Secur. 2013, 39, 2–16. [CrossRef] 156. Ichise, H.; Jin, Y.; Iida, K. Analysis of via-resolver DNS TXT queries and detection possibility of botnet communications. In Proceedings of the IEEE Paciﬁc Rim Conference on Communications, Computers, and Signal Processing, Victoria, BC, Canada, 24–26 August 2015; p. 7334837. [CrossRef] Future Internet 2021, 13, 198 40 of 43 40 of 43 157. Jin, Y.; Ichise, H.; Iida, K. Design of Detecting Botnet Communication by Monitoring Direct Outbound DNS Queries. In Proceedings of the 2nd IEEE International Conference on Cyber Security and Cloud Computing, Cscloud 2015—IEEE International Symposium of Smart Cloud, IEEE SSC 2015, New York, NY, USA, 3–5 November 2015; p. 7371456. [CrossRef] 158. Nguyen, T.D.; Dung, T.C.; Nguyen, L.G. DGA botnet detection using collaborative ﬁltering and density-based clustering. In Proceedings of the Sixth International Symposium on Information and Communication Technology, Hue, Vietnam, 3–4 December 2015; pp. 203–209. [CrossRef] pp 159. Mohd Safar, N.; Abdullah, N.; Kamaludin, H.; Abd Ishak, S.; Isa, M. Characterising and detection of botnet in P2P network for UDP protocol. Indones. J. Electr. Eng. Comput. Sci. 2020, 18, 1584–1595. [CrossRef] 160. Tsai, M.H.; Chang, K.C.; Lin, C.C.; Mao, C.H.; Lee, H.M. C&C tracer: Botnet command and control behavior tracing. In Proceedings of the IEEE International Conference on Systems, Man and Cybernetics, Anchorage, AK, USA, 9–12 October 2011; p. 6083942. [CrossRef] p 161. References Lysenko, S.; Bobrovnikova, K.; Savenko, O.; Kryshchuk, A. BotGRABBER: SVM-Based Self-Adaptive System for the Network Resilience against the Botnets Cyberattacks. Commun. Comput. Inf. Sci. 2019, 1039, 127–143. [CrossRef] 162. Tong, V.; Nguyen, G. A method for detecting DGA botnet based on semantic and cluster analysis. In Proceedings of the Seventh Symposium on Information and Communication Technology, Ho Chi Minh, Vietnam, 8–9 December 2016; pp. 272–277. [CrossRef] 163. Kelley, T.; Furey, E. Getting Prepared for the Next Botnet Attack: Detecting Algorithmically Generated Domains in Botnet Command and Control. In Proceedings of the 2018 29th Irish Signals and Systems Conference (ISSC), Belfast, UK, 21–22 June 2018; pp. 1–6. [CrossRef] pp 164. Vishvakarma, D.K.; Bhatia, A.; Riha, Z. Detection of Algorithmically Generated Domain Names in Botnets. In Advanced Information Networking and Applications; Barolli, L., Takizawa, M., Xhafa, F., Enokido, T., Eds.; Springer International Publishing: Cham, Switzerland, 2020; pp. 1279–1290. pp 165. Truong, D.T.; Cheng, G. Detecting domain-ﬂux botnet based on DNS trafﬁc features in managed network. Secur. Commun. Netw. 2016, 9, 2338–2347. [CrossRef] 166. Natarajan, V.; Sheen, S.; Anitha, R. Multilevel analysis to detect covert social botnet in multimedia social networks. Comput. J. 2015, 58, 679–687. [CrossRef] 167. Alhomoud, A.; Awan, I.; Pagna Disso, J.F.; Younas, M. A next-generation approach to combating botnets. Computer 2013, 46, 6459493. [CrossRef] 168. Sharafaldin, I.; Gharib, A.; Lashkari, A.H.; Ghorbani, A.A. BotViz: A memory forensic-based botnet detection and visualization approach. In Proceedings of the 2017 International Carnahan Conference on Security Technology (ICCST), Madrid, Spain, 23–26 October 2017; pp. 1–8. [CrossRef] pp 169. Virustotal.com. 2020. Available online: Virustotal.com (accessed on 18 December 2020). pp 169. Virustotal.com. 2020. Available online: Virustotal.com (accessed on 18 December 2020). M.; Martinez Perez, G. UMUDGA: A dataset for proﬁling algorithmically generated domain names in botnet f 2020, 30, 105400. [CrossRef] 170. Zago, M.; Gil Perez, M.; Martinez Perez, G. UMUDGA: A dataset for proﬁling algorithmically gener detection. Data Brief 2020, 30, 105400. [CrossRef] 171. k. Idriss, H. Mirai Botnet In Lebanon. In Proceedings of the 2020 8th International Symposium on Digital Forensics and Security (ISDFS), Beirut, Lebanon, 1–2 June 2020; pp. 1–6. [CrossRef] pp 172. Balasubramanian, Y.; Baggam, D.S.; Venkatraman, S.; Ramaswamy, V. Quantum IDS for mitigation of DDoS attacks by mirai botnets. Commun. Comput. Inf. Sci. 2018, 828, 488–501. [CrossRef] 173. Tzagkarakis, C.; Petroulakis, N.; Ioannidis, S. Botnet Attack Detection at the IoT Edge Based on Sparse Representation. References Wang, P.; Wang, F.; Lin, F.; Cao, Z. Identifying Peer-to-Peer Botnets Through Periodicity Behavior Analysis. In Proceedings of the 17th IEEE International Conference on Trust, Security and Privacy in Computing and Communications and 12th IEEE International Conference on Big Data Science and Engineering, Trustcom/bigdatase 2018, New York, NY, USA, 1–3 August 2018; p. 8455919. [CrossRef] p 183. Barthakur, P.; Dahal, M.; Ghose, M.K. A framework for P2P botnet detection using SVM. In Proceedings of the 2012 International Conference on Cyber-enabled Distributed Computing and Knowledge Discovery, Cyberc 2012, Sanya, China, 10–12 October 2012; p. 6384967. [CrossRef] p 184. Chen, Z.; Yu, X.; Zhang, C.; Zhang, J.; Lin, C.; Song, B.; Gao, J.; Hu, X.; Yang, W.; Yan, E. Fast botnet detection from streaming logs using online lanczos method. In Proceedings of the 2017 IEEE International Conference on Big Data (Big Data), Boston, MA, USA, 11–14 December 2017; pp. 1408–1417. [CrossRef] 185. Ersson, J.; Moradian, E. Botnet Detection with Event-Driven Analysis. Procedia Comput. Sci. 2013, 2 185. Ersson, J.; Moradian, E. Botnet Detection with Event-Driven Analysis. Procedia Comput. Sci. 2013, 22, 662–671. [CrossRef] 186. Almutairi, S.; Mahfoudh, S.; Almutairi, S.; Alowibdi, J.S. Hybrid Botnet Detection Based on Host and Network Analysis. J. Comput. Netw. Commun. 2020, 2020, 9024726. [CrossRef] 186. Almutairi, S.; Mahfoudh, S.; Almutairi, S.; Alowibdi, J.S. Hybrid Botnet Detection Based on Host and Network Analysis. J. Comput. Netw. Commun. 2020, 2020, 9024726. [CrossRef] p 187. Lu, W.; Miller, M.; Xue, L. Detecting Command and Control Channel of Botnets in Cloud. In Intelligent, Secure, and Dependable Systems in Distributed and Cloud Environments; Traore, I., Woungang, I., Awad, A., Eds.; Springer International Publishing: Cham, Switzerland, 2017; pp. 55–62. pp 188. Zeng, Y.; Yan, G.; Eidenbenz, S.; Shin, K.G. Measuring the effectiveness of infrastructure-level detection of large-scale botnets. In Proceedings of the 2011 IEEE Nineteenth IEEE International Workshop on Quality of Service, San Jose, CA, USA, 6–7 June 2011; p. 5931312. [CrossRef] p 189. François, J.; Wang, S.; Bronzi, W.; State, R.; Engel, T. BotCloud: Detecting botnets using MapReduce. In Proceedings of the 2011 IEEE International Workshop on Information Forensics and Security, Wifs 2011, Iguacu Falls, Brazil, 29 November–2 December 2011; p. 6123125. [CrossRef] p 190. Saraubon, K.; Limthanmaphon, B. Fast Effective Botnet Spam Detection. In Proceedings of the 2009 Fourth International Conference on Computer Sciences and Convergence Information Technology, Seoul, Korea, 24–26 November 2009; pp. 1066–1070. [CrossRef] 191. Crespo, B.G.; Garwood, A. References In Proceedings of the 2019 Global IoT Summit (GIoTS), Aarhus, Denmark, 17–21 June 2019; pp. 1–6. [CrossRef] 174. Prokoﬁev, A.O.; Smirnova, Y.S.; Surov, V.A. A method to detect Internet of Things botnets. In Proceedings of the 2018 IEEE Conference of Russian Young Researchers in Electrical and Electronic Engineering, Elconrus 2018, Moscow, Russia; St. Petersburg, Russia, 29 January–1 February 2018; pp. 105–108. [CrossRef] 175. Giachoudis, N.; Damiris, G.P.; Theodoridis, G.; Spathoulas, G. Collaborative agent-based detection of DDoS IoT botnets. In Proceedings of the 15th Annual International Conference on Distributed Computing in Sensor Systems, DCOSS 2019, Santorini Island, Greece, 29–31 May 2019; p. 8804480. [CrossRef] , , y ; p [ ] 176. Spathoulas, G.; Giachoudis, N.; Damiris, G.P.; Theodoridis, G. Collaborative blockchain-based detection of distributed denial of service attacks based on internet of things botnets. Future Internet 2019, 11, 226. [CrossRef] g 177. Cui, P.; Guin, U. Countering Botnet of Things using Blockchain-Based Authenticity Framework. In Proceedings of the 2019 IEEE Computer Society Annual Symposium on VLSI (ISVLSI), Miami, FL, USA, 15–17 July 2019; p. 8839425. [CrossRef] 178. Zareh, A.; Shahriari, H.R. BotcoinTrap: Detection of Bitcoin Miner Botnet Using Host Based Approach. In Proceedings of the 2018 15th International ISC (Iranian Society of Cryptology) Conference on Information Security and Cryptology, ISCISC 2018, Tehran, Iran, 28–29 August 2018; p. 8546867. [CrossRef] g p 179. Zhuang, D.; Chang, J.M. PeerHunter: Detecting peer-to-peer botnets through community behavior analysis. In Proceedings of the 2017 IEEE Conference on Dependable and Secure Computing, Taipei, Taiwan, 7–10 August 2017; p. 8073832. [CrossRef] the 2017 IEEE Conference on Dependable and Secure Computing, Taipei, Taiwan, 7–10 August 2017; p. 8073832. [CrossRef] 180. Priyanka; Dave, M. PeerFox: Detecting parasite P2P botnets in their waiting stage. In Proceedings of the 2015 International Conference on Signal Processing, Computing and Control, ISPCC 2015, Solan, India, 24–26 September 2015; p. 7375054. [CrossRef] 181. Obeidat, A.A.; Al-Kofahi, M.M.; Bawaneh, M.J.; Hanandeh, E.S. A novel botnet detection system for P2P networks. J. Comput. Sci. 2017, 13, 329–336. [CrossRef] 180. Priyanka; Dave, M. PeerFox: Detecting parasite P2P botnets in their waiting stage. In Proceedings of the 2015 International Conference on Signal Processing, Computing and Control, ISPCC 2015, Solan, India, 24–26 September 2015; p. 7375054. [CrossRef] 181. Obeidat, A.A.; Al-Kofahi, M.M.; Bawaneh, M.J.; Hanandeh, E.S. A novel botnet detection system for P2P networks. J. Comput. Sci. 2017, 13, 329–336. [CrossRef] 41 of 43 Future Internet 2021, 13, 198 182. References Fighting Botnets with Cyber-Security Analytics: Dealing with Heterogeneous Cyber-Security Information in New Generation SIEMs. In Proceedings of the 2014 Ninth International Conference on Availability, Reliability and Security, Fribourg, Switzerland, 8–12 September 2014; pp. 192–198. [CrossRef] 192. Derhab, A.; Bouras, A.; Muhaya, F.B.; Khan, M.K.; Xiang, Y. Spam Trapping System: Novel security framework to ﬁght against spam botnets. In Proceedings of the 2014 21st International Conference on Telecommunications (ICT), Lisbon, Portugal, 4–7 May 2014; pp. 467–471. [CrossRef] pp 193. Tang, Y.; Cheng, G.; Yu, J.T.; Zhang, B. Catching modern botnets using active integrated evidential reasoning. J. Internet Serv. Appl. 2013, 4, 1–10. [CrossRef] 194. Yusof, M.; Saudi, M.M.; Ridzuan, F. A New Android Botnet Classiﬁcation for GPS Exploitation Based on Permission and API Calls. Lect. Notes Electr. Eng. 2018, 465, 27–37. [CrossRef] g 195. Mongkolluksamee, S.; Visoottiviseth, V.; Fukuda, K. Robust Peer to Peer Mobile Botnet Detection by Using Communication Patterns. In Proceedings of the AINTEC ’18, AINTEC Asian Internet Engineering Conference, Bangkok, Thailand, 12–14 November 2018; Association for Computing Machinery: New York, NY, USA, 2018; pp. 38–45. [CrossRef] p g y pp [ ] 196. Karim, A.; Salleh, R.; Khan, K. SMARTbot: A Behavioral Analysis Framework Augmented with Machine Learning to Identify Mobile Botnet Applications. PLoS ONE 2016, 11, e0150077. [CrossRef] [PubMed] 197. Tidke, S.K.; Karde, P.; Thakare, V. Identiﬁcation of Botnet hidden behind smartphone applications. In Proceedings of the 2017 International Conference on Energy, Communication, Data Analytics and Soft Computing (ICECDS), Chennai, India, 1–2 August 2017; pp. 420–424. [CrossRef] 198. Garip, M.T.; Reiher, P.; Gerla, M. RIoT: A Rapid Exploit Delivery Mechanism against IoT Devices Using Vehicular Botnets. In Proceedings of the 2019 IEEE 90th Vehicular Technology Conference (VTC2019-Fall), Honolulu, HI, USA, 22–25 September 2019; pp. 1–6. [CrossRef] pp 199. Baltazar, J.; Costoya, J.; Flores, R. Steep rise in Koobface variants is boosted by social networking. Comput. Fraud. Secur. 2009, 2009, 19–20. [CrossRef] 200. Cybersecurity & Infrastructure Security Agency—“Alert (TA14-150A)—GameOver Zeus P2P Malware”. Available online: https://us-cert.cisa.gov/ncas/alerts/TA14-150A (accessed on 4 August 2016). 201. Hatzivasilis, G.; Soultatos, O.; Chatziadam, P.; Fysarakis, K.; Askoxylakis, I.; Ioannidis, S.; Alaxandris, G.; Katos, V.; Spanoudakis, G. WARDOG: Awareness detection watchbog for Botnet infection on the host device. IEEE Trans. Sustain. Comput. 2019, 6, 4–18. [CrossRef] 202. Ahmed, Z.; Danish, S.M.; Qureshi, H.K.; Lestas, M. Protecting IoTs from mirai botnet attacks using blockchains. References In Proceedings of the 2019 IEEE 24th International Workshop on Computer Aided Modeling and Design of Communication Links and Networks (CAMAD), Limassol, Cyprus, 11–13 September 2019; p. 8858484. [CrossRef] yp p p 203. Yin, D.; Zhang, L.; Yang, K. A DDoS Attack Detection and Mitigation with Software-Deﬁned Internet of Things Framework. IEEE Access 2018, 6, 24694–24705. [CrossRef] Future Internet 2021, 13, 198 42 of 43 204. Sadeghian, A.; Zamani, M. Detecting and preventing DDoS attacks in botnets by the help of self triggered black holes. In Proceedings of the 2014 Asia-Paciﬁc Conference on Computer Aided System Engineering (APCASE), Bali, Indonesia, 10–12 February 2014; pp. 38–42. [CrossRef] y pp 205. De Donno, M.; Donaire Felipe, J.M.; Dragoni, N. ANTIBIOTIC 2.0: A Fog-based Anti-Malware for Internet of Things. In Proceedings of the 2019 IEEE European Symposium on Security and Privacy Workshops (EuroS PW), Stockholm, Sweden, 17–19 June 2019; pp. 11–20. [CrossRef] pp 206. Wang, P.; Wu, L.; Cunningham, R.; Zou, C.C. Honeypot Detection in Advanced Botnet Attacks. Int. J. Inf. Comput. Secur. 2010, 4, 30–51. [CrossRef] 207. Jerkins, J.A. Motivating a market or regulatory solution to IoT insecurity with the Mirai botnet code. In Proceedings of the 2017 IEEE 7th Annual Computing and Communication Workshop and Conference, CCWC 2017, Las Vegas, NV, USA, 9–11 January 2017; p. 7868464. [CrossRef] p 208. Oliveri, A.; Lauria, F. Sagishi: An undercover software agent for inﬁltrating IoT botnets. Netw. Secur p 208. Oliveri, A.; Lauria, F. Sagishi: An undercover software agent for inﬁltrating IoT botnets. Netw. Secur. 2019, 2019, 9–14. [CrossRef] 209. Tetarave, S.K.; Tripathy, S.; Kalaimannan, E.; John, C.; Srivastava, A. A Routing Table Poisoning Model for Peer-to-Peer (P2P) Botnets. IEEE Access 2019, 7, 67983–67995. [CrossRef] , ; , g g g , , [ ] 209. Tetarave, S.K.; Tripathy, S.; Kalaimannan, E.; John, C.; Srivastava, A. A Routing Table Poisoning Model for Peer-to-Peer (P2P) Botnets. IEEE Access 2019, 7, 67983–67995. [CrossRef] 210. Davis, C.R.; Fernandez, J.M.; Neville, S. Optimising sybil attacks against P2P-based botnets. In Proceedings of the 2009 4th International Conference on Malicious and Unwanted Software, Malware 2009, Montreal, QC, Canada, 13–14 October 2009; p. 5403016. [CrossRef] p 211. Kelly, C.; Pitropakis, N.; McKeown, S.; Lambrinoudakis, C. Testing and Hardening IoT Devices against the Mirai Botnet. In Proceedings of the 2020 International Conference on Cyber Security and Protection of Digital Services (Cyber Security), Dublin, Ireland, 15–19 June 2020; p. 9138887. [CrossRef] 212. References Hallman, R.; Bryan, J.; Palavicini, G.; Divita, J.; Romero-Mariona, J. IoDDoS—The internet of distributed denial of sevice attacks A case study of the mirai malware and IoT-Based botnets. In Proceedings of the IOTBDS 2017—2nd International Conference on Internet of Things, Big Data and Security, Porto, Portugal, 24–26 April 2017; pp. 47–58. [CrossRef] 213. Huan, L.; Yu, Y.; Lv, L.; Li, S.; Xia, C. A botnet-oriented collaborative defense scheme description language. In Proceedings of the 9th International Conference on Computational Intelligence and Security, CIS 2013, Emeishan, China, 14–15 December 2013; p. 6746511. [CrossRef] p 214. Bradbury, D. Fighting botnets with sinkholes. Netw. Secur. 2012, 2012, 12–15. [CrossRef] h C b h C h k h lk G S S S l A l l f p 214. Bradbury, D. Fighting botnets with sinkholes. Netw. Secur. 2012, 2012, 12–15. [CrossRef] 214. Bradbury, D. Fighting botnets with sinkholes. Netw. Secur. 2012, 2012, 12–15. [CrossRef] 215. Kanich, C.; Kreibich, C.; Levchenko, K.; Enright, B.; Voelker, G.M.; Paxson, V.; Savage, S. Spamalytics: An empirical analysis of spam marketing conversion. Commun. ACM 2009, 52, 99–107. [CrossRef] 215. Kanich, C.; Kreibich, C.; Levchenko, K.; Enright, B.; Voelker, G.M.; Paxson, V.; Savage, S. Spamalyt spam marketing conversion. Commun. ACM 2009, 52, 99–107. [CrossRef] 216. Watkins, L.; Kawka, C.; Corbett, C.; Robinson, W.H. Fighting banking botnets by exploiting inherent command and control vulnerabilities. In Proceedings of the 9th IEEE International Conference on Malicious and Unwanted Software, Malcon 2014, Fajardo, PR, USA, 28–30 October 2014; p. 6999411. [CrossRef] 217. Stone-Gross, B.; Cova, M.; Gilbert, B.; Kemmerer, R.; Kruegel, C.; Vigna, G. Analysis of a botnet takeov 9, 5560627. [CrossRef] 218. The Spamhaus Project. Available online: https://www.spamhaus.org/bcl/ (accessed on 23 June 2020). 219. IBM X-Force exchange. Available online: https://exchange.xforce.ibmcloud.com/collection/Botnet-Command-and-Control- Servers-7ac6c4578facafa0de50b72e7bf8f8c4 (accessed on 23 June 2020). 220. Li, J.; Xue, Z. Distributed Threat Intelligence Sharing System: A New Sight of P2P Botnet Detection. In Proceedings of the 2nd International Conference on Computer Applications and Information Security, ICCAIS 2019, Riyadh, Saudi Arabia, 1–3 May 2019; p. 8769511. [CrossRef] p 221. Dog, S.E.; Tweed, A.; Rouse, L.; Chu, B.; Qi, D.; Hu, Y.; Yang, J.; Al-Shaer, E. Strategic cyber threat intelligence sharing: A case study of IDS logs. In Proceedings of the 2016 25th International Conference on Computer Communications and Networks, ICCCN 2016, Waikoloa, HI, USA, 1–4 August 2016; p. 7568578. [CrossRef] g p 222. References Eslahi, M.; Rostami, M.R.; Hashim, H.; Tahir, N.M.; Naseri, M.V. A data collection approach for Mobile Botnet analysis and detection. In Proceedings of the 2014 IEEE Symposium on Wireless Technology and Applications (ISWTA), Kota Kinabalu, Malaysia, 28 September–1 October 2014; pp. 199–204. [CrossRef] y p pp 223. Garcia, S.; Erquiaga, M.J.; Shirokova, A.; Garcia Garino, C. Geost Botnet. Operational Security Failures of a New Android Banking Threat. In Proceedings of the 2019 IEEE European Symposium on Security and Privacy Workshops (EuroS PW), Stockholm, Sweden, 17–19 June 2019; pp. 406–409. [CrossRef] 224. Vidal-González, S.; García-Rodríguez, I.; Aláiz-Moretón, H.; Benavides-Cuéllar, C.; Benítez-Andrades, J.A.; García-Ordás, M.T.; Novais, P. Analyzing IoT-Based Botnet Malware Activity with Distributed Low Interaction Honeypots. In Trends and Innovations in Information Systems and Technologies; Rocha, Á., Adeli, H., Reis, L.P., Costanzo, S., Orovic, I., Moreira, F., Eds.; Springer International Publishing: Cham, Switzerland, 2020; pp. 329–338. International Publishing: Cham, Switzerland, 2020; 225. Safaei Pour, M.; Mangino, A.; Friday, K.; Rathbun, M.; Bou-Harb, E.; Iqbal, F.; Samtani, S.; Crichigno, J.; Ghani, N. On data-driven curation, learning, and analysis for inferring evolving internet-of-Things (IoT) botnets in the wild. Comput. Secur. 2020, 91, 101707. [CrossRef] 226. Borgaonkar, R. An analysis of the asprox botnet. In Proceedings of the 4th International Conference on Emerging Security Information, Systems and Technologies, Securware 2010, Venice, Italy, 18–25 July 2010; p. 5633693. [CrossRef] 43 of 43 Future Internet 2021, 13, 198 227. Li, X.; Duan, H.; Liu, W.; Wu, J. The growing model of Botnets. In Proceedings of the 2010 International Conference on Green Circuits and Systems, Shanghai, China, 21–23 June 2010; p. 5543027. [CrossRef] y g J p 228. Patsakis, C.; Casino, F.; Katos, V. Encrypted and covert DNS queries for botnets: Challenges and countermeasures. Comput. Secur. 2020, 88, 101614. [CrossRef] 229. Stone-Gross, B.; Cova, M.; Cavallaro, L.; Gilbert, B.; Szydlowski, M.; Kemmerer, R.; Kruegel, C.; Vigna, G. Your Botnet is My Botnet: Analysis of a Botnet Takeover. In Proceedings of the 16th ACM Conference on Computer and Communications Security, CCS 09, Chicago, IL, USA, 9–13 November 2009; Association for Computing Machinery: New York, NY, USA, 2009; pp. 635–647. [CrossRef] 230. Barford, P.; Yegneswaran, V. An Inside Look at Botnets. In Malware Detection; Christodorescu, M., Jha, S., Maughan, D., Song, D., Wang, C., Eds.; Springer: Boston, MA, USA, 2007; pp. 171–191. 231. Shahrestani, A.; Feily, M.; Masood, M.; Muniandy, B. References Visualization of invariant bot behavior for effective botnet trafﬁc detection. In Proceedings of the 2012 International Symposium on Telecommunication Technologies, ISTT 2012, Kuala Lumpur, Malaysia, 26–28 November 2012; p. 6481606. [CrossRef] p 232. Rowe, B.; Wood, D.; Reeves, D. How the public views strategies designed to reduce the threat of botnets. In Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artiﬁcial Intelligence and Lecture Notes in Bioinformatics); Springer: Berlin/Heidelberg, Germany, 2010; Volume 6101, pp. 337–351. [CrossRef] g y pp 233. Khattak, S.; Ramay, N.R.; Khan, K.R.; Syed, A.A.; Khayam, S.A. A Taxonomy of Botnet Behavior, Detection, and Defense. IEEE Commun. Surv. Tutor. 2014, 16, 898–924. [CrossRef] 234. Peterson, A. Yes, Terrorists Could Have Hacked Dick Cheneys Heart. Available online: https://www.washingtonpost.com/ news/the-switch/wp/2013/10/21/yes-terrorists-could-have-hacked-dick-cheneys-heart/ (accessed on 14 December 2020).
https://openalex.org/W3120038096	https://jenci.springeropen.com/track/pdf/10.1186/s43046-020-00056-y	English	null	Renal angiomyoadenomatous tumor (RAT): a rare distinct entity with diagnostic challenges—a case report	Journal of Egyptian National Cancer Institute/Journal of the Egyptian National Cancer Institute	2,021	cc-by	3,331	© The Author(s). 2021 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. CASE REPORTS Open Access Open Access Background ducts, a stroma that is leiomyomatosis in nature and in- terspersed by abortive vascular channels [1, 2]. Renal angiomyoadenomatous tumor (RAT) is a very rare neoplasm with fewer than 15 cases reported in literature till now. Newer entities are being added continuously to the already existing database of tumors and RAT is one such entity that has not yet found its place in the World Health Organization (WHO) classification of kidney tu- mors. It has been microscopically described as a neo- plasm containing an epithelial component in the form of Here, we present a case of RAT in a 21-year-old male with a distinctive presentation and gross features in an attempt to include diversity to the pathological profile of this, particularly rare neoplasm. Abstract Background: Renal angiomyoadenomatous tumor (RAT) is a recently described rare renal neoplasm with variations in the presentation, gross, and microscopic findings, and having a benign course and good prognosis. It is characterized microscopically by the admixture of three components—epithelial cells arranged in tubules and nests, angiomyomatous stroma, and capillary-sized interconnecting vascular channels in close association with the epithelial cell clusters. Microscopically, these tumors can be confused with clear cell carcinoma, papillary carcinoma, mixed epithelial and stromal tumors, and angiomyolipoma. RAT differs from conventional clear cell carcinomas, which can rarely be associated with an identical leiomyomatosis stroma occasionally forming abortive vascular structures. RAT is a distinct morphologic entity, being different morphologically, immunohistochemically, and genetically from all renal tumors including conventional clear cell carcinoma and mixed epithelial and stromal tumor of the kidney. Case presentation: Here, we report a case of a 21-year-old man with renal angiomyoadenomatous tumor, a rare neoplasm with only a few previous cases reported in the literature. Unlike our case, most tumors have been identified in middle-aged males; they present as well-circumscribed, encapsulated tan-brown masses with variably prominent cystic areas. Conclusion: Diagnosis of RAT is challenging because of the rarity of the disease and common presenting symptoms to other renal pathology and is supplemented with histopathology and immunohistochemistry. A multidisciplinary team approach for diagnosis and management along with long-term follow-up are warranted. Keywords: Rare renal tumor, Renal angiomyoadenomatous tumor, Renal cell carcinoma, Mixed epithelial and stromal tumor, Case report Renal angiomyoadenomatous tumor (RAT): a rare distinct entity with diagnostic challenges—a case report Ankur Majumder1, Ravi Hari Phulware2* , Arvind Ahuja1, Anurag Singla3 and Pawan Kumar4 (2021) 33:1 (2021) 33:1 Majumder et al. Journal of the Egyptian National Cancer Institute (202 https://doi.org/10.1186/s43046-020-00056-y Majumder et al. Journal of the Egyptian National Cancer Institute https://doi.org/10.1186/s43046-020-00056-y Journal of the Egyptian National Cancer Institute * Correspondence: ravipaarti@gmail.com 2Department of Pathology, All India Institute of Medical Sciences (AIIMS), Room no. C-2, Level 3, Rishikesh, India Full list of author information is available at the end of the article Case presentation A 21-year-old male patient presented to the Ram Manohar Lohia (RML) Hospital, Post Graduate Institute of Medical Sciences (PGIMER), New Delhi, India, with pain in left flank pain intermittent hematuria. Non-contrast computed tomography (NCCT) scan of the abdomen showed gross * Correspondence: ravipaarti@gmail.com 2Department of Pathology, All India Institute of Medical Sciences (AIIMS), Room no. C-2, Level 3, Rishikesh, India Full list of author information is available at the end of the article Majumder et al. Journal of the Egyptian National Cancer Institute (2021) 33:1 Page 2 of 5 (2021) 33:1 Page 2 of 5 hydronephrosis of the left kidney (white arrows) with severe parenchymal thinning with pelvic ureteric junction obstruc- tion (PUJO) (Fig. 1). A left radical nephrectomy was done with hilar lymph node dissection with a clinical diagnosis of non-functioning kidney secondary to left PUJO. CD10 (Fig. 2f) and HMB-45 (Fig. 2g), the stromal com- ponent was positive for smooth muscle actin (SMA) (Fig. 2h) and was immunonegative for CD34, estrogen receptor (ER); thereby confirming the smooth muscle differentiation. Therefore, based on the histomorpholo- gical and immunohistochemical staining pattern diagno- sis of renal angiomyoadenomatous tumor was given. Gross examination showed a specimen of the kidney measuring 14 × 8 × 6 cm. The external surface was bos- selated. The capsule was easily stripped off. No scars were noted externally. Serial slicing revealed multiple interconnecting cysts with few thickened areas replacing the entire parenchyma of the kidney. The cysts ranged in size from 1.5 to 4 cm. Cortico-medullary differenti- ation could not be made out (Fig. 2a). A small part of the ureter was seen measuring 1 cm. Hilar lymph nodes measuring 2.5 × 1 was received separately. Discussion RAT is a rare and distinct neoplasm. The average age group in reported cases was 46 years. No sex predilec- tion has been noted in the tumor. Rare studies have shown a tumor with cystic changes. Common differen- tials for mixed renal carcinomas to be kept in mind are mixed epithelial and stromal tumor of the kidney (MESTK), angiomyolipomas, clear cell renal cell carcin- omas with angioleiomyomatous stroma, and clear cell papillary renal cell carcinomas (ccpRCC) [3–5]. Microscopic examinations from multiple sections showed a tumor composed of epithelial, smooth muscle and vascular components. The epithelial component was arranged in the form of tubules. The tumor cells had a moderate amount of clear to eosinophilic cytoplasm with a basally placed round to oval vesicular nucleus and ap- ical snouting. Intervening stromal areas showed smooth muscle differentiation arranged in fascicles and bundles (Fig. 2c, d). For the vascular component, both thick and thin blood vessels were seen which were lined by plump endothelial cells. The resected end of the ureter and the hilar structures were free of the tumor. The specimen la- beled as hilar lymph node showed features of reactive lymphoid hyperplasia. Immunohistochemistry (IHC) was done for the tumor. The epithelial component was posi- tive for pan-cytokeratins (Fig. 2e) and was negative for y According to Michal et al. [2], RAT epithelial compo- nent consists of adenomatous structures composed of cells that are secretory having basophilic nuclei alienated along the basal membrane and prominent apical snouts, resulting in a characteristic appearance of “Shark’s smile.” This epithelial component is usually shown immunopositivity for all cytokeratins like CK-7 more than CK-20, CAM 5.2, and cytokeratins AE1-AE3. In addition to cytokeratins positivity, epithelial membrane antigen (EMA) and vimentin are also positive. While IHC for CD10, Melan-A, and HMB-45 is immunonega- tive. The present case showed immunopositivity for pan- cytokeratin, while IHC foe CD10 and HMB-45 were Fig. 1 Axial (a, b) and coronal (c) non-contrast-enhanced CT (NCCT) scan images showing gross hydronephrosis of the left kidney (white arrows) with severe parenchymal thinning (curved red arrows) Fi 1 A i l ( b) d l ( ) h d CT (N Fig. 1 Axial (a, b) and coronal (c) non-contrast-enhanced CT (NCCT) scan images showing gross hydronephrosis of the left kidney (white arrows) with severe parenchymal thinning (curved red arrows) Majumder et al. Discussion Journal of the Egyptian National Cancer Institute (2021) 33:1 (2021) 33:1 Page 3 of 5 Majumder et al. Journal of the Egyptian National Cancer Institute Majumder et al. Journal of the Egyptian National Cancer Institute (2021) 33:1 Page 3 o Fig. 2 Histopathological examination of renal angiomyoadenomatous tumor (RAT): gross image demonstrating multiple interconnecting cystic spaces with intervening thickened areas with patchy hemorrhagic areas (a). Hematoxylin and eosin (H&E) stained lower magnification image showing cystic spaces (epithelial component) along with thick-walled blood vessels (blue arrow) in renal parenchyma (b, h&e, × 40). Epithelial component arranged in adenomatous/glandular/tubular pattern with basally located bland small beaded nuclei and intimately surrounded by thin vascular channels. The cytoplasm of the cells is optically clear with blisters like apical snouts giving the appearance of “Shark’s Smile/moth- eaten” (green arrow). Stromal smooth muscle component (red arrow) (H&E; × 100, c; and × 400, d). The epithelial component is immunopositive for pan-cytokeratin (e, × 400), immunonegative for CD10 (f, × 400), and immunonegative for HMB-45 (g, × 40). The stromal smooth muscle was Fig. 2 Histopathological examination of renal angiomyoadenomatous tumor (RAT): gross image demonstrating multiple interconnecting cystic spaces with intervening thickened areas with patchy hemorrhagic areas (a). Hematoxylin and eosin (H&E) stained lower magnification image showing cystic spaces (epithelial component) along with thick-walled blood vessels (blue arrow) in renal parenchyma (b, h&e, × 40). Epithelial component arranged in adenomatous/glandular/tubular pattern with basally located bland small beaded nuclei and intimately surrounded by thin vascular channels. The cytoplasm of the cells is optically clear with blisters like apical snouts giving the appearance of “Shark’s Smile/moth- eaten” (green arrow). Stromal smooth muscle component (red arrow) (H&E; × 100, c; and × 400, d). The epithelial component is immunopositive for pan-cytokeratin (e, × 400), immunonegative for CD10 (f, × 400), and immunonegative for HMB-45 (g, × 40). The stromal smooth muscle was immunopositive for smooth muscle actin (SMA) (h, × 100) Fig. 2 Histopathological examination of renal angiomyoadenomatous tumor (RAT): gross image demonstrating multiple interconnecting cystic spaces with intervening thickened areas with patchy hemorrhagic areas (a). Hematoxylin and eosin (H&E) stained lower magnification image showing cystic spaces (epithelial component) along with thick-walled blood vessels (blue arrow) in renal parenchyma (b, h&e, × 40). Epithelial component arranged in adenomatous/glandular/tubular pattern with basally located bland small beaded nuclei and intimately surrounded by thin vascular channels. Discussion The cytoplasm of the cells is optically clear with blisters like apical snouts giving the appearance of “Shark’s Smile/moth- eaten” (green arrow). Stromal smooth muscle component (red arrow) (H&E; × 100, c; and × 400, d). The epithelial component is immunopositive for pan-cytokeratin (e, × 400), immunonegative for CD10 (f, × 400), and immunonegative for HMB-45 (g, × 40). The stromal smooth muscle was immunopositive for smooth muscle actin (SMA) (h, × 100) resistant, and mucicarmine negative [2–4]. Our case shows similar morphology. negative. Focal solid and clear cell areas may be seen. It may resemble conventional clear cell carcinoma Fuhr- man grade 1. These secretory cells usually contain glyco- gen which is periodic acid-Schiff (PAS) positive, diastase RAT shows a unique relation between the capillary net- work and the epithelial component. The capillaries tightly Majumder et al. Journal of the Egyptian National Cancer Institute Page 4 of 5 (2021) 33:1 Page 4 of 5 (2021) 33:1 surround the basal membrane of the adenomatous struc- tures. The identification of these endothelial cells of the capillary network is possible mostly by the immunohisto- chemistry for CD34. This intimate capillary network is not seen elsewhere [2, 6]. The stromal muscular component is made up of strands that grow among the epithelial com- ponent, resembling abortive vessels without the elastic layer. Occasionally myxoid, hyaline, or metaplastic change (ossification) is seen. This leiomyomyomatous stroma can also be seen in conventional clear cell renal cell carcin- omas [2]. of ccpRCC will be evident with a minor component of the RAT-like area [10]. Immunohistochemical feature of RAT may overlap with ccpRCC, but morphologically ccpRCC will be having protuberant papillary architecture with thick cellular core and the large, generous clear cells lining the papillary structures so that the cells of one papilla may touch the cells of the adjacent papilla. In RAT, papillary structures are absent and the clear cell component is less prominent [6, 7, 10]. p p RAT usually a solid tumor with some microcystic areas. The presence of macro-cystic areas in RAT is a very rare occurrence. Michal et al. [2] studied five cases of RAT in his initial study out of which only one showed marked cystic areas. The present case tumor replaced the whole kidney and showed marked cystic changes which is an uncommon finding in RAT and not reported before. However, some studies like Deml et al. Acknowledgements Acknowledgements N/A Discussion [11] have postulated that RAT and clear cell papillary renal cell carcinoma (ccpRCC) are two entities of the same spectrum of disease and it is difficult to distinguish on the grounds of morphology, immunohistochemistry markers, and molecular changes. They have described that RAT is a tumor with “varying amounts of tubular, papillary, and cystic architecture.” Other differentials in- clude Xp11 and transcription factor E3 (TFE3) trans- location cancer [8, 9, 11]. Leiomyomatous components in the tumor stroma of RAT are usually immunopositive for SMA, vimentin, and h-caldesmon while negative for HMB-45 and Melan-A. In the index case, the leiomyomatous compo- nent in the tumor stroma was positive for SMA. Accord- ing to literature, few studies have reported renal cell tumors with glandular elements and leiomyomatosis stroma as a metachronous renal cell carcinoma with “an abnormally large quantity of smooth muscle, not related to the pelvis or calyces, nor to blood vessels” or de- scribes them as “hamartomas” or “fibroleiomuscular” component. Kuhn et al. reported five cases of renal cell carcinomas with angioleiomyomas-like components and a desmoplastic reaction in the stroma, unlike what we see in RAT [1–3, 5]. Precise diagnosis is crucial since this neoplasm has an excellent prognosis. Fluorescence in situ hybridization studies in four cases by Kuroda et al. have revealed that monosomy of chromosomes 1, 11, and 16 can be consid- ered to be diagnostic in RAT. The preferred treatment is surgical resection and there have no reported cases of recurrence or death due to the neoplasm [10]. MESTK is usually seen in middle-aged, peri- menopausal women and is related to estrogen. It was earlier grouped under the broader term of “Cystic nephroma.” The stroma in MESTK is identical to ovar- ian stroma with few leiomyomatosis areas. There can be various Müllerian epithelial type differentiations, e.g., tubal, endometrial, squamous. Intestinal mucinous glan- dular epithelium and Paneth cells may be seen. These features are not seen in RATs [6, 7]. Angiomyolipomas usually occur in association with tuberous sclerosis. They contain adipose tissue with thick blood vessels de- void of the elastic layer, which can sometimes be seen in RATs. However, they have a typical arrangement of myoid stromal cells which are perpendicular to vascular lumens. Also, angiomyolipomas, tumors stain positive for melanocytic markers like HMB45 and Melan-A, which is not seen in RATs. None of the melanocytic markers tested positive in the angioleiomyomatous stroma of RATs [7–9]. Conclusion T i To summarize, RAT is a rare renal neoplasm with varia- tions in the presentation, gross, and microscopic find- ings, and having a benign course and good prognosis. However, owing to its distinct morphological, immuno- histochemical, and genetic profile, a correct diagnosis needs to be made. Abbreviations RAT R l RAT: Renal angiomyoadenomatous tumor; WHO: World Health Organization; NCCT: Non-contrast computed tomography; PUJO: Pelvic ureteric junction obstruction; IHC: Immunohistochemistry; SMA: Smooth muscle actin; MESTK: Mixed epithelial and stromal tumor of the kidney; CcpRCC: Clear cell papillary renal cell carcinomas; PAS: Periodic acid-Schiff Conventional clear cell carcinomas may rarely show leiomyomatosis stroma. However, the characteristic Shark’s Smile is not seen. The VHL gene mutation and CD10 marker positivity are seen consistently in clear cell carcinomas and not found in RATs. Finally, we need to differentiate RATs from ccpRCC. Grossly both the tumor may show either cystic or papillary architecture. Rare cases of clear cell ccpRCC with RAT-like areas have been reported in the literature. In these cases, areas RAT: Renal angiomyoadenomatous tumor; WHO: World Health Organization; NCCT: Non-contrast computed tomography; PUJO: Pelvic ureteric junction obstruction; IHC: Immunohistochemistry; SMA: Smooth muscle actin; MESTK: Mixed epithelial and stromal tumor of the kidney; CcpRCC: Clear cell papillary renal cell carcinomas; PAS: Periodic acid-Schiff References 1. Michal M, Hes O, Havlicek F. Benign renal angiomyoadenomatous tumor: a previously unreported renal tumor. Ann DiagnPathol. 2000;4:311–5. 1. Michal M, Hes O, Havlicek F. Benign renal angiomyoadenomatous tumor: a previously unreported renal tumor. Ann DiagnPathol. 2000;4:311–5. 2. Michal M, Hes O, Nemcova J, Sima R, Kuroda N, Bulimbasic S, et al. Renal angiomyoadenomatous tumor: morphologic, immunohistochemical, and molecular genetic study of a distinct entity. Virchows Arch. 2009;454:89–99. 3. Verine J. Renal angiomyoadenomatous tumor: morphologic, immunohistochemical, and molecular genetic study of a distinct entity. Virchows Arch. 2009;454:479–80. 4. Singh C, Kendi AT, Manivel JC, et al. Renal angiomyoadenomatous tumor. Ann DiagnPathol. 2012;16:470–6. 5. Kuhn E, De Anda J, Manoni S, Netto G, Rosai J. Renal cell carcinoma associated with prominent angioleiomyoma-like proliferation. Report of 5 cases and review of the literature. Am JSurgPathol. 2006;30:1372–81. 5. Kuhn E, De Anda J, Manoni S, Netto G, Rosai J. Renal cell carcinoma associated with prominent angioleiomyoma-like proliferation. Report of 5 cases and review of the literature. Am JSurgPathol. 2006;30:1372–81. 6. Petersson F, Grossmann P, Hora M, Superga M, Perez D, Martinek P. Renal cell carcinoma with areas mimicking renal angiomyoadenomatous tumor/ clear cell papillary renal cell carcinoma. Hum Pathol. 2013;44:1412–20. 7. Mohanty SK, Parwani AV. Mixed epithelial and stromal tumors of the kidney: an overview Arch Pathol Lab Med 2009;133:1483–6 6. Petersson F, Grossmann P, Hora M, Superga M, Perez D, Martinek P. Renal cell carcinoma with areas mimicking renal angiomyoadenomatous tumor/ clear cell papillary renal cell carcinoma. Hum Pathol. 2013;44:1412–20. 7 Mohanty SK Parwani AV Mixed epithelial and stromal tumors of the kidney: y 7. Mohanty SK, Parwani AV. Mixed epithelial and stromal tumors of the kidney: an overview. Arch Pathol Lab Med. 2009;133:1483–6. 8. Tan G, Liu L, Qiu M, Chen L, Cao J, Liu J. Clinicopathologic features of renal epithelioid angiomyolipoma: report of one case and review of the literature. Int JClinExpPathol. 2015;8:1077–80. 9. Williamson SR, Eble JN, Cheng L, et al. Clear cell papillary renal cell carcinoma: differential diagnosis and extended immunohistochemical profile. Mod Pathol. 2013;26:697–708. 10. Kuroda N, Michal M, Hes O, et al. Renal angiomyoadenomatous tumor: fluorescence in situ hybridization. PatholInt. 2009;59:689–91. 11. Deml KF, Schildhaus HU, Comperat E, et al. Clear cell papillary renal cell carcinoma and renal angiomyoadenomatous tumor: two variants of a morphologic, immunohistochemical, and genetic distinct entity of renal cell carcinoma. Am J Surg Pathol. 2015;39:889–901. Consent for publication Written informed consent was obtained from the patient included in the study. Funding f d g No funds were received. Competing interests The authors declare that they have no conflicts of interest. Ethics approval and consent to participate Ethics approval and consent to participate Informed written consent was taken from the patient. Internal review board approval is not needed as it is a case report. Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Funding No funds were received. Funding No funds were received. Author details 1D f 1Department of Pathology, ABVIMS, PGIMER, RML Hospital, New Delhi, India. 2Department of Pathology, All India Institute of Medical Sciences (AIIMS), Room no. C-2, Level 3, Rishikesh, India. 3Department of Urology and Renal Transplant, PGIMER, ABVIMS, RML Hospital, New Delhi, India. 4Department of Radiology, Goa Medical College, Bambolim, Goa, India. Received: 26 September 2020 Accepted: 15 December 2020 Authors’ contributions AM i i i AM: manuscript writing and data collection. RP: manuscript revision and editing. PK: radiological supervision. AA: pathology revision. AS: manuscript revision and supervision. All authors had read and approved the manuscript. Page 5 of 5 Page 5 of 5 Majumder et al. Journal of the Egyptian National Cancer Institute Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
https://openalex.org/W1506499959	https://revistas.udc.es/index.php/rgf/article/download/rgf.2004.5.0.5335/g5335_pdf	es		E ja quites son los Mans. Nota paleográfica sobre o v. 4 da cantiga B 454 de D. Garcia Mendiz d’Eixo	Revista galega de filoloxía	2,004	cc-by-sa	2,477	DOI: https://doi.org/10.17979/rgf.2004.5.0.5335 E ja quites son los Mans. Nota paleográfica sobre o v. 4 da cantiga B 454 de D. Garcia Mendiz d’Eixo Xosé Bieito Arias Freixedo Universidade de Vigo Son de todos coñecidas as dificultades que presenta para os editores a cantiga B 454 de D. Garcia Mendiz d’Eixo, non tanto polo feito de que estea escrita en provenzal, ou de que en aparencia tente reproducir esta lingua, senón porque probabelmente por esta razón nalgún momento da súa transmisión manuscrita un copista tivo dificultades para interpretala e a versión que nos foi transmitida chegounos deturpada e presenta grandes problemas de lectura e de interpretación. Abonda con remitir ás varias edicións para comprobar que as interpretacións son totalmente diverxentes. Por exemplo, a dúas edicións realizadas con detemento como son a de J. Marie D’Heur (1973: 93-104) –demasiado afastada do texto manuscrito– e a recente edición de Graça Videira Lopes (nº 12), das que reproducimos a primeira estrofa, onde se atopa o noso verso, o que nos serve para situalo en contexto e ademais para exemplificar a considerábel diverxencia entre ambas versións. A la mazo on e[s] la corona, e li parent son tan, e la terra es tro bona, e ja quyte sson li chan, c’ara me volh tornar a Sousa, a lo mon logar que m’adola e ma sauda (D’Heur, p. 100) Alá nazque la Torona e los pavous son tan[s] e la terra é tro bona! Eia! quites son los chans! C’ora me volho tornar a Sousa, a lo mon logar, que me adota e me saudona. (Lopes, p. 34) Revista Galega de Filoloxía, ISSN 1576-2661 \| e-ISSN 2444-9121, 2004, 5: 165-169 165 Xosé Bieito Arias Freixedo E ja quites son los Mans. Nota paleográfica sobre o v. 4 da cantiga B 454... Non imos realizar aquí unha edición de todo o texto; simplemente chamaremos a atención sobre unha deficiente lectura do verso cuarto da composición, que se repite de editor en editor, e que coidamos se pode corrixir se se analizan as características paleográficas do manuscrito un pouco pormenorizadamente. Se o compararmos agora coa grafía que se rexistra na palabra final do cuarto verso da primeira estrofa, veremos que estamos perante outra grafía diferente (ofrecemos a imaxe da estrofa íntegra para que o lector ou lectora poida cotexar con ela as edicións críticas ofrecidas máis arriba): D’Heur le este verso no manuscrito do seguinte xeito: Eia. quytes fõ los chaus G. V. Lopes le tamén como chaus a última palabra do verso, aínda que edita chans e propón interpretala como ‘cantos’ ou talvez como ‘prantos’. Aínda na sección final de notas (p. 553) propón outra interpretación posíbel mantendo a lección do manuscrito, que ela le chaus, co sentido de ‘calores’: “(acabaram-se os calores)”. É sobre este último termo que queremos chamar a atención, pois no manuscrito non se le chaus, senón Maus, como trataremos de demostrar, cotexando as grafías con outras grafías da mesma man, do mesmo folio e da mesma cantiga. O cotexo da grafía inicial deste termo con outras grafías que aparecen na mesma cantiga, copiadas pola mesma man, para representar o fonema africado palatal xordo, o dígrafo ch, permítenos concluír que neste verso cuarto estamos perante unha grafía diferente. Na columna b do folio 99 (p. 225 da ed. facsímile), onde continúa a copia da cantiga, rexístrase este dígrafo en tres ocasións, e nas tres presenta o mesmo trazado: o c unido ao h sen erguer o trazo, de xeito que o h presenta no hástil o lazo característico da cursividade da escrita. Os tres exemplos aparecen en liñas seguidas, o primeiro deles riscado, porque se debe a un salto de liña do copista, como pode comprobarse na seguinte imaxe: Parece claro que a grafía non é a mesma que a da outra imaxe. É importante ter en conta o usus scribendi do amanuense, que parece indicar que o hábito do copista era o de escribir o dígrafo ch sen erguer o trazo e con bucle ou lazo no hástil do h. En moi última instancia o h podería selo, mais o trazo que o precedería en ningún caso podería corresponder a un c. Cal é a grafía do v. 4, pois? O certo é que a aplicación dunhas nocións básicas de paleografía permítenos propor a hipótese de que nos atopamos perante unha das variantes do M maiúsculo que presenta a gótica cursiva. É mais, no mesmo folio 99, o mesmo copista emprega unha grafía similar como inicial en dous versos da cantiga seguinte, aos que corresponden as seguintes imaxes: Na última palabra das liñas primeira e terceira temos o dígrafo ch (no primeiro caso o copista riscou logo a palabra ao decatarse de que pertencía ao verso seguinte). Despois do a inicial da segunda liña volve aparecer este dígrafo. 166 167 Xosé Bieito Arias Freixedo E ja quites son los Mans. Nota paleográfica sobre o v. 4 da cantiga B 454... Non imos realizar aquí unha edición de todo o texto; simplemente chamaremos a atención sobre unha deficiente lectura do verso cuarto da composición, que se repite de editor en editor, e que coidamos se pode corrixir se se analizan as características paleográficas do manuscrito un pouco pormenorizadamente. Se o compararmos agora coa grafía que se rexistra na palabra final do cuarto verso da primeira estrofa, veremos que estamos perante outra grafía diferente (ofrecemos a imaxe da estrofa íntegra para que o lector ou lectora poida cotexar con ela as edicións críticas ofrecidas máis arriba): D’Heur le este verso no manuscrito do seguinte xeito: Eia. quytes fõ los chaus G. V. Lopes le tamén como chaus a última palabra do verso, aínda que edita chans e propón interpretala como ‘cantos’ ou talvez como ‘prantos’. Aínda na sección final de notas (p. 553) propón outra interpretación posíbel mantendo a lección do manuscrito, que ela le chaus, co sentido de ‘calores’: “(acabaram-se os calores)”. É sobre este último termo que queremos chamar a atención, pois no manuscrito non se le chaus, senón Maus, como trataremos de demostrar, cotexando as grafías con outras grafías da mesma man, do mesmo folio e da mesma cantiga. O cotexo da grafía inicial deste termo con outras grafías que aparecen na mesma cantiga, copiadas pola mesma man, para representar o fonema africado palatal xordo, o dígrafo ch, permítenos concluír que neste verso cuarto estamos perante unha grafía diferente. Na columna b do folio 99 (p. 225 da ed. facsímile), onde continúa a copia da cantiga, rexístrase este dígrafo en tres ocasións, e nas tres presenta o mesmo trazado: o c unido ao h sen erguer o trazo, de xeito que o h presenta no hástil o lazo característico da cursividade da escrita. Os tres exemplos aparecen en liñas seguidas, o primeiro deles riscado, porque se debe a un salto de liña do copista, como pode comprobarse na seguinte imaxe: Parece claro que a grafía non é a mesma que a da outra imaxe. É importante ter en conta o usus scribendi do amanuense, que parece indicar que o hábito do copista era o de escribir o dígrafo ch sen erguer o trazo e con bucle ou lazo no hástil do h. En moi última instancia o h podería selo, mais o trazo que o precedería en ningún caso podería corresponder a un c. Cal é a grafía do v. 4, pois? O certo é que a aplicación dunhas nocións básicas de paleografía permítenos propor a hipótese de que nos atopamos perante unha das variantes do M maiúsculo que presenta a gótica cursiva. É mais, no mesmo folio 99, o mesmo copista emprega unha grafía similar como inicial en dous versos da cantiga seguinte, aos que corresponden as seguintes imaxes: Na última palabra das liñas primeira e terceira temos o dígrafo ch (no primeiro caso o copista riscou logo a palabra ao decatarse de que pertencía ao verso seguinte). Despois do a inicial da segunda liña volve aparecer este dígrafo. 166 167 Xosé Bieito Arias Freixedo E ja quites son los Mans. Nota paleográfica sobre o v. 4 da cantiga B 454... mandatos’. Sería o mesmo significado que se rexistra na cansó do trobador provenzal Marcabrú A la fontana del vergier (vid. Riquer, nº 20, v. 27): que fai los mans e los prezicx: “que dispón as ordes e as predicacións”. No noso texto este significado tamén tería sentido se o termo mans se referise ás ordes que obrigaban ao protagonista a estar lonxe da súa terra (Sousa), á que agora quere tornar, porque as tales ordes ou disposicións xa foron revocadas: e ja quites son los mans. Polo tanto haberá que ler o rimante do noso verso como Maus. Mais a interpretación do termo maus como forma plural substantivizada do adxectivo mau (mao), derivado de MALUS, que é a que xorde a primeira vista e que concordaría co sentido do conxunto da cantiga (o poeta, que está exilado, ve chegada a hora de tornar á súa patria, porque os maus, ou sexa, os ‘maos’, os que llo impedían, agora son quites, isto é, desapareceron, ou foron desprazados do poder), non é a correcta. En efecto, se tivermos en conta o esquema de rimas da cantiga, vemos que é o mesmo na primeira e segunda estrofas: ababcca. Na segunda estrofa a rima b é -al, e nesta primeira estrofa é preciso corrixir no v. 2 (tan[s]) engadindo un -s para emparellala coa rima do v. 4, en que á súa vez hai que corrixir u por n, para que a rima b sexa -ans nos dous versos. Atopámonos así co rimante Mans. Mais, cal é o valor desta palabra no texto? Non resulta doado decantarse por un significado concreto, xa que este varía dependendo do étimo a que se remonte o termo. Tendo en conta o sentido que dun xeito global parece percibirse no texto, relativo ao desexo do protagonista de tornar ao seu lugar patrio, e considerando as circunstancias históricas reais do autor, que viviu en persoa a experiencia do desterro, coidamos que o termo provenzal mans podería remontarse ao baixo latín MANSUS, termo empregado abundantemente na literatura documental medieval en latín (cf. Du Cange, Glossarium Mediae et infimae latinitatis, s. v.) cun valor equivalente ao de villa, ou sexa, ‘casa de campo, coas construcións adxacentes e terreos de cultivo’, ou tamén equivalente a vicus, isto é, ‘lugar, aldea’. Coidamos, en definitiva, que calquera das dúas interpretacións aquí propostas melloran as existentes. En todo o caso, son máis xustificábeis desde o punto de vista paleográfico. Referencias Bibliográficas: CANCIONEIRO DA BIBLIOTECA NACIONAL (Colocci-Brancuti) Cod. 10991, I. (1992). Reprodução facsimilada. Edição de L. f. Lindley Cintra (Lisboa: Imprensa Nacional / Casa da Moeda / Biblioteca Nacional). HEUR, J. M. d’ (1973): Troubadours d’oc et troubadours galiciens-portugais. Recherches sur quelques échanges dans la littérature de l’Europe au Moyen Age (París: Fundação Calouste Gulbenkian). LOPES, G. Videira (1994): A sátira nos cancioneiros medievais galego-portugueses (Lisboa: Imprensa Universitaria / Editorial Estampa). RIQUER, M. de (1993): Los trovadores. Historia literaria y textos (Barcelona: Ariel). Este valor, que se podería xeneralizar como ‘propiedades, posesións’, cadra perfectamente co sentido xeral da composición, do que se desprende o desexo do protagonista de ‘tornar / a Sousa’, ao seu lugar familiar, agora que as súas posesións, os seus mansos (mans) xa estaban libres (quites). Aínda sería posíbel outra interpretación do termo mans se o entendésemos como un substantivo derivado regresivo de mandar (<MANDARE), co sentido de ‘ordes, 168 169 Xosé Bieito Arias Freixedo E ja quites son los Mans. Nota paleográfica sobre o v. 4 da cantiga B 454... mandatos’. Sería o mesmo significado que se rexistra na cansó do trobador provenzal Marcabrú A la fontana del vergier (vid. Riquer, nº 20, v. 27): que fai los mans e los prezicx: “que dispón as ordes e as predicacións”. No noso texto este significado tamén tería sentido se o termo mans se referise ás ordes que obrigaban ao protagonista a estar lonxe da súa terra (Sousa), á que agora quere tornar, porque as tales ordes ou disposicións xa foron revocadas: e ja quites son los mans. Polo tanto haberá que ler o rimante do noso verso como Maus. Mais a interpretación do termo maus como forma plural substantivizada do adxectivo mau (mao), derivado de MALUS, que é a que xorde a primeira vista e que concordaría co sentido do conxunto da cantiga (o poeta, que está exilado, ve chegada a hora de tornar á súa patria, porque os maus, ou sexa, os ‘maos’, os que llo impedían, agora son quites, isto é, desapareceron, ou foron desprazados do poder), non é a correcta. En efecto, se tivermos en conta o esquema de rimas da cantiga, vemos que é o mesmo na primeira e segunda estrofas: ababcca. Na segunda estrofa a rima b é -al, e nesta primeira estrofa é preciso corrixir no v. 2 (tan[s]) engadindo un -s para emparellala coa rima do v. 4, en que á súa vez hai que corrixir u por n, para que a rima b sexa -ans nos dous versos. Atopámonos así co rimante Mans. Mais, cal é o valor desta palabra no texto? Non resulta doado decantarse por un significado concreto, xa que este varía dependendo do étimo a que se remonte o termo. Tendo en conta o sentido que dun xeito global parece percibirse no texto, relativo ao desexo do protagonista de tornar ao seu lugar patrio, e considerando as circunstancias históricas reais do autor, que viviu en persoa a experiencia do desterro, coidamos que o termo provenzal mans podería remontarse ao baixo latín MANSUS, termo empregado abundantemente na literatura documental medieval en latín (cf. Du Cange, Glossarium Mediae et infimae latinitatis, s. v.) cun valor equivalente ao de villa, ou sexa, ‘casa de campo, coas construcións adxacentes e terreos de cultivo’, ou tamén equivalente a vicus, isto é, ‘lugar, aldea’. Coidamos, en definitiva, que calquera das dúas interpretacións aquí propostas melloran as existentes. En todo o caso, son máis xustificábeis desde o punto de vista paleográfico. Referencias Bibliográficas: CANCIONEIRO DA BIBLIOTECA NACIONAL (Colocci-Brancuti) Cod. 10991, I. (1992). Reprodução facsimilada. Edição de L. f. Lindley Cintra (Lisboa: Imprensa Nacional / Casa da Moeda / Biblioteca Nacional). HEUR, J. M. d’ (1973): Troubadours d’oc et troubadours galiciens-portugais. Recherches sur quelques échanges dans la littérature de l’Europe au Moyen Age (París: Fundação Calouste Gulbenkian). LOPES, G. Videira (1994): A sátira nos cancioneiros medievais galego-portugueses (Lisboa: Imprensa Universitaria / Editorial Estampa). RIQUER, M. de (1993): Los trovadores. Historia literaria y textos (Barcelona: Ariel). Este valor, que se podería xeneralizar como ‘propiedades, posesións’, cadra perfectamente co sentido xeral da composición, do que se desprende o desexo do protagonista de ‘tornar / a Sousa’, ao seu lugar familiar, agora que as súas posesións, os seus mansos (mans) xa estaban libres (quites). Aínda sería posíbel outra interpretación do termo mans se o entendésemos como un substantivo derivado regresivo de mandar (<MANDARE), co sentido de ‘ordes, 168 169
https://openalex.org/W3208526359	https://europepmc.org/articles/pmc8583406?pdf=render	English	null	Implementation and Knowledge of the Clinical Practice Guide for Palliative Care in the Ecuadorian Primary Care Level	International journal of environmental research and public health/International journal of environmental research and public health	2,021	cc-by	9,103	Citation: Rodríguez Quintana, T.; Dávalos-Batallas, V.; Vargas-Martínez, A.-M.; López, L.; Bonilla-Sierra, P.; Lomas-Campos, M.M.; Leon-Larios, F. Implementation and Knowledge of the Clinical Practice Guide for Palliative Care in the Ecuadorian Primary Care Level. Int. J. Environ. Res. Public Health 2021, 18, 11573. https://doi.org/10.3390/ ijerph182111573 Academic Editors: María José Cabañero-Martínez, Manuel Fernández Alcántara and Rafael Montoya Juárez Keywords: clinical practice guide; palliative care; knowledge Received: 23 September 2021 Accepted: 31 October 2021 Published: 4 November 2021 Int. J. Environ. Res. Public Health 2021, 18, 11573. https://doi.org/10.3390/ijerph182111573 Article Implementation and Knowledge of the Clinical Practice Guide for Palliative Care in the Ecuadorian Primary Care Level intana 1 , Viviana Dávalos-Batallas 1, Ana-Magdalena Vargas-Martínez 2,* , cia Bonilla-Sierra 1 , María-de-las-Mercedes Lomas-Campos 2 and Fatima Leon-Larios 2 Tamara Rodríguez Quintana 1 , Viviana Dávalos-Batallas 1, Ana-Magdalena Vargas-Martínez 2,* , Lucelly López 3, Patricia Bonilla-Sierra 1 , María-de-las-Mercedes Lomas-Campos 2 and Fatima Leon-Larios 2 Tamara Rodríguez Quintana 1 , Viviana Dávalos-Batallas 1, Ana-Magdalena Vargas-Martí Lucelly López 3, Patricia Bonilla-Sierra 1 , María-de-las-Mercedes Lomas-Campos 2 and 1 Department of Health Sciences, Universidad Técnica Particular de Loja (UTPL), Loja 110107, Ecuador; trodriguez3@utpl.edu.ec (T.R.Q.); vddavalos@utpl.edu.ec (V.D.-B.); pbonilla65@utpl.edu.ec (P.B.-S.) 2 N i D S h l f N i Ph i h d P di U i i f S ill y * Correspondence: avargas5@us.es Abstract: Ecuador assumed the commitment of including Palliative Care (PC) in its health policies. In 2014, the Ministry of Public Health (Ministerio de Salud Pública, MSP) approved the Clinical Practice Guide for Palliative Care (Guía de Práctica Clínica sobre Cuidados Paliativos, GPCCP), with application at the national level, as a mandatory internal regulation in all institutions belonging to the National Health System. In 2021, there is no evidence about the degree of implementation. The objective was to evaluate the implementation (I) of the GPCCP guide and the knowledge (C) of the health personnel working in the Zone 7 Health Centers (HCs). This is a cross-sectional, descriptive, and prospective study. A total of 292 professionals were interviewed: managers (38), physicians (150), and nurses (104). Three surveys based on the GPCCP guide were elaborated: one for the implementation, which was applied to the individuals in charge, and the others to assess the health professionals’ knowledge. The SPSS program was used, version 25. In the three groups, more than half of the participants had no training in PC, 91.2% of the HCs have the GPCCP guide, there is PC medical history (MH) in 38.2%, and morphine is used in 14.7%. The implementation of the GPCCP guide was inadequate in 52.9% of the cases. Only 25% treat the agony symptoms and 30%, delirium; 4.4% acknowledge the use of morphine in dyspnea, and 13.3% identify the subcutaneous route as the ﬁrst choice for hydration at the end-of-life phase. Strategies to implement the GPCCP guide and to improve the health personnel’s knowledge must be implemented in Zone 7 centers. International Journal of Environmental Research and Public Health International Journal of Environmental Research and Public Health 1. Introduction Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- iations. The view of palliative care (PC) in the world has changed over the years. It is acknowl- edged that it relieves suffering, controls symptoms, and improves quality of life and care quality [1]. Comprehensive care for patients with diseases that restrict life during their evolution offers an effective outcome–cost ratio for the patient, the family, and the health systems [2]. The Political Statement of the High-Level Meeting of the United Nations’ General Assembly on prevention and control of non-communicable diseases (2014) acknowledges the need to improve access to palliative care. The recommendation is to develop and implement palliative care policies to strengthen health systems [3]. Cost-effective and equitable palliative care services must be integrated in all levels, with an emphasis on primary, community, and home care [4]. Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). p y y The World Health Organization (WHO) proposes four key elements as a strategy to improve PC, namely: policy, education, availability of medication, and implementation of services [5]. The policy is considered indispensable to satisfy the other three elements; Int. J. Environ. Res. Public Health 2021, 18, 11573. https://doi.org/10.3390/ijerph182111573 https://www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2021, 18, 11573 2 of 14 however, the existence of national PC laws or plans does not ensure that there will be more PC services, at least in Latin America (LA). however, the existence of national PC laws or plans does not ensure that there will be more PC services, at least in Latin America (LA). In Latin America, with a population of more than 600 million inhabitants, 75% of the population dies due to non-communicable diseases. In 2018, more than 3.5 million people living with some severe health problems were reported, reason why it is indispensable to continue developing this care. Despite this increase, the care provided is still insufﬁcient and unequal [6,7]. Currently, coverage of PC needs in Latin America is only 7.6% and only 3.5% in Ecuador [8]. 2. Materials and Methods This was a cross-sectional, descriptive, and cross-sectional study. It was developed during the 2019–2020 period in Health Zone 7, which encompasses three provinces: Loja, El Oro, and Zamora Chinchipe. 1. Introduction Important changes have been implemented in the public and health policies, favoring the integration of PC in the National Health System and assisting in the creation of public policies aimed at developing and implementing PC in Ecuador. In 2014, and as part of these strategies, the Ministry of Public Health approved the Clinical Practice Guide for Palliative Care and established its application in the entire National Health System as a mandatory internal regulation [9]. In relation to the training of future health professionals, although inclusion of the PC component in the Health Sciences courses is mandatory, previous studies have found that it is still necessary to advance in training, as a number of knowledge gaps have been identiﬁed [10,11]. The purpose of this article is to assess the implementation degree of the MSP Palliative Care Guide in Zone 7 Health Centers, as well as the level of knowledge about palliative care of the health personnel working in these centers, six years after approval of the Palliative Care National Plan in Ecuador 2.1. Population and Sample There are 1554 health centers that depend on the Public Health Ministry (Ministerio de Salud Pública, MSP) in Ecuador. Health Zone 7, comprised by the provinces of Loja, El Oro, and Zamora Chinchipe, has 153 health centers, 66 of them in urban areas. The total population of this zone is 1,341,334 inhabitants who receive care, some in MSP health units and others in units from the public health network such as the different health centers belonging to the various social insurance systems in the country. To assess the implementation of the GPCCP guide, the administrative managers of 38 health centers in urban areas where the Family and Community Medicine specialization students were doing their internships were interviewed. To assess the health personnel’s level of knowledge about palliative care, all the administrative personnel, physicians, and nurses who met the inclusion criteria were interviewed. The following were established as inclusion criteria: for the health professionals in managerial positions, holding such positions for at least three months; and, for the physicians and nurses, working in the health center for at least three months; there were no exclusion criteria. Non-probabilistic convenience sampling was performed. 2.2. Ethical Considerations All the participants were informed about the study objectives and were asked to sign a written informed consent of their participation in the study. They were guaranteed anonymity of the information provided and data conﬁdentiality. Approval was obtained from the Committee of Ethics in Research with Human Beings of UTPL under Edict No. UTPL-CEISH-2020-RI05; in addition, authorization from the Zone 7 Health Coordina- tion Ofﬁce and the participants’ written and signed informed consent were also obtained. 3 of 14 Int. J. Environ. Res. Public Health 2021, 18, 11573 2.4. Instrument Information was collected about sociodemographic and labor variables, as well about experience and training in palliative care. Speciﬁcally, the variable related to the training hours, which was asked categorically, was transformed into a numerical variable for certain analyses that are detailed later in this article (“0 h” = 0; “less than 20 h” = 10; “between 20 and 50 h” = 35; “between 50 and 100 h” = 75; and “more than 100 h” = 100). Questions were asked about the existence of the GPCCP guide, the annual training plan including the theme, number of training sessions, written performance protocol for coordination with the reference hospital, palliative care medical history, essential PC medications, population that required palliative care, trained interdisciplinary team, time available for the PC team to evaluate the ﬁrst appointment of a patient with palliative needs in external consultation, coordination between the different assistance services to ensure care continuity, opioid availability, and other techniques that can help relieve the symptoms. A number of variables were studied, such as: instruments used in PC, medication use, criteria to include patients in the program, number of symptoms that are assessed with the Edmonton scale, use of a prognostic scale to predict survival in a patient on PC, use of another scale to assess pain, criteria to include a patient with an advance-stage disease and a life expectancy of less than one year in the PC program, ICD-10 code used for palliative care, criteria to refer a patient, use of morphine at the end-of-life phase, ﬁrst choice route for home hydration, aspects included in the psychosocial evaluation of the patient and the family, information provided by the family physician about the terminal phase and the need to transfer the patient to palliative care, about the changes in evolution of the disease, about palliative sedation and identiﬁcation of pathological grief, among others. 2.3. Procedures and Data Collection 2.3. Procedures and Data Collection 2.3. Procedures and Data Collection To assess the implementation of the Palliative Care Guide, an ad hoc survey was applied, based on the guidelines set forth in the MSP guide, consisting of 27 questions: 11 collecting general information and the rest being speciﬁc about implementation of the guide. This instrument was applied to the health center administrative manager, and was self-administered with the interviewer present. From the speciﬁc questions, inadequate implementation was only considered when one of the answers was negative. Whereas, to assess the knowledge about PC among the health personnel, separate surveys for the physicians and for the nursing personnel were elaborated. Both included 34 items, 9 for collecting general data of the interviewee and the rest to assess knowledge about palliative care. The two surveys were based on the guidelines set forth in Ecuador’s GPCCP guide, designed by the research authors and applied in the health centers. 3. Results Descriptive characteristics of the sample are shown in Table 1. The total of subjects studied was 292, divided as follows: 38 health center administrative managers, 150 physi- cians, and 104 nurses. cians, and 104 nurses. Table 1. Characteristics of the sample (n = 292). Variables Total Managers (n = 38) Nurses (n = 104) Physicians (n = 150) Sociodemographic Gender [female] 199 (69.1) 19 (55.9) 83 (79.8) 97 (64.7) * Age x(SD) 37.49 (10.80) 35.32 (9.52) 37.38 (10.50) 38.05 (11.27) Canton Loja 20 (52.63) 20 (52.6) N.A. N.A. Machala 3 (7.89) 3 (7.89) N.A. N.A. Pasaje 1 (2.63) 1 (2.63) N.A. N.A. Yacuambi 2 (5.26) 2 (5.26) N.A. N.A. Yantzaza 2 (5.26) 2 (5.26) N.A. N.A. Zamora 5 (13.16) 5 (13.16) N.A. N.A. Zaruma 5 (13.16) 5 (13.16) N.A. N.A. 3. Results Cycle I 13 (6.19) 13 (39.4) 0 (0.0) 0 (0.0) * IV 11 (5.24) 2 (6.1) 3 (4.3) 6 (5.6) V 9 (4.29) 1 (3.0) 4 (5.8) 4 (3.7) VI 120 (57.14) 8 (24.2) 43 (62.3) 69 (63.9) IX 57 (27.14) 9 (27.3) 19 (27.5) 29 (26.9) Parrish Catamayo 35 (13.89) 1 (2.6) 9 (11.0) 25 (18.9) * Machala 56 (22.22) 1 (2.6) 25 (30.5) 30 (22.7) Pasaje 18 (7.14) 1 (2.6) 7 (8.5) 10 (7.6) Zamora 53 (21.04) 3 (7.9) 19 (23.2) 31 (23.5) Others 130 (35.71) 32 (84.2) 22 (26.8) 36 (27.3) Professional training and experience Professional experience (years) x (SD) 11.08 (9.54) 9.73 (8.61) 10.19 (8.53) 12.02 (10.33) Time as administrative manager (years) x (SD) 3.34 (7.24) 3.34 (7.24) NA NA Experience in PC [Yes] 116 (40.28) 13 (38.2) 47 (45.2) 56 (37.3) Experience in PC (years) 1.89 (3.74) 1.90 (4.39) 2.54 (4.91) 1.46 (2.43) Training in PC (hours) 0 h 17 (8.72) 2 (7.7) 0 (0.0) 15 (16.0) * Less than 20 h 102 (52.31) 14 (53.8) 50 (66.7) 38 (40.4) Between 20 and 50 h 53 (27.18) 7 (26.9) 17 (22.7) 29 (30.9) Between 50 and 100 h 13 (6.67) 1 (3.8) 4 (5.3) 8 (8.5) More than 100 h 10 (5.13) 2 (7.7) 4 (5.3) 4 (4.3) Graduate Courses in PC [Yes] 15 (7.11) 2 (7.1) 3 (4.0) 10 (9.3) Higher Education in PC [Yes] 149 (65.07) 10 (34.5) 32 (38.6) 38 (32.5) Continuing Education in PC [Yes] 31 (14.09) 3 (10.3) 11 (13.6) 17 (15.5) Higher Education in PC + Continuing Education in PC [Yes] 15 (7.32) 4 (15.4) 8 (10.7) 3 (2.9) Note: x (SD), the mean values and standard deviation are shown in between round brackets when the variable is numerical. The frequency and percentage are shown in between round brackets when the variable is categorical. * and represent 1% and 5% statistically signiﬁcant differences between the variables by professional category (administrative clerk, nurse and physician). N.A. (Not applicable). In relation to the sociodemographic characteristics, women prevailed in the three hi h d h h lf f h i i hi i l Table 1. Characteristics of the sample (n = 292). Note: x (SD), the mean values and standard deviation are shown in between round brackets when the variable is numerical. The frequency and percentage are shown in between round brackets when the variable is categorical. 2.5. Statistical Analyses In data description, absolute and percentage frequencies were used for the qualitative variables; and means and standard deviations were employed for the quantitative variables. In addition, a comparison of means and frequencies by professional group (managers, physicians, and nurses) was carried out, showing statistical signiﬁcance after analyses by means of the t-test and ANOVA or Pearson’ chi-square tests, depending on the type of variable. Likewise, an analysis by means of Pearson’s chi-square was performed to compare the physicians’ and nurses’ knowledge about Palliative Care regarding certain aspects of the Clinical Practice Guide. The relationship between this knowledge and the years of general professional experience in palliative care and training in this area for each professional group (physicians and nurses) was studied using t-test, except for the variable related to ﬁrst choice hydration route for patients at the end-of-life phase in their homes for which the ANOVA analysis was used. Data processing was performed in R statistical software (version R-4.0.5, Free Software Foundation’s GNU General Public License: https://www.r-project.org/about.html, access date: 18 May 2021). A 5% statistical signiﬁcance level was adopted in all the analyses. 4 of 14 Int. J. Environ. Res. Public Health 2021, 18, 11573 3. Results * and represent 1% and 5% statistically signiﬁcant differences between the variables by professional category (administrative clerk, nurse and physician). N.A. (Not applicable). Note: x (SD), the mean values and standard deviation are shown in between round brackets when the variable is numerical. The frequency and percentage are shown in between round brackets when the variable is categorical. * and ** represent 1% and 5% statistically signiﬁcant differences between the variables by professional category (administrative clerk, nurse and physician). N.A. (Not applicable). In relation to the sociodemographic characteristics, women prevailed in the three groups, which represented more than half of the participants, reaching approximately 80% in the group of nurses. The overall mean age of the sample was around 37 years old. Among the administrative managers of the centers participating in the study, generalist physicians were a majority with 76.5%, and only 14.7% were specialists in Family and Community Medicine. Among the physicians, 38% were generalists, 28% were specialists in Family and Community Medicine, and 14.67% were surgeons. 5 of 14 Int. J. Environ. Res. Public Health 2021, 18, 11573 The general professional experience presented a mean of 11 years, exceeding a mean of 12 years in the group of physicians. Less than 50% of the subjects interviewed acknowl- edged having experience in palliative care, and the nurses were those who most reported having had experience and whose mean time of experience in this care ﬁeld was longer. Regarding the type of training received, medical professionals were those who most underwent graduate training and continuing education, whereas nursing professionals asserted having mostly undergone university undergraduate training. 34.5% of the ad- ministrative managers who stated having undergone training did so at the undergraduate level, 10.3% in continuing education and only 7% at the graduate level. In the case of the physicians, 33% state having undergone undergraduate training, 15.5% continuing education and only 9.3% graduate education, whereas 38.6% of the nurses underwent PC training at the undergraduate level, 13.6% in continuing education, and only 4% at the graduate level. In addition, 52.3% attended less than 20 h of training in this topic (Table 1). 3.1. Implementation of the Clinical Practice Guide for Palliative Care in the HCs 3.1. Implementation of the Clinical Practice Guide for Palliative Care in the HCs When analyzing the criteria to assess the implementation of the Clinical Practice Guide for Palliative Care based on the information provided by the administrative managers (Table 2), it was found that 91.18% of the health centers have this Guide; whereas there is a training program that includes PC only in 35.29% of the cases. There is a written performance protocol for coordination with the reference hospital in 36.36% of the HCs. It is noteworthy that there is PC medical history only in 38.24% of the cases. Table 2. Implementation of the Clinical Practice Guide for Palliative Care. Table 2. Implementation of the Clinical Practice Guide for Palliative Care. Table 2. Implementation of the Clinical Practice Guide for Palliative Care. N % CPG for Palliative Care [Yes] 31 91.18 Annual training program in the HC (including PC) [Yes] 12 35.29 Number of annual training sessions 1 (SD: 1.98) Coordination protocol with reference hospital [Yes] 12 36.36 Medical History in PC [Yes] 13 38.24 Availability of essential PC medications suggested by the WHO [Yes] Paracetamol/Ibuprofeno/Diclofenac/Ketorolac/Naproxen 33 97.06 Codeine/Tramadol 18 52.94 Morphine/Buprenorphine 5 14.71 Butylscopolamine/Hyoscine Bromide 24 72.73 Lactulose/Glycerol 25 73.53 Loperamide 1 3.03 Metroclopramide 32 94.12 Furosemide/Spironolactone 27 81.82 Omeprazole 33 97.06 Metronidazole 34 100 Prednisone/Dexamethasone 26 78.79 Fluconazole 34 100 Fenitoine/Phenobarbital/Clonazepam/Carbamazepine/Valproic Acid/Alprazolam/Diazepam 27 79.41 Amitriptyline 9 26.47 Haloperidol 7 21.21 Salbutamol 32 94.12 Opioid availability [Yes] 14 41.18 HC population that required PC (2018) Less than 50 31 93.94 Between 50 and 100 1 3.03 Between 100 and 150 1 3.03 More than 150 0 0.00 Interdisciplinary PC team Physician 19 57.58 Nurse 11 34.38 Psychologist 6 18.75 Physiotherapist 2 6.25 Social Worker 6 18.75 Care time available, ﬁrst appointment in external consultation 15 min 1 3.03 30 min 13 39.39 45 min 7 21.21 60 min 12 36.36 Opioid availability [Yes] HC population that required PC (2018) Care time available, ﬁrst appointment in external consultation 6 of 14 Int. J. Environ. Res. Public Health 2021, 18, 11573 Table 2. Cont. N % Coordination between different 24 h care services [Yes] 17 51.52 Other techniques to be considered Hypnosis 4 12.50 Relaxation therapy 17 53.12 Aromatherapy 1 3.12 Homeopathy 1 3.12 Acupuncture 7 21.88 Ozone therapy 1 3.12 Reiki 1 3.12 Table 2. Cont. 3.1. Implementation of the Clinical Practice Guide for Palliative Care in the HCs Coordination between different 24 h care services [Yes] Other techniques to be considered Coordination between different 24 h care services [Yes] Other techniques to be considered In relation to the essential PC medications that must be included in the basic medica- tion chart in primary care units, it was found that there is Tramadol in 52.9% of the HCs and, regarding Morphine, it is present in only 14.7%. Other medications were also found in proportions below 50%, such as Amitriptyline (26.5%) and Haloperidol (21.2%). However, there is opioid availability in slightly more than 40% of the health centers. The interdisciplinary teams for this type of care have physicians trained in PC (57.58%), with trained nurses (34.38%), a psychologist, and a social worker (18.75%) of the teams and only one physiotherapist (6.25%); whereas only 36.36% acknowledges that the regulated time available for the team to evaluate the ﬁrst appointment is 60 min, according to the MSP’s PC care standard. Slightly more than half of the health centers coordinate care with other secondary care level hospital institutions 24 h a day, considering that the study area does not have tertiary-level hospitals or 24 h home care teams. y p The following were mentioned among the difﬁculties acknowledged to implement the MSP’s palliative care model ﬂowchart, with integration of all the resources: lack of training in the theme, deﬁcit in human resources and medications/medical inputs, limited time availability to provide care, and the physicians’ instability in contractual terms, including professionals who comply with the one-year rural service and others with ﬁxed contracts. 3.2. Knowledge and Handling of the CPG Guide for PC by Physicians and Nurses Approximately 50% of the physicians use both the ESAS and Pfeiffer’s and Karnofsky’s scales when approaching patients in palliative care. 24.83% uses some survival prediction scale; whereas 60.42% of the physicians interviewed use a scale to assess pain and more than one-third employ survival time and the symptoms as criteria to include a patient with a chronic disease in an advanced stage and lower life expectancy in the program. Nearly one-third of the physicians know the ICD-10 code corresponding to palliative care and 62.5% refer the patient to a PC unit when diagnosing an incurable disease and shortened life prognosis, always based on the disease, the life prognosis, and the expected evolution (Table 3). The end-of-life symptoms that are treated by almost all the physicians are severe pain, constipation, and vomiting. It draws the attention that only 20.91% treat agony symptoms and 25.89%, delirium. In relation to the use of morphine at the end-of- life phase, 76.3% use it for pain; however, only 5.41% acknowledge the need to use it in dyspnea and 27.12% state using it for palliative sedation although it is a non-hypnotic analgesic drug. In relation to the ﬁrst-choice route to hydrate patients at the end-of-life phase in their homes, 71.32% prefer the intravenous route and only 16.96% has a preference for the subcutaneous route, which is the one recognized and recommended in the CPG guide (Table 3). This is similar among nurses, referring to the intravenous route as the ﬁrst choice for home rehydration. The differences between physicians and nurses regarding the proportions in their answers were statistically signiﬁcant. Int. J. Environ. Res. Public Health 2021, 18, 11573 7 of 14 Table 3. Use and knowledge about instruments, inclusion criteria, and prescription in PC by the physicians. 3.2. Knowledge and Handling of the CPG Guide for PC by Physicians and Nurses 8 of 14 Int. J. Environ. Res. Public Health 2021, 18, 11573 Table 4. Comparison between physicians and nurses regarding knowledge/beliefs related to palliative care. Physicians Nurses p-Value n % N % Knowledge/Beliefs First choice hydration route for patients at the end-of-life phase in their homes Oral 37 31.09 15 14.4 0.003 Subcutaneous 19 16.96 3 2.9 0.001 Intravenous 92 71.32 86 82.7 0.003 Morphine is the standard used to compare the analgesic effect of other opioids [Yes] 101 68.71 51 49.0 0.002 Adjuvant therapies are important in pain management [Yes] 132 89.80 90 87.4 0.551 Somnolence associated with electrolyte imbalance can reduce the effect of sedation [Yes] 73 49.66 52 50.5 0.898 People who take opioids should adopt certain measures to improve bowel elimination [Yes] 116 78.91 60 58.3 0.001 The drugs that can cause respiratory depression are appropriate to treat severe dyspnea [Yes] 32 21.77 15 14.6 0.154 At high doses, codeine causes more nausea and vomiting than morphine [Yes] 68 46.26 48 47.1 0.901 Dolantine is not an effective analgesic in the control of chronic pain [Yes] 29 19.73 25 24.5 0.369 Note: Pearson’ chi-square tests were used to compare the different variables related to knowledges/beliefs between doctors and nurses. Numbers in bold indicates that p-value is less than 0.05, being statistically signiﬁcant. Comparison between physicians and nurses regarding knowledge/beliefs related to palliative care. -square tests were used to compare the different variables related to knowledges/beliefs between doctors and nurses. ndicates that p-value is less than 0.05, being statistically signiﬁcant. When investigating how physicians assess the psychosocial sphere of the patients in palliative care and of the families, it was found that approximately 90% address the impact of the disease, although less than 70% evaluate the spiritual resources. More than 90% of the interviewees inform the family about the patient’s terminal phase and about the transfer to palliative care, as well as about evolution and palliative sedation. 67.59% is able to recognize pathological grief. However, around 25% of the physicians feel trained in the diagnosis and management of urgencies in PC and 55.78% feel trained to provide psychosocial and spiritual support to patients and families (Table 5). Table 5. Management of the psychosocial sphere by physicians. 3.2. Knowledge and Handling of the CPG Guide for PC by Physicians and Nurses Physicians n % n % Instruments ESAS [knows it] 19 30.16 Barthel [knows it] 27 30.00 [Uses it] 31 49.21 [Uses it] 30 33.33 [Knows it and uses it] 13 20.63 [Knows it and uses it] 33 36.67 NECPAL [knows it] 24 35.82 Symptoms assessed by the Edmonton scale [Uses it] 31 46.27 0–3 28 22.95 [Knows it and uses it] 12 17.91 4–6 60 49.18 Karnofsky [knows it] 46 42.59 7–10 34 27.87 [Uses it] 22 20.37 Use of the survival prognosis scale [Yes] 36 24.83 [No] 52 35.86 [Knows it and uses it] 40 37.04 [Does not know any scale] 57 39.31 Pfeiffer [knows it] 18 26.87 Use of a scale to assess pain [Yes] 87 60.42 [No] 35 24.31 [Uses it] 33 49.25 [Does not know any scale] 22 15.28 [Knows it and uses it] 16 23.88 Inclusion criteria Patient with advanced-stage disease and life expectancy of less than one year [it does include it] 124 83.78 When a patient is referred to a PC unit Criterion/Criteria to include it When diagnosing an incurable disease and shortened life prognosis, always based on the disease, the life prognosis and the evolution expected 90 62.50 Time 15 10.87 The patient and/or family requires so 6 4.17 Presence of symptoms 11 7.97 Progressive and irreversible deterioration, with increase in the number of complications and/or needs 42 29.17 Time and presence of symptoms 112 81.16 In agony 3 2.08 ICD-10 code Never 3 2.08 Z21 11 8.27 Difﬁculty referring a patient to a PC unit Z50.4 10 7.52 Yes, I don’t know any unit nearby 50 34.72 Z51.5 96 72.18 Yes, the family gets scared at the word “palliative” 11 7.64 Z52.1 16 12.03 Yes, I’d rather be treated in PHC 11 7.64 No, I refer the patient so that they can be followed-up in both services (PHC and Hospital) 61 42.36 No, I refer the patient so that they are followed-up in a PC unit 11 7.64 Drug prescription Dyspnea 51 44.35 Agony 23 20.91 Delirium 29 25.89 Use of morphine as end-of-life treatment for: Vomiting 74 63.79 Pain 103 76.3 Constipation 80 67.23 Palliative sedation 32 27.12 Severe pain 81 66.39 Dyspnea 6 5.41 Asthenia-Anorexia-Cachexia 38 32.48 Pain and dyspnea 37 31.36 Regarding the knowledge investigated in physicians and nurses, it can be seen (Tables 3 and 4) that there are no statistically signiﬁcant differences in their answers, except for the assertions related to: “Morphine is the standard used to compare the analgesic effect of other opioids” and “People who take opioids should adopt certain measures to improve Regarding the knowledge investigated in physicians and nurses, it can be seen (Tables 3 and 4) that there are no statistically signiﬁcant differences in their answers, except for the assertions related to: “Morphine is the standard used to compare the analgesic effect of other opioids” and “People who take opioids should adopt certain measures to improve bowel elimination”, where the physicians showed greater knowledge. 3.2. Knowledge and Handling of the CPG Guide for PC by Physicians and Nurses N % Training to provide psychosocial and spiritual support to the patient and the family [Yes] 82 55.78 Training to diagnose and manage urgencies in PC [Yes] 37 25.17 Psychosocial assessment (aspects included) Impact of the disease [Yes] 131 89.73 Coping styles [Yes] 116 81.12 Spiritual resources [Yes] 97 68.31 Information provided to the family About the terminal phase and the need to refer to PC 139 94.56 About the changes in evolution of the disease 142 96.6 About palliative sedation 133 91.10 Identiﬁcation of pathological grief Clinical case about pathological grief (situation that this person is going through) Depression 43 29.66 Anxiety 4 2.76 Pathological grief 98 67.59 Table 5. Management of the psychosocial sphere by physicians. Table 5. Management of the psychosocial sphere by physicians. Int. J. Environ. Res. Public Health 2021, 18, 11573 9 of 14 Regarding the nurses’ knowledge about different aspects included in the GCPCP guide, it is worth noting that 48% relates PC to deterioration or worsening of the clinical condition, which is associated with the fact that 73% consider that the pain treatment method is based on disease extension and not on its severity, and that only 53% of the interviewees consider that the PC philosophy is compatible with active treatments. In this line, they consider addiction as a problem for the use of morphine in patients who require long-term pain treatment in the PC context. Seventy-ﬁve percent of the nurses consider that it is appropriate to use a placebo in pain treatment. On the other hand, less than 50% acknowledge the impact of the disease and of fatigue on the manifestation of physical pain. More than 50% and 75% of the interviewees ignore the secondary effects of codeine and of meperidine, respectively. Regarding the psychoemotional aspects and team care, it draws the attention that 60% believe that the accumulation of losses makes development of the Burnout syndrome inevitable in health personnel and that 34% did not answer this question. In relation to grief or loss, 72% consider that they are easier to solve when there is a distant or conﬂictive relationship. 3.3. Relationship between General Knowledge about the Different Aspects of PC and the Physicians’ and Nurses’ Training and Experience 3.3. Relationship between General Knowledge about the Different Aspects of PC and the Physicians’ and Nurses’ Training and Experience General professional experience showed a statistically signiﬁcant difference in the group of physicians with the knowledge related to “Morphine is the standard used to compare the analgesic effect of other opioids”, “Adjuvant therapies are important in pain management”, and “The drugs that can cause respiratory depression are appropriate to treat severe dyspnea”. Among the nurses, a statistically signiﬁcant relationship was only observed for the following assertion: “Meperidine is not an effective analgesic in the control of chronic pain” (Table 6). Table 6. Physicians’ and nurses’ knowledge according to general professional experience and experience in palliative care. Professional Experience (Years) Experience in Palliative Care (Years) Medicine Mean (SD) p-Value Nursing Mean (SD) p-Value Medicine Mean (SD) p-Value Nursing Mean (SD) p-Value Morphine is the standard used to compare the analgesic effect of other opioids 0.016 0.466 0.368 0.171 Yes No Prefers not to answer 11.67 (10.22) 5.93 (4.86) 15.42 (11.26) 9.69 (7.79) 9.22 (7.98) 11.80 (10.09) 1.63 (2.45) 0.58 (0.79) 1.38 (2.81) 2.96 (5.07) 3.43 (6.62) 0.91 (1.08) Adjuvant therapies are important in pain management 0.005 0.300 0.861 0.580 Yes No Prefers not to answer 11.11 (9.77) 12.00 (0.00) 20.43 (12.58) 9.77 (8.06) 7.00 (0.00) 13.75 (11.82) 1.52 (2.54) 1.00 (0.00) 1.15 (1.34) 2.41 (4.25) 0.00 (0.00) 1.20 (1.87) Somnolence associated with electrolyte imbalance can reduce the effect of sedation 0.851 0.311 0.638 0.021 Yes No Prefers not to answer 11.96 (10.64) 11.41 (10.09) 12.79 (10.31) 10.35 (8.00) 8.33 (6.86) 12.00 (11.10) 1.68 (2.79) 1.24 (2.26) 1.32 (1.76) 3.34 (5.07) 1.30 (2.27) 0.68 (1.00) People who take opioids should adopt certain measures to improve bowel elimination 0.129 0.518 0.430 0.661 Yes No Prefers not to answer 11.39 (10.24) 10.00 (7.70) 16.05 (11.36) 10.28 (8.60) 7.93 (6.15) 11.14 (9.53) 1.34 (2.26) 1.89 (3.79) 2.05 (2.67) 2.56 (4.89) 1.50 (1.29) 2.04 (3.12) ans’ and nurses’ knowledge according to general professional experience and experience in palliative care. Table 6. Physicians’ and nurses’ knowledge according to general professional experience and experien Int. J. Environ. Res. Public Health 2021, 18, 11573 10 of 14 Table 6. Cont. 3.3. Relationship between General Knowledge about the Different Aspects of PC and the Physicians’ and Nurses’ Training and Experience Professional Experience (Years) Experience in Palliative Care (Years) Medicine Mean (SD) p-Value Nursing Mean (SD) p-Value Medicine Mean (SD) p-Value Nursing Mean (SD) p-Value The drugs that can cause respiratory depression are appropriate to treat severe dyspnea 0.006 0.437 0.422 0.246 Yes No Prefers not to answer 7.38 (6.12) 14.25 (11.34) 11.49 (10.03) 9.10 (6.91) 9.31 (8.27) 11.58 (9.47) 0.97 (1.66) 1.57 (2.21) 1.72 (3.26) 1.46 (2.85) 3.02 (5.29) 1.63 (2.28) At high doses, codeine causes more nausea and vomiting than morphine 0.943 0.515 0.346 0.194 Yes No Prefers not to answer 12.31 (9.55) 11.58 (11.06) 11.83 (11.16) 9.39 (7.75) 9.93 (8.10) 11.50 (9.66) 1.75 (3.10) 1.61 (1.73) 1.08 (1.67) 3.02 (5.07) 1.23 (2.77) 1.57 (2.33) Dolantine is not an effective analgesic in the control of chronic pain 0.650 0.028 0.225 0.599 Yes No Prefers not to answer 11.75 (9.82) 10.71 (9.73) 12.64 (10.85) 7.80 (6.63) 6.38 (6.29) 11.92 (9.30) 1.32 (1.77) 0.91 (1.28) 1.78 (2.93) 1.95 (2.94) 1.38 (1.94) 2.57 (4.76) Note: p-value refers to the association studied between professional experience measured in years and the rest of the variables described in the ﬁrst column of the table, as well as between the experience in palliative care and the rest of the variables, for each healthcare professional (doctors and nurses) separately. The statistical analysis used was an ANOVA analysis. Numbers in bold indicates that p-value is less than 0.05, being statistically signiﬁcant. Table 6. Cont. Professional Experience (Years) Experience in Palliative Care (Years) Note: p-value refers to the association studied between professional experience measured in years and the rest of the variables described in the ﬁrst column of the table, as well as between the experience in palliative care and the rest of the variables, for each healthcare professional (doctors and nurses) separately. The statistical analysis used was an ANOVA analysis. Numbers in bold indicates that p-value is less than 0.05, being statistically signiﬁcant. Regarding the speciﬁc professional experience in PC, it only showed a statistically signiﬁcant relationship among the nurses for the following assertion: “Somnolence associ- ated with electrolyte imbalance can reduce the effect of sedation”, noticing more in-depth knowledge as the years of experience increase (Table 6). Training in PC did not prove to be a determining factor in the improvement of knowl- edge about the different aspects contemplated in the GPCCP guide, both in physicians and in nurses. 4. Discussion This study intended to analyze the impact of implementing the Clinical Practice Guide for Palliative Care in Ecuador’s Health Zone 7, since its approval in 2014. To such end, physicians and nurses involved in palliative care were asked about their knowledge and professional experience. Likewise, the managers of the health centers involved were also included. Insufﬁcient implementation of the CPCCP guide is evidenced, despite approval of the 2015–2017 Palliative Care National Plan, where all health establishments are compelled to apply it, which does not necessarily lead to putting it into practice to ensure provision of the service. This phenomenon has been observed in several studies [12,13] and, perhaps, the most notorious is the one by the WHO in 2015, which concludes that, in order to have successful Palliative Care programs, universal access to the essential PC medications is re- quired, as well as generalized education and implementation, in line with the results found in this study. In addition, adequate funding and political commitment are required [10,14]. y q g p q Nine out of 10 administrative managers stated that the demand for PC among the population of their health centers was less than or equal to 50 individuals, although chronic diseases in advanced stages are the most prevalent in Ecuador [15]. This can be due to the fact that they are not properly identiﬁed or that they are only considered at the end of the chronic process, instead of at early stages, as currently recommended [16]. Clinical practice guides are important for decision-making, especially in primary care [17,18]; however, their insufﬁcient implementation precludes timely care to the pa- tients with progressive diseases at advanced stages, thus contributing to quality-of-life deterioration and to increased suffering [1]. Usually, one of the main causes of insufﬁcient implementation is lack of knowledge about their existence [10,19]; however, not only their knowledge by the health professionals involved must be weighted, but also receiving training on how to implement them and that access to the necessary resources is ensured. The fact that the participants possess little training in PC is an expected result, as other authors have acknowledged that, in the Ecuadorian context, training is still limited, both at the undergraduate and graduate levels, thus translating into low identiﬁcation of in- dividuals in need of this type of care and highlighting the need to maintain continuing education [10,11,20]. 3.3. Relationship between General Knowledge about the Different Aspects of PC and the Physicians’ and Nurses’ Training and Experience However, in most of the assertions, there is coincidence of a greater number of training hours with the fact of knowing these aspects, when compared to those who gave negative answers to these questions (Table 7). Table 7. Physicians’ and nurses’ knowledge according to hours of training in palliative care. Table 7. Physicians’ and nurses’ knowledge according to hours of training in palliative care. Training in Palliative Care (Years) Medicine Mean (SD) p-Value Nursing Mean (SD) p-Value Morphine is the standard used to compare the analgesic effect of other opioids Yes No Prefer not to answer 27.21 (27.57) 20.83 (15.94) 21.00 (23.37) 0.588 26.25 (23.16) 18.25 (16.80) 25.53 (33.62) 0.491 Adjuvant therapies are important in pain management Yes No Prefer not to answer 27.08 (26.83) 35.00 (0.00) 9.44 (10.74) 0.146 25.55 (26.19) 35.00 (0.00) 12.78 (8.33) 0.325 Somnolence associated with electrolyte imbalance can reduce the effect of sedation Yes No Prefer not to answer 25.91 (26.79) 22.41 (22.70) 28.81 (29.58) 0.690 25.24 (26.85) 26.94 (21.22) 17.67 (23.67) 0.522 Int. J. Environ. Res. Public Health 2021, 18, 11573 11 of 14 Table 7. Cont. Training in Palliative Care (Years) Medicine Mean (SD) p-Value Nursing Mean (SD) p-Value People who take opioids should adopt certain measures to improve bowel elimination Yes No Prefer not to answer 27.30 (27.53) 22.50 (13.69) 17.14 (21.10) 0.398 25.63 (28.58) 20.45 (20.67) 23.26 (19.98) 0.818 The drugs that can cause respiratory depression are appropriate to treat severe dyspnea Yes No Prefer not to answer 29.00 (29.89) 27.50 (24.15) 19.64 (26.31) 0.364 29.17 (34.43) 20.00 (15.28) 26.96 (29.39) 0.414 At high doses, codeine causes more nausea and vomiting than morphine Yes No Prefer not to answer 30.35 (26.65) 27.81 (27.32) 18.43 (23.91) 0.123 26.94 (26.76) 13.85 (9.39) 26.04 (27.27) 0.250 Dolantine is not an effective analgesic in the control of chronic pain Yes No Prefer not to answer 25.88 (26.53) 24.75 (21.24) 25.61 (27.86) 0.989 16.39 (16.70) 30.00 (18.59) 25.47 (28.84) 0.290 Note: p-value refers to the association studied between training in palliative care measured in years and the rest of the variables described in the ﬁrst column of the table for each healthcare professional (doctors and nurses) separately. The statistical analysis used was an ANOVA analysis. Table 7. Cont. Training in Palliative Care (Years) 4. Discussion It is necessary to develop a speciﬁc training program on palliative Int. J. Environ. Res. Public Health 2021, 18, 11573 12 of 14 12 of 14 care, both at the undergraduate and graduate levels for nurses and physicians, in order to optimize the care provided [21]. Even if Ecuador has recognized the palliative care medical specialty, it is necessary to keep advancing in the development of homogeneous contents that ensure quality care [20]. Not all the health professionals involved in palliative care who participated in the study had undergone training or had the medical specialty. Certain labor instability in these professionals to be able to do their job should be added to the aforementioned, since stability is sometimes hindered because they need to change work centers many times in a brief period of time [10]. Nevertheless, it was indeed noticed that those individuals who had received previous training knew more aspects of the guide and concepts related to palliative care, a reason why the knowledge acquired exerts a positive impact on palliative care management [22,23]. It is worrying that health personnel have insufﬁcient knowledge about PC. In the literature that was consulted, the importance of this personnel possessing adequate knowl- edge to improve quality of life and reduce anguish has been highlighted [19]. Few of them treat agony, dyspnea, and delirium, being necessary to know correct management of the opioids that help alleviate suffering [24,25]; for them, training should be addressed that removes the myths and erroneous concepts around the use of these substances, such as those related to addiction to morphine derived from its use [26]. During the 67th World Health Assembly, the WHO urged the countries to ensure funding and allocate resources that include availability of essential medications for symptom relief [10]. It is necessary to sustain advances in the provision of resources and in the involvement of governments in policies that ensure palliative care. Regarding the psychosocial approach, it was noticed in the study that the psychologi- cal and social sphere were taken into account, but to the detriment of the spiritual, despite the beneﬁts associated with quality of life when the spiritual sphere is included [27,28]. The professionals felt less prepared for the spiritual approach to the patient and the family [29], as well as for diagnoses and urgencies in palliative care [30]. Limitations The authors consider that this study allows having an idea of how the implementation process of the palliative care guide was carried out, although it is not to be forgotten that it was conducted in a single Ecuadorian zone, which is why the results cannot be generalized to the entire country. In addition, it should be analyzed if the situation is even worse in the country’s rural and remote areas, where health services are even scarcer. As this is a cross-sectional study, no causality of what is detailed can be established. However, this study focuses attention on the need to know the degree of implementation of the palliative care guide in other health areas, since it is necessary to keep researching policies and strategies that turn out to be successful so that the implementation of palliative care attains high standards in the country. 4. Discussion Therefore, the spiritual sphere should be taken into account and strategies should be implemented to improve coping in health professionals. 5. Conclusions The implementation of the GPCCP guide in Zone 7 Health Centers is insufﬁcient. Among the main difﬁculties encountered are lack of information about the guide, speciﬁc training on its application, and limited availability of essential PC medications and opioids. The health professionals possess insufﬁcient knowledge about PC and limited professional experience. They acknowledge that they do not feel trained in the diagnosis and man- agement of PC urgencies, in the use of morphine for pain treatment, and in its secondary effects. It becomes necessary to keep working on the training of the health professionals involved in palliative care, improving their care skills, especially in the spiritual sphere, as well as to enhance accessibility to the drugs which are required for an effective approach. It would be desirable to reproduce this work in other areas of the country, to know the degree of implementation of the guide at the national level, and improve the degree of Int. J. Environ. Res. Public Health 2021, 18, 11573 13 of 14 13 of 14 knowledge and training about palliative care. It seems necessary to include training in palliative care in undergraduate degree programs in the health sciences, as well as in postgraduate education. Author Contributions: Conceptualization, methodology, and investigation, T.R.Q., V.D.-B., P.B.-S. and L.L.; software, A.-M.V.-M.; validation, T.R.Q. and L.L.; formal analysis, A.-M.V.-M.; data curation, A.-M.V.-M.; writing—original draft preparation, F.L.-L., P.B.-S.; T.R.Q., M.M.L.-C., and A.-M.V.-M.; writing—review and editing, F.L.-L., P.B.-S.; T.R.Q., M.M.L.-C. and A.-M.V.-M. All authors have read and agreed to the published version of the manuscript. Author Contributions: Conceptualization, methodology, and investigation, T.R.Q., V.D.-B., P.B.-S. and L.L.; software, A.-M.V.-M.; validation, T.R.Q. and L.L.; formal analysis, A.-M.V.-M.; data curation, A.-M.V.-M.; writing—original draft preparation, F.L.-L., P.B.-S.; T.R.Q., M.M.L.-C., and A.-M.V.-M.; writing—review and editing, F.L.-L., P.B.-S.; T.R.Q., M.M.L.-C. and A.-M.V.-M. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by Universidad Técnica Particular de Loja (Ecuador), grant number 2522. Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki, and approved on 3 December 2020 by the Ethics Committee of Universidad Técnica Particular de Loja under protocol code 2522. Informed Consent Statement: Written informed consent has been obtained from the patient(s) to publish this paper. Data Availability Statement: The data that support the ﬁndings of this study are available from the corresponding author (A.-M.V.-M.) upon reasonable request. References 1. Dans, M.; Kutner, J.S.; Agarwal, R.; Baker, J.N.; Bauman, J.R.; Beck, A.C.; Campbell, T.C.; Carey, E.C.; Case, A.A.; Dalal, S.; et al. NCCN Guidelines® Insights: Palliative Care, Version 2.2021. J. Natl. Compr. Cancer Netw. 2021, 19, 780–788. [CrossRef] [PubMed] 2. Aldridge, M.D.; Hasselaar, J.; Garralda, E.; Van Der Eerden, M.; Stevenson, D.; McKendrick, K.; Centeno, C.; Meier, D.E. Education, implementation, and policy barriers to greater integration of palliative care: A literature review. Palliat. Med. 2016, 30, 224–239. 1. Dans, M.; Kutner, J.S.; Agarwal, R.; Baker, J.N.; Bauman, J.R.; Beck, A.C.; Campbell, T.C.; Carey, E.C.; Case, A.A.; Dalal, S.; et al. NCCN Guidelines® Insights: Palliative Care, Version 2.2021. J. Natl. Compr. Cancer Netw. 2021, 19, 780–788. [CrossRef] [PubMed] 1. Dans, M.; Kutner, J.S.; Agarwal, R.; Baker, J.N.; Bauman, J.R.; Beck, A.C.; Campbell, T.C.; Carey, E.C.; Case, A.A.; Dalal, S.; et al. NCCN Guidelines® Insights: Palliative Care, Version 2.2021. J. Natl. Compr. Cancer Netw. 2021, 19, 780–788. [CrossRef] [PubMed] 2. Aldridge, M.D.; Hasselaar, J.; Garralda, E.; Van Der Eerden, M.; Stevenson, D.; McKendrick, K.; Centeno, C.; Meier, D.E. Education, implementation, and policy barriers to greater integration of palliative care: A literature review. Palliat. Med. 2016, 30, 224–239. [CrossRef] g 2. Aldridge, M.D.; Hasselaar, J.; Garralda, E.; Van Der Eerden, M.; Stevenson, D.; McKendrick, K.; Centeno, C.; Meier, D.E. Education, implementation, and policy barriers to greater integration of palliative care: A literature review. Palliat. Med. 2016, 30, 224–239. [CrossRef] 3. United Nations General Assembly. United Nations 66/2. Political Declaration of the High-level Meeting of the General Assembly on the Prevention and Control of Non-Communicable Diseases. 2012. Available online: http://www.who.int/publications/en/ (accessed on 5 March 2021). ( ) 4. Murray, S.A.; Boyd, K.; Sheikh, A.; Thomas, K.; Higginson, I.J. Developing primary palliative care. BMJ 2004, 329, 1056–1057. [CrossRef] [PubMed] 5. Stjernswärd, J.; Foley, K.M.; Ferris, F.D. The Public Health Strategy for Palliative Care. J. Pain Symptom Manag. 2007, 33, 486–493. [CrossRef] [PubMed] 6. Bonilla, P. Cuidados paliativos en Latinoamérica. Rev. Nutr. Clin. Metab. 2021, 4, 4–13. [CrossRef] p 7. Pastrana, T.; Torres, I.; De Lima, L. Palliative care development in Latin America: An analysis using macro indicators. Palliat. Med. 2014, 28, 1231–1238. [CrossRef] [PubMed] 8. Pastrana, T.; Lima, L.; Sánchez-Cárdenas, M.; Steijn, D.; Garralda, E.; Pons-Izquierdo, J.J.; Centeno, C. Atlas de Cuidados Paliativos de Latinoamérica 2020, 2nd ed.; IAHPC Press: Houston, TX, USA, 2021. 9. Ministerio de Salud Pública. Cuidados Paliativos. 5. Conclusions Acknowledgments: We are grateful to economist Lizeth Ramon Jaramillo who created the database and provided guidance in collection of the information. Likewise, the work of the graduate students attending the Family and Community Medicine course at UTPL is recognized. Conﬂicts of Interest: The authors declare no conﬂict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. References Guía de Práctica Clínica; MSP: Quito, Ecuador, 2014. Available online: https://aplicaciones.msp.gob.ec/salud/archivosdigitales/documentosDirecciones/dnn/archivos/GPC%20Cuidados%20 paliativos%20FINAL.pdf (accessed on 5 March 2021). p p 10. Dávalos-Batallas, V.; Mahtani-Chugani, V.; López-Núñez, C.; Duque, V.; Leon-Larios, F.; Lomas-Campos, M.-D.-L.-M.; Sanz, E. Knowledge, Attitudes and Expectations of Physicians with Respect to Palliative Care in Ecuador: A Qualitative Study. Int. J. Environ. Res. Public Health 2020, 17, 3906. [CrossRef] [PubMed] 11. Hidalgo-Andrade, P.; Mascialino, G.; Miño, D.; Mendoza, M.; Marcillo, A. Knowledge of Palliative Care in Ecuador. Int. J. Environ. Res. Public Health 2021, 18, 4840. [CrossRef] [PubMed] 12. May, P.; Hynes, G.; McCallion, P.; Payne, S.; Larkin, P.; McCarron, M. Policy analysis: Palliative care in Ireland. Heal. Policy 2014, 115, 68–74. [CrossRef] [PubMed] [ ] [ ] 13. Clelland, D.; Van Steijn, D.; Whitelaw, S.; Connor, S.; Centeno, C.; Clark, D. Palliative Care in Public Policy: Results from a Global Survey. Palliat. Med. Rep. 2020, 1, 183–190. [CrossRef] [PubMed] 14 of 14 14 of 14 Int. J. Environ. Res. Public Health 2021, 18, 11573 14. Sharkey, L.; Loring, B.; Cowan, M.; Riley, L.; Krakauer, E.L. National palliative care capacities around the world: Results from the World Health Organization Noncommunicable Disease Country Capacity Survey. Palliat. Med. 2018, 32, 106–113. [CrossRef] 15. Bonilla-Sierra, P.; Vargas-Martínez, A.-M.; Davalos-Batallas, V.; Leon-Larios, F.; Lomas-Campos, M.-D.-L.-M. Chronic Diseases 14. Sharkey, L.; Loring, B.; Cowan, M.; Riley, L.; Krakauer, E.L. National palliative care capacities around the world: Results from the World Health Organization Noncommunicable Disease Country Capacity Survey. Palliat. Med. 2018, 32, 106–113. [CrossRef] 15. Bonilla-Sierra, P.; Vargas-Martínez, A.-M.; Davalos-Batallas, V.; Leon-Larios, F.; Lomas-Campos, M.-D.-L.-M. Chronic Diseases and Associated Factors among Older Adults in Loja, Ecuador. Int. J. Environ. Res. Public Health 2020, 17, 4009. [CrossRef] [PubMed] 14. Sharkey, L.; Loring, B.; Cowan, M.; Riley, L.; Krakauer, E.L. National palliative care capacities around the world: Results from the World Health Organization Noncommunicable Disease Country Capacity Survey. Palliat. Med. 2018, 32, 106–113. [CrossRef] 15. Bonilla-Sierra, P.; Vargas-Martínez, A.-M.; Davalos-Batallas, V.; Leon-Larios, F.; Lomas-Campos, M.-D.-L.-M. Chronic Diseases and Associated Factors among Older Adults in Loja, Ecuador. Int. J. Environ. Res. Public Health 2020, 17, 4009. [CrossRef] [PubMed] 16. Zimmermann, C.; Swami, N.; Krzyzanowska, M.; Hannon, B.; Leighl, N.; Oza, A.; Moore, M.; Rydall, A.; Rodin, G.; Tannock, I.; et al. Early palliative care for patients with advanced cancer: A cluster-randomised controlled trial. Lancet 2014, 383, 1721–1730. [CrossRef] 17. References Kovacs, E.; Strobl, R.; Phillips, A.; Stephan, A.-J.; Müller, M.; Gensichen, J.; Grill, E. Systematic Review and Meta-analysis of the Effectiveness of Implementation Strategies for Non-communicable Disease Guidelines in Primary Health Care. J. Gen. Intern. Med. 2018, 33, 1142–1154. [CrossRef] [PubMed] 18. Jeba, J.; Ponissery, J.; Ramaswamy, A.; Johnson, J.R.; Thelly, A.S.; Bilimoria, P.J. Developing evidence-based clinical guidelines in palliative care for home care setting in India. Indian J. Palliat. Care 2020, 26, 319–322. [CrossRef] [PubMed] 19. Hidalgo-Andrade, P.; Mascialino, G. Beliefs About Morphine in Palliative Care: Results From an Ecuadorian Sample. J. Palliat. Care 2021, 8258597211026398. [CrossRef] 20. Pastrana, T.; De Lima, L.; Stoltenberg, M.; Peters, H. Palliative Medicine Specialization in Latin Amer J. Pain Symptom Manag. 2021, 62, 960–967. [CrossRef] [PubMed] ana, T.; De Lima, L.; Stoltenberg, M.; Peters, H. Palliative Medicine Specialization in Latin America: A Compar n Symptom Manag. 2021, 62, 960–967. [CrossRef] [PubMed] 21. Boucher, N.A.; Dries, E.; Franzione, A.; Burton-Chase, A.M.; Morris, D.; Sautter, J. Developing the Future End-of-Life Health Care Workforce: Lessons Learned From a Survey of Advanced Health Professions Students. Am. J. Hosp. Palliat. Med. 2021, 10499091211035711. [CrossRef] 22. Aparna, S.; Gopichandran, L.; Garg, R.; Sinha, A.P. Impact of Standardized Educational Intervention on Improvement in Analgesic Knowledge and Its Compliance among Patients of Advanced Cancer Patients Attending Palliative Clinic: Prospective Randomized Study. J. Pain Palliat. Care Pharmacother. 2021, 3, 1–8. [CrossRef] [PubMed] 23. McCallan, T.; Daudt, H. Engaging Family Physicians in the Provision of Palliative and End-of-Life Care: Can We Do Better? Palliat. Med. Rep. 2021, 2, 207–211. [CrossRef] 24. Heijltjes, M.T.; Morita, T.; Mori, M.; Heckel, M.; Klein, C.; Stiel, S.; Miccinesi, G.; Deliens, L.; Robijn, L.; Stone, P.; et al. Physicians’ Opinion and Practice With the Continuous Use of Sedatives in the Last Days of Life. J. Pain Symptom Manag. 2021. [CrossRef] [PubMed] 25. Reedy, F.; Pearson, M.; Greenley, S.; Clark, J.; Currow, D.C.; Bajwah, S.; Fallon, M.; Johnson, M.J. Professionals’, patients’ and families’ views on the use of opioids for chronic breathlessness: A systematic review using the framework method and pillar process. Palliat. Med. 2021, 35, 1421–1433. [CrossRef] p 26. Allende-Perez, S.R.; Peña-Nieves, A.; Gómez, M.Á.H.; Nicolau, E.D.L.P.; Verástegui-Avilés, E. Proﬁle, Treatment, and Survival of Palliative Cancer Patients in a Middle Income Country’s Cancer Center. Am. J. Hosp. Palliat. Med. 2021, 10499091211034772. [CrossRef] 27. References Hiratsuka, Y.; Suh, S.-Y.; Kim, S.-H.; Cheng, S.-Y.; Yoon, S.-J.; Koh, S.-J.; Park, S.A.; Seo, J.-Y.; Kwon, J.H.; Park, J.; et al. Factors related to spiritual well-being in the last days of life in three East Asian countries: An international multicenter prospective cohort study. Palliat. Med. 2021, 35, 1564–1577. [CrossRef] 28. Yang, Y.; Zhao, X.; Cui, M.; Wang, S.; Wang, Y. Longitudinal changes in spiritual well-being and associations with emotional distress, pain, and optimism–pessimism: A prospective observational study of terminal cancer patients admitted to a palliative care unit. Support. Care Cancer 2021, 29, 7703–7714. [CrossRef] [PubMed] pp 29. Chahrour, W.H.; Hvidt, N.C.; Hvidt, E.A.; Viftrup, D.T. Learning to care for the spirit of dying patients: The impact of spiritual care training in a hospice-setting. BMC Palliat. Care 2021, 20, 115. [CrossRef] [PubMed] 30. Cardoso, C.; Prazeres, F.; Xavier, B.; Gomes, B. Family Physicians’ Perspectives on Their Role in Palliative Care: A Double Focus Group in Portugal. Int. J. Environ. Res. Public Health 2021, 18, 7282. [CrossRef]
https://openalex.org/W2786755026	https://europepmc.org/articles/pmc5830642?pdf=render	English	null	A symbolic network-based nonlinear theory for dynamical systems observability	Scientific reports	2,018	cc-by	17,782	A symbolic network-based nonlinear theory for dynamical systems observability Received: 6 October 2017 Accepted: 6 February 2018 Published: xx xx xxxx Received: 6 October 2017 Accepted: 6 February 2018 Published: xx xx xxxx Christophe Letellier 1, Irene Sendiña-Nadal 2,3, Ezequiel Bianco-Martinez4 & Murilo S. Baptista4 Christophe Letellier 1, Irene Sendiña-Nadal 2,3, Ezequiel Bianco-Martinez4 & Murilo S. Baptista4 When the state of the whole reaction network can be inferred by just measuring the dynamics of a limited set of nodes the system is said to be fully observable. However, as the number of all possible combinations of measured variables and time derivatives spanning the reconstructed state of the system exponentially increases with its dimension, the observability becomes a computationally prohibitive task. Our approach consists in computing the observability coefficients from a symbolic Jacobian matrix whose elements encode the linear, nonlinear polynomial or rational nature of the interaction among the variables. The novelty we introduce in this paper, required for treating large- dimensional systems, is to identify from the symbolic Jacobian matrix the minimal set of variables (together with their time derivatives) candidate to be measured for completing the state space reconstruction. Then symbolic observability coefficients are computed from the symbolic observability matrix. Our results are in agreement with the analytical computations, evidencing the correctness of our approach. Its application to efficiently exploring the dynamics of real world complex systems such as power grids, socioeconomic networks or biological networks is quite promising. Variables spanning the state space of a dynamical system which is irreducible to a few smaller subsystems are always dependent on each other through linear and nonlinear interactions. Consequently, one may expect to be able to determine an adequate subset of variables together with their well-selected Lie derivatives to get a full observability of the underlying dynamics, that is, for distinguishing all possible states of the network1,2. With the emergence of the Science of Complexity, complex networks are more and more often considered in various fields as well exemplified by power grids3, socio-economics networks4–6, or biological systems7–10. To allow a reliable monitoring, dynamical analysis or control of these high-dimensional systems, suitable and systematic techniques are required to identify the subset of variables providing the best (if not the full) observability of their underlying dynamics. A related problem is how to unfold the whole dynamics by completing this subset of variables to recon- struct a space whose dimension is at least equal to the dimension of the original state space. www.nature.com/scientificreports www.nature.com/scientificreports www.nature.com/scientificreports Received: 6 October 2017 Accepted: 6 February 2018 Published: xx xx xxxx Results On rare occasions, nonlinear systems are fully observable from just a single scalar time series20 as previously investigated for many chaotic systems21–23. Since full (global) observability warrants that every distinct point of the original state space x d ∈ can be univocally identified, there is a great interest to target the minimal set of variables to measure for accomplishing such a full observability condition. As shown in Section Methods, to assess the (local or global) observability through a given measurement vector s, both a subset of m variables–sometimes designated as “sensors”24 –and the Lie derivatives have to be provided. Our aim is therefore to provide a method that can indeed solve the problem of determining the min- imum set of variables to measure for observing a large complex system. In order to avoid testing the rank of the observability matrix via algebraic computation, we propose to use a technique based on a symbolic computation of the observability matrix in which the terms are not explicitely expressed but only their linear, nonlinear poly- nomial or rational character2,13.h The general and systematic procedure developed by Bianco-Martinez and coworkers2 and compiled in the Methods section, is in fact very time consuming since the computation of the observability coefficients corre- sponding to all the 5.2 ⋅ 106 possible vectors spanning a reconstructed space of a 13-dimensional system would require intense and long computational time (more than 18 days). One optimization strategy is to reduce the number of possible combinations by identifying candidate variables that should be disregarded as members of the measurement set. A lack of observability has its origin in the existence of a singular observability manifold, a domain in the original state space where the determinant Det  of the observability matrix  is zero25. Let us note here that there is one very special case in which Det  ≈ 0 and, consequently, practical problems in the state esti- mation may occur. This usually happens when Det  depends on some parameter(s) which may be arbitrarily small26. In general, a linear system may be (rarely and practically) non-observable with a nonzero determinant of the observability matrix. By construction, a null or non-constant determinant Det  is rooted in a null or a non- linear component in the Jacobian matrix. www.nature.com/scientificreports/ www.nature.com/scientificreports/ nonlinear systems to serve as testbed of our approach’s performance and reliability. In addition, we will show that our approach correctly identifies whether a nonlinear dynamical system is fully or only partially observable, an information not accurately obtained by previously proposed methodological approaches16–19. y y p y p p g pp Since we are dealing with dynamical systems in general, the Jacobian matrix will be used to access the nature of how variables interact, allowing us to optimize our assessment of the symbolic observability coefficients. For high-dimensional complex systems this is a quite demanding computation since the number of cases to investi- gate increases with the system’s dimension and exact analytical computations are prohibitive. Indeed, in practice, monitoring all the variables defining the system’s state is experimentally infeasible or inefficient, and it is of utmost importance to develop a methodological framework addressing the problem of targeting those variables yielding full observability. Despite several approaches have been proposed16,17, most of them neglect the nonlinear nature typically exhibited by complex systems and/or do not provide the space reconstructed from the measured variables. On the one hand, since nonlinearities are most often related to a lack of observability, linear approaches cannot properly address this problem. On the other hand, finding the appropriate combination of sensors (and time deriva- tives) spanning the reconstructed space is a very time demanding computational task for large dimensional systems. p g p y g p g y Here, we adopted a nonlinear symbolic approach taking into account the nature of the interactions among variables and analyze the distribution of the linear and nonlinear load of the variables in the symbolic Jacobian matrix of the system. By means of two easy-to-implement criteria we are able to successfully identify the minimal set of variables (and their time derivatives) candidate to be measured for completing the reconstructed space. Our results are in full agreement with the analytical prediction of getting a no null determinant of the observa- bility matrix and the technique drastically reduces the search for candidate variables, thus providing a key step to observe and model natural and man-made complex systems of large dimension.h p y g The subsequent part of the paper is organized as follows. Section Results is devoted to illustrate how our pro- posed approach works considering a few large dimensional dynamical systems. www.nature.com/scientificreports/ Section Methods briefly intro- duces the observability theory, and the current challenges for the determination of a system’s observability in nonlinear systems. Finally, the Discussion section provides some conclusions to this work. A symbolic network-based nonlinear theory for dynamical systems observability Dealing with multivariate time series, specially those produced by high-dimensional dynamical networks, is not a trivial problem11–13. Attempts to estimate network observability using symbolic techniques14,15 were made to overcome the large computational times associated with the exact analytical calculations. In those approaches, a dimension reduction is performed in real time on a symbolic observability matrix until state estimation is possi- ble from the selected measurements. However, linear and nonlinear interactions among variables are considered on an equal footing while it is strongly required to distinguish them for a reliable assessment of the observability of a system2,13. In order to tackle such a challenging task, we propose a methodological approach that will be applied to reaction networks derived from dynamical systems with appropriately large dimension to corroborate our assessments with rigorous analytical calculations, and yet provide a framework making also possible the veri- fication of observability in networked dynamical systems. The chosen reaction networks are models of interesting biological and physical systems: the circadian oscillation in the Drosophila period protein, the Rayleigh-Bénard convection, and the DNA replication. They also represent nonlinear systems with increasing nonlinear complex- ity, commonly observed in other natural and man-made systems. Therefore, they are an appropriate subset of 1CORIA-UMR 6614 Normandie Université, Campus Universitaire du Madrillet, F-76800, Saint-Etienne du Rouvray, France. 2Complex Systems Group, Universidad Rey Juan Carlos, 28933, Móstoles, Madrid, Spain. 3Center for Biomedical Technology, Universidad Politécnica de Madrid, 28223, Pozuelo de Alarcón, Madrid, Spain. 4Institute for Complex Systems and Mathematical Biology, SUPA, University of Aberdeen, Old Aberdeen, AB24 3UE, United Kingdom. Correspondence and requests for materials should be addressed to C.L. (email: Christophe.Letellier@coria.fr) Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w 1 Results Thus, the in-strength of the ith variable provided by the non-measured ones is defined as non-measured is influenced by the other candidate variables. Let be {xk} the set of variables candidate to be non measured with k ∈ Vnm ⊂ {1, 2, ..., m} and Vnm the set of integers indexing the non-measured variables. Thus, the in-strength of the ith variable provided by the non-measured ones is defined as i J ( ) (2) k V k i ik in nm nm ∑ σ = ∈ ≠  i J ( ) k V k i ik in nm nm ∑ σ = ∈ ≠  (2) where = ≡ J 1 1 ik , = ≡ J 1 2 ik , and J 1 3 ik= ≡. Using this correspondence for the symbolic terms, we can assume that the larger is σ i( ) in nm , the less observable through the ith variable the system is. The rationale is as fol- lows: the more nonlinearly coupled is the ith function fi(x) with the non-measured variables, the larger the degree of the determinant of the observability matrix, and the less observable the dynamics through the measurements is22. Therefore, we should preferably remove that variable with the largest non-measured in-strength σin nm.h h p y g g in Therefore, the minimal set of variables to measure for reconstructing a state space with a full observability can be automatically determined from (i) the symbolic Jacobian matrix ∼  of the system under study, (ii) the linear out-strength out lin σ , and (iii) the in-strength in nm σ provided by the non-measured variables. Note that the knowledge of the exact functional dependence of the coupling between variables is not necessary, only its polynomial or fractional nature2. From the vector of state variables x, those components xi having the largest σout lin are discarded as candidates to be sensors after having checked they are present at least once in the equations governing the dynamics of a sensor variable. All remaining possible embeddings s that can be constructed from the final set of variables candidate to become sensors are then tested using a Matlab® algorithm and ranked according to the corresponding estimated symbolic observability coefficient ηs. Results Our technique relies precisely on tracking those nonlinear terms in the Jacobian matrix and, therefore, taking into account both linear and nonlinear interactions between variables becomes so relevant in assessing observability. g y By analogy with what is done for chemical reactions27, it is possible to consider any dynamical system as a reaction network, whose associated weighted adjacency matrix is the symbolic Jacobian matrix ∼ . Using the terminology from graph theory, we define the linear out-strength of the node i, σ i( ) out lin , as the number of times the ith variable appears in linear terms in the governing equations, that is, i J ( ) (1) j i j J ji out lin 1 ji ∑ σ = . ≠ \| =   (1) The larger is σ i( ) out lin , the higher the probability the ith variable needs not to be measured because it is related to other variables via linear couplings which will not induce nonlinear terms in the determinant of the observability matrix.h   lin lin The larger is σ i( ) out lin , the higher the probability the ith variable needs not to be measured because it is related to other variables via linear couplings which will not induce nonlinear terms in the determinant of the observability matrix.h   li li The situation in which   = = J J 1 ij ji and i j ( ) ( ) 1 out lin out lin σ σ = = means that variables i and j are exclusively linearly coupled with no other variables involved. Consequently, full observability of the ith variable can only be accessed by measuring the jth variable and vice versa. It is thus necessary (and sufficient) to measure at least one of them because they cannot be simultaneously excluded from the set of measured variables. A second criterion to decide which variable to choose between these two is needed. The idea is built on how a variable candidate to be Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w 2 www.nature.com/scientificreports/ non-measured is influenced by the other candidate variables. Let be {xk} the set of variables candidate to be non measured with k ∈ Vnm ⊂ {1, 2, ..., m} and Vnm the set of integers indexing the non-measured variables. A 5D rational model for the circadian PER oscillations in Drosophila. In our attempt to consider biological or physically motivated systems, let us start with the model A 5D rational model for the circadian PER oscillations in Drosophila. In our attempt to consider biological or physically motivated systems, let us start with the model      = + − + = − + + + = + + + −    + + +    = + −    + + + +   + = − x vK K x v x K x x k x Vx K x V x K x x Vx K x V x K x x V K x V K x x Vx K x x V K x k v K x k x x k x k x s I I m m s d d 1 4 4 5 4 1 1 2 1 1 2 1 2 2 3 2 3 3 1 2 1 2 4 4 4 4 3 2 2 3 3 3 3 4 3 3 3 3 4 4 4 4 1 4 2 5 5 1 4 2 5 ˙ ˙ ˙ ˙ ˙      = + − + = − + + + = + + + −    + + +    = + −    + + + +   + = − x vK K x v x K x x k x Vx K x V x K x x Vx K x V x K x x V K x V K x x Vx K x x V K x k v K x k x x k x k x (3) s I I m m s d d 1 4 4 5 4 1 1 2 1 1 2 1 2 2 3 2 3 3 1 2 1 2 4 4 4 4 3 2 2 3 3 3 3 4 3 3 3 3 4 4 4 4 1 4 2 5 5 1 4 2 5 ˙ ˙ ˙ ˙ ˙   = − x k x k x (3) 5 1 4 2 5 ˙ (3) proposed by Goldbeter for the circadian oscillation in the Drosophila period protein29. This is a five-dimensional rational model which produces a limit cycle for the parameter values initially reported29. Results fi In order to validate whether our proposed method is in agreement with algebraic computations, we will con- sider three dynamical systems of increasing dimension (d = 5, 9 and 13) describing complex systems coming from biology or physics. As it is known that the presence of symmetries in the state space can affect the assesment of observability22, we will also consider the case of equivariant dynamical systems obeying f(Γ ⋅ x) = Γ ⋅ f(x), where Γ defines a discrete symmetry like a rotation or an inversion28. A 5D rational model for the circadian PER oscillations in Drosophila. In our attempt to consider biological or physically motivated systems, let us start with the model It turns out that only two combinations yielded full observability: ( 1) x x x 2 2 3 4 2 η = and η = ( 1) x x x 2 2 3 5 2 where the derivatives of x2 and x4 and of x2 and x5 are, respectively, the ones completing each set of sensors. These results are algebraically confirmed by the constant determinants of the observability matrix (or equiva- lently of the Jacobian matrix of the coordinate transformation between the original state sapce and the reconstructed space30) and equal to k k x x x s 2 2 2 3 4 2 ∆ = − and k k x x x s 1 2 2 3 5 2 ∆ = , respectively. The relevance of correctly chosing the deriv- atives is exemplified by replacing in s x x x x x ( , , , , ) 2 2 3 5 5 =   the derivative of the fifth variable by the derivative of the third one: the coordinate transformation x x x 2 2 3 2 5 Φ yields a symbolic observability coefficient η = .0 70 x x x 2 2 3 2 5 , thus reflect- ing a significant lack of observability. This is further supported by the corresponding determinant ∆ = − + k V K K x ( ) (6) x x x s 4 4 4 4 2 2 2 3 2 5 (6) which is now no longer constant as it depends on variable x4. There is thus a singular observabiliy manifold. On the other hand, if the derivative of the second variable is substituted with the derivative of the third one, the Jacobian matrix J of such a transformation Φx x x 2 3 2 5 2 can be rank deficient with a null determinant for some domain of the original state space.i which is now no longer constant as it depends on variable x4. There is thus a singular observabiliy manifold. On the other hand, if the derivative of the second variable is substituted with the derivative of the third one, the Jacobian matrix J of such a transformation Φx x x 2 3 2 5 2 can be rank deficient with a null determinant for some domain of the original state space. A 5D rational model for the circadian PER oscillations in Drosophila. In our attempt to consider biological or physically motivated systems, let us start with the model This is in a rather good agreement with the analytical determinants Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w 3 www.nature.com/scientificreports/      ∆ = + + + ∆ ∆ ∆ ∆ = − + + v K k V K VK K x K x K x x k V K V K k K x K x 256 ( ) ( ) ( ) ; , and where complexity exceedsour computational abilities; ( ) ( ) , (5 x s I I x x x x s 4 16 1 3 3 2 3 2 1 1 4 5 4 8 3 3 4 1 2 2 5 12 1 4 3 3 3 3 1 2 1 2 3 3 6 1 2 4 1 5 2 5 3 5 4 5 5 5 = − + + k V K V K k K x K x ( ) ( ) , s 1 4 3 3 3 3 1 2 1 2 3 3 6 1 2 4 3 4 (5) since the simpler determinant (with singularity of the 10th degree) is obtained for variable x5 providing the best observ- ability, then variable x1 is associated with a determinant with a singularity of the 14th degree, and the three variables x2, x3 and x4 providing the poorest observability are associated with determinants too complicated to be computed with MAPLE®. The number (125) of all possible combinations of dimension 5 is still sufficiently small for allowing a system- atic computation of the corresponding symbolic observability coefficients ηs. Prior to carry out those computations, we conducted our a priori analysis to target the candidate variables to be discarded. The linear out-strengths are (1) (4) (5) 1 out lin out lin out lin σ σ σ = = = , the two others being null. Among the off-diagonal terms of the symbolic Jacobian matrix which are equal to 1, we have J45 = J54 = 1, meaning that at least one of the two variables x4 and x5 has to be measured. Therefore, this suggests that the sets with the minimum number of sensors providing full observability com- prise at least three variables, either (x2, x3, x4) or (x2, x3, x5). In order to have a five dimensional space, these two sets have to be completed with two Lie derivatives. A 5D rational model for the circadian PER oscillations in Drosophila. In our attempt to consider biological or physically motivated systems, let us start with the model g p Finally, we wanted to assess the observability of the couple of variables {x1,x5}, identified as sensors of the sys- tem (3) in a previous work16. Surprisingly enough, we got 0 48 x x 1 5 4 η = . and 0 x x x x x x 1 2 5 3 1 3 5 2 1 4 5 η η η = = = for all the possible 5 dimensional vectors constructed with those two variables. These symbolic observability coefficients are also in agreement with the determinants of the Jacobian matrices of the corresponding transformations, ∆ = + + k V K VK K x K x ( ) ( ) (7) x x 1 3 3 2 3 2 1 1 3 3 4 1 2 2 1 5 4 (7) whose rational dependence on variables x2 and x3 defines a singular observability manifold associated with the transformation x x 1 5 4 Φ , and the other three determinants ∆ = ∆ = ∆ = 0 x x x x x x 1 2 5 3 1 3 5 2 1 4 5 , characterizing a rank defi- cient observability matrix . This is therefore a first evidence that our method to reduce the number of sensor variables correctly assesses the observability of this rather complex reaction network. whose rational dependence on variables x2 and x3 defines a singular observability manifold associated with the transformation x x 1 5 4 Φ , and the other three determinants ∆ = ∆ = ∆ = 0 x x x x x x 1 2 5 3 1 3 5 2 1 4 5 , characterizing a rank defi- cient observability matrix . This is therefore a first evidence that our method to reduce the number of sensor variables correctly assesses the observability of this rather complex reaction network. A 9D system for the Rayleigh-Bénard convection. Let us consider now a nine-dimensional system describing the dynamics of three-dimensional fluid cells with a square plateform in a Rayleigh-Bénard convec- tion31. It was obtained by applying a triple second-order Fourier series ansatz to the governing hydrodynamic equations. Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w A 5D rational model for the circadian PER oscillations in Drosophila. In our attempt to consider biological or physically motivated systems, let us start with the model This system is interesting in the sense that its complexity already presents a big challenge from the analytical point of view. The correspond- ing symbolic Jacobian matrix reads as =       . ~ 1 0 0 0 1 1 1 1 0 0 0 1 1 1 0 0 0 1 1 1 0 0 0 1 1 (4) 5D =       . ~ 1 0 0 0 1 1 1 1 0 0 0 1 1 1 0 0 0 1 1 1 0 0 0 1 1 (4) 5D =       . ~ 1 0 0 0 1 1 1 1 0 0 0 1 1 1 0 0 0 1 1 1 0 0 0 1 1 (4) 5D (4) The symbolic observability coefficients corresponding to a univariate measurement s = xi are η = .0 17 x1 5 , 0 08 x2 5 η = . , 0 02 x3 5 η = . , η = .0 09 x4 5 and, 0 30 x5 5 η = . , where the notation xi 5 refers to the vector  ̈ ⃛ .... x x x x x ( , , , , ) i i i i i whose observability to span the state space of the original system is estimated. According to the observability coefficient values, the ranking of the variables providing better observability is     x x x x x 5 1 4 2 3. www.nature.com/scientificreports/ This system is equivariant under the rotation (see page 506 in Gilmore and Letellier’s book32) where R is the reduced Rayleigh number and parameters bi (i = 1, …, 6) define the geometry of the square cell31 This system is equivariant under the rotation (see page 506 in Gilmore and Letellier’s book32) 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 (9) Γ =       . 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 (9) Γ =       . (9) In fact, eight variables are symmetry-related by pairs, namely (x1 − x3), (x2 − x4), (x5 − x9), and (x7 − x8), and variable x6 is the single one left invariant under the symmetry (9). Up to four co-existing chaotic attractors were observed in this system31. An example of one of those chaotic attractors is shown in Fig. 1 for the bi-values In fact, eight variables are symmetry-related by pairs, namely (x1 − x3), (x2 − x4), (x5 − x9), and (x7 − x8), and variable x6 is the single one left invariant under the symmetry (9). Up to four co-existing chaotic attractors were observed in this system31. An example of one of those chaotic attractors is shown in Fig. www.nature.com/scientificreports/ m x1 x2 x3 x4 x5 x6 x7 x8 x9 η 8 1 2 1 1 1 1 1 1 — 1.00 8 1 1 2 1 1 1 1 — 1 1.00 8 1 1 1 2 1 1 1 — 1 1.00 8 1 1 1 2 1 1 1 1 — 1.00 8 2 1 1 1 1 1 — 1 1 1.00 8 1 1 — 1 1 1 1 2 1 1.00 8 1 1 — 1 1 1 2 1 1 1.00 8 — 1 1 1 1 1 2 1 1 1.00 Table 1. All possible subsets with m = 8 measured variables and one Lie derivative (of the variable for which “2” is reported) providing a full observability of the state space associated with the 9-dimensional system (8). Those variables not affecting the full observability when not measured are highlighted in bold face. Table 1. All possible subsets with m = 8 measured variables and one Lie derivative (of the variable for which “2” is reported) providing a full observability of the state space associated with the 9-dimensional system (8). Those variables not affecting the full observability when not measured are highlighted in bold face. Table 1. All possible subsets with m = 8 measured variables and one Lie derivative (of the variable for which “2” is reported) providing a full observability of the state space associated with the 9-dimensional system (8). Those variables not affecting the full observability when not measured are highlighted in bold face. where R is the reduced Rayleigh number and parameters bi (i = 1, …, 6) define the geometry of the square cell31. A 5D rational model for the circadian PER oscillations in Drosophila. In our attempt to consider biological or physically motivated systems, let us start with the model The equations read as ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ σ σ σ σ σ σ σ σ σ      = − − + + − = − + − + − = − + − − + = − − − + + = − + − = − + − = − − + − = − + − + = − − + − + + − x b x x x b x b x x b x x x x x x x x x x x b x x x b x b x x b x x x x x x x x x x x b x x x x b x x x x x x b x Rx x x x x x b x Rx x x x x x x Rx Rx x x x x x x x x 2 2 2 2 2 2 2 2 (8) 1 1 1 2 4 4 4 2 3 3 5 2 7 2 2 1 4 2 5 4 5 9 3 1 3 2 4 4 2 2 3 1 5 2 8 4 4 2 3 2 5 4 5 9 5 5 5 2 2 4 2 6 6 6 2 9 4 9 7 1 7 1 5 8 4 9 8 1 8 3 5 7 2 9 9 9 2 4 2 6 4 6 4 7 2 8 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ σ σ σ σ σ σ σ σ σ      = − − + + − = − + − + − = − + − − + = − − − + + = − + − = − + − = − − + − = − + − + = − − + − + + − x b x x x b x b x x b x x x x x x x x x x x b x x x b x b x x b x x x x x x x x x x x b x x x x b x x x x x x b x Rx x x x x x b x Rx x x x x x x Rx Rx x x x x x x 2 2 2 2 2 2 2 2 1 1 1 2 4 4 4 2 3 3 5 2 7 2 2 1 4 2 5 4 5 9 3 1 3 2 4 4 2 2 3 1 5 2 8 4 4 2 3 2 5 4 5 9 5 5 5 2 2 4 2 6 6 6 2 9 4 9 7 1 7 1 5 8 4 9 8 1 8 3 5 7 2 9 9 9 2 4 2 6 4 6 4 7 ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ σ σ σ σ σ σ σ σ σ      = − − + + − = − + − + − = − + − − + = − − − + + = − + − = − + − = − − + − = − + − + = − − + − + + − x b x x x b x b x x b x x x x x x x x x x x b x x x b x b x x b x x x x x x x x x x x b x x x x b x x x x x x b x Rx x x x x x b x Rx x x x x x x Rx Rx x x x x x x x x 2 2 2 2 2 2 2 2 (8) 1 1 1 2 4 4 4 2 3 3 5 2 7 2 2 1 4 2 5 4 5 9 3 1 3 2 4 4 2 2 3 1 5 2 8 4 4 2 3 2 5 4 5 9 5 5 5 2 2 4 2 6 6 6 2 9 4 9 7 1 7 1 5 8 4 9 8 1 8 3 5 7 2 9 9 9 2 4 2 6 4 6 4 7 2 8 (8) www.nature.com/scientificreports/ www.nature.com/scientificreports/ -0,2 0,0 0,2 0,4 0,6 x5 -0,5 -0,4 -0,3 -0,2 -0,1 0,0 0,1 0,2 0,3 x5 . Figure 1. A chaotic attractor produced by the 9-dimensional dynamical network (8). Parameter values: σ = 0.5 and R = 14.22, and rest of parameter values are listed in (10). -0,2 0,0 0,2 0,4 0,6 x5 -0,5 -0,4 -0,3 -0,2 -0,1 0,0 0,1 0,2 0,3 x5 . Figure 1. A chaotic attractor produced by the 9-dimensional dynamical network (8). Parameter values: σ = 0.5 and R = 14.22, and rest of parameter values are listed in (10). m x1 x2 x3 x4 x5 x6 x7 x8 x9 η 8 1 2 1 1 1 1 1 1 — 1.00 8 1 1 2 1 1 1 1 — 1 1.00 8 1 1 1 2 1 1 1 — 1 1.00 8 1 1 1 2 1 1 1 1 — 1.00 8 2 1 1 1 1 1 — 1 1 1.00 8 1 1 — 1 1 1 1 2 1 1.00 8 1 1 — 1 1 1 2 1 1 1.00 8 — 1 1 1 1 1 2 1 1 1.00 m x1 x2 x3 x4 x5 x6 x7 x8 x9 η 8 1 2 1 1 1 1 1 1 — 1.00 8 1 1 2 1 1 1 1 — 1 1.00 8 1 1 1 2 1 1 1 — 1 1.00 8 1 1 1 2 1 1 1 1 — 1.00 8 2 1 1 1 1 1 — 1 1 1.00 8 1 1 — 1 1 1 1 2 1 1.00 8 1 1 — 1 1 1 2 1 1 1.00 8 — 1 1 1 1 1 2 1 1 1.00 Table 1. All possible subsets with m = 8 measured variables and one Lie derivative (of the variable for which “2” is reported) providing a full observability of the state space associated with the 9-dimensional system (8). Those variables not affecting the full observability when not measured are highlighted in bold face. Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w www.nature.com/scientificreports/ (1) (3) (7) (8) (9) 1 out lin out lin out lin out lin out lin These values indicate that variables x2, x4, x5, x6 are necessarily to be included in the list of variables to be measured for obtaining a full observability since none of them linearly affect the dynamics of the rest. The other five are candidate variables to be removed from the measurements. Their respective in-strengths coming from the non measured variables are: σ σ σ σ σ = = = = = . (1) (3) 3, (7) (8) 5, and (9) 4 in nm in nm in nm in nm in nm To determine whether all the candidate variables could be removed, we checked if there are pairs of exclusive varia- bles, that is, when two candidate variables are linearly coupled each other (one being linearly “seen” by the other). Among the off-diagonal elements Jij equal to 1, we have   J J 17 71 = , J29, J J 38 83   = , and J49 . Variables (x1, x7) and (x3, x8) thus form two pairs of mutually exclusive variables. Variable x9 is the single one not involved in an exclusive pair and can be removed from the set of measured variables. To decide which variable from each pair can be safely removed, we check which variables have the largest in-strength from the candidate variables to be non measured. The comparison returns that variables x7 and x8 are the ones to be removed since σ σ σ σ = = > = = (7) (8) 5 (1) (3) 3 in nl in nm in nm in nm .hi in in in in The first test to assess the accuracy in selecting the minimal set of variables providing the highest observability consists in systematically investigating those combinations where all the variables are measured except one. The symbolic observability coefficients are reported in Table 1. In all cases providing full observability with just a sin- gle variable not being considered, the discarded measure matches one of the candidate variables x1, x3, x7, x8 and x9 (marked in bold face in the table) confirming our preselection analysis.fi i g p y Our systematic computation of the symbolic coefficients allows us to quantify the number of times Ni(η) the variable xi is not part of an embedding providing a given observability value η. www.nature.com/scientificreports/ And a = 0.5. This system is irreducible in the sense that it cannot be split in lower-dimensional independent systems.h y This system is an interesting example because it constitutes a highly connected reaction network for which a graphical approach as the one developed by Liu and coworkers17 leads to only measure one of its nine variables to estimate its states (see the Supplementary Section S1).The variable that least influences the others (or equiva- lently, the one least “seen” by the rest) is x6 (σout(6) = 2) since it only affects nonlinearly the derivative of x9. From a symmetry point of view, variable x6 must be measured to recover the right symmetry property: without this variable, the reconstructed state space would be necessarily associated with an inversion symmetry (a symmetry the original system does not have). g y ) The symbolic Jacobian matrix of system (8) is g y The symbolic Jacobian matrix of system (8) is  =       ~ 1 1 1 1 1 0 1 0 0 1 1 0 1 1 0 0 0 1 1 1 1 1 1 0 0 1 0 0 1 1 1 1 0 0 0 1 0 1 0 1 1 0 0 0 0 0 1 0 1 0 0 0 0 1 1 0 0 1 1 1 1 1 1 0 1 1 0 0 1 1 1 1 0 1 0 1 1 1 1 1 1 (11) 9D (11) For this 9-dimensional system, there are 24309 possible combinations of variables and their derivatives candi- dates for providing full observability (see the Supplementary Section S2). Dealing with all these potential solu- tions is still afordable with our symbolic technique but it would take a rather long computational time (about 2 h). In order to reduce the number of combinations to test, we computed the linear out-strength i( ) out lin σ of the 9 varia- bles which are (2) (4) (5) (6) 0 out lin out lin out lin out lin σ σ σ σ = = = = (2) (4) (5) (6) 0 out lin out lin out lin out lin σ σ σ σ = = = = and σ σ σ σ σ = = = = = . www.nature.com/scientificreports/ 1 for the bi-values b a a b a a b a a b a a b a a b a 4 1 1 2 1 2 2(1 ) 21 1 1 8 1 2 4 1 2 (10) 1 2 2 2 2 2 3 2 2 4 2 2 5 2 2 6 2      = + + = + + = − + = + = + = + . (10) Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w 5 www.nature.com/scientificreports/ www.nature.com/scientificreports/ In particular, the values of Ni(η = 1.0) for the variables potentially candidate to be non measured (N1(1.0) = 4, N3(1.0) = 2, N7(1.0) = 7, N8(1.0) = 5, and N9(1.0) = 9, see the first part of Table 2), support our initial choice for not measuring x7, x8, and x9 but measuring x1 and x3, since x7, x8 and x9 seem to be less essential for providing full observability. Consequently, as long as full observability is required, our two network-based criteria correctly identify those variables whose absence from the set of sensors does not affect the full observability of the system. Indeed, when the minimal number of variables, that is, m = 6, is measured, the two possible combinations providing a full observability correspond to a space recon- structed from variables x1, x2, x3, x4, x5 and x6 (see the Supplementary Table S1. We therefore correctly assessed the best variables to measure for getting full observability with the minimum of variables. Of course, it is also possible to get full observability by measuring more than 6 variables. In that case, we searched for them among the 8 preselected variables using the linear out-strength. From the 354 possible combinations, we obtained 6 combinations using 7 measured variables and 2 with 8 measured variables. Performing a full blind search, with no preselection, from a total number of 1080 combinations with 7 or 8 measured variables, we found 2 and 6 additional combinations pro- viding full observability, respectively. All of them are reported in the Supplementary Table S1. www.nature.com/scientificreports/ Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w 6 www.nature.com/scientificreports/ m x1 x2 x3 x4 x5 x6 x7 x8 x9 η 6 2 2 2 1 1 1 — — — 1.00 6 2 2 2 1 1 1 — — — 1.00 7 1 2 2 1 1 1 1 — — 1.00 7 1 2 — 1 1 1 1 2 — 1.00 7 — 1 1 2 1 1 2 1 — 1.00 7 — 1 1 2 1 1 2 1 — 1.00 7 — 1 1 1 1 1 2 1 1 1.00 7 — 1 — 1 1 1 2 2 1 1.00 7 2 2 1 1 1 1 — 1 — 1.00 7 2 1 2 2 1 1 — — — 1.00 7 2 1 2 1 1 1 — — 1 1.00 7 2 1 1 2 1 1 — 1 — 1.00 7 2 1 1 1 1 1 — 1 1 1.00 5 2 3 2 1 1 — — — — 0.90 5 2 1 2 3 1 — — — — 0.90 6 3 2 1 1 1 1 — — — 0.90 6 3 1 2 1 1 1 — — — 0.90 6 3 1 1 2 1 1 — — — 0.90 6 2 3 1 1 1 1 — — — 0.90 6 2 3 1 1 1 — — 1 — 0.90 6 2 1 3 1 1 1 — — — 0.90 6 2 1 1 3 1 1 — — — 0.90 6 2 1 1 3 1 — — 1 — 0.90 6 1 3 2 1 1 1 — — — 0.90 6 1 3 2 1 1 — 1 — — 0.90 6 1 3 — 1 1 — 1 1 — 0.90 6 1 2 3 1 1 1 — — — 0.90 6 1 2 — 1 1 — 1 3 — 0.90 6 1 1 3 2 1 1 — — — 0.90 6 1 1 2 3 1 1 — — — 0.90 6 1 1 2 3 1 — 1 — — 0.90 6 — 1 1 3 1 — 2 1 — 0.90 6 — 1 1 2 1 — 3 1 — 0.90 6 — 1 — 1 1 — 3 2 1 0.90 6 — 1 — 1 1 — 2 3 1 0.90 6 — 1 — 1 1 — 2 3 1 0.90 5 4 2 1 1 1 — — — — 0.80 5 4 1 2 1 1 — — — — 0.80 5 4 1 1 2 1 — — — — 0.80 5 2 1 4 1 1 — — — — 0.80 5 1 2 4 1 1 — — — — 0.80 5 1 1 4 2 1 — — — — 0.80 6 4 1 1 1 1 — — 1 — 0.80 6 4 1 1 1 1 — — — 1 0.80 6 1 1 4 1 1 — 1 — — 0.80 6 1 1 4 1 1 — — — 1 0.80 Table 2. www.nature.com/scientificreports/ List of the different possible combinations of mea providing a symbolic observability coefficient η ≥0 75 of t 7 — 1 1 2 1 1 2 1 — 1.00 7 — 1 1 2 1 1 2 1 — 1.00 7 — 1 1 1 1 1 2 1 1 1.00 7 — 1 — 1 1 1 2 2 1 1.00 7 2 2 1 1 1 1 — 1 — 1.00 7 2 1 2 2 1 1 — — — 1.00 7 2 1 2 1 1 1 — — 1 1.00 7 2 1 1 2 1 1 — 1 — 1.00 7 2 1 1 1 1 1 — 1 1 1.00 5 2 3 2 1 1 — — — — 0.90 5 2 1 2 3 1 — — — — 0.90 6 3 2 1 1 1 1 — — — 0.90 6 3 1 2 1 1 1 — — — 0.90 6 3 1 1 2 1 1 — — — 0.90 6 2 3 1 1 1 1 — — — 0.90 6 2 3 1 1 1 — — 1 — 0.90 6 2 1 3 1 1 1 — — — 0.90 6 2 1 1 3 1 1 — — — 0.90 6 2 1 1 3 1 — — 1 — 0.90 6 1 3 2 1 1 1 — — — 0.90 6 1 3 2 1 1 — 1 — — 0.90 6 1 3 — 1 1 — 1 1 — 0.90 6 1 2 3 1 1 1 — — — 0.90 6 1 2 — 1 1 — 1 3 — 0.90 6 1 1 3 2 1 1 — — — 0.90 6 1 1 2 3 1 1 — — — 0.90 6 1 1 2 3 1 — 1 — — 0.90 6 — 1 1 3 1 — 2 1 — 0.90 6 — 1 1 2 1 — 3 1 — 0.90 6 — 1 — 1 1 — 3 2 1 0.90 6 — 1 — 1 1 — 2 3 1 0.90 6 — 1 — 1 1 — 2 3 1 0.90 5 4 2 1 1 1 — — — — 0.80 5 4 1 2 1 1 — — — — 0.80 5 4 1 1 2 1 — — — — 0.80 5 2 1 4 1 1 — — — — 0.80 5 1 2 4 1 1 — — — — 0.80 5 1 1 4 2 1 — — — — 0.80 6 4 1 1 1 1 — — 1 — 0.80 6 4 1 1 1 1 — — — 1 0.80 6 1 1 4 1 1 — 1 — — 0.80 6 1 1 4 1 1 — — — 1 0.80 Table 2. www.nature.com/scientificreports/ List of the different possible combinations of measured variables and their Lie derivative orders providing a symbolic observability coefficient η ≥ 0.75 of the state space associated with the 9-dimensional system (8). Table 2. List of the different possible combinations of measured variables and their Lie derivative orders providing a symbolic observability coefficient η ≥ 0.75 of the state space associated with the 9-dimensional system (8). Table 2. List of the different possible combinations of measured variables and their Lie derivative orders providing a symbolic observability coefficient η ≥ 0.75 of the state space associated with the 9-dimensional system (8). We also investigated the combinations with less than 6 measured variables and providing the largest symbolic observability coefficients whose dependency on m is shown in Fig. 2. For m = 5 (m = 4, 3, 2 and 1) there are 4 (2, 6, 4, and 4, respectively) combinations with η = 0.90 (η = 0.73, 0.36, 0.14, 0.04, respectively). All those combina- tions are made up of the six preselected variables identified by solely using the symbolic Jacobian matrix and σout lin and in nm σ to rank them.h in The non-preselected variables can be involved in reconstructed vectors when a good but not a full observabil- ity is desired or when m > 6 (a good observability is considered when η > 0.75 as reported in a previous work33). Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w 7 www.nature.com/scientificreports/ 0 1 2 3 4 5 6 7 8 9 Number of measured variables m 0 0,2 0,4 0,6 0,8 1 η Figure 2. Largest symbolic observability coefficient η versus the number m of measured variables for the 9D Rayleigh-Bénard model (8). 0 1 2 3 4 5 6 7 8 9 Number of measured variables m 0 0,2 0,4 0,6 0,8 1 η Figure 2. Largest symbolic observability coefficient η versus the number m of measured variables for the 9D Rayleigh-Bénard model (8). For instance, this is exemplified in the middle and lower part of Table 2 with a systematic computation of the symbolic observability coefficients η for embeddings built from 5 or 6 measured variables. In this case, we observe that a very good observability (η = 0.90) can be obtained with only 5 variables being measured, namely (x1, x2, x3, x4, x5). www.nature.com/scientificreports/ The under- lying mechanisms are described by the set of thirteen differential equations              β α β β β β µ      = − + + + + + = − − + + + − + − + + + + + + = − + + + = − + + + = + − + − − + = + − + − − + = − + − − + + = − + + + + = − + + = − + + + = − + − − + = + − + − − + = ′ ′ ′ ′ ′ x k k k k x x k x k k x x k k x k x x x x m K x k x x x k x x k k x k k k x x x k k k x x k k x x k x x k k k k x x k x x x K x k x K x x k x x x K x k x K x x k x K x k x x x K x x k x k k k x x k k x x k x x k k k x x k x x k k k k x x k x x K x k x K x x k x x x K x k x K x m m ( ) ( ) ( ) ( ) ( ) ( )( ) ( ) ( ) ( ) ( )(1 ) 1 ( )(1 ) 1 (1 ) 1 ( ) ( ) ( ) ( ) ( ) (1 ) 1 ( )(1 ) 1 (12) ee r p r r r r i r ir u u r ur wr wr w ee r r r u r ur c r cr 1 1 2 w 7 2 1 25 8 7 4 4 2 3 4 2 p 2 1 8 3 m 2 7 2 1 8 8 2 3 8 6 9 7 2 2 4 10 3 5 6 8 2 3 8 4 9 4 7 2 1 7 4 2 2 4 5 i 1 8 5 m 5 i 5 m 5 6 2 1 8 6 m 2 6 u 2 6 m 2 6 7 7 m 7 w 1 8 7 m 7 8 w 1 25 2 7 2 8 7 4 10 9 8 2 3 8 4 6 9 10 7 2 8 7 4 2 2 10 11 u 5 11 m 11 u 11 m 11 12 c 1 8 12 m 12 c 12 m 12 where x1 = G2K, x2 = R, x3 = G1K, x4 = G2R, x5 = IE, x6 = UbE2, x7 = Wee1, x8 = PG2, x9 = G1R, x10 = PG2R, x11 = UbE, and x12 = Cdc25, are concentration variables, see Novak and Tyson34 for a more detailed explanation of the meaning of these variables and values of the rate constants. www.nature.com/scientificreports/              β α β β β β µ      = − + + + + + = − − + + + − + − + + + + + + = − + + + = − + + + = + − + − − + = + − + − − + = − + − − + + = − + + + + = − + + = − + + + = − + − − + = + − + − − + = ′ ′ ′ ′ ′ x k k k k x x k x k k x x k k x k x x x x m K x k x x x k x x k k x k k k x x x k k k x x k k x x k x x k k k k x x k x x x K x k x K x x k x x x K x k x K x x k x K x k x x x K x x k x k k k x x k k x x k x x k k k x x k x x k k k k x x k x x K x k x K x x k x x x K x k x K x m m ( ) ( ) ( ) ( ) ( ) ( )( ) ( ) ( ) ( ) ( )(1 ) 1 ( )(1 ) 1 (1 ) 1 ( ) ( ) ( ) ( ) ( ) (1 ) 1 ( )(1 ) 1 (12) ee r p r r r r i r ir u u r ur wr wr w ee r r r u r ur c r cr 1 1 2 w 7 2 1 25 8 7 4 4 2 3 4 2 p 2 1 8 3 m 2 7 2 1 8 8 2 3 8 6 9 7 2 2 4 10 3 5 6 8 2 3 8 4 9 4 7 2 1 7 4 2 2 4 5 i 1 8 5 m 5 i 5 m 5 6 2 1 8 6 m 2 6 u 2 6 m 2 6 7 7 m 7 w 1 8 7 m 7 8 w 1 25 2 7 2 8 7 4 10 9 8 2 3 8 4 6 9 10 7 2 8 7 4 2 2 10 11 u 5 11 m 11 u 11 m 11 12 c 1 8 12 m 12 c 12 m 12 where x1 = G2K, x2 = R, x3 = G1K, x4 = G2R, x5 = IE, x6 = UbE2, x7 = Wee1, x8 = PG2, x9 = G1R, x10 = PG2R, x11 = UbE, and x12 = Cdc25, are concentration variables, see Novak and Tyson34 for a more detailed explanation of the meaning of these variables and values of the rate constants. www.nature.com/scientificreports/ The fact that variable x6 is not included, prevents from a full observability and, in particular, its absence induces a lack of symmetry, as previously discussed. Again, by looking at the distribution of Ni(0.9) we have: N1(0.9) = 5, N3(0.9) = 5, N7(0.9) = 14, N8(0.9) = 14, and N9(0.9) = 20. Therefore, for this level of observability (η = 0.9) the last three variables, x7,x8, and x9 can be again chosen to be removed from the set of observations. If we accept a slightly lower observability coefficient (η = 0.80), other possibilities emerge in which variable x6 is systematicaly removed from the set of measured variables (last part of Table 2). Our two criteria are thus very efficient to detect those variables not really impacting the access to a full observability measure, but discard some possibilities offering a good (but not full until m ≤ 6) observability with severe consequences on the symmetry properties of the reconstructed attractor. A 13D model for DNA replication. A third and even more challenging case is now considered, a 13-dimensional model for the DNA replication in fission yeast. Fission yeast cells are carrying two mutant genes, wee1− and cdc25OP, which initiate mitosis in eukaryotic cells before the end of their DNA replication. A second feature is that DNA synthesis can be restarted without intervening mitoses. Novak and Tyson proposed a model for cell cycle in fission yeast taking into account these two properties in Schizosuccharomyces pombe34. Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w This 13-dimensional rational model is characterized by the symbolic Jacobian matrix 8 Following the same procedure as with the two previous examples, we performed our a priori analysis by sys- tematically computing the symbolic observability coefficients when a single variable is removed and collecting only those combinations providing either full or null observability (see Table 3). As expected, when one of the variables x1, x4, x8, x9 or x10 is not included in the observation set of 12 variables plus one derivative, the observ- ability is full. On the contrary, if one of the variables x6, x7, x11 and x12 is removed, the symbolic observability coefficient drops to zero for any possible choice of the first derivative. These variables are therefore essential and need to be measured. This is due to the fact that these variables have no out-connection other than to themselves as shown in the corresponding columns of the symbolic Jacobian matrix in Eq. (13). Let us now validate whether it is possible to retrieve a full observability when either the set {x1, x4, x9, x10} or {x4, x8, x9, x10} are removed from the list of variables to measure. This was performed by systematically computing the symbolic observability coefficients for all the combinations reconstructing a 13-dimensional space without taking into account those two sets of variables. We found that for this DNA model, there are not too many possi- bilities to reconstruct a space providing full observability of the original state space (see the Supplementary Table S1). For instance, when removing two of them, x8 and x9 the reconstructed state vector    x x x x x x x x x x x x x ( , , , , , , , , , , , , ) 1 1 2 2 3 3 4 5 6 7 11 12 13 is the only one providing full observability. This 13-dimensional rational model is characterized by the symbolic Jacobian matrix ~ =       . J 1 1 0 1 0 0 0 1 0 0 0 0 0 1 1 1 1 0 0 0 1 1 1 0 0 1 0 1 1 0 0 0 0 0 1 0 0 0 0 1 1 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 1 0 0 0 0 0 1 0 0 0 0 1 0 1 0 0 0 0 0 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 0 1 0 0 0 0 1 1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 (13) 13D (13) According to the linear out-strengths σout lin, we have five candidate variables eligible to be excluded from the observations since having not null σout lin values: σ σ σ σ σ = = < = = = . (1) (8) 1 (4) (9) (10) 2 out lin out lin out lin out lin out lin The linear off-diagonal elements = J( 1) ij are J14, J18, J24 , J29 , J2,10  , J39, J81 , and J8,10  . There is therefore a single pair of exclusive candidate variables, that is, variables x1 and x8 (J J 1 18 81 = =   ). Variables x4, x9, and x10 can therefore be safely removed from the set of measured variables and x1 and x8 can not be simultaneously removed. The in-strength in nm σ from the set of potentially non-measured variables is equal to 2 for all the candidate variables except for x9 which is σ = (9) 0 in nm . Our criteria does not allow us this time to resolve the uncertainty between x1 and x8. www.nature.com/scientificreports/ (12) where x1 = G2K, x2 = R, x3 = G1K, x4 = G2R, x5 = IE, x6 = UbE2, x7 = Wee1, x8 = PG2, x9 = G1R, x10 = PG2R, x11 = UbE, and x12 = Cdc25, are concentration variables, see Novak and Tyson34 for a more detailed explanation of the meaning of these variables and values of the rate constants. where x1 = G2K, x2 = R, x3 = G1K, x4 = G2R, x5 = IE, x6 = UbE2, x7 = Wee1, x8 = PG2, x9 = G1R, x10 = PG2R, x11 = UbE, and x12 = Cdc25, are concentration variables, see Novak and Tyson34 for a more detailed explanation of the meaning of these variables and values of the rate constants. where x1 = G2K, x2 = R, x3 = G1K, x4 = G2R, x5 = IE, x6 = UbE2, x7 = Wee1, x8 = PG2, x9 = G1R, x10 = PG2R, x11 = UbE, and x12 = Cdc25, are concentration variables, see Novak and Tyson34 for a more detailed explanation of the meaning of these variables and values of the rate constants. where x1 = G2K, x2 = R, x3 = G1K, x4 = G2R, x5 = IE, x6 = UbE2, x7 = Wee1, x8 = PG2, x9 = G1R, x10 = PG2R, x11 = UbE, and x12 = Cdc25, are concentration variables, see Novak and Tyson34 for a more detailed explanation of the meaning of these variables and values of the rate constants. Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w 8 www.nature.com/scientificreports/ This 13-dimensional rational model is characterized by the symbolic Jacobian matrix If we discard three variables (x8, x9, and x10), there are two combinations allowing for a full observabiltiy embedding:   x x x x x x x x x x x x x ( , , , , , , , , , , , , ) 1 1 2 2 3 4 5 6 7 10 11 12 13 and   x x x x x x x x x x x x x ( , , , , , , , , , , , , ) 1 1 2 3 3 4 5 6 7 10 11 12 13 . A system- atic analysis of the symbolic observability coefficients as the number of variables are removed from the set of observations indicate that the coefficient already drops to 0.93 when m = 9 variables are measured (see Fig. 3). As the estimated threshold for an optimal observability is 0.7533, it is worthless to investigate sets of size smaller than m = 7.fi To actually check the results accounted for the symbolic observability coefficients ηs, we computed all the determinants Det s corresponding to ηs = 1 (results are reported in the Supplementary Table S1). In the 14 cases for which the symbolic observability coefficients are equal to one, the determinant Det s was always nonzero (for the whole state space): our technique always correctly identify reconstructed vectors providing full observability of the original space. As another example, as shown in Fig. 3, full observability is never achieved for m = 9 and the largest symbolic observability coefficient is 0.93, which still provides a good observability. www.nature.com/scientificreports/ www.nature.com/scientificreports/ Non-measured Derivative retained η x1 x8 1.00 x4 x1 or x2 1.00 x8 x1 1.00 x9 x2 or x3 1.00 x10 x2 or x8 1.00 x6 ∀xi 0.00 x7 xi ∀ 0.00 x11 ∀xi 0.00 x12  ∀xi 0.00 Table 3. Symbolic observability coefficients when twelve (out of thirteen) variables of the DNA model (12) are measured. The derivative used for reconstructing a 13-dimensional state space is also reported. Table 3. Symbolic observability coefficients when twelve (out of thirteen) variables of the DNA model (12) are measured. The derivative used for reconstructing a 13-dimensional state space is also reported. 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Number of measured variables m 0 0,2 0,4 0,6 0,8 1 η Figure 3. Largest symbolic observability coefficient η versus the number m of measured variables for the DNA model (12). 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Number of measured variables m 0 0,2 0,4 0,6 0,8 1 η Figure 3. Largest symbolic observability coefficient η versus the number m of measured variables for the DNA model (12). k k k k k k k k k x x K x K x Det ( ) ( ) ( ) ( 1) ( 1) ( 1 ) ( 1 ) (15) s 7r 2 2p 8r 4 7r 4 u c 11 12 mc 12 11 β = − + + + + − − + − + − µ  (15) associated with the reconstructed space       x x x x x x x x x x x x x x ( , , , , , , , , , , , , , ) 13 1 1 2 2 3 3 6 7 10 11 11 12 12 13 providing a slightly smaller observability (η = 0.86) in agreement with the singular observability manifold (15) of second order, defined by Det s = 0, that is, by (x11 − 1)(x12 − 1) = 0. Due to a too large complexity, it was not possible to analytically compute the observability matrix when a single variable is measured. As detailed in the Supplementary Section S1, Liu and coworkers’ graphical technique shows that by measuring the four variables x6, x7, x11, and x12 it is possible to estimate the states of the system. This 13-dimensional rational model is characterized by the symbolic Jacobian matrix By using the recon- structed space      x x x x x x x x x x x x x x ( , , , , , , , , , , , , , ) 13 1 1 2 2 3 3 5 6 7 10 11 12 12 13 , (one of the cases reported in Table 4) the determinant  k k k k k k k k x K x Det ( ) ( ) ( ) ( 1) 1 (14) r 7 2 2p 8r 4 7r 4 c 12 mc 12 β = − + + + + − + − (14) is zero for x12 = 1, a singular observability manifold of first order, explaining why the observability coefficient is no longer equal to 1 but close to it. To further show how the observability coefficient decreases when Det s van- ishes for a singularity of higher degree22, we computed the determinant is zero for x12 = 1, a singular observability manifold of first order, explaining why the observability coefficient is no longer equal to 1 but close to it. To further show how the observability coefficient decreases when Det s van- ishes for a singularity of higher degree22, we computed the determinant Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w 9 www.nature.com/scientificreports/ Clearly, our results are in strong disagreement. At this point, it is relevant to explain why our results are so different from those reported in previous works16,17. The first reason for the discrepancy is that Liu’s algorithm uses a linear theory, only taking into account whether the ith variable participates or not in the differential equation of variable j and not con- cerned on how is that dependence. The latter is just equivalent to use a symbolic Jacobian matrix equal to 1 1 0 1 0 0 0 1 0 0 0 0 0 1 1 1 1 0 0 0 1 1 1 0 0 1 0 1 1 0 0 0 0 0 1 0 0 0 0 1 1 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 1 0 0 0 0 0 1 0 0 0 0 1 0 1 0 0 0 0 0 1 0 0 0 0 0 1 1 0 0 0 0 0 1 1 0 0 0 0 0 1 0 1 0 0 0 0 1 1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 (16) 13D lin ~ =       . (16) and not the one defined in Eq. (13). Despite there are more than 2100 combinations with 6 or 7 measured varia- bles resulting in full observability, there is none with a single variable. However, with a linear approach, it is still possible to show that the combinations providing full observability correctly identify the variables which must necessarily be used (Table 5), that is, variables x6, x7, x11, and x12. Nevertheless, the observability is obviously and not the one defined in Eq. (13). Despite there are more than 2100 combinations with 6 or 7 measured varia- bles resulting in full observability, there is none with a single variable. Discussionh The observability of a complex system refers to the property of being able to infer its whole state space by meas- uring the dynamics of a limited set of its variables. Determining the conditions that guarantee the full observ- ability of a system involves testing a number of possibilities that increases exponentially with the dimension of that system and, for each case, it is required to compute the determinant of the observability matrix defining the singular observability manifold, that is, the subset of the state space that cannot be observed from the measure- ment25. It was shown in one of our previous works2 that for a five-dimensional rational system, the analytical computation of such a determinant may already exceed the capacity of softwares like Maple® or Mathematica®. Therefore, alternative approaches to investigate large complex systems are needed. Those proposed for instance by Sedoglavic16 or Liu and coworkers17 remain yet unsatisfactory as discussed by Wang and coworkers35, mainly because they do not provide a method to select which Lie derivatives accompany the measured variables and, more importantly, they do not consider a nonlinear observability theory appropriate to deal with nonlinear sys- tems, nonlinearities occuring in the node dynamics or nonlinearly coupled units. g y y p Actually, the treatment proposed by Sedoglavic16 is only probabilistic and tests local observability, not the global one. On the other hand, the graphical approach developed by Liu and co-workers17 is based on a linear description of the system which can only lead, by definition, to approximated results since, as previously dis- cussed, the lack of observability mainly originates in the location (in the fluence graph) of nonlinear terms. We here investigated the three systems considered by Liu and coworkers (see the Supplementary Section S1) and showed that, in contrast with our results, theirs are not in agreement with the analytical predictions. In our previous work2, we investigated the same five-dimensional rational system considered by Sedoglavic16. While in the latter reference, the algorithm developed by the author identifies the first variable as the one providing (in fact local) observability, it is only poorly the case when the symbolic algorithm developed in the former one is applied to the possible combinations using this variable, even combined with other variables. www.nature.com/scientificreports/ The first part corresponds to the case where variables {x4, x8, x9, x10} are not measured, the middle part to the case where variables {x1, x4, x9, x10} are not. Only the cases where the symbolic coefficient is non-zero and for which only a first derivative is used (to avoid too many possibilities) are reported. overestimated (1 compared to 0.43 with a nonlinear theory) and, in addition, this approach does not allow to select what derivatives to use for spanning the reconstructed space. Assessing the observability of a network with a linear theory thus provides very poor and misleading results. www.nature.com/scientificreports/ However, with a linear approach, it is still possible to show that the combinations providing full observability correctly identify the variables which must necessarily be used (Table 5), that is, variables x6, x7, x11, and x12. Nevertheless, the observability is obviously Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w 10 www.nature.com/scientificreports/ m x1 x2 x3 x4 x5 x6 x7 x8 x9 x10 x11 x12 x13 η 9 2 2 2 — 1 1 1 — — — 1 2 1 0.93 9 2 2 2 — 1 1 2 — — — 1 1 1 0.93 9 2 2 2 — 1 2 1 — — — 1 1 1 0.93 9 2 2 2 — 2 1 1 — — — 1 1 1 0.93 8 2 2 2 — — 1 1 — — — 2 2 1 0.86 8 2 2 2 — — 1 2 — — — 2 1 1 0.86 8 2 2 2 — — 2 1 — — — 2 1 1 0.86 9 — 2 2 — 1 1 1 2 — — 1 2 1 0.93 9 — 2 2 — 1 1 2 2 — — 1 1 1 0.93 9 — 2 2 — 1 2 1 2 — — 1 1 1 0.93 9 — 2 2 — 2 1 1 2 — — 1 1 1 0.93 8 — 2 2 — — 1 1 2 — — 2 2 1 0.86 8 — 2 2 — — 1 2 2 — — 2 1 1 0.86 8 — 2 2 — — 2 1 2 — — 2 1 1 0.86 7 — 1 2 — — 1 1 3 — — 2 3 — 0.72                Table 4. Symbolic observability coefficients for the DNA system. The first part corresponds to the case where variables {x4, x8, x9, x10} are not measured, the middle part to the case where variables {x1, x4, x9, x10} are not. Only the cases where the symbolic coefficient is non-zero and for which only a first derivative is used (to avoid too many possibilities) are reported. Table 4. Symbolic observability coefficients for the DNA system. Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w Discussionh m N M1 M2 M3 M4 M5 M6 6 230 9 87 111 62 0 230 7 1896 307 879 981 728 327 1892 m M7 M8 M9 M10 M11 M12 M13 6 230 9 110 63 230 230 9 7 1892 303 978 722 1891 1891 470 Table 5. Number N of combinations providing a full observability—according to a linear theory—when m variables are measured. The numbers Mi in which the ith variable is involved in a vector spanning the reconstructed state space providing a full observability are also reported. In bold, the four variables which are the most often involved. Table 5. Number N of combinations providing a full observability—according to a linear theory—when m variables are measured. The numbers Mi in which the ith variable is involved in a vector spanning the reconstructed state space providing a full observability are also reported. In bold, the four variables which ar the most often involved. Table 5. Number N of combinations providing a full observability—according to a linear theory—when m variables are measured. The numbers Mi in which the ith variable is involved in a vector spanning the reconstructed state space providing a full observability are also reported. In bold, the four variables which are the most often involved. the observability matrix. For the 42 combinations providing η = 1, the success rate is 100%. While we were able to identify and algebraically check all the resulting combinations for the 5D and 9D models, it was impossible for the 13D model due to the large amount of them. In the case of the 9D model, for which we obtained mp = 6 pre- selected variables, our procedure missed 2 out of 8 combinations with m = 7, and 4 out of 6 with m = 8 (see the Supplementary Table S1). The missed combinations involve at least one variable which was not preselected and, consequently, not considered in our computations. When m = 6, some combinations are associated with a sym- bolic observability coefficient equal to 0.90: in that case, 36 out of 54 corresponding to this value of the symbolic observability coefficient were made up of the preselected variables. When m = 5 < mp, 100% of the combinations (34) associated with the largest symbolic observability coefficient (0.90) involved the preselected variables. Methods Introduction to observability theory. Our framework to quantify the observability of a dynamical sys- tem is here introduced with some definitions. Let us consider a d-dimensional dynamical system represented by the state vector x ∈ d whose components are given by x f x x x x i d ( , , , , ), 1, 2, 3, , (17) i i d 1 2 3 = … = … x f x x x x i d ( , , , , ), 1, 2, 3, , i i d 1 2 3 = … = … (17) x f x x x x i d ( , , , , ), 1, 2, 3, , (17) i i d 1 2 3 = … = … fi is the ith component of the vector field f. m where fi is the ith component of the vector field f. where fi is the ith component of the vector field f. i Let us introduce the vector ∈ s m  whose m components are the time series of measured variables given by the measurement function Let us introduce the vector ∈ s m  whose m components are the time series of measured variables given by the measurement function s h x ( ) (18) = . (18) s h x ( ) = . One of the formal ways to define the observability of a system is as follows40. We provide such a definition in the case where a single scalar time series is measured, s = h(x), but a generalization to the case of m measured variables is straighforward. The dynamical system (17) is said to be state observable at time tf if every initial state x(0) can be uniquely determined from the knowledge of a finite time series of the measured variable s(τ), 0 ≤ τ ≤ tf. In practice, it is possible to test whether the dynamical system (17) is observable through a measurement function by computing the rank of the observability matrix20, that is, the Jacobian matrix of the Lie derivatives of s. Discussionh It is important to note that, when m ≤ 6, all combinations providing the largest symbolic observability coefficient (see Fig. 2) are made up of the preselected variables (and are actually found). This means that the preselected variables are indeed the revelant ones for estimating the system states and, that all combinations using these variables are correctly identified. To the best of our knowledge, we have a single case for which the full observability was not detected by our procedure33: it corresponds to the rare case for which two nonlinear terms cancel each other in the computation of the determinant of the observability matrix.i p y Finally, as firstly reported by Parlitz and coworkers36, the observality of a system could be addressed by using delay coordinates. As shown by Gibson and coworkers37, delay coordinates are related to derivatives by a rotation and a rescaling. Consequently, any result valid for derivative coordinates (not affected by rotation and rescal- ing) holds for delay coordinates. The reduced sets of m measured variables (m < d) are not dependent on the use of delay or derivative coordinates, only on the choice of the complementary coordinates to reconstruct a d-dimensional space. Therefore, the extension of the technique proposed in this work to networks of discrete time systems (discretization of continuous-time systems) and iterated maps seems to be rather straightforward according, for instance, to Sarachik and Kreindler38 and to Nijmeijer39, respectively. Discussionh And what is even more questionable, it is that when x1 is combined with one of the four other variables, x2, x3, x4 and x5, the largest symbolic observability coefficient is still very small, that is, 0.30, 0.18, 0.30, and 0.48, respectively.fi y yfi y p y We have shown how the efficiency of the algorithm initially proposed by Bianco-Martinez and coworkers2 is improved by identifying the minimal set of measured variables providing full observability before any search for the corresponding Lie derivatives. The reduced sets of candidate variables capable of fully reconstructing large reaction networks was correctly determined by analyzing the way the variables interact, by only applying two sim- ple criteria on the symbolic Jacobian matrix of the networked system. For the 13 DNA model, there are 5.2 ⋅ 106 possible combinations to test (see the Supplementary Section S2 for the details), requiring more than 18 days of computations with a 2.5 GHz Intel Core i5 processor. With our preselection of variables, only 2870 combinations are needed to be tested lowering the computation time to about 4 min, that is, by a factor greater than 1800! These criteria reduce drastically the time spent for searching candidate variables, thus providing the grounds to observe natural and man-made complex systems.i p y In order to evaluate the reliability of our procedure, we computed a success rate defined as the number of times a symbolic observability coefficient equal to 1 actually corresponds to getting a constant analytical determinant of 11 www.nature.com/scientificreports/ m N M1 M2 M3 M4 M5 M6 6 230 9 87 111 62 0 230 7 1896 307 879 981 728 327 1892 m M7 M8 M9 M10 M11 M12 M13 6 230 9 110 63 230 230 9 7 1892 303 978 722 1891 1891 470 Table 5. Number N of combinations providing a full observability—according to a linear theory—when m variables are measured. The numbers Mi in which the ith variable is involved in a vector spanning the reconstructed state space providing a full observability are also reported. In bold, the four variables which are the most often involved. Methods Showing that a measurement function is generic is not a trivial problem which is out of the scope of the present work. There is, therefore, no guarantee that the Takens theorem applies here. Moreover, our aim is to select the minimal set of measurements providing the best observability of the system. When a higher-dimensional reconstructed space is considered, this means that a global diffeomorphism perhaps may be obtained but it also means that the meas- urements provide information that is non-optimal and from which the analysis is most likely problematic and tricky42,43. Consequently, investigating higher-dimensional reconstructed spaces has a rather limited interest in the present context.h The fact the system is fully observable from the two measured variables considered in the matrix (20) depends also on the particular choice of the pair (k, l), the numbers of successive derivatives computed from xi and xj, respectively. Therefore, it is crucial to specify how the measured variables and their derivatives are used to recon- struct the state space. An approach–as the ones developed in other works16–18–missing this necessary condition cannot indeed properly address the problem of full (or even good) observability. Symbolic observability formalism. The procedure to calculate the symbolic observability coefficients is implemented in four steps as follows: i) Construction of the symbolic Jacobian matrix (∼ ). The Jacobian matrix , composed of elements Jij, of the system (17) is transformed into a symbolic Jacobian matrix ∼  by replacing each linear element Jij by 1, each non-linear polynomial element Jij by, and each rational element Jij by when the j th variable is present in the denominator or by 1 otherwise. This is more or less equivalent to the so-called influence (or fluence) diagram as used by Letellier and Aguirre23 where linear and nonlinear coupling terms are associated with solid and dashed arrows, respectively, and as used by Liu and coworkers17 where coupling terms are labelled with arrows (without distinguishing linear from nonlinear couplings). In the present approach, rational terms are distinguished from nonlinear polynomial terms since they strongly reduce the observability2.  ii) Construction of the symbolic observability matrix (Os). Let us consider for simplicity a univariate measurement s = h(x) = xi. For this particular case, the first row of Os is just defined by the derivative of the measurement function dh, that is, O 1 j1 = if j = i and 0 otherwise. Methods Differentiating s(t) yields   x x x x f x x s t t h h h h ( ) d d ( ) ( ) ( ), f  = = ∂ ∂ = ∂ ∂ =   x x x x f x x s t t h h h h ( ) d d ( ) ( ) ( ), f  = = ∂ ∂ = ∂ ∂ = where f h(x) is the Lie derivative of h along the vector field f. The jth order Lie derivative is given by   = ∂ ∂ − x x f x h h x ( ) ( ) ( ), f j f j 1   = ∂ ∂ − x x f x h h x ( ) ( ) ( ), f j f j 1 being the zero order Lie derivative the measured variable itself, x x h h ( ) ( ) f 0  = . Therefore, the observability matrix  ∈ × s d d  can be written as being the zero order Lie derivative the measured variable itself, x x h h ( ) ( ) f 0  = . Therefore, the observability matrix  ∈ × s d d  can be written as Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w 12 www.nature.com/scientificreports/  =       − O L L x x x x h h h ( ) d ( ) d ( ) d ( ) (19) s f f d 1 (19) where d x ≡ ∂ ∂ and the dynamical system (17) is said to be state observable if and only if the observability matrix has full rank, that is, rank (s) = d. Notice that, the full observability of a system is determined by the space spanned not only by the measured variables but also by their appropriate Lie derivatives1.h The observability matrix s corresponds in fact to the Jacobian matrix of the change of coordinates Φs: x → X where ∈ X d  is the reconstructed state vector from the m measured variables and their adequately chosen d − m Lie derivatives30. Methods Let us make explicit the observability matrix x x i j  for the situation where two arbitrary variables xi and xj are measured directly, that is, when s = (h1(x),h2(x)) = (xi, xj), and h1 and h2 are two measurement func- tions. In this case, the observability matrix reads as, O L L L L x x x x x x h h h h h h d ( ) d ( ) d ( ) d ( ) d ( ) d ( ) (20 x x f f k f f l 1 1 1 2 2 2 i j i i j j   =       (20) where the order Lie derivatives k and l are such that k + l = d − 2, that is, there are d − 2 + 1 possibilities for choosing k and l. According to the Takens theorem41, it is possible to increase the dimension up to 2dH + 1 where dH is ideally the Haussdorff dimension of the attractor to ensure a global diffeomorphism between the original state space and the reconstructed one, as long as the measurement function is generic. Showing that a measurement function is generic is not a trivial problem which is out of the scope of the present work. There is, therefore, no guarantee that the Takens theorem applies here. Moreover, our aim is to select the minimal set of measurements providing the best observability of the system. When a higher-dimensional reconstructed space is considered, this means that a global diffeomorphism perhaps may be obtained but it also means that the meas- urements provide information that is non-optimal and from which the analysis is most likely problematic and tricky42,43. Consequently, investigating higher-dimensional reconstructed spaces has a rather limited interest in the present context.h where the order Lie derivatives k and l are such that k + l = d − 2, that is, there are d − 2 + 1 possibilities for choosing k and l. According to the Takens theorem41, it is possible to increase the dimension up to 2dH + 1 where dH is ideally the Haussdorff dimension of the attractor to ensure a global diffeomorphism between the original state space and the reconstructed one, as long as the measurement function is generic. Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w Second, the resulting symbolic Jacobian matrix ∼′ is thus reduced into a row where each element O J kj i ij = ∑′  is just the sum of the elements of the jth column according to the addition law2 y iv) Selecting the miminal set of variables to measure and the adequate Lie derivatives for providing a full observability. The symbolic observability coefficients ηs for each one of the sets of m measured variables and their selected d − m Lie derivatives are ranked versus the decreasing value of ηs and increasing m. Those featuring ηs = 1 and the smallest m can be selected as the minimal sets of variables to measure. References 1. Aguirre, L. A. & Letellier, C. Observability of multivariate differential embeddings. J. Phys. A 38, 6311 (2005). 2. Bianco-Martinez, E., Baptista, M. S. & Letellier, C. Symbolic computations of non-linear observability. Phys. Rev. E (2015).h ( ) 3. Pagani, G. A. & Aiello, M. The Power Grid as a complex network: A survey. Physica A 392, 2688–2700 (2013).hf gh y y 4. Ajrouch, K. J., Blandon, A. Y. & Antonucci, T. C. Social networks among men and women: The effects of age and socioeconomic status. J. Gerontol. B 60, S311–S317 (2005). 5. Moore, C., Cumming, G. S., Slingsby, J. & Grewar, J. Tracking socioeconomic vulnerability using network analysis: Insights from an avian influenza outbreak in an ostrich production network. PLoS One 9, e86973 (2014). l p 6. Varela, L. M., Rotundo, G., Ausloos, M. & Carrete, J. Complex network analysis in socioeconomic models. Complexity Geographical Economics, Dynamic Modeling and Econometrics in Economics and Finance 19, 209–245 (2015). Geographical Economics, Dynamic Modeling and Econometrics in Economics and Finance 19, 209 245 (2015). 7. Colcombet, J. & Hirt, H. Arabidopsis MAPKs: a complex signalling network involved in multiple biological processes. Biochem. J 413, 217–226 (2008). 7. Colcombet, J. & Hirt, H. Arabidopsis MAPKs: a complex signalling network involved in multiple biolo 413, 217–226 (2008). 8. Rubinov, M. & Sporns, O. Complex network measures of brain connectivity: Uses and interpretations. Neuroimage 52, 1059–1069 (2010).i ( ) 9. Raue, A., Becker, V., Klingmüller, U. & Timmer, J. Identifiability and observability analysis for experimental design in nonlinear dynamical models. Chaos 20, 045105 (2010). y 0. Sendiña-Nadal, I. et al. Unveiling protein functions through the dynamics of the interaction network. PLoS One 6, e17679 (2011). 10. Sendiña Nadal, I. et al. Unveiling protein functions through the dynamics of the interaction network. PLoS One 6, e17679 (2011). 11. Allefeld, C. & Kurths, J. An approach to multivariate phase synchronization analysis and its application to event-related potentials. Int. J. Bifurcat. Chaos 14, 417–426 (2004). f 2. Donner, R. Multivariate analysis of spatially heterogeneous phase synchronisation in complex systems: application to self-organised control of material flows in networks. Eur. Phys. J. B 63, 349–361 (2008).fi l 13. Letellier, C. & Aguirre, L. A. Symbolic observability coefficients for univariate and multivariate analysis. Phys. Rev. E 79, 066210 (2009). 14. Slutsker, I. W. & Scudder, J. M. Network observability analysis through measurement Jacobian matrix reduction. Second, the resulting symbolic Jacobian matrix ∼′ is thus reduced into a row where each element O J kj i ij = ∑′  is just the sum of the elements of the jth column according to the addition law2 Second, the resulting symbolic Jacobian matrix ∼′ is thus reduced into a row where each elemen O J kj i ij = ∑′  is just the sum of the elements of the jth column according to the addition law2 a a a a a a a a a 0 , 1 for 0, 1 for 0, 1, 1 1 (22 ⊕ = ⊕ = ≠ ⊕ = ≠ ⊕ = . (22) In the case m variables are measured, the construction of Os is performed by blocks of size (di + 1) × d, being di the number of derivatives of si and d m d i m i 1 ∑ + = = , and the construction of each block follows the same rules described above for univariate measures.fih  In the case m variables are measured, the construction of Os is performed by blocks of size (di + 1) × d, being di the number of derivatives of si and d m d i m i 1 ∑ + = = , and the construction of each block follows the same rules described above for univariate measures. iii) Computation of the symbolic observability coefficients. The determinant of Os is computed according to the symbolic product rule defined in (21) and expressed as products and addends of the symbolic terms 1, 1 and 1, whose number of occurrences are stored in variables N1, N and N, respectively. A special condition is required for rational systems such that, if N = 0 and N ≠ 0 then N = N. The symbolic observability coefficient for the measurement s is then equal to η = + + D N D N D N 1 1 1 (23) s 1 2 1 3 1 (23) with D = max(1, N1) + N + N and 0 ≤ ηs ≤ 1, being ηs = 1 for a combination providing full observability. y iv) Selecting the miminal set of variables to measure and the adequate Lie derivatives for providing a full observability. The symbolic observability coefficients ηs for each one of the sets of m measured variables and their selected d − m Lie derivatives are ranked versus the decreasing value of ηs and increasing m. Those featuring ηs = 1 and the smallest m can be selected as the minimal sets of variables to measure. Methods The second row is directly obtained from ∼ by copying its ith row, that is, O J j ij 2   = ∀j, being i the index of the measured variable. The kth row is obtained as follows. First, each element of the ith row of the symbolic Jacobian observability matrix ∼  is multiplied by the corresponding ith component of the vector = v O O ( , , ) d 1 ℓ ℓ where = −  k 1 refers to the (k − 1)th row of the symbolic observability matrix Os. The rules to perform the symbolic product ⊗ J v ij i are such that2 Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w 13 www.nature.com/scientificreports/ ⊗ = ⊗ = ⊗ = = ⊗ = ≠ . a a a a a a a a 0 0, 1 , 1 for 1, 1, 1 1 for 0 (2 (21) Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w References Detecting strange attractors in turbulence. Lecture Notes in Mathematics 898, 366–381 (1981). 42. Letellier, C. Estimating the Shannon entropy: recurrence plots versus symbolic dynamics. Physical Review Letters 96, 254102 (2006).i 42. Letellier, C. Estimating the Shannon entropy: recurrence plots versus symbolic dynamics. Physical Review Letters 96, 254102 (2006). 43. Portes, L. L., Benda, R. N., Ugrinowitsch, H. & Aguirre, L. A. Impact of the recorded variable on recurrence quantification analysis of flows. Physics Letters A 378, 2382–2388 (2014). 42. Letellier, C. Estimating the Shannon entropy: recurrence plots versus symbolic dynamics. Physical Review Letters 96, 254102 (2006). 43 Portes L L Benda R N Ugrinowitsch H & Aguirre L A Impact of the recorded variable on recurrence quantification analysis Author Contributions C.L. and M.S.B. conceived the study. E.B.-M. and I.S.-N. prepared the MATLAB codes and made the symbolic calculations. CL performed the analytical computations. C.L., M.S.B. and I.S.-N. wrote the Manuscript. Acknowledgements g EBM and MSB acknowledge the Engineering and Physical Sciences Research Council (EPSRC), grant Ref. EP/ I032608/1. ISN acknowledges partial support from the Ministerio de Economía y Competitividad of Spain under project FIS2013-41057-P and from the Group of Research Excelence URJC-Banco de Santander. References IEEE Transactions on Power Systems 2(2), 331–336 (1987).i y 5. Bretas, N. G. & London, J. B. A. Network observability: the critical measurement identification using the symbolic Jacobian matrix IEEE Proceedings of the International Conference on Power System Technology 2, 1222–1226 (1998). g f f y gy 6. Sedoglavic, A. A probabilistic algorithm to test local algebraic observability in polynomial time. J. Symb. Comput. 33, 735–755 (2002). ( ) 7. Liu, Y.-Y., Slotine, J.-J. & Barabási, A.-L. Observability of complex systems. Proc. Natl. Acad. Sci. (USA) 110, 2460–2465 (2013). Liu, Y.-Y., Slotine, J.-J. & Barabási, A.-L. Observability of complex sy 8. Whalen, A. J., Brennan, S. N., Sauer, T. D. & Schiff, S. J. Observability and controllability of nonlinear networks: The role o symmetry. Phys. Rev. X 5, 011005 (2015).f y y y 9. Leitold, D., Vathy-Fogarassy, Á. & Abonyi, J. Controllability and observability in complex networks – the effect of connection types Scientific Reports 7, 151 (2017). fi p ( ) 20. Hermann, R. & Krener, A. Nonlinear controllability and observability. IEEE Trans. Autom. Control 22, 728–740 (1977). y y 21. Letellier, C., Maquet, J., Le Sceller, L., Gouesbet, G. & Aguirre, L. A. On the non-equivalence of observables in phase space reconstructions from recorded time series. J. Phys. A 31, 7913–7927 (1998).hl y 2. Letellier, C. & Aguirre, L. A. Investigating nonlinear dynamics from time series: The influence of symmetries and the choice o observables. Chaos 12, 549–558 (2002). Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w 14 www.nature.com/scientificreports/ 3. Letellier, C. & Aguirre, L. A. A graphical interpretation of observability in terms of feedback circuits. Phys. Rev. E 72, 056202 (2005) 4. Yang, Y., Wang, J. & Motter, A. E. Network observability transitions. Phys. Rev. Lett. 109, 258701 (2012).l g g p p y y 24. Yang, Y., Wang, J. & Motter, A. E. Network observability transitions. Phys. Rev. Lett. 109, 258701 (2012).l 25. Frunzete, M., Barbot, J.-P. & Letellier, C. Influence of the singular manifold of non-observable states in reconstructing c attractors. Phys. Rev. E 86, 026205 (2012). y ( ) 26. Aguirre, L. A., Portes, L. L. & Letellier, C. Observability and synchronization of neuron models. Chaos 27, 103103 (2017). 26. Aguirre, L. A., Portes, L. L. & Letellier, C. Observability and synchronization of neuron models. Chaos 27, 103103 (2017). 27. Banaji, M., Donnell, P. & Baigent, S. Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w References P matrix properties, injectivity and stability in chemical reaction systems. SIAM J. Appl. Math. 67, 1523–1547 (2007). 7. Banaji, M., Donnell, P. & Baigent, S. P matrix properties, injectivity and stability in chemical reaction systems. SIAM J. Appl. Math 67, 1523–1547 (2007). 28. Letellier, C. & Gilmore, R. Covering dynamical systems: Two-fold covers. Phys. Rev. E 63, 016206 (2001). g y y y 29. Goldbeter, A. A model for circadian oscillations in the Drosophila period protein. Proc. Royal Soc. B 261, 319–324 (1995).f 30. Letellier, C., Aguirre, L. A. & Maquet, J. Relation between observability and differential embeddings for nonlinear dynamics. Rev. E 71, 066213 (2005). 31. Reiterer, P., Lainscsek, C., Schürrer, F., Letellier, C. & Maquet, J. A nine-dimensional Lorenz system to study high-dimensional chaos. J. Phys. A 31, 7121–7139 (1998). G l R & L ll C Th S f Ch (O f d U P ) y 32. Gilmore, R. & Letellier, C. The Symmetry of Chaos (Oxford University Press, 2007).fi h 33. Sendiña-Nadal, I., Boccaletti, S. & Letellier, C. Observability coefficients for predicting the class of synchronizability from the algebraic structure of the local oscillators. Phys. Rev. E 94, 042205 (2016).i g y ( ) 34. Novak, B. & Tyson, J. J. Modeling the control of DNA replication in fission yeast. Proc. Natl. Acad. Sci. 94, 9147–9152 (1997). G G C ll b l d b b l l f d l d 35. Wang, B., Gao, L., Gao, Y., Deng, Y. & Wang, Y. Controllability and observability analysis for vertex domination centrality in dir networks. Sci. Rep. 4, 5399 (2014).f 36. Parlitz, U., Schumann-Bischoff, J. & Luther, S. Local observability of state variables and parameters in nonlinear modeling quan by delay reconstruction. Chaos 24, 024411 (2014). y y 7. Gibson, J. F., Farmer, J. D., Casdagli, M. & Eubank, S. An analytic approach to practical state space reconstruction. Physica D 57 1–30 (1992). 38. Sarachik, P. E. & Kreindler, E. Controllability and observability of linear discrete-time systems. International Journal of Control 1(5), 419–432 (1965). ( ) 39. Nijmeijer, H. Observability of autonomous discrete time non-linear systems: a geometric approach. International Journal of Control 36(5), 867–874 (1982). 40. Kailath, T. Linear Systems (Prentice Hall, 1980). 40. Kailath, T. Linear Systems (Prentice Hall, 1980). y 41. Takens, F. Detecting strange attractors in turbulence. Lecture Notes in Mathematics 898, 366–381 (1981). 41. Takens, F. Additional Information upplementary information accompanies this paper at https://doi.org/10.1038/s41598-018-21967-w. Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-018-21967-w. Competing Interests: The authors declare no competing interests. Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre- ative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not per- mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Scientific REPOrTS \| (2018) 8:3785 \| DOI:10.1038/s41598-018-21967-w 15
https://openalex.org/W4253837667	https://newprairiepress.org/cgi/viewcontent.cgi?article=2145&context=kaesrr	English	null	Rate and extent of losses from top spoilage in pilot-scale, horizontal silos	Kansas Agricultural Experiment Station research reports	1,992	cc-by	1,955	Rate and extent of losses from top spoilage in pilot-scale, Rate and extent of losses from top spoilage in pilot-scale, horizontal silos horizontal silos Follow this and additional works at: https://newprairiepress.org/kaesrr Part of the Other Animal Sciences Commons Part of the Other Animal Sciences Commons ansas Agricultural Experiment Station Research Reports ansas Agricultural Experiment Station Research Reports ansas Agricultural Experiment Station Research Reports ansas Agricultural Experiment Station Research Reports Volume 0 Issue 1 Cattleman's Day (1993-2014) Volume 0 Issue 1 Cattleman's Day (1993-2014) Article 742 This research report is available in Kansas Agricultural Experiment Station Research Reports: https://newprairiepress.org/kaesrr/vol0/iss1/742 Recommended Citation Recommended Citation Dickerson, J.T.; Bolsen, K.K.; Brent, B.E.; Lin, C.; and Boyer, John E. (1992) "Rate and extent of losses from top spoilage in pilot-scale, horizontal silos," Kansas Agricultural Experiment Station Research Reports: Vol. 0: Iss. 1. https://doi.org/10.4148/2378-5977.2145 This report is brought to you for free and open access by New Prairie Press. It has been accepted for inclusion in Kansas Agricultural Experiment Station Research Reports by an authorized administrator of New Prairie Press. Copyright 1992 the Author(s). Contents of this publication may be freely reproduced for educational purposes. All other rights reserved. Brand names appearing in this publication are for product identification purposes only. No endorsement is intended, nor is criticism implied of similar products not mentioned. K-State Research and Extension is an equal opportunity provider and employer. Summary silages, particularly below the top 12 in., where OM losses were quite low in both crops. Corn and forage sorghum silages were stored in pilot-scale silos for 180 days, and dry matter (DM) and organic matter (OM) recover- ies and estimated OM recovery were measured at three depths within the top 3 ft. of silage. The unsealed silages deteriorated badly in the top 12 in. Actual DM and OM losses in the top 24 in. were higher in unsealed than sealed silages at each successive storage period (7 to 180 days). (Key Words: Silage, Top Spoilage, Pilot- scale, Ash.) Financial assistance was provided by Kemin Industries Inc., Des Moines, Iowa and Mr. Richard 1 Porter, Porter Farms, Reading, Kansas. Former garduate student. Current address: Biotal, Inc., Ft. Dodge, Iowa. 2 Former graduate student. Current address: Agri-King, Inc., Fulton, Illinois. 3 Department of Statistics. 4 Experimental Procedures of the corn and forage sorghum silages are presented in Table 1. Experiment 1: whole-plant corn. On August 28 and 29, 1990, 45 polyethylene- lined, 55-gallon capacity drums were packed to equal densities with whole-plant corn similar to that used to fill the farm-scale, bunker silos described on page 132 of this report. Each drum was divided horizontally with plastic netting into thirds to partition the fresh material at 12 and 24 in. below the original surface. A perforated 1.0 in. PVC pipe at the bottom of each drum drained off percolated water. Treatments were 1) left unsealed; 2) sealed with .4 mm polyethylene sheeting; 3) sealed with .4 mm polyethylene sheeting after a topical application of a commercial mold inhibitor, Top Savor®, at 1.0 lb/10 sq. ft. (provided by Kemin Industries, Inc., Des Moines, Iowa); and 4) left unsealed for 7 days post-filling, then sealed as described for treat- ment 3. Drums were stored outside and opened at 7, 21, 90, and 180 days post-filling (3 drums/treatment/opening time). The silage was weighed, mixed, and sampled at each location and processed as described for the farm-scale silages. Experiment 1: whole-plant corn. Un- sealed corn silage in the top 12 in. continued to loose DM and OM as storage time advanced (P<.05). Estimated OM recoveries were lower (P<.05) at both 90 and 180 days post- filling than at 7 and 21 days. The estimated corn silage OM recoveries tended to be higher than the actual OM recoveries in unsealed silage less than 24 in. deep. After 180 days, DM and OM recoveries of the unsealed corn silage at 12- to 24-in. depth were lower (P<.05) than the three previous storage times. However, at the 24-to 36-in. depth, DM and OM recoveries remained constant as storage time advanced. Both corn silages sealed immediately (TRT 2 and 3) had similar DM and OM recoveries as storage time advanced. Delayed-seal silage (TRT 4) had similar (P>.05) DM and OM recoveries above 12 in. after 90 and 180 days. In addition, the DM and OM recoveries of the delayed-seal corn silage below 12 in. were not affected by storage time and had values that were similar to those of the two corn silages that were sealed immediately. Experiment 2: forage sorghum. Experimental Procedures On September 27 and 28, 1990, 45 drums were packed to equal densities with whole-plant forage sorghum similar to that used to fill the farm-scale, bunker silos described on page 132 of this report. Procedures were the same as in Experiment 1. Experiment 2: forage sorghum. As was observed for the corn silages, unsealed forage sorghum silage in the top 12 in. continued to loose DM and OM as storage time increased (P<.05). In the second 12 in., DM and OM recoveries were lower at 90 and 180 days post- filling than at 7 and 21 days (P<.05), but the loss was much less pronounced. The DM recoveries for the silages sealed immediately (TRT 2 and 3) were higher (P<.05) in the top 12 in. at 7 and 21 days post-filling than at 90 and 180 days. The DM and estimated OM re- coveries were similar (P>.05) for the delayed- seal silage stored above 12 in. by 21 days post- filling, and storage time did not affect DM or OM recoveries below 12 in. Chemical analysis of the samples and statistical analysis of the data. All samples were analyzed as described on page 132 of this report. Data were treated by analysis of vari- ance. Correlation coefficients and estimates of linear regression parameters were determined from actual and estimated OM recovery values from the pilot-scale silos. Introduction Large horizontal silos (i.e., bunkers, trenches, and stacks) are economical for stor- ing large volumes of ensiled feeds, but much of the silage is exposed to the environment. In shallow structures, 20 to 25% of the original ensiled volume can be within the top 3 ft. Past research with alfalfa and corn has shown DM losses of 30 to 60% in the top 2 ft. of silage. Controlled experiments have not adequately characterized the losses occurring in this top layer. The unsealed silages began to deteriorate immediately in the top 12 in. in both crops, and deterioration progressed to the second 12 in. by 90 days post-filling. Sealing immedi- ately after filling preserved more DM and OM after 180 days in the top 12 in. than delayed sealing. Silages from both crops, when sealed immediately and treated with a mold inhibitor, Top Savor®, had the highest DM and OM recoveries in the 0- to 12-in. depth at 7 days post-filling. Therefore, our objectives were to deter- mine the rate and extent of losses in the top 3 ft. in pilot-scale, horizontal silos and to verify a method designed to estimate these losses using silage ash content. Organic matter recoveries estimated by an equation using silage ash content were highly correlated (r>.93) to actual OM recoveries in all unsealed silages. Estimated and actual OM recoveries were not highly correlated in sealed 137 Results and Discussion The effects of sealing treatment, depth from the original surface, and day post-filling on the DM content and DM and OM recoveries Estimating OM recovery from silage ash content. When OM losses were large (top 24 in. in unsealed silage), OM recoveries esti- 138 in.). We conclude that in situations where serious silage OM loss is occurring, those losses can be estimated from changes in silage ash content. low (sealed silage and silage deeper than 24 ash content. Table 1. Effects of Sealing Treatment, Depth from the Original Surface (Depth) and Day Post- filling on the DM Content and DM and OM Recoveries (Rec.) of the Corn and Forage Sorghum Silages in Experiments 1 and 2 Sealing treatment1 Depth, inches Day post- filling DM, % Actual DM rec.2 Actual OM rec.2 Est. OM rec.3 DM, % Actual DM rec.2 Actual OM rec.2 Est. Results and Discussion OM rec.3 ------ Corn silages------ --- Forage sorghum silages --- TRT 1 0 to 12 7 33.6 86.1a 78.6a 83.0a 30.2 85.9a 85.2a 90.0a 90 28.7 45.6c 40.8c 41.4b 25.3 46.9c 44.1c 50.4c 180 36.0 35.7d 31.5d 36.9b 22.1 37.7d 36.6d 42.3c 12 to 24 7 31.4 87.0a 82.6a 91.2a 29.4 92.6a 92.3a 95.9a 90 26.9 81.5a 76.8a 81.1a 21.6 67.9b 62.3b 69.7b 180 23.3 62.1b 57.7b 64.5b 23.0 65.8b 64.0b 69.5b 24 to 36 7 32.2 88.8a,b 84.2a,b 89.7a 29.1 93.1a 92.6a 92.6 90 29.0 91.1a 86.1a 85.4a 25.3 88.3b 85.9b 90.9 180 27.7 83.4b 78.9b 84.2a 25.5 92.6a,b 91.8a,b 87.5 TRT 2 0 to 12 7 32.8 92.7a 88.1a 94.0a 29.4 91.4a 88.7a,b 88.7 90 31.9 86.4b 81.6b 85.1b 28.1 87.5a,b 86.6a,b 88.5 180 33.2 85.2b 80.6b 86.2b 29.4 86.5b 85.8b 89.5 12 to 24 7 33.8 94.1a 89.6a 97.7a 29.9 95.6a,b 92.9 89.1 90 31.9 87.2b 82.6b 88.5b 29.0 93.6a,b 93.0 91.8 180 32.3 87.0b 82.5b 90.5b 28.9 92.1b 91.5 90.7 24 to 36 7 33.6 93.9a 89.3a 96.4a 29.9 96.2 93.7 93.7 90 32.5 92.2a,b 86.3a,b 84.9b 28.7 95.5 95.3 93.1 180 31.0 86.1b 81.3b 83.7b 29.3 94.6 94.0 90.9 TRT 3 0 to 12 7 31.8 96.9a 92.3a 97.0a 29.6 93.4a 92.9a 93.4a 90 32.7 88.0b 83.1b 83.7b 28.6 89.3b 85.4b 86.5b 180 32.9 87.7b 83.0b 87.4b 29.4 90.0b 89.6a,b 94.5a 12 to 24 7 34.4 95.5a 90.9a 96.6a 29.3 96.4 95.8a 92.6a,b 90 32.3 89.6b 84.7b 84.7b 28.7 91.4 88.7b 89.1b 180 32.1 87.5b 82.7b 85.2b 29.1 90.8 90.4a,b 93.2a,b 24 to 36 7 33.5 93.8 89.0 91.8 29.8 96.9a 96.4a 92.7a,b 90 31.2 88.1 83.3 85.9 28.0 92.7b 90.1b 90.5b 180 30.9 86.3 81.5 83.9 29.3 96.4a 95.8a 92.0a,b TRT 4 0 to 12 7 33.6 86.1a 78.6a,b 83.0a 30.2 85.9a 85.2a 90.0a 90 32.8 77.0b 72.4b 75.9b 26.3 80.5b 82.7a 85.6b 180 33.9 77.1b 72.6b 76.3b 30.5 78.1b 73.0b 79.9b 12 to 24 7 31.4 87.0a,b 82.6a,b 91.2a 29.4 92.6 92.3 95.9 90 31.1 85.1b 80.1b 78.4b 27.9 89.3 88.5 87.8 180 32.4 87.8a,b 83.2a,b 87.1a 29.9 91.9 91.2 89.6 24 to 36 7 32.2 88.8 84.2 89.7a 29.1 93.1 92.6 92.6 90 30.7 86.6 81.6 80.2b 28.9 92.7 90.2 91.9 180 30.2 88.1 83.1 81.6b 28.5 95.6 95.0 91.4 Treatment (TRT) 1 = unsealed; TRT 2 = sealed immediately; TRT 3 = sealed immediately plus Top Savor®; 1 and TRT 4 = sealed 7 days post-filling plus Top Savor®. ( ) y q p g p Means within day post-filling at each depth and sealing treatment in the same column with different a,b,c,d superscripts differ (P<.05). d as a % of the DM or OM ensiled. d (est.) OM recovery calculated from the equation on page 128 of this report. Results and Discussion Expressed as a % of the DM or OM ensiled. 2 Estimated (est.) OM recovery calculated from the equation on page 128 of this report. 3 Means within day post-filling at each depth and sealing treatment in the same column with different a,b,c,d Effects of Sealing Treatment, Depth from the Original Surface (Depth) and Day Post- filling on the DM Content and DM and OM Recoveries (Rec.) of the Corn and Forage Sorghum Silages in Experiments 1 and 2 ( ) y q p g p Means within day post-filling at each depth and sealing treatment in the same column with different a,b,c,d superscripts differ (P<.05). y q p g p Means within day post-filling at each depth and sealing treatment in the same column with different a,b,c,d superscripts differ (P<.05). 139
https://openalex.org/W4389052839	https://link.springer.com/content/pdf/10.1007/s10765-023-03297-w.pdf	English	null	Modeling Thermodynamic Properties of Mixtures of CO2 + O2 in the Allam Cycle by Equations of State	International journal of thermophysics	2,023	cc-by	11,380	International Journal of Thermophysics (2023) 44:182 https://doi.org/10.1007/s10765-023-03297-w International Journal of Thermophysics (2023) 44:182 https://doi.org/10.1007/s10765-023-03297-w Abstractf Different equations of state (EOS) were applied for describing thermodynamic properties of the system CO2 + O2, which is important for the Allam cycle: cubic EOS (Soave–Redlich–Kwong, Peng–Robinson), molecular-based EOS (PC-SAFT, PCP-SAFT, SAFT-VR Mie, polar soft-SAFT, BACKONE, sCPA), and multiparam- eter EOS (GERG-2008, EOS-CG). The pure component models were taken from the literature. The results for the mixture were compared to experimental data from the literature for two cases: (i) the thermodynamic properties of the mixture were modeled using predictive mixing and combination rules; (ii) additional binary inter- action parameters were fitted to experimental data for improving the performance. In the predictive mode (i), the best results were obtained with molecular-based EOS. In the adjusted mode (ii) the best results were obtained with the multi-parameter GERG-2008 EOS and EOS-CG. Keywords Allam cycle · Carbon capture · Carbon dioxide · Equation of state · Oxygen Keywords Allam cycle · Carbon capture · Carbon dioxide · Equation of state · Oxygen Special Issue in Honor of Professor Roland Span’s 60th Birthday. Jens Staubach1 · Gerhard Schwarz2 · Stephan Möbius2 · Hans Hasse1 · Simon Stephan1 Received: 12 September 2023 / Accepted: 12 November 2023 / Published online: 27 November 2023 © The Author(s) 2023 * Simon Stephan Simon.Stephan@rptu.de 1 Laboratory of Engineering Thermodynamics (LTD), RPTU Kaiserslautern, 67663 Kaiserslautern, Germany 2 KSB SE & Co. KGaA, 67227 Frankenthal, Germany Modeling Thermodynamic Properties of Mixtures of CO2 + O2 in the Allam Cycle by Equations of State Jens Staubach1 · Gerhard Schwarz2 · Stephan Möbius2 · Hans Hasse1 · Simon Stephan1 * Simon Stephan Simon.Stephan@rptu.de 1 Laboratory of Engineering Thermodynamics (LTD), RPTU Kaiserslautern, 67663 Kaiserslautern, Germany 2 KSB SE & Co. KGaA, 67227 Frankenthal, Germany 1 Introduction Power plants based on the Allam cycle generate electric energy as well as CO2 of high purity at high pressure from burning hydrocarbon fuels with pure oxygen [1]. The process is attractive, as it combines a high thermal efficiency with efficient CO2 capture [2, 3]. In the simplest picture, the Allam cycle can be described as a 0123456789) 1 3 456789) 3 182 Page 2 of 23 182 Page 2 of 23 International Journal of Thermophysics (2023) 44:182 Joule-Brayton cycle with internal heat recuperation in which the main working fluid is CO2 . Additionally, oxygen and water are present in the working fluid. Water is removed before the compression and most of the CO2 is removed after the com- pression. Hence, for modeling the compression, the knowledge of the properties of mixtures of CO2 and O2 is essential. To some extent, also other components can be present such as argon, nitrogen, and trace elements from the fuel. In this work, we focus on the main components CO2 and O2 of the working fluid. l For the design of power plants based on the Allam cycle, information on the thermal and caloric properties as well as phase equilibria of the working fluid is needed. The interest in the Allam cycle has led to a number of experimental studies of thermophysical properties of mixtures relevant for the Allam cycle in recent years [4–10] that can be used as basis for the thermodynamic modeling by equations of state (EOS). In principle, EOS can be used to predict mixture properties from pure component properties using predictive mixing and combination rules. Nevertheless, for many applications, such predictions are not accurate enough, so that adjustable binary parameters are introduced that are fitted to selected experimental data of the mixture. In this work, the mixture CO2 + O2 was modeled using different EOS based on pure component models that were taken from the literature. Both, predictive mix- ing and combination rules as well as mixing and combination rules with adjustable parameters were tested. Different types of EOS have been used for modeling mixtures of CO2 with other components, see, e.g. Refs. [4, 11–13]. In 2011, Diamantonis et al. [11] have assessed predictions of cubic EOS and EOS based on statistical associa- tion fluid theory (SAFT) for phase equilibria of binary mixtures of CO2 with ( CH4, N2, O2, SO2, Ar, and H2S ). 1 3 1 Introduction The studied EOS included the Redlich–Kwong [14], Soave–Redlich–Kwong (SRK) [15], Peng-Robinson (PR) [16], SAFT [17], and perturbed chain-SAFT (PC-SAFT) [18] EOS. Diamantonis et al. [11] concluded that the PC-SAFT EOS is overall the most accurate EOS for predicting CO2 + gas mix- ture properties. In 2014, Mazzoccoli et al. [12] studied different EOS regarding their predictions of phase equilibria and homogeneous state density data of binary mix- tures of CO2 with ( N2, O2, and Ar ). The studied EOS included the Benedict-Webb- Rubin (BWR) [19], PC-SAFT, and GERG-2008 [20] EOS. The GERG-2008 and PC-SAFT EOS were found to perform well in most cases. In 2016, Lasala et al. [4] compared different cubic EOS with different combination rules for predicting phase equilibria of binary mixtures of CO2 with ( N2, Ar, and O2 ) and found that using a temperature-dependent parameter in the combination rule improved the accuracy significantly. In 2017, Perez et al. [13] studied the modeling of vapor–liquid equi- libria and homogeneous state densities of 108 binary mixtures related to carbon cap- ture and storage using the SRK, PR, PC-SAFT, and SAFT-VR Mie [21] EOS using a temperature-dependent combination rule parameter. Perez et al. [13] concluded that the SAFT-VR Mie EOS is the most accurate of the studied EOS.i Overall, the findings reported in the literature give no clear picture, which EOS to use for modeling mixture properties of CO2 + O2 . Furthermore, a significant amount of new data has become available only recently and was not included in most studies discussed above [4–10]. Therefore, in this work, we have studied the modeling of mixture properties of CO2 + O2 using different EOS. First, the available literature 1 3 International Journal of Thermophysics (2023) 44:182 182 Page 3 of 23 data are reviewed and compiled in a database. This includes vapor–liquid equilib- rium data Tpx′x′′ as well as homogeneous state property data (density 휌 , speed of sound w, and isothermal compressibility 훽 ) for a given composition, pressure, and temperature. Ten EOS were used for modeling the mixture properties, namely the SRK [15], PR [16], PC-SAFT [18], PCP-SAFT [18, 22], SAFT-VR Mie [21], polar soft-SAFT [23, 24], BACKONE [25, 26], sCPA [27, 28], GERG-2008 [20], and EOS-CG [29] EOS. The EOS-CG as well as pure component EOS models for CO2 and O2 [30, 31] were developed by Roland Span and co-workers. 1 Introduction Two cases were considered: (i) the thermodynamic properties of the mixture were modeled using fully predictive mixing and combination rules (“predictive mode”); (ii) temperature-dependent binary interaction parameters were fitted to a training dataset of experimental data for improving the performance of the models (“adjusted mode”). The binary interaction parameters were fitted to the same training dataset for all studied EOS; except the EOS-CG [29]. The binary interaction parameters for the mixture CO2 + O2 were already fitted in the original publication by Gernert and Span [29] and they are adopted here for the EOS-CG in the “adjusted mode”. The results are compared for both cases (i) and (ii) for the complete literature data, a training, and a test dataset. This allows a comparison of the different EOS on an equal basis.i This paper is structured as follows: first, the experimental data basis for mixtures of CO2 + O2 that was established from literature data is presented; then, the stud- ied EOS are described, focusing in the modeling of mixture properties. Finally, the results of the application of the EOS in the predictive and adjusted mode are pre- sented and critically compared. 2.1 Experimental Data Basis Tables 1 and 2 give an overview of the available experimental data on the vapor–liq- uid equilibrium and homogeneous state property data, respectively, in the system CO2 + O2 . In total, 255 phase equilibrium data points and 1866 homogeneous state data points have been reported. The phase equilibrium datasets report pressure p and the compositions of both phases x’, x″ for a given temperature T. For the homogene- ous states, data on the density 휌, the speed of sound w, and the isothermal compress- ibility 훽 at a given pressure and temperature are available. The database compiled within this work is provided in the electronic Supplementary Information. Phase equilibrium data (cf. Table 1) have been reported in the lit- erature in the range 218 K ≤T ≤298 K , 0.9 MPa ≤p ≤14.4 MPa, and 0.0 mol⋅mol−1 < xO2 ≤0.55 mol·mol−1. The phase equilibrium data cov- ers almost the complete range between the temperature of the tri- ple point and the critical temperature of pure CO2 ( Ttr,CO2 = 216.58 K [32, 33], Tc,CO2 = 304.15 K [34]). The data on homogeneous state points lie in the region 250 K ≤T ≤423 K, 0.5 MPa ≤p ≤47.8 MPa , and 0.0 mol⋅mol−1 < xO2 ≤0.75 mol·mol−1. The critical temperature of oxygen is 3 3 International Journal of Thermophysics (2023) 44:182 182 Page 4 of 23 Table 1 Overview of studies reporting experimental data on vapor–liquid equilibrium properties of the system CO2 + O2 N is the number of data points References Year N Ranges T ∕K p ∕MPa x′ O2 / mol ⋅ mol−1 Lasala et al. [4, 5] 2016 67 223–293 1.7–13.9 0.01–0.45 Westman et al. [6] 2016 61 218–298 2.0–14.4 0.01–0.55 Ahmad et al. [61] 2014 11 277–298 4.1–8.1 0.02–0.05 Fredenslund et al. [62] 1972 11 224 0.9–14.2 0.01–0.45 Fredenslund and Sather [63] 1970 71 223–283 1.0–13.2 0.01–0.39 Kaminishi et al. [64] 1966 16 233–298 3.7–12.7 0.03–0.37 Chueh et al. 2.1 Experimental Data Basis [65] 1965 4 273 5.5–11.7 0.05–0.31 Zenner and Dana [66] 1963 25 218–273 2.2–14.3 0.03–0.50 ble 1 Overview of studies reporting experimental data on vapor–liquid equilibrium properties of the stem CO2 + O2 The last three columns give information on the range of states in which data are provided: T and p are temperature and pressure, x′ O2 is the liquid phase mole fraction of O2 ble 2 Overview of studies reporting experimental data on homogeneous state point properties of the stem CO2 + O2 Table 2 Overview of studies reporting experimental data on homogeneous state point properties of the system CO2 + O2 N is the number of data points The last three columns give information on the range of states in which data are provided: T and p are temperature and pressure, xO2 is the mole fraction of O2 References Year N Ranges T ∕K p ∕MPa xO2 / mol ⋅ mol−1 pρT Lozano-Martin et al. [8] 2021 274 250–375 0.5–19.8 0.50–0.75 Ahamada et al. [10] 2020 820 277–416 1.0–20.5 0.13–0.48 Lozano-Martin et al. [7] 2020 162 250–375 0.5–13.1 0.05–0.20 Commodore et al. [9] 2018 112 324–400 2.3–35.3 0.01 Al-Siyabi [67] 2013 72 283–423 7.7–47.8 0.05 Mantovani et al. [34] 2012 196 303–383 1.0–20.0 0.06–0.13 Mazzoccoli et al. [68] 2012 106 273–293 1.0–20.0 0.04–0.15 Gururaja et al. [69] 1967 7 297–298 1.3–1.6 0.08–0.81 w, β Al-Siyabi [67] 2013 62 268–301 8.9–41.0 0.07 The last three columns give information on the range of states in which data are provided: T and p are temperature and pressure, xO2 is the mole fraction of O2 Tc,O2 = 154.77 K [34]. Hence, oxygen is supercritical in the entire studied tempera- ture range. The mixture CO2 + O2 is a type I mixture according to the scheme of van Konynenburg and Scott [35]. Tc,O2 = 154.77 K [34]. Hence, oxygen is supercritical in the entire studied tempera- ture range. The mixture CO2 + O2 is a type I mixture according to the scheme of van Konynenburg and Scott [35]. The literature data were checked for clear outliers by compari- son of each data point to the rest of the data along isotherms and iso- bars at equal compositions. 2.1 Experimental Data Basis Overall, seven data points for VLE and 12 data 1 3 International Journal of Thermophysics (2023) 44:182 Page 5 of 23 182 Table 3 Number of experimental data for homogeneous state point properties and vapor–liquid equilib- rium properties of the system CO2 + O2 used in the comparison of the EOS The number of data points N is reported for the density 휌 , isothermal compressibility 훽 , speed of sound w, liquid phase composition x′ and vapor phase composition x′′ Property N휌 N훽 Nw Nx′ Nx′′ Region Liquid Supercritical Vapor Supercritical Supercritical All data 41 1371 325 62 62 259 259 Training data 41 391 105 32 32 91 89 Test data 0 980 220 30 30 168 170 Table 3 Number of experimental data for homogeneous state point properties and vapor–liquid equili rium properties of the system CO2 + O2 used in the comparison of the EOS The number of data points N is reported for the density 휌 , isothermal compressibility 훽 , speed of soun w, liquid phase composition x′ and vapor phase composition x′′ The number of data points N is reported for the density 휌 , isothermal compressibility 훽 , speed of sound w, liquid phase composition x′ and vapor phase composition x′′ Fig. 1 Pressure–temperature diagram with the homogeneous state data and phase equilibrium data compiled in the database, cf. Tables 1 and 2. Additionally, the vapor pressure curves of CO2 and O2 are shown (computed from the reference EOS models from Refs. [30, 43]). The star indicates the critical points. The area of the training data are marked by the red box (Color figure online) Fig. 1 Pressure–temperature diagram with the homogeneous state data and phase equilibrium data compiled in the database, cf. Tables 1 and 2. Additionally, the vapor pressure curves of CO2 and O2 are shown (computed from the reference EOS models from Refs. [30, 43]). The star indicates the critical points. The area of the training data are marked by the red box (Color figure online) points for homogeneous state points were removed for the EOS evalua- tion; for details see Supplementary Information. The remaining experimen- tal data were split into a training dataset and a test dataset. The training data- set comprises of data between 273 K ≤T ≤333 K, 5 MPa ≤p ≤30 MPa , and 0.0 mol⋅mol−1 < xO2 ≤0.5 mol⋅mol−1 . 2.1 Experimental Data Basis The test dataset includes the remaining data. An overview of the number of experimental data points in each dataset is given in Table 3. Additionally, Fig. 1 gives an overview of the temperatures and pressures for which data are available. 1 3 International Journal of Thermophysics (2023) 44:182 182 Page 6 of 23 Table 4 Overview of the equation of states (EOS) used in this work including information on the basic publication of the EOS and its year, and the publications from which the pure component models were taken together with the number of parameters (indicated by #) EOS Year Reference EOS CO2 O2 Cubic SRK 1972 [15] [34] #3 [34] #3 PR 1976 [16] [34] #3 [34] #3 Molecular-based PC-SAFT 2001 [18] [39] #3 [39] #3 PCP-SAFT 2005 [18, 22] [22] #4 [39] #3 SAFT-VR Mie 2013 [21] [38] #4 [38] #4 polar soft-SAFT 2020 [23, 24] [37] #5 [36] #3 BACKONE 1996 [25, 26] [25] #4 [25] #3 sCPA 1999 [27, 28] [40] #5 [41] #3 Multi-parameter GERG-2008 2012 [20] [42] #22 [31] #12 EOS-CG 2016 [29] [30] #42 [43] #32 2.2 Equations of State In this work, ten EOS were used for the modeling of mixture properties for the sys- tem CO2 + O2 . Table 4 gives an overview of the EOS used in this work. For all cal- culations carried out in this work, an in-house code was used that was validated by comparing the results with data from the EOS developers. All EOS were implemented using the Helmholtz energy formulation. The Helm- holtz energy per particle a = A∕N can be written as a sum of the ideal gas contribu- tion and a configurational contribution as (1) a kBT = ̃a = ̃aid + ̃aconﬁg. (1) The configurational contribution ̃aconﬁg is calculated by the respective EOS. For the ideal gas contribution ̃aid , the pure component models from Kunz and Wagner [20] were used in all cases. The ideal gas contribution of the total Helmholtz energy of a mixture is calculated as The configurational contribution ̃aconﬁg is calculated by the respective EOS. For the ideal gas contribution ̃aid , the pure component models from Kunz and Wagner [20] were used in all cases. The ideal gas contribution of the total Helmholtz energy of a mixture is calculated as (2) ̃aid = N ∑ i=1 xi[̃aid,i(𝜌, T) + ln xi], (2) where ̃aid,i is the ideal gas contribution of the pure components i = CO2 and O2 and xi indicates the mole fraction of component i. The pure component EOS models for CO2 and O2 were taken from the literature in all cases. In all cases, VLE data were used for their parametrization [20, 22, 25, 34, 36–41]—in some cases, additionally, homogeneous state property data were used [20, 22]. The pure component model parameters are summarized in the Supplementary Material. International Journal of Thermophysics (2023) 44:182 Page 7 of 23 182 182 2.2.1 Multi‑parameter EOS The GERG-2008 [20] EOS and the EOS-CG [29] are empirical multi-parameter EOS. The pure component parameters have no direct physical meaning. They are substance-specific and were correlated in the original works [30, 31, 42, 43] to a large number of experimental data for the respective substance. This leads to very accurate models for the pure substances regarding the data that was used for the fit- ting. Yet, using multi-parameter EOS for extrapolation, i.e. outside the state range that was considered for the model parametrization, has to be done with caution [44–48]. The two multi-parameter EOS models used here for describing the mixture CO2 + O2 use different pure component models, cf. Table 4. The GERG-2008 EOS uses the pure component EOS models from Refs. [31, 42], whereas the EOS-CG uses those from Refs. [30, 43]. For multi-parameter EOS, the extension to mixtures is usually carried out by defining a reducing function for the mixture density 휌r(x) and mixture temperature Tr(x) . Thereby, the configurational contribution to the Helmholtz energy of the mix- ture can be written as (3) ̃amulti-parameter conﬁg = ̃aconﬁg ( 𝜌 𝜌r , Tr T , x ) . (3) The reducing functions are calculated according to Ref. [20] as The reducing functions are calculated according to Ref. [20] as (4) 1 휌r = N i=1 N j=1 xixj훽v,ij훾v,ij ⋅ xi + xj 훽2 v,ijxi + xj ⋅1 8 ⎛ ⎜ ⎜⎝ 1 휌1∕3 c,i + 1 휌1∕3 c,j ⎞ ⎟ ⎟⎠ 3 , (4) and and (5) Tr = N ∑ i=1 N ∑ j=1 xixj훽T,ij훾T,ij ⋅ xi + xj 훽2 T,ijxi + xj ⋅(Tc,iTc,j)0.5, (5) with the critical temperature Tc,i, Tc,j and critical density 휌c,i, 휌c,j of component i and j. Both, the GERG-2008 EOS and the EOS-CG use the mixture ansatz outlined in Eqs. 3–5. The parameters ( 훽v,ij, 훽T,ij, 훾v,ij, 훾T,ij) are adjustable mixture parameters in the GERG-2008 and EOS-CG model. They obey the relations [20] with the critical temperature Tc,i, Tc,j and critical density 휌c,i, 휌c,j of component i and j. Both, the GERG-2008 EOS and the EOS-CG use the mixture ansatz outlined in Eqs. 3–5. The parameters ( 훽v,ij, 훽T,ij, 훾v,ij, 훾T,ij) are adjustable mixture parameters in the GERG-2008 and EOS-CG model. They obey the relations [20] (6) 훽v,ij = 1∕훽v,ji, 훽T,ij = 1∕훽T,ji, (6) and and (7) 훾v,ij = 훾v,ji, 훾T,ij = 훾T,ji. 2.2.2 Cubic EOS The SRK [15] and PR [16] EOS were implemented according to Ref. [50] in the Helmholtz energy form. The pure component models are specified by the numbers for the critical temperature Tc , the critical pressure pc , and the acentric factor 휔 . The corresponding numbers are given in the Supplementary Information. The stand- ard one-fluid mixing rules were applied for the calculation of the mixture attraction energy parameter am and the mixture co-volume parameter bm: (8) am = N ∑ i=1 N ∑ j=1 xixjaij(T), (8) (9) bm = N ∑ i=1 N ∑ j=1 xixjbij. (9) For the calculation of the cross-interaction parameters between the components i and j, the modified Lorentz-Berthelot combination rules were used, i.e. For the calculation of the cross-interaction parameters between the components i and j, the modified Lorentz-Berthelot combination rules were used, i.e. (10) aij(T) = 휉ij √ aii(T)ajj(T), (10) and and (11) bij = 1 2(bii + bjj), (11) where 휉ij is a binary interaction parameter. where 휉ij is a binary interaction parameter. 2.2.1 Multi‑parameter EOS (7) An additional departure function can be regressed to mixture properties for multi- parameter EOS to improve their accuracy [20, 49] which was, however, not used here. 1 3 3 182 Page 8 of 23 International Journal of Thermophysics (2023) 44:182 2.2.3 Molecular‑based EOS Six molecular-based EOS were used in this work: PC-SAFT [18], PCP-SAFT [18, 22], SAFT-VR Mie [21], polar soft-SAFT [23, 24], BACKONE [25, 26], and sCPA [27, 28] EOS. In these EOS the different molecular features are represented by inde- pendent terms in the Helmholtz energy. The configurational Helmholtz energy is formulated as (12) ̃aconﬁg = ̃arep + ̃adisp + ̃achain + ̃apolar + ̃aassoc, (12) with the repulsion ̃arep , dispersion ̃adisp , chain ̃achain , polar ̃apolar , and association ̃aassoc contribution. Table 5 gives an overview of the terms used in the different molecular- based EOS. For O2 , the basic modeling approach was the same for all EOS, i.e. dis- persive and repulsive interactions as well as the molecule elongation are modeled. For CO2 , in addition to these terms, in some cases, the polarity of the molecule was explicitly described by an additional term. Different approaches are used for this in the EOS shown in Table 5: a polar term is used for the PCP-SAFT, polar soft-SAFT and BACKONE and an association term in the sCPA. 1 3 International Journal of Thermophysics (2023) 44:182 Page 9 of 23 1 182 Table 5 Overview of the configurational Helmholtz energy terms used in the modeling of CO2 and O2 for each of the studied molecular-based EOS models The pure component model parameters were taken from the literature EOS Substance Repulsion Dispersion Chain (elongation) Polar Association PC-SAFT O2 X X X CO2 X X X PCP-SAFT O2 X X X CO2 X X X X SAFT-VR Mie O2 X X X CO2 X X X Polar soft-SAFT O2 X X X CO2 X X X X BACKONE O2 X X X CO2 X X X X sCPA O2 X X X CO2 X X X X The pure component model parameters were taken from the literature The pure component models used for the PC-SAFT [18], PCP-SAFT [18, 22], SAFT-VR Mie [21], and polar soft-SAFT [23, 24] have (at least) three parameters: the size parameter 휎 , dispersion energy parameter 휀 , and the chain length param- eter m. The SAFT-VR Mie [21] models have an additional adjustable parameter 휆r , which indicates the repulsion exponent for the Mie potential [51]. Moreover, the quadrupole terms of the PCP-SAFT and polar soft-SAFT have the quadrupole moment as an additional parameter. 2.2.3 Molecular‑based EOS The quadrupole term used in the polar soft- SAFT EOS has an additional parameter xp, which describes the number of quad- rupole elements in the molecule (modeling the delocalization of the quadrupole). The BACKONE [25, 26] EOS incorporates the parameter 훼 (instead of the chain length parameter m) modeling the elongation of the molecule. The sCPA [27, 28] EOS combines the Helmholtz energy term from the SRK EOS ̃aSRK for modeling dispersive and repulsive interactions with an association term ̃aassoc . Since CO2 is modeled with the association term, it has two additional parameters: the asso- ciation energy parameter ̂𝜀 and the association volume parameter ̂𝛽 . Details and the numeric values of the parameters of the pure component models, that were taken from the literature [22, 24, 25, 36, 38–41], are given in the Supplementary Information. The mixing rules used in the molecular-based EOS were directly adopted from the original works [18, 21–28]. In the sCPA EOS, the cross-interaction parame- ters between the components were computed according to Eqs. 8–11 (see above). For the PC-SAFT, the PCP-SAFT, polar soft-SAFT, and the BACKONE EOS, the modified Lorentz-Berthelot combination rules were used, i.e. (13) 휀ij = 휉ij √휀ii휀jj, 휀ij = 휉ij √휀ii휀jj, (13) 1 3 3 International Journal of Thermophysics (2023) 44:182 182 Page 10 of 23 (14) 휎ij = 1 2(휎ii + 휎jj). (14) In the SAFT-VR Mie EOS, the energy cross-interaction parameter was computed with the combination rule In the SAFT-VR Mie EOS, the energy cross-interaction parameter was computed with the combination rule (15) 휀ij = 휉ij 휎3 ii휎3 jj 휎3 ij √휀ii휀jj, (15) while 휎ij was calculated according to Eq. 14. Therein, the binary interaction param- eter 휉ij is treated the same as the binary interaction parameter of the cubic EOS, cf. Eq. 10. For the chain term, polar term, and association term, no combination rule applies. while 휎ij was calculated according to Eq. 14. Therein, the binary interaction param- eter 휉ij is treated the same as the binary interaction parameter of the cubic EOS, cf. Eq. 10. For the chain term, polar term, and association term, no combination rule applies. International Journal of Thermophysics (2023) 44:182 Page 11 of 23 182 Table 6 Binary interaction parameters for the mixture CO2 + O2 obtained from the regression to experimental data, cf. Eq. 16 EOS o1 o2 SRK 0.2970 469 2.3560 240 × 10−3 PR 1.1025 889 − 7.3000 922 × 10−4 PC-SAFT 1.0819 439 − 4.6704 942 × 10−4 PCP-SAFT 1.3391 420 − 1.1561 565 × 10−3 SAFT-VR Mie 1.0749 257 − 2.2777 264 × 10−4 polar soft-SAFT 1.2515 353 − 8.9082 688 × 10−4 BACKONE 1.3151 601 − 1.1994 463 × 10−3 sCPA 1.2425 680 − 9.0735 012 × 10−4 Table 7 Binary interaction parameters for the mixture CO2 + O2 obtained from the regression to experi- mental data for the GERG-2008 EOS, cf. Eqs. 4 and 5 EOS 훽v,CO2,O2 훽T,CO2,O2 훾v,CO2,O2 훾T,CO2,O2 GERG-2008 0.976180 838 1.011636 925 1.213083 507 1.008384 202 ble 7 Binary interaction parameters for the mixture CO2 + O2 obtained from the regression to experi- ental data for the GERG-2008 EOS, cf. Eqs. 4 and 5 (17) min ̄ok F(̄ok) = 1 N𝜌 N𝜌 ∑ t=1 [ 𝜌EOS t −𝜌Ref t 𝜌Ref t ]2 + 1 100 ⋅N𝛽 N𝛽 ∑ t=1 [ 𝛽EOS t −𝛽Ref t 𝛽Ref t ]2 + 1 10 ⋅Nw Nw ∑ t=1 [ wEOS t −wRef t wRef t ]2 + 0.5 Nx O2 Nx CO2 ∑ t=1 [ x,EOS O2,t −x,Ref O2,t ]2 + 0.5 Nx O2 Nx O2 ∑ t=1 [ x,EOS O2,t −x,Ref O2,t ]2 where the superscript Ref indicates the experimental data points and the superscript EOS indicates the value computed by a given EOS and Nl indicates the number of data points for the properties l = 휌, 훽, w, x O2, and x O2 . Only experimental data from the training set were considered in the fitting procedure, cf. Fig. 1. The weights used in the fit, cf. Eq. 17, were determined in an iterative approach such that a good com- promise between the individual objectives was obtained. The isothermal compressi- bility data and speed of sound data were weighted considerably less (1/100 and 1/10 times less, respectively) compared to the homogeneous state density data since they were only available for a single composition of the mixture ( xO2 = 0.0652 mol⋅ mol−1). The adjusted parameters o1 and o2 for the cubic and molecular-based EOS are given in Table 6. 2.2.4 Regression of Binary Interaction Parameters First, all EOS were used in a purely predictive mode, i.e. no mixture parameters were adjusted. In the predictive mode, we have used the multi-parameter EOS with 훽v,CO2,O2 = 훽T,CO2,O2 = 훾v,CO2,O2 = 훾T,CO2,O2 = 1 , cf. Eqs. 4 and 5. The cubic and molecular-based EOS were used in the predictive mode with 휉CO2,O2 = 1 , cf. Eqs. 10, 13, and 15. For the multi-parameter EOS, in the predictive mode, the Lor- entz–Berthelot combination rules were applied. For the GERG-2008 EOS, the pre- dictive mode corresponds to the EOS model as originally proposed by Kunz and Wagner [20], which was not adjusted to CO2 + O2 mixture data. For the EOS-CG, the predictive mode reflects a simplified version of the original EOS-CG model that was fitted to CO2 + O2 mixture data by Gernert and Span [29]. i Additionally, the binary interaction parameters were regressed to the training set of the mixture data in an adjusted mode. The obtained models were compared to the predictive mode models. For the cubic and molecular-based EOS, the cross-interac- tion energy parameter 휉CO2,O2 was adjusted. Therefore, a linear temperature-depend- ent function was used (16) 휉CO2,O2 = o1 + o2 ⋅(T∕K), (16) where o1 and o2 are the adjustable parameters. For the GERG-2008 EOS, the four binary interaction parameters o = 훽v,CO2,O2, 훽T,CO2,O2, 훾v,CO2,O2, 훾T,CO2,O2 were adjusted. For the EOS-CG, the binary interaction parameters 훾v,CO2,O2, 훾T,CO2,O2 were adapted for the adjusted mode from the original work by Gernert and Span [29] with 훽v,CO2,O2, 훽T,CO2,O2 = 1 . Hence, for the cubic, molecular-based EOS, and the EOS- CG two adjustable parameters were used, whereas four adjustable parameters were used for the GERG-2008 EOS. This follows the usual way these model classes are used for mixture modeling [20, 29, 49, 52–55]. The binary interaction parameters ok for the EOS k were obtained by minimizing a least-squares objective function F given by 1 3 1 3 International Journal of Thermophysics (2023) 44:182 2.2.5 Assessment of the EOS The performance of the different EOS used in both the predictive and adjusted mode was evaluated in a systematic way. For all EOS, the performance for the homogene- ous state properties and the phase equilibrium properties was evaluated. Therefore, mean deviations from a given model and the experimental data were computed. For the homogeneous state properties, the mean absolute relative deviation 훿rel,k lm was cal- culated for each of the studied EOS k for the properties l = 휌, 훽, and w for the fluid region m = vapor, liquid, and supercritical by (18) 훿rel,k lm = 1 Nlm Nlm ∑ t=1 \|\|\|\|\| 100 % YEOS lmt −YRef lmt YRef lmt \|\|\|\|\| , (18) where YRef indicates the value of the experimental data point taken from the litera- ture. For the homogeneous state point properties, the assignment of the data points from the literature to the fluid regions was carried out by estimating the critical tem- peratures depending on the composition from the experimental phase boundary data (see electronic Supplementary Information). The mean absolute relative deviation 훿rel,k lm is therefore a measure for the performance of the EOS k for the property l in the fluid region m. l The performance of the EOS models for describing the vapor–liquid equilibrium phase boundary was evaluated by the mean absolute deviation 훿abs,k n for phase com- position for a given temperature and pressure defined as (19) 훿abs,k n = 1 Nn Nn ∑ t=1 \|xEOS O2,nt −xRef O2,nt\|, (19) where n represents the liquid (n = liq) and vapor (n = vap) side of the phase boundary. where n represents the liquid (n = liq) and vapor (n = vap) side of the phase boundary. l k Both mean deviations 훿rel,k lm and 훿abs,k n were calculated for three cases: (i) for all data points available for the mixture CO2 + O2 , (ii) for the data points considered for the mixture parametrization, i.e. the training data (cf. Fig. 1), and (iii) the test data. International Journal of Thermophysics (2023) 44:182 The adjusted binary interaction parameters for the GERG-2008 [20] EOS 훽v,CO2,O2, 훽T,CO2,O2, 훾v,CO2,O2, 훾T,CO2,O2 are given in Table 7. Note, that the 1 3 182 Page 12 of 23 International Journal of Thermophysics (2023) 44:182 182 mixture parameters for the EOS-CG as proposed in Ref. [29] were obtained by regression to a wider temperature range compared to the training dataset used in this work. 3.1 Vapor–Liquid Equilibrium The results for the mean absolute deviation in the liquid and vapor phase mole frac- tion 훿abs liq and 훿abs vap , respectively, are given in Table 8. Additionally, Fig. 2 shows the 1 3 International Journal of Thermophysics (2023) 44:182 Page 13 of 23 182 Page 13 of 23 182 182 Table 8 Overview of the quality of the description of experimental VLE data for the system CO2 + O2 by different EOS. Results from predictions based on pure component models are compared to results that were obtained after adjusting binary parameters (given in brackets) to the training data. Results are reported for each EOS from top to bottom for all data, the training data, and the test data. For details, see text EOS No phase split 훿abs liq / mol ⋅ mol−1 훿abs vap / mol ⋅ mol−1 PR 28 (0) 0.075 (0.025) 0.018 (0.022) 9 (0) 0.044 (0.009) 0.020 (0.023) 19 (0) 0.092 (0.034) 0.017 (0.022) SRK 21 (0) 0.071 (0.040) 0.020 (0.032) 3 (0) 0.042 (0.024) 0.024 (0.026) 18 (0) 0.088 (0.048) 0.018 (0.035) PC-SAFT 3 (0) 0.051 (0.024) 0.026 (0.026) 0 (0) 0.020 (0.012) 0.036 (0.030) 3 (0) 0.069 (0.031) 0.021 (0.024) PCP-SAFT 0 (2) 0.035 (0.036) 0.029 (0.025) 0 (0) 0.019 (0.011) 0.024 (0.024) 0 (2) 0.044 (0.050) 0.031 (0.025) SAFT-VR Mie 0 (0) 0.028 (0.024) 0.019 (0.018) 0 (0) 0.026 (0.020) 0.019 (0.019) 0 (0) 0.029 (0.026) 0.019 (0.018) polar soft-SAFT 0 (0) 0.042 (0.026) 0.050 (0.045) 0 (0) 0.027 (0.025) 0.063 (0.063) 0 (0) 0.050 (0.027) 0.043 (0.036) BACKONE 0 (0) 0.041 (0.033) 0.065 (0.063) 0 (0) 0.021 (0.024) 0.078 (0.076) 0 (0) 0.051 (0.037) 0.058 (0.056) sCPA 0 (0) 0.018 (0.030) 0.044 (0.043) 0 (0) 0.011 (0.012) 0.053 (0.053) 0 (0) 0.023 (0.040) 0.039 (0.037) GERG-2008 0 (26) 0.069 (0.017) 0.039 (0.016) 0 (0) 0.038 (0.007) 0.048 (0.017) 0 (26) 0.085 (0.024) 0.035 (0.015) EOS-CG 0 (11) 0.069 (0.014) 0.039 (0.026) 0 (0) 0.038 (0.008) 0.047 (0.036) 0 (11) 0.085 (0.017) 0.035 (0.020) phase envelopes obtained from the EOS in comparison with experimental data for three different temperatures. Both, Table 8 and Fig. 2 show the results for the predic- tive mode and the adjusted mode. phase envelopes obtained from the EOS in comparison with experimental data for three different temperatures. Both, Table 8 and Fig. 2 show the results for the predic- tive mode and the adjusted mode. 3.1 Vapor–Liquid Equilibrium The critical pressure obtained from the predictive calculations by most studied EOS significantly overestimates the highest pressures reported by the experimental data for the studied temperatures, cf. Fig. 2. The only exceptions are the PR and SRK EOS, which in return yield high deviations for the liquid phase composition, cf. Table 8. A good prediction of the studied phase envelopes (top row in Fig. 2) is obtained for the SAFT-VR Mie and the PCP-SAFT EOS, which is also reflected in the mean deviations given in Table 8. The inclusion of the quadrupole in the model 1 3 International Journal of Thermophysics (2023) 44:182 1 3 International Journal of Thermophysics (2023) 44:182 182 Page 14 of 23 Fig. 2 Phase envelope of the system CO2 + O2 calculated by different EOS (lines) compared to litera- ture data (symbols) [4, 6, 61, 63–66]. Results from predictions based on pure component models (top) are compared to results that were obtained after adjusting binary parameters (bottom) to the training data. The GERG-2008 and EOS-CG results collapse to one line in the top figures (predictive mode). For details, see text (Color figure online) Fig. 2 Phase envelope of the system CO2 + O2 calculated by different EOS (lines) compared to litera- ture data (symbols) [4, 6, 61, 63–66]. Results from predictions based on pure component models (top) are compared to results that were obtained after adjusting binary parameters (bottom) to the training data. The GERG-2008 and EOS-CG results collapse to one line in the top figures (predictive mode). For details, see text (Color figure online) of CO2 of the PCP-SAFT EOS compared to the PC-SAFT model reduces the devia- tions in the liquid phase composition and overall improves the description of the phase boundary for the temperatures T ∕K = 288.15, 298.15 , cf. Fig. 2. The multi- parameter GERG-2008 EOS yields high deviations for the liquid side composition 훿abs liq and also significantly overestimates the critical point. The EOS-CG yields almost identical results for the predictive calculations as the GERG-2008 EOS. This is interesting, since different pure component models are used in the GERG-2008 EOS and EOS-CG.i For the adjusted mode, the performance of all models significantly improves for the training dataset—as expected. In particular, the results obtained from the adjusted GERG-2008 EOS are in very good agreement with the experimental data. One reason for this may be that four parameters were used for the adjustment. The PR EOS also yields significantly lower deviations compared to the predictive mode and is on a par with the adjusted GERG-2008 EOS with only two parameters. For the liquid phase concentration deviations, the EOS-CG and the adjusted GERG- 2008 EOS yield practically identical results (especially for the training dataset). On the contrary, for the vapor phase, the adjusted GERG-2008 EOS yields signifi- cantly lower concentration deviations compared to the EOS-CG, cf. Table 8. International Journal of Thermophysics (2023) 44:182 For 1 3 International Journal of Thermophysics (2023) 44:182 182 Page 15 of 23 the molecular-based EOS, the SAFT-VR Mie and the PCP-SAFT EOS yield a good description of the phase envelope in the adjusted mode (only two adjusted param- eters). Both EOS overestimate the critical point, but the SAFT-VR Mie EOS over- estimates the critical point stronger compared to the PCP-SAFT EOS, cf. Fig. 2. A more detailed discussion on the overestimation of the critical point by the molecu- lar-based EOS is given in the Supplementary Information. Interestingly, the descrip- tion of the phase envelope by the polar soft-SAFT, BACKONE, and sCPA show almost no improvement by the adjustment of the binary interaction parameters. For the sCPA, this could be the result of modeling CO2 as an associating fluid. The asso- ciation contribution is not influenced by the binary interaction parameter since only CO2 exhibits associating sides in the mixture. The models (both predictive mode and adjusted mode) were also tested for describing the phase behavior at low temperatures, i.e. for the test dataset. The parameter adjustment also improves the performance of the PR, SRK, and PC-SAFT for the test data liquid phase composition significantly. On the other hand, the mean deviations for the vapor phase composition 훿abs vap increase only slightly. The tempera- ture-dependent binary interaction parameter also results in an improvement in the description of the phase envelope at lower temperatures for the more complex mod- els PCP-SAFT and SAFT-VR Mie. For the adjusted GERG-2008 EOS, some devia- tions occur for the phase equilibria at T ⪅240 K . This is not surprising considering the empirical character of the GERG-2008 model type. The EOS-CG on the other hand, yields low deviations even for T ⪅240 K , which is probably due to the fact that the mixture model parameters were adjusted to a wider temperature range [29]. Adjusting the mixture parameters of the GERG-2008 model to a wider temperature range most likely would improve the model performance. 3.2 Homogeneous State Properties The mean average relative deviation 훿rel results for homogeneous state properties are reported in Table 9. Additionally, deviation plots for the homogeneous state den- sities are given in the Supplementary Information. In the predictive mode, i.e. not using adjusted mixture parameters, the GERG-2008 EOS and (simplified) EOS-CG yield overall the best results, especially for the speed of sound and the liquid density. The molecular-based PCP-SAFT and SAFT-VR Mie EOS also yield low deviations for the density. For most molecular-based and both cubic EOS, relatively large devi- ations are obtained for the 2nd-order derivative properties (isothermal compressibil- ity 훽 and speed of sound w). This is in line with results reported in the literature [56–59] and probably due to the fact that the pure component models were in most cases not adjusted to homogeneous state property data. The isothermal compressibil- ity 훽 is predicted with mean deviations of approximately 5% by the PC-SAFT EOS. Interestingly, the PCP-SAFT EOS, which uses additionally a quadrupole contribu- tion for CO2 , yields lower deviations for the density, but almost equivalent mean deviations for the 2nd-order derivative properties (isothermal compressibility 훽 and speed of sound w) compared to the PC-SAFT EOS. This is likely a result of different fitting strategies used for the development of the pure component models [22, 60]. 1 3 3 International Journal of Thermophysics (2023) 44:182 182 Page 16 of 23 Table 9 Overview of the quality of the description of experimental homogeneous state data for the sys- tem CO2 + O2 by different EOS Results from predictions based on pure component models are compared to results that were obtained after adjusting binary parameters (given in brackets) to the training data. Results for 훿rel are reported for each EOS from top to bottom for all data, the training data, and the test data. 3.2 Homogeneous State Properties For details, see text EOS Property 휌 훽 w Region Liquid Supercritical Vapor Supercritical Supercritical PR 3.11 (2.42) 3.37 (1.93) 3.20 (2.11) 28.33 (32.55) 13.51 (14.35) 3.11 (2.42) 5.38 (2.45) 5.82 (3.69) 32.83 (38.39) 11.97 (12.99) – 2.57 (1.73) 1.95 (1.36) 23.53 (26.33) 15.15 (15.80) SRK 8.73 (9.28) 2.95 (2.76) 1.60 (1.42) 30.06 (31.06) 10.84 (11.06) 8.73 (9.28) 4.60 (4.46) 3.18 (2.76) 33.44 (34.44) 9.00 (9.20) – 2.29 (2.08) 0.84 (0.78) 26.46 (27.47) 12.81 (13.04) PC-SAFT 1.72 (1.01) 2.95 (1.68) 2.40 (1.91) 4.42 (4.29) 4.98 (4.03) 1.72 (1.01) 4.57 (2.08) 4.50 (3.67) 4.95 (4.76) 7.96 (6.62) – 2.30 (1.52) 1.39 (1.06) 3.87 (3.79) 1.81 (1.26) PCP-SAFT 1.38 (1.79) 1.90 (1.62) 1.58 (1.72) 5.96 (6.53) 5.95 (6.11) 1.38 (1.79) 2.00 (2.03) 3.05 (3.20) 6.61 (7.20) 7.79 (7.92) – 1.86 (1.46) 0.87 (1.01) 5.26 (5.82) 3.99 (4.18) SAFT-VR Mie 1.75 (2.00) 1.59 (1.62) 2.07 (2.15) 10.95 (11.42) 14.96 (15.20) 1.75 (2.00) 2.06 (2.10) 3.92 (4.01) 11.72 (12.25) 16.92 (17.21) – 1.41 (1.43) 1.19 (1.25) 10.13 (10.54) 12.87 (13.06) polar soft-SAFT 2.34 (2.41) 2.58 (3.09) 4.56 (4.56) 15.10 (14.95) 12.83 (12.77) 2.34 (2.41) 3.68 (3.90) 8.88 (9.02) 18.69 (18.39) 16.13 (16.00) – 2.14 (2.76) 2.49 (2.45) 11.26 (11.29) 9.31 (9.33) BACKONE 3.97 (3.53) 2.20 (2.80) 2.20 (2.65) 11.69 (10.18) 6.18 (5.42) 3.97 (3.53) 4.30 (3.73) 4.34 (5.45) 12.06 (10.05) 6.62 (5.54) – 1.36 (2.43) 1.18 (1.33) 11.28 (10.31) 5.70 (5.29) sCPA 1.78 (1.70) 1.76 (2.02) 2.52 (2.64) 10.41 (10.87) 11.45 (11.56) 1.78 (1.70) 2.49 (2.42) 5.00 (5.25) 11.15 (11.80) 9.53 (9.68) – 1.47 (1.87) 1.34 (1.39) 9.62 (9.87) 13.50 (13.57) GERG-2008 0.56 (2.59) 1.31 (1.36) 2.25 (1.42) 3.56 (3.27) 0.41 (0.72) 0.56 (2.59) 1.90 (2.05) 4.78 (2.78) 3.45 (2.96) 0.48 (0.84) – 1.08 (1.08) 1.04 (0.77) 3.67 (3.60) 0.34 (0.60) EOS-CG 0.56 (0.67) 1.31 (1.34) 2.26 (1.37) 3.32 (4.04) 0.63 (0.67) 0.56 (0.67) 1.90 (2.01) 4.79 (2.70) 3.34 (3.85) 0.60 (1.04) – 1.08 (1.07) 1.05 (0.74) 3.30 (4.25) 0.66 (0.28) Results from predictions based on pure component models are compared to results that were obtained after adjusting binary parameters (given in brackets) to the training data. Results for 훿rel are reported for each EOS from top to bottom for all data, the training data, and the test data. For details, see text A detailed discussion is given in the Supplementary Information. 3.2 Homogeneous State Properties The speed of sound w is the only studied quantity, where the rotational and vibrational degrees of freedom, included in the ideal gas term ̃aid , cf. Eq. 1, influence the results. The GERG-2008 EOS yields the lowest deviations ≤0.5% followed by the (simplified) EOS-CG with deviations ≤0.7% . From the remaining studied EOS, the PC-SAFT EOS yields the lowest deviations for the speed of sound ≤5%. A detailed discussion is given in the Supplementary Information. The speed of sound w is the only studied quantity, where the rotational and vibrational degrees of freedom, included in the ideal gas term ̃aid , cf. Eq. 1, influence the results. The GERG-2008 EOS yields the lowest deviations ≤0.5% followed by the (simplified) EOS-CG with deviations ≤0.7% . From the remaining studied EOS, the PC-SAFT EOS yields the lowest deviations for the speed of sound ≤5%. 1 3 International Journal of Thermophysics (2023) 44:182 182 Page 17 of 23 The results obtained from the adjusted models provide further interesting insights. While in most cases, significant improvements were observed for the modeling of the VLE phase behavior by adjusting binary interaction parameters (cf. Table 8 and Fig. 2), a more ambivalent picture evolves for the homogeneous state proper- ties (cf. Table 9). In some cases, significant improvements are observed, e.g. for the adjusted GERG-2008 EOS for modeling the isothermal compressibility β as well as for the PC-SAFT EOS for most properties. Yet, in several cases, the model accuracy slightly decreases using an adjusted binary mixture parameter. This is likely due to the fact, that the description of the phase equilibrium and homogeneous state prop- erties are conflicting objectives. Accordingly, improvements in the description of the phase equilibrium are bought at the expense of the decreasing accuracy for the mod- eling of the homogeneous state properties. This is particularly prominent for liquid state densities and speed of sound described by the adjusted GERG-2008 EOS and for the 2nd-order derivatives β and w described by the PCP-SAFT EOS. The EOS- CG yields slightly higher deviations for the liquid state densities and isothermal compressibility but lower deviations for the vapor state densities compared to the corresponding predictive mode results. For other properties and EOS, this conflict- ing objectives effect is also present, but less prominent. Overall, the EOS-CG and adjusted GERG-2008 yield the lowest deviations in the adjusted mode for homo- geneous state properties. 3.2 Homogeneous State Properties Also, the adjusted PCP-SAFT and PC-SAFT EOS yield a reasonably accurate description of the homogeneous state properties—in many cases well competitive to the multi-parameter EOS. The deviations for the test data are in most cases lower compared to the devia- tions for the training data, cf. Table 9. This is likely due to the fact, that the data in the vicinity of the critical point of CO2 is only included in the training data. Mod- eling this data is a challenging task for EOS and therefore yields larger deviations compared to the remaining data. In most cases, the adjusted mode results for the test data show the same trend com- pared to the adjusted mode results for the training data. A lower deviation 훿rel for a given property in the training dataset leads to a lower deviation for this property in the test dataset, when compared to the predictive mode results. The adjusted param- eters can, therefore, also be used to extrapolate to lower (or higher) temperatures and pressures for all studied EOS. Yet, these extrapolations should be done with caution. 4 Conclusions Accurate models for the thermodynamic properties of the mixture CO2 + O2 are crucial for designing Allam cycle processes. In this work, different EOS were used for modeling the mixture CO2 + O2 . First, the available literature data were critically reviewed and compiled in a database, which is provided in the electronic Supple- mentary Information. Two modeling approaches were used: (i) the EOS were used in a purely predictive mode and (ii) in an adjusted mode by parametrizing binary interaction parameters to experimental mixture data. In total, ten EOS were studied: two cubic [15, 16], six molecular-based [18, 21–28], and two multi-parameter EOS [20, 29]. 1 3 182 Page 18 of 23 International Journal of Thermophysics (2023) 44:182 Using adjusted mixture parameters does not automatically result in an improve- ment in the accuracy of all considered properties due to the conflicting objects dur- ing the fit, i.e. correct phase envelope and homogeneous state property description. The phase equilibrium predictions of the GERG-2008 EOS (i.e. no mixture param- eters adjusted) show significant deviations to the experimental phase equilibrium data. The regression of the binary interaction parameter reduces the deviations of the compositions of the coexisting phases obtained from the GERG-2008 EOS dras- tically. Yet, this goes in hand with a decrease of the accuracy for some homogene- ous state properties. Also, using the adjusted GERG-2008 model for extrapolation to lower temperatures results in significant deviations for the phase envelope. i The classical cubic PR EOS yields low deviations for the phase equilibrium predictions for the adjusted mode, but comparably large deviations for homogene- ous state properties. The molecular-based EOS models, on the other hand, are an attractive alternative to the multi-parameter EOS. They provide robust extrapolation capabilities, but yield slightly less accurate description of the thermodynamic prop- erties. In particular, the PCP-SAFT and SAFT-VR Mie models considered in this work yield a good performance. In both cases (cubic and molecular-based EOS), the extrapolation by the temperature-dependent binary interaction parameter results in lower deviations for the phase envelope at low temperatures ( T ⪅240 K ) compared to the GERG-2008 EOS. The EOS-CG (where the binary interaction parameters were regressed using data from a wider temperature range [29]) yields overall the lowest deviations for the phase envelopes at T ⪅240 K . 4 Conclusions In this work, reliable mod- els for describing the mixture behavior CO2 + O2 were developed and compared to the EOS-CG model that was adjusted to mixture data in the literature. In particular, the adjusted GERG-2008, the PCP-SAFT, and the SAFT-VR Mie EOS as well as the EOS-CG from the literature provide accurate models for the thermodynamic proper- ties of the mixture CO2 + O2 . For future work, it would be interesting to apply these thermodynamic models for describing the actual Allam cycle in an exemplaric use case. Also, for a future work, testing the effects of more advanced mixing rules for the different EOS would be interesting.l f The working fluid in the Allam cycle primarily consists of CO2 + O2 . Yet, to some extent also water, nitrogen, argon, and other residue components are present in the process [3]. Extending the models to include these components in the parametri- zation and studying the influence of the obtained models on the description of the Allam cycle would be interesting. Declarations Competing interests The authors have no competing interests to declare that are relevant to the content of this article. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permis- sion directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/ licenses/by/4.0/. 5 Supplementary Information Supplementary data can be found at https://doi.org/10.1007/s10765-023-03297- w. The numerical values for the experimental data for the CO2 + O2 mixture, cf. Tables 1 and 2, are reported as an electronic spreadsheet. The pure component parameters for CO2 and O2 for the studied EOS are reported. Furthermore, results for the phase envelope of temperatures below T < 273 K predicted and calculated by the studied EOS from this work are reported. Furthermore, the overestimation of 1 3 International Journal of Thermophysics (2023) 44:182 Page 19 of 23 182 the critical point by molecular-based EOS and the poor description of the 2nd-order derivatives by the cubic and molecular-based EOS is discussed. the critical point by molecular-based EOS and the poor description of the 2nd-order derivatives by the cubic and molecular-based EOS is discussed. Supplementary Information The online version contains supplementary material available at https://doi. org/10.1007/s10765-023-03297-w. Author Contributions JS: Data curation, formal analysis, methodology (equal), visualization, writing/ original draft. GS: Writing/review and editing (supporting), validation (equal). SM: Writing/review and editing (supporting), validation (equal). HH: funding (equal), writing/review and editing (equal). SiSt: Funding (equal), conceptualization, supervision, methodology (equal), writing/review and editing (equal). Funding Open Access funding enabled and organized by Projekt DEAL. S. Stephan gratefully acknowl- edges financial support of the present work by the IRTG 2057 (252408385). Data Availability All literature data of the mixture CO2 + O2 used in this work are provided in the electronic Supplementary Information in a spreadsheet file. An explanation of that data are further- more given in the Supplementary Information PDF file that can be found here https://doi.org/10.1007/ s10765-023-03297-w. References Ertesvåg, Vapor–liquid equilibrium data for the carbon dioxide and oxygen (CO2 + O2) system at the temperatures 218, 233, 253, 273, 288 and 298 K and pressures up to 14 MPa. Fluid Phase Equilib. 421, 67–87 (2016). https://doi.org/10.1016/j.fluid.2016.04.002 l 7. D. Lozano-Martín, G.U. Akubue, A. Moreau, D. Tuma, C.R. Chamorro, Accurate experimental ( p , 휌 , T ) data of the (CO2 + O2) binary system for the development of models for CCS processes. J. Chem. Thermodyn. 150, 106210 (2020). https://doi.org/10.1016/j.jct.2020.106210 8. D. Lozano-Martín, D. Vega-Maza, M.C. Martín, D. Tuma, C.R. Chamorro, Thermodynamic char- acterization of the (CO2 + O2) binary system for the development of models for CCS processes: Accurate experimental ( p , 휌 , T ) data and virial coefficients. J. Supercrit. Fluids 169, 105074 (2021). https://doi.org/10.1016/j.supflu.2020.105074 l 9. J.A. Commodore, C.E. Deering, R.A. Marriott, Volumetric properties and phase behavior of sulfur dioxide, carbon disulfide and oxygen in high-pressure carbon dioxide fluid. Fluid Phase Equilib. 477, 30–39 (2018). https://doi.org/10.1016/j.fluid.2018.08.012 l 10. S. Ahamada, A. Valtz, S. Chabab, L. Blanco-Martín, C. Coquelet, Experimental density data of three carbon dioxide and oxygen binary mixtures at temperatures from 276 to 416 K and at pres- sures up to 20 MPa. J. Chem. Eng. Data 65, 5313–5327 (2020). https://doi.org/10.1021/acs.jced. 0c00484 11. N.I. Diamantonis, G.C. Boulougouris, E. Mansoor, D.M. Tsangaris, I.G. Economou, Evaluation of cubic, SAFT, and PC-SAFT equations of state for the vapor–liquid equilibrium modeling of CO2 mixtures with other gases. Ind. Eng. Chem. Res. 52, 3933–3942 (2013). https://doi.org/10.1021/ ie303248q q 12. M. Mazzoccoli, B. Bosio, E. Arato, S. Brandani, Comparison of equations-of-state with p–휌 –T experimental data of binary mixtures rich in CO2 under the conditions of pipeline transport. J. Supercrit. Fluids 95, 474–490 (2014). https://doi.org/10.1016/j.supflu.2014.09.047 l 13. A.G. Perez, C. Coquelet, P. Paricaud, A. Chapoy, Comparative study of vapour–liquid equilibrium and density modelling of mixtures related to carbon capture and storage with the SRK, PR, PC- SAFT and SAFT-VR Mie equations of state for industrial uses. Fluid Phase Equilib. 440, 19–35 (2017). https://doi.org/10.1016/j.fluid.2017.02.018 l 14. O. Redlich, J.N.S. Kwong, On the thermodynamics of solutions. V. An equation of state. Fugacities of gaseous solutions. Chem. Rev. 44, 233–244 (1949). https://doi.org/10.1021/cr60137a013i 15. G. Soave, Equilibrium constants from a modified Redlich–Kwong equation of state. Chem. Eng. Sci. 27, 1197–1203 (1972). https://doi.org/10.1016/0009-2509(72)80096-4 16. D.-Y. Peng, D.B. Robinson, A new two-constant equation of state. Ind. Eng. Chem. Fundam. 1 59–64 (1976). https://doi.org/10.1021/i160057a011 17. References 1. R.J. Allam, M. Palmer, C. Hill, G.W. Brown, Jr., System and method for high efficiency power generation using a carbon dioxide circulating working fluid. Palmer Labs, LLC, assignee. Patent 2011/0179799, 31 August 2010 (2011) 2. R.J. Allam, M.R. Palmer, G.W. Brown, J. Fetvedt, D. Freed, H. Nomoto, M. Itoh, N. Okita, C. Jones, High efficiency and low cost of electricity generation from fossil fuels while eliminating atmospheric emissions. Including carbon dioxide. Energy Procedia 37, 1135–1149 (2013). https:// doi.org/10.1016/j.egypro.2013.05.211 2. R.J. Allam, M.R. Palmer, G.W. Brown, J. Fetvedt, D. Freed, H. Nomoto, M. Itoh, N. Okita, C. Jones, High efficiency and low cost of electricity generation from fossil fuels while eliminating atmospheric emissions. Including carbon dioxide. Energy Procedia 37, 1135–1149 (2013). https:// doi.org/10.1016/j.egypro.2013.05.211 3. R. Allam, S. Martin, B. Forrest, J. Fetvedt, X. Lu, D. Freed, G.W. Brown, T. Sasaki, M. Itoh, J. Manning, Demonstration of the Allam cycle: an update on the development status of a high effi- ciency supercritical carbon dioxide power process employing full carbon capture. Energy Procedia 114, 5948–5966 (2017). https://doi.org/10.1016/j.egypro.2017.03.1731 3. R. Allam, S. Martin, B. Forrest, J. Fetvedt, X. Lu, D. Freed, G.W. Brown, T. Sasaki, M. Itoh, J. Manning, Demonstration of the Allam cycle: an update on the development status of a high effi- ciency supercritical carbon dioxide power process employing full carbon capture. Energy Procedia 114, 5948–5966 (2017). https://doi.org/10.1016/j.egypro.2017.03.1731 4. S. Lasala, P. Chiesa, R. Privat, J.-N. Jaubert, VLE properties of CO2-based binary systems contain- ing N2, O2 and Ar: experimental measurements and modelling results with advanced cubic equa- tions of state. Fluid Phase Equilib. 428, 18–31 (2016). https://doi.org/10.1016/j.fluid.2016.05.015l 4. S. Lasala, P. Chiesa, R. Privat, J.-N. Jaubert, VLE properties of CO2-based binary systems contain- ing N2, O2 and Ar: experimental measurements and modelling results with advanced cubic equa- tions of state. Fluid Phase Equilib. 428, 18–31 (2016). https://doi.org/10.1016/j.fluid.2016.05.015l l 5. S. Lasala, Advanced cubic equations of state for accurate modelling of fluid mixtures. application to CO2 capture systems. PhD thesis, Politecnico di Milano Milan, Italy (2016). https://doi.org/10. 13140/RG.2.1.2948.9041 l 5. S. Lasala, Advanced cubic equations of state for accurate modelling of fluid mixtures. application to CO2 capture systems. PhD thesis, Politecnico di Milano Milan, Italy (2016). https://doi.org/10. 13140/RG.2.1.2948.9041 1 3 182 Page 20 of 23 International Journal of Thermophysics (2023) 44:182 6. S.F. Westman, H.G.J. Stang, S.W. Løvseth, A. Austegard, I. Snustad, I.S. References W.G. Chapman, K.E. Gubbins, G. Jackson, M. Radosz, SAFT: equation-of-state solution model for associating fluids. Fluid Phase Equilib. 52, 31–38 (1989). https://doi.org/10.1016/0378-3812(89) 80308-5 18. J. Gross, G. Sadowski, Perturbed-chain SAFT: an equation of state based on a perturbation theory for chain molecules. Ind. Eng. Chem. Res. 40, 1244 (2001). https://doi.org/10.1021/ie0003887i 19. K.E. Starling, M.S. Han, Thermo data refined for LPG, part 14: mixtures. Hydrocarb. Process. 5 129–132 (1972) 20. O. Kunz, W. Wagner, The GERG-2008 wide-range equation of state for natural gases and other mix- tures: an expansion of GERG-2004. J. Chem. Eng. Data 57, 3032–3091 (2012). https://doi.org/10. 1021/je300655b j 21. T. Lafitte, A. Apostolakou, C. Avendaño, A. Galindo, C.S. Adjiman, E.A. Müller, G. Jackson, Accu- rate statistical associating fluid theory for chain molecules formed from Mie segments. J. Chem. Phys. 139, 154504 (2013). https://doi.org/10.1063/1.4819786 y , ( ) p g 22. J. Gross, An equation-of-state contribution for polar components: quadrupolar molecules. AIChE J. 51, 2556–2568 (2005). https://doi.org/10.1002/aic.10502 23. F.J. Blas, L.F. Vega, Thermodynamic behaviour of homonuclear and heteronuclear Lennard–Jones chains with association sites from simulation and theory. Mol. Phys. 92, 135–150 (1997). https:// doi.org/10.1080/002689797170707 g 24. I.I.I. Alkhatib, L.M.C. Pereira, J. Torne, L.F. Vega, Polar soft-SAFT: theory and comparison with molecular simulations and experimental data of pure polar fluids. Phys. Chem. Chem. Phys. 22, 13171–13191 (2020). https://doi.org/10.1039/D0CP00846J ( ) p g 25. A. Müller, J. Winkelmann, J. Fischer, Backone family of equations of state: 1. Nonpolar and polar pure fluids. AIChE J. 42, 1116–1126 (1996). https://doi.org/10.1002/aic.690420423 1 3 International Journal of Thermophysics (2023) 44:182 182 Page 21 of 23 26. U. Weingerl, M. Wendland, J. Fischer, A. Müller, J. Winkelmann, Backone family of equations of state: 2. nonpolar and polar fluid mixtures. AIChE J. 47, 705–717 (2001). https://doi.org/10.1002/ aic.690470317 27. G.M. Kontogeorgis, I.V. Yakoumis, H. Meijer, E. Hendriks, T. Moorwood, Multicomponent phase equilibrium calculations for water–methanol–alkane mixtures, Fluid Phase Equilib. 158–160, 201– 209 (1999). https://doi.org/10.1016/S0378-3812(99)00060-6 28. G.M. Kontogeorgis, M.L. Michelsen, G.K. Folas, S. Derawi, N. Solms, E.H. Stenby, Ten years with the CPA (cubic-plus-association) equation of state. Part 1. Pure compounds and self-associating sys- tems. Ind. Eng. Chem. Res. 45, 4855–4868 (2006). https://doi.org/10.1021/ie051305v 29. J. Gernert, R. Span, EOS-CG: a Helmholtz energy mixture model for humid gases and CCS mix- tures. J. Chem. Thermodyn. 93, 274–293 (2016). https://doi.org/10.1016/j.jct.2015.05.015l 30. R. Span, W. References Wagner, A new equation of state for carbon dioxide covering the fluid region from the triple-point temperature to 1100 K at pressures up to 800 MPa. J. Phys. Chem. Ref. Data 25, 1509–1596 (1996). https://doi.org/10.1063/1.555991 31. R. Span, W. Wagner, Equations of state for technical applications. II. results for nonpolar fluids. In J. Thermophys. 24, 41–109 (2003). https://doi.org/10.1023/A:1022310214958 32. S. Angus, B. Armstrong, R.B. Reuck, International Thermodynamic Tables of the Fluid State-3 Carbon Dioxide (Pergamon, New York, 1976) 33. P.J. Linstrom, W.G. Mallard, NIST Chemistry WebBook, NIST Standard Reference Database Num- ber 69. National Institute of Standards and Technology, Gaithersburg (retrieved June 2023) (2023). https://doi.org/10.18434/T4D303 34. M. Mantovani, P. Chiesa, G. Valenti, M. Gatti, S. Consonni, Supercritical pressure–density–tem- perature measurements on CO2–N2, CO2–O2 and CO2–Ar binary mixtures. J. Supercrit. Fluids 61, 34–43 (2012). https://doi.org/10.1016/j.supflu.2011.09.001 p g j pl 35. P.H. Konynenburg, R.L. Scott, Critical lines and phase equilibria in binary van der Waals mixtures. Philos. Trans. R. Soc. Lond. 298, 495–540 (1980). https://doi.org/10.1098/rsta.1980.0266 36. A.M.A. Dias, J.C. Pàmies, Ja.A.P. Coutinho, I.M. Marrucho, L.F. Vega, SAFT modeling of the sol- ubility of gases in perfluoroalkanes. J. Phys. Chem. B 108, 1450–1457 (2004). https://doi.org/10. 1021/jp036225o 37. I.I.I. Alkhatib, F. Llovell, L.F. Vega, Assessing the effect of impurities on the thermophysical prop- erties of methane-based energy systems using polar soft-SAFT. Fluid Phase Equilib. 527, 112841 (2021). https://doi.org/10.1016/j.fluid.2020.112841i l 38. S. Dufal, T. Lafitte, A. Galindo, G. Jackson, A.J. Haslam, Developing intermolecular-potential mod- els for use with the SAFT-VR Mie equation of state. AIChE J. 61, 2891–2912 (2015). https://doi. org/10.1002/aic.14808l 39. N.I. Diamantonis, I.G. Economou, Evaluation of statistical associating fluid theory (SAFT) and per- turbed chain-SAFT equations of state for the calculation of thermodynamic derivative properties of fluids related to carbon capture and sequestration. Energy Fuels 25, 3334–3343 (2011). https://doi. org/10.1021/ef200387p 40. I. Tsivintzelis, G.M. Kontogeorgis, M.L. Michelsen, E.H. Stenby, Modeling phase equilibria for acid gas mixtures using the CPA equation of state. Part II: binary mixtures with CO2. Fluid Phase Equilib. 306, 38–56 (2011). https://doi.org/10.1016/j.fluid.2011.02.006 l 41. L. Sun, X. Liang, N. Solms, G.M. Kontogeorgis, Solubility modeling of air in aqueous electrolyte solutions with the e-CPA equation of state. Ind. Eng. Chem. Res. 59, 18693–18704 (2020). https:// doi.org/10.1021/acs.iecr.0c03164 42. R. Klimeck, Entwicklung einer Fundamentalgleichung für Erdgase für das Gas-und Flüssigkeitsge- biet sowie das Phasengleichgewicht. PhD thesis, Fakultät für Maschinenbau, Ruhr-Universität Bochum, Bochum (2000) ( ) 43. R. Schmidt, W. References Wagner, A new form of the equation of state for pure substances and its application to oxygen. Fluid Phase Equilib. 19, 175–200 (1985). https://doi.org/10.1016/0378-3812(85)87016-3 44. R. Span, W. Wagner, Equations of state for technical applications. I. Simultaneously optimized functional forms for nonpolar and polar fluids. Int. J. Thermophys. 24, 1–39 (2003). https://doi.org/ 10.1023/A:1022390430888 45. R. Span, W. Wagner, On the extrapolation behavior of empirical equations of state. Int. J. Thermo- phys. 18, 1415–1443 (1997). https://doi.org/10.1007/BF02575343l 46. S. Stephan, U.K. Deiters, Characteristic curves of the Lennard–Jones fluid. Int. J. Thermophys. 41, 147 (2020). https://doi.org/10.1007/s10765-020-02721-9 1 3 182 Page 22 of 23 International Journal of Thermophysics (2023) 44:182 47. J. Staubach, S. Stephan, Prediction of thermodynamic properties of fluids at extreme conditions: assessment of the consistency of molecular-based models. In: Aurich, J.C., Garth, C., Linke, B.S. (eds.) Proceedings of the 3rd Conference on Physical Modeling for Virtual Manufacturing Systems and Processes, pp. 170–188 (2023). https://doi.org/10.1007/978-3-031-35779-4_10 48. R. Span, Multiparameter Equations of State. Springer, Berlin (2000). https://doi.org/10.1007/ 978-3-662-04092-8 49. E.W. Lemmon, R.T. Jacobsen, A generalized model for the thermodynamic properties of mixtures. Int J Thermophys 20 825 835 (1999) https://doi org/10 1023/A:1022627001338 49. E.W. Lemmon, R.T. Jacobsen, A generalized model for the thermodynamic properties of Int. J. Thermophys. 20, 825–835 (1999). https://doi.org/10.1023/A:1022627001338 50. I.H. Bell, A. Jäger, Helmholtz energy transformations of common cubic equations of state for use with pure fluids and mixtures. J. Res. Natl Inst. Stand. Technol. 121, 238–263 (2016). https://doi. org/10.6028/jres.121.011 g j 51. G. Mie, Zur kinetischen Theorie der einatomigen Körper. Ann. Phys. 316, 657–697 (1903). https:// doi.org/10.1002/andp.19033160802 52. E. Schäfer, G. Sadowski, S. Enders, Interfacial tension of binary mixtures exhibiting azeotropic behavior: measurement and modeling with PCP-SAFT combined with density gradient theory. Fluid Phase Equilib. 362, 151 (2014). https://doi.org/10.1016/j.fluid.2013.09.042 l 53. N. Solms, M.L. Michelsen, G.M. Kontogeorgis, Prediction and correlation of high-pressure gas sol- ubility in polymers with simplified PC-SAFT. Ind. Eng. Chem. Res. 44, 3330–3335 (2005). https:// doi.org/10.1021/ie049089y 54. M. Yarrison, W.G. Chapman, A systematic study of methanol+n-alkane vapor–liquid and liquid– liquid equilibria using the CK-SAFT and PC-SAFT equations of state. Fluid Phase Equilib. 226, 195–205 (2004). https://doi.org/10.1016/j.fluid.2004.09.024 l 55. I. Antolović, J. Staubach, S. Stephan, J. Vrabec, Phase equilibria of symmetric Lennard–Jones mix- tures and a look at the transport properties near the upper critical solution temperature. Phys. Chem. Chem. Phys. 25, 17627–17638 (2023). https://doi.org/10.1039/D3CP01434G y p g 56. F. Llovell, C.J. Peters, L.F. References Vega, Second-order thermodynamic derivative properties of selected mixtures by the soft-SAFT equation of state. Fluid Phase Equilib. 248, 115–122 (2006). https://doi. org/10.1016/j.fluid.2006.07.018 l 57. X. Liang, K. Thomsen, W. Yan, G.M. Kontogeorgis, Prediction of the vapor–liquid equilibria and speed of sound in binary systems of 1-alkanols and n-alkanes with the simplified PC-SAFT equation of state. Fluid Phase Equilib. 360, 222–232 (2013). https://doi.org/10.1016/j.fluid.2013.09.037 q ( ) p g jl 58. M. Khammar, J.M. Shaw, Speed of sound prediction in 1-n-alcohol+n-alkane mixtures using a translated SAFT-VR-Mie equation of state. Fluid Phase Equilib. 288, 145–154 (2010). https://doi. org/10.1016/j.fluid.2009.10.022 g jl 59. S. Stephan, J. Staubach, H. Hasse, Review and comparison of equations of state for the Lennard Jones fluid. Fluid Phase Equilib. 523, 112772 (2020). https://doi.org/10.1016/j.fluid.2020.112772 ll 60. J. Gross, G. Sadowski, Application of the perturbed-chain SAFT equation of state to associating systems. Ind. Eng. Chem. Res. 41, 5510–5515 (2002). https://doi.org/10.1021/ie010954df y g g 61. M. Ahmad, J. Gernert, E. Wilbers, Effect of impurities in captured CO2 on liquid-vapor equilibrium. Fluid Phase Equilib. 363, 149–155 (2014). https://doi.org/10.1016/j.fluid.2013.11.009 , , ,f p p 2 q p q Fluid Phase Equilib. 363, 149–155 (2014). https://doi.org/10.1016/j.fluid.2013.11.009 l 62. A. Fredenslund, M. Jørgen, O. Persson, Gas–liquid equilibrium of oxygen–ca Chem. Eng. Data 17, 440–443 (1972). https://doi.org/10.1021/je60055a019 l 62. A. Fredenslund, M. Jørgen, O. Persson, Gas–liquid equilibrium of oxygen–carbon dioxide system. 62. A. Fredenslund, M. Jørgen, O. Persson, Gas liquid equilibrium of oxygen ca Chem. Eng. Data 17, 440–443 (1972). https://doi.org/10.1021/je60055a019 63. A. Fredenslund, G. Sather, Gas–liquid equilibrium of the oxygen–carbon dioxide system. Eng Data 15 17 22 (1970) https://doi org/10 1021/je60044a024 63. A. Fredenslund, G. Sather, Gas–liquid equilibrium of the oxygen–carbon dioxide system. J. Chem 63. A. Fredenslund, G. Sather, Gas–liquid equilibrium of the oxygen–c Eng. Data 15, 17–22 (1970). https://doi.org/10.1021/je60044a024 63. A. Fredenslund, G. Sather, Gas liquid equilibrium of the oxygen c Eng. Data 15, 17–22 (1970). https://doi.org/10.1021/je60044a024 Eng. Data 15, 17–22 (1970). https://doi.org/10.1021/je60044a024 64. G.-I. Kaminishi, T. Toriumi, Gas–liquid equilibrium under high pressures. VI. Vapor–liquid phase equilibrium in the CO2-H2, CO2-N2, and CO2-O2 systems. J. Soc. Chem. Ind. Jpn. 69, 175–178 (1966). https://doi.org/10.1246/nikkashi1898.69.2_175 65. P.L. Chueh, N.K. Muirbrook, J.M. Prausnitz, Multicomponent vapor–liquid equilibria at high pres- sures: Part II. Thermodynamic analysis. AIChE J. 11, 1097–1102 (1965). https://doi.org/10.1002/ aic.690110625 66. G.H. Zenner, L.I. Dana, Liquid–vapor equilibrium compositions of carbon dioxide–oxygen–nitro- gen mixtures. Chem. Eng. Prog. Symp. Ser. 44, 36–41 (1963)f 67. I. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 69. G.J. Gururaja, M.A. Tirunarayanan, A. Ramachandran, Dynamic viscosity of gas mixtures. J. Chem. Eng. Data 12, 562–567 (1967). https://doi.org/10.1021/je60035a024 69. G.J. Gururaja, M.A. Tirunarayanan, A. Ramachandran, Dynamic viscosity of gas mixtures. J. Chem . G.J. Gururaja, M.A. Tirunarayanan, A. Ramachandran, Dynamic viscosity of gas mixtures. J. Chem. Eng. Data 12, 562–567 (1967). https://doi.org/10.1021/je60035a024 References Al-Siyabi, Effect of impurities on CO2 stream properties. PhD thesis, Heriot-Watt University (2013). http://hdl.handle.net/10399/2643 68. M. Mazzoccoli, B. Bosio, E. Arato, Pressure–density–temperature measurements of binary mix- tures rich in CO2 for pipeline transportation in the CCS process. J. Chem. Eng. Data 57, 2774–2783 (2012). https://doi.org/10.1021/je300590v 1 3 3 International Journal of Thermophysics (2023) 44:182 Page 23 of 23 18 Page 23 of 23 182 182 69. G.J. Gururaja, M.A. Tirunarayanan, A. Ramachandran, Dynamic viscosity of gas mixtures. J. Chem. Eng. Data 12, 562–567 (1967). https://doi.org/10.1021/je60035a024 Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 1 3
https://openalex.org/W4394766158	https://link.springer.com/content/pdf/10.1007/s00415-024-12302-7.pdf	English	null	[18F]DPA-714-PET-MRI reveals pronounced innate immunity in human anti-LGI1 and anti-CASPR2 limbic encephalitis	Journal of neurology	2,024	cc-by	4,273	[18F]DPA‑714‑PET‑MRI reveals pronounced innate immunity in human anti‑LGI1 and anti‑CASPR2 limbic encephalitis Wolfgang Roll1 · Jan Bauer2 · Andre Dik3 · Christoph Mueller3 · Philipp Backhaus1,4 · Saskia Räuber5 · Bastian Zinnhardt4,6 · Marco Gallus3,7 · Catriona Wimberley8 · Peter Körtvelyessy9 · Philipp Schindler10 · Werner Stenzel11 · Christian E. Elger12 · Albert Becker13 · Jan Lewerenz14 · Heinz Wiendl3 · Sven G. Meuth5 · Michael Schäfers1,4 · Nico Melzer5 Received: 8 February 2024 / Revised: 29 February 2024 / Accepted: 1 March 2024 / Published online: 12 March 2024 © The Author(s) 2024 8 Edinburgh Imaging, University of Edinburgh, Edinburgh, UK Dear Sirs, cell-based assays may yield false-positive and -negative results [5]. Hence, novel biomarkers directly addressing the parenchymal immune response are urgently warranted to enhance diagnostic accuracy together with AAB testing. [18F]DPA-714 is a second-generation PET-tracer target- ing the 18 kDa translocator-protein (TSPO), overexpressed on the mitochondrial membrane of activated microglia and other innate immune cells [6, 7]. Previous studies provided data on increased TSPO expression, suggesting ongoing inflammation, in mesial temporal seizure foci and in con- tralateral mesial temporal lobe [8] with similar inflammatory changes found in brain tissue specimen of ALE [9]. cell-based assays may yield false-positive and -negative results [5]. Hence, novel biomarkers directly addressing the parenchymal immune response are urgently warranted to enhance diagnostic accuracy together with AAB testing. A major cause of mesial temporal lobe seizures and epilepsy, memory disturbance and psychiatric symptoms is autoim- mune limbic encephalitis (ALE), mediated by adaptive and concomitant innate autoimmune inflammatory processes. Standard clinical workup comprises magnetic resonance imaging (MRI), electroencephalography (EEG), CSF analy- sis, and neuropsychological assessment [1]. [18F]DPA-714 is a second-generation PET-tracer target- ing the 18 kDa translocator-protein (TSPO), overexpressed on the mitochondrial membrane of activated microglia and other innate immune cells [6, 7]. Previous studies provided data on increased TSPO expression, suggesting ongoing inflammation, in mesial temporal seizure foci and in con- tralateral mesial temporal lobe [8] with similar inflammatory changes found in brain tissue specimen of ALE [9]. ALE with autoantibodies (AABs), against leucine-rich, glioma inactivated (LGI1) and contactin-associated protein- like 2 (CASPR2) belong to the most frequent subtypes of ALE. In these ALE entities, MRI and EEG often display unspecific and very subtle changes, routine CSF analysis is often unremarkable [2, 3], and no specific pattern of cognitive dysfunction exists [4]. * Nico Melzer nico.melzer@med.uni-duesseldorf.de Journal of Neurology (2024) 271:3653–3659 https://doi.org/10.1007/s00415-024-12302-7 Journal of Neurology (2024) 271:3653–3659 https://doi.org/10.1007/s00415-024-12302-7 LETTER TO THE EDITORS Dear Sirs, Two patients with AABs against surface membrane neural antigens underwent [18F]DPA-714-PET-MRI. Both patients were high affinity binders without TSPO polymor- phism. Both patients received high-dose corticosteroid ther- apy (patient #1: 500 mg/d for 5 days; patient #2: 1000 mg/d for 3 days) followed by tapering and immunoabsorption therapy (5 cycles). Afterwards patients received an induc- tion therapy with 1000 mg Rituximab. Patient #1 received additional maintenance therapy with 1000 mg Rituximab 6 months and 1 year after initial diagnosis. Patient #2 received an additional dose of 1000 mg Rituximab 2 weeks after the first cycle, followed by 4 more cycles of Rituximab during the following 2 years. Standard 10–20 surface electrode systems with additional anterior temporal electrodes for short-term- and basal tem- poral electrodes for long-term-EEG were used. The EEG records were rated regarding interictal epileptic discharges/ slowing or ictal events confined to anterior temporal elec- trodes in an unilateral or bilateral fashion as previously described [4]. Cell counts, protein and immunoglobulin levels as well as the presence of IgG AAB against intracellular and neural surface membrane antigens in serum and CSF were ana- lyzed as previously described [4]. Viral, fungal and bac- terial pathogens, and rheumatological-vasculitic disorders were ruled out. A 71-year-old patient with seropositive ALE (patient #1) was hospitalized following the occurrence of memory defi- cits and impulsiveness. LGI1 AABs were detected in serum (titer 1:100) but not in CSF. CSF routine analysis results were as follows: cell count 0/µl; glucose: 68.4 mg/dl, oligo- clonal bands: type 5; IgG Index: <0.7. [18F]DPA-714 was prepared automatically in a GE TRACERlab MX module as described in detail previously [7]. [18F]DPA-714-PET-MRI (Fig. 1) before initiation of immunotherapy showed asymmetrically elevated tracer uptake with punctum maximum in the left amygdala (SUVR; left: 1.431; right: 1.364) and hippocampus (SUVR; left: 1.287; right: 1.227) compared to the contralateral hemi- sphere (Fig. 1). Uptake was above cerebellar gray matter, used as reference region for the calculation of SUVR. Asym- metrical uptake correlated with FLAIR signal alterations with temporomesial edema in the left hemisphere. Consist- ently EEG revealed anterior temporal sharp-slow-waves and slowing on the left hemisphere. No significant mesial tempo- ral cognitive dissociation with asymmetrical mesial temporal dysfunction was found (z-score left: – 2.28; right: – 2.32). Hybrid Imaging was performed on a 3 T PET-MRI (mMR; Siemens Healthcare). Dear Sirs, Even AAB testing using Here, we aimed at corroborating the potential of TSPO- PET-MRI as a novel diagnostic imaging marker for the assessment of innate immunity in human ALE with AABs against LGI1 and CASPR2 given the fact that antigen-bound * Nico Melzer nico.melzer@med.uni-duesseldorf.de 1 Department of Nuclear Medicine, University of Münster, Münster, Germany 2 Department of Neuroimmunology, Center for Brain Research, Medical University of Vienna, Vienna, Austria 3 Department of Neurology with Institute of Translational Neurology, University of Münster, Münster, Germany 4 European Institute for Molecular Imaging, University of Münster, Münster, Germany 5 Department of Neurology, Medical Faculty and University Hospital, Heinrich Heine University of Düsseldorf, Moorenstraße 5, 40225 Düsseldorf, Germany 6 Biomarkers and Translational Technologies (BTT), Pharma Research & Early Development (pRED), F.Hoffmann-La Roche Ltd., Basel, Switzerland 7 Department of Neurological Surgery, University of California San Francisco, San Francisco, USA * Nico Melzer nico.melzer@med.uni-duesseldorf.de 1 Department of Nuclear Medicine, University of Münster, Münster, Germany 2 Department of Neuroimmunology, Center for Brain Research, Medical University of Vienna, Vienna, Austria 3 Department of Neurology with Institute of Translational Neurology, University of Münster, Münster, Germany 4 European Institute for Molecular Imaging, University of Münster, Münster, Germany 5 Department of Neurology, Medical Faculty and University Hospital, Heinrich Heine University of Düsseldorf, Moorenstraße 5, 40225 Düsseldorf, Germany 6 Biomarkers and Translational Technologies (BTT), Pharma Research & Early Development (pRED), F.Hoffmann-La Roche Ltd., Basel, Switzerland 7 Department of Neurological Surgery, University of California San Francisco, San Francisco, USA 8 Edinburgh Imaging, University of Edinburgh, Edinburgh, UK 9 Department of Neurology, Campus Benjamin Franklin, Charité-Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Berlin, Germany 10 Department of Clinical Radiology, University of Münster, Münster, Germany 11 Department of Neuropathology, and Berlin Institute of Health (BIH), Charité-Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, Berlin, Germany 12 Department of Epileptology, University of Bonn, Bonn, Germany 13 Section for Translational Epilepsy Research, Department of Neuropathology, University of Bonn, Bonn, Germany 14 Department of Neurology, University of Ulm, Ulm, Germany 9 Department of Neurology, Campus Benjamin Franklin, Charité-Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Berlin, Germany 1 Department of Nuclear Medicine, University of Münster, Münster, Germany 10 Department of Clinical Radiology, University of Münster, Münster, Germany 11 Department of Neuropathology, and Berlin Institute of Health (BIH), Charité-Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, Berlin, Germany 12 Department of Epileptology, University of Bonn, Bonn, Germany 13 Section for Translational Epilepsy Research, Department of Neuropathology, University of Bonn, Bonn, Germany 6 Biomarkers and Translational Technologies (BTT), Pharma Research & Early Development (pRED), F.Hoffmann-La Roche Ltd., Basel, Switzerland 7 Department of Neurological Surgery, University of California San Francisco, San Francisco, USA :(0123 3456789) 3654 Journal of Neurology (2024) 271:3653–3659 known to affect the binding affinity of TSPO-PET-tracers, as described previously [7]. Dear Sirs, AABs have been shown to yield a microglia response in the brain parenchyma [10]. A focus of this work was on the immunohistochemical crossvalidation of the TSPO-PET signal on the cellular level. Sections from control (Autopsy brain from male, 71 years without neurological disease) and two different ALE patients with identical AAB (LGI1, CASPR2) were stained for TSPO as previously described using anti-TSPO (Abcam, ab109497, 1:1.000) antibodies [11]. Two ALE patients underwent combined [18F]DPA-714- PET-MRI as compassionate use. Patients underwent routine clinical evaluation in the University Hospital of Münster, Germany. Diagnosis of ALE was based on current consen- sus criteria [1, 4]. Retrospective analysis was approved by local ethics committee (Ethikkommission der Ärztekam- mer Westfalen-Lippe; reference number 2013–350-f-S and 2021–144-f-S, and from the Medical University of Vienna, EK 1206/2013) and was performed in accordance with the principles of the 1964 Declaration of Helsinki and its later amendments. All patients gave written informed consent. Multiplex immunofluorescent labeling was performed with antibodies against TSPO (Abcam, ab109497, 1:10.000), neurons (NeuN; Merck MAB377, 1:2500), oligodendro- cytes [12] (TPPP/p25 (1:5000, kind gift from Romana Höftberger), astrocytes (GFAP, Thermo Scient. #MS- 1376, 1:1000), and microglia/macrophages (Iba-1, Wako #019–19741, 1:10.000) by utilizing the Akoya Fluorescent Multiplex kit according to the manufacturer’s protocol [11]. Two patients with AABs against surface membrane neural antigens underwent [18F]DPA-714-PET-MRI. Both patients were high affinity binders without TSPO polymor- phism. Both patients received high-dose corticosteroid ther- apy (patient #1: 500 mg/d for 5 days; patient #2: 1000 mg/d for 3 days) followed by tapering and immunoabsorption therapy (5 cycles). Afterwards patients received an induc- tion therapy with 1000 mg Rituximab. Patient #1 received additional maintenance therapy with 1000 mg Rituximab 6 months and 1 year after initial diagnosis. Patient #2 received an additional dose of 1000 mg Rituximab 2 weeks after the first cycle, followed by 4 more cycles of Rituximab during the following 2 years. Multiplex immunofluorescent labeling was performed with antibodies against TSPO (Abcam, ab109497, 1:10.000), neurons (NeuN; Merck MAB377, 1:2500), oligodendro- cytes [12] (TPPP/p25 (1:5000, kind gift from Romana Höftberger), astrocytes (GFAP, Thermo Scient. #MS- 1376, 1:1000), and microglia/macrophages (Iba-1, Wako #019–19741, 1:10.000) by utilizing the Akoya Fluorescent Multiplex kit according to the manufacturer’s protocol [11]. Neuropsychological assessments were performed after recovery from seizures. Verbal memory scores were extracted for left temporal lobe cognitive function and visual memory scores for right temporal lobe, as described previ- ously [4]. Dear Sirs, Although AABs in anti-LGI1 and anti-CASPR2 ALE are often not detectable in CSF [2, 3], they can be retrieved as monoclonal AABs from intrathecal antibody secreting B-cell populations and exert functional effects consistent with their pathogenic parenchymal effect [15]. region in T2/FLAIR-MRI and non-detectable serum AAB against LGI1. region in T2/FLAIR-MRI and non-detectable serum AAB against LGI1. region in T2/FLAIR-MRI and non-detectable serum AAB against LGI1. A 65-year-old patient with seropositive ALE (patient #2) and AABs against CASPR2 in serum (titer 1:3200) and CSF (titer 1:320) was hospitalized following recurrent tem- poral lobe seizures and associated memory deficits. CSF routine analysis results were as follows: cell count 6/µl; glucose: 61.3 mg/dl, oligoclonal bands: type 1; IgG Index: <0.7. In [18F]DPA-714-PET-MRI (Fig. 2) before initiation of immunotherapy quantitative uptake values were above cer- ebellar reference region for amygdala (SUVR; left: 1.222; right: 1.222) and hippocampus (SUVR; left: 1.155; right: 1.114), however, not asymmetrical as in patient #1. FLAIR images revealed nearly symmetrical signal alterations in mesial temporal lobes of both hemispheres. Consistently in EEG, abnormalities occurred in both hemispheres with ante- rior temporal slowing. Neuropsychological testing showed cognitive dissociation with dominant dysfunction in visual memory in comparison to verbal memory, indicating right mesial temporal impairment (z-score left: -0.27; right: -2.12) in contrast to symmetrical alterations in [18F]DPA-714-PET, FLAIR and EEG. Indeed, imaging TSPO expression in infiltrating and parenchymal immune cells allows for the detection of Iba-1 + phagocytes as the main source not only in ALE, but also in the myeloid tumor microenvironment [6], and cerebral vasculitis [7]. GFAP + astrocytes contribute to the TSPO-PET signal to a lower amount [6, 7]. Moreover, in the previous studies, the [18F]DPA-714-PET signal exceeded the MRI abnormalities [7]. Thus, [18F]DPA-714-PET-MRI might allow for the imaging of key pathological processes in inflammatory CNS diseases. Larger studies have to define the role of [18F]DPA-714-PET compared to standard MRI in the clinical setting. It is important to address not only imaging at initial diagnosis, but also for response assess- ment during/after immunotherapy. Questions arising on the specificity of the signal over the time of the disease should be addressed in dedicated preclinical models, in comparison to changes in FLAIR/T2 signal alterations. Relative T1 volume of amygdala (left: 0.130; right: 0.144) and hippocampus (left: 0.258; right: 0.268) again did not show relevant lateralization. Dear Sirs, Dynamic PET was acquired in list mode after injection of 237/258 MBq [18F]DPA-714 for 60 min after injection. MRI included non-contrast enhanced sequences: isotropic (1 mm) 3D structural T1-weighted, axial T2-weighted-sequences and axial/coronar FLAIR. 3D T1-weighted MR images were processed with Free- surfer (http://surfer.nmr.mgh.harvard.edu/), as previously described [11]. Relative volume of hippocampus and amyg- dala to intracerebral volume were used for further analysis. After coregistration with segmented T1-weighted MR images, regional standardized uptake value ratio (SUVR) of the [18F]DPA-714 PET were extracted, using the cerebel- lar grey matter as reference region [11]. Relative T1 volume of amygdala (left: 0.120; right: 0.101) and hippocampus (left: 0.257; right: 0.252) did not show relevant lateralization. One year after TSPO-PET patient #1 reported a subjectively complete recovery with normaliza- tion of the volume/signal increase of the left temporomesial Patients’ blood samples were analyzed for single nucleo- tide polymorphism c.439A > G (rs6971, p.Thr147Ala), 3655 Journal of Neurology (2024) 271:3653–3659 Fig. 1 FLAIR-MRI, [18F] DPA-714 PET and fused images (upper left) of patient #1 with anti-LGI1 autoimmune limbic encephalitis. Quantification of the SUVR and relative T1 volumes of amygdala and hip- pocampus of patient #1 (upper right). Multiplex staining, in brain tissue samples of an inde- pendent patient with anti-LGI1 autoimmune limbic encephalitis obtained from epilepsy surgery for seizure control, for TSPO together with GFAP (A), TPPP/ p25 (B), Iba-1 (C) and NeuN (D) in the hippocampus. Strong TSPO mitochondrial reactivity is only seen in the Iba-1+ micro- glial cells. GFAP+ astrocytes and TPPP/p25+ oligodendro- cytes show much weaker TSPO reactivity. No double labeling for TSPO is seen in NeuN+ neurons. The insets in A-D show higher magnifications of the single cells indicated by the arrowheads Journal of Neurology (2024) 271:3653–3659 3656 that ALE is a bilateral albeit often asymmetric disease [1]. In accordance with findings in seronegative and seroposi- tive ALE with AABs against intracellular neural antigens, the [18F]DPA-714 signal correlated with FLAIR-MRI and EEG alterations but not with neuropsychological assess- ments and T1-volumetry [11]. In this study, clear evidence argued for FLAIR signal increase and EEG abnormalities being down-stream effects of an antibody-mediated neural effector mechanism [13, 14]. Consistently, it has recently been demonstrated that antigen-bound parenchymal AABs via Fc receptor signaling also elicit such a parenchymal microglia response [10]. Dear Sirs, Two years after TSPO-PET, patient #2 had a normaliza- tion of previously observed EEG changes and of the volume/ signal increase of the temporomesial region in T2/FLAIR- MRI. CASPR2 AABs remained positive. In control brain, moderate expression of TSPO was equally detected in oligodendrocytes, astrocytes and micro- glial cells. Neuronal cell bodies were negative (Fig. 3). In both anti-LGI1 (Fig. 1C) and anti-CASPR2 (Fig. 2C) ALE brain (staining for TSPO was stronger in activated glial cells although neurons remained negative (Figs. 1D and 2D). Multiplex immunofluorescence imaging showed that in anti-LGI1 ALE brain (Fig. 1) especially the Iba-1 + microglial cells showed strong expression of TSPO. Here, GFAP + astrocytes and TPPP/p25 + oligodendro- cytes showed weaker expression of TSPO. NeuN + neurons showed absence of TSPO reactivity. A comparable TSPO reactivity was seen in the anti-CASPR2 ALE case (Fig. 2) with strong TSPO expression in microglial cells, less intense reactivity in astrocytes and oligodendrocytes and absence of TSPO expression in neurons. Limitations of our study include method inherent disad- vantages of [18F]DPA-714-PET-MRI discussed in previous publications [6, 7]. First to mention is limited availability of PET-MR systems and tracers as [18F]DPA-714 [8, 11]. Small patient number and matched PET-MRI and histopathologi- cal specimen from different patients are a further limitation. A major limitation of [18F]DPA-714 is reliable quantifica- tion and specificity of tracer binding. Gold standard for the assessment of specific binding to the target is kinetic mod- eling out of dynamic PET-datasets. In a previous analysis, we were able to show that binding potentials calculated by kinetic modeling showed a very high correlation to SUVR with cerebellar grey matter as reference region used in this study [11]. Comparison to healthy controls would further strengthen our results. However, preclinical evaluation of [18F]DPA-714-PET-MRI in an experimental setup with his- topathological correlation underlined specificity of the PET signal in human ALE [11]. Potential spillover from the cho- roid plexus is another disadvantage of [18F]DPA-714-PET, Our data from two ALE patients with AABs against sur- face membrane neural antigens are in line with previously published results showing elevated temporomesial [18F] DPA-714 uptake in patients with temporal lobe epilepsy in a bilateral symmetric or asymmetric fashion suggesting ongoing inflammation inside and outside of current seizure foci [8]. Our findings are further consistent with the notion 3657 Journal of Neurology (2024) 271:3653–3659 Fig. allied diseases within the German NEtwork for Research on AuToim- mune Encephalitis—CONNECT GENERATE; 01GM1908). allied diseases within the German NEtwork for Research on AuToim- mune Encephalitis—CONNECT GENERATE; 01GM1908). but can be limited when using automatic brain segmentation as in this study [8, 11]. Neuropsychological assessment is hampered by the fact that nonverbal memory performance is associated with both temporal lobes [4]. Future studies should include 3D FLAIR allowing for reliable volumetry in a larger patient cohort. Availability of data and material All data generated or analyzed during the current study are included in this published article. Declarations To conclude, we provide preliminary data on the poten- tial [18F]DPA-714-PET-MRI as a direct imaging maker of neuroinflammation in ALE with antibodies against surface membrane neural antigens. Larger studies are needed to define the abilities of [18F]DPA-714-PET-MRI for clinical and treatment monitoring purposes in comparison to stand- ard of care. Conflicts of interest On behalf of all the authors, the corresponding author states that there is no conflict of interest. Ethics approval and consent to participate This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the local Ethics Committee (Ethikkommission der Ärztekammer Westfalen-Lippe; reference number 2013–350-f-S and 2021–144-f-S, and from the Medical University of Vienna, EK 1206/2013). All patients gave written informed consent. Acknowledgements We thank all technical imaging staff from the Department of Nuclear Medicine and the Department of Radiology, University of Münster, Münster, Germany, especially Anne Exler and Stan Milachowski for performing human DPA-714-PET-MRI scans. We acknowledge Michael Kassiou, School of Chemistry, The Univer- sity of Sydney, Australia, for providing data on [18F]DPA-714 toxicol- ogy, and Frank Tüttelmann and Albrecht Röpke, Institute of Human Genetics, University of Münster, Münster, Germany, for the analyses of TSPO binding affinities. Consent for publication Written informed consent for publication was obtained. Dear Sirs, 2 FLAIR-MRI, [18F] DPA-714 PET and fused images (upper left) of patient #2 with anti-CASPR2 autoimmune limbic encephalitis. Quantifica- tion of the SUVR and relative T1 volumes of amygdala and hippocampus of patient #2 (upper right). Multiplex stain- ing, in brain tissue samples of an independent patient with anti-CASPR2 autoimmune lim- bic encephalitis, obtained from epilepsy surgery for seizure control, for TSPO together with TPPP/p25 (A), GFAP (B), Iba-1 (C) and quadruple staining for TSPO, TPPP/p25, GFAP and Iba-1 (D). In addition, here, strong TSPO mitochondrial reactivity is only seen in Iba-1+ microglial cells. GFAP+ astro- cytes and TPPP/p25+ oligoden- drocytes show much weaker TSPO reactivity. The arrowhead (enlarged in the inset) here points at a large neuron that is negative for TSPO. The insets in A-D show higher magni- fications of the single cells indicated by the arrowheads 3658 Journal of Neurology (2024) 271:3653–3659 Fig. 3 TSPO staining was performed in A control brain (71 year with no neurological disease), B anti-LGI1 autoimmune limbic encepha- litis brain and C anti-CASPR2 autoimmune limbic encephalitis brain. Whereas in control brain, a moderate expression of TSPO in all glial cells is seen, in anti-LGI1 and anti-CASPR2 autoimmune limbic encephalitis brain, activated glial cells show an increased reactivity for TSPO. Neurons in control as well as LGI1 and CASPR encephalitis brain are negative for TSPO all glial cells is seen, in anti-LGI1 and anti-CASPR2 autoimmune limbic encephalitis brain, activated glial cells show an increased reactivity for TSPO. Neurons in control as well as LGI1 and CASPR encephalitis brain are negative for TSPO all glial cells is seen, in anti-LGI1 and anti-CASPR2 autoimmune limbic encephalitis brain, activated glial cells show an increased reactivity for TSPO. Neurons in control as well as LGI1 and CASPR encephalitis brain are negative for TSPO Fig. 3 TSPO staining was performed in A control brain (71 year with no neurological disease), B anti-LGI1 autoimmune limbic encepha- litis brain and C anti-CASPR2 autoimmune limbic encephalitis brain. Whereas in control brain, a moderate expression of TSPO in Consent for publication Written informed consent for publication was obtained. Consent for publication Written informed consent for publication was obtained. Open Access This article is licensed under a Creative Commons Attri- bution 4.0 International License, which permits use, sharing, adapta- tion, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Author contributions WR, JB, AD, CM, HW, SGM, MS and NM con- tributed to the study conception and secured funding. WR, JB, SR, PK, SGM, MS and NM designed the study. Material preparation and data collection were performed by WR, AD, PB, SR, BZ, MG, PK, PS, WS, CEE, AB and JL. WR, JB, AD, BZ, CW, PS and NM contributed to data analysis. The first draft of the manuscript was written by WR, NM, MS und SGM. All the authors commented on previous versions of the manuscript. All the authors read and approved the final manuscript. Funding Open Access funding enabled and organized by Projekt DEAL. This project was supported by the German Federal Ministry of Science and Education (Comprehensive, Orchestrated, National Net- work to Explain, Categorize and Treat autoimmune encephalitis and Journal of Neurology (2024) 271:3653–3659 3659 References emission tomographic imaging of translocator protein. JAMA Neurol 72:882. https://doi.org/10.1001/jamaneurol.2015.0941 9. Bien CG, Vincent A, Barnett MH et al (2012) Immunopathology of autoantibody-associated encephalitides: clues for pathogenesis. Brain 135:1622–1638. https://doi.org/10.1093/brain/aws082 1. Graus F, Titulaer MJ, Balu R et al (2016) A clinical approach to diagnosis of autoimmune encephalitis. Lancet Neurol 15:391–404. https://doi.org/10.1016/S1474-4422(15)00401-9 1. Graus F, Titulaer MJ, Balu R et al (2016) A clinical approach to diagnosis of autoimmune encephalitis. Lancet Neurol 15:391–404. https://doi.org/10.1016/S1474-4422(15)00401-9 10. Pellerin K, Rubino SJ, Burns JC et al (2021) MOG autoantibodies trigger a tightly-controlled FcR and BTK-driven microglia prolif- erative response. Brain 144:2361–2374. https://doi.org/10.1093/ brain/awab231 p g 2. Lancaster E, Huijbers MGM, Bar V et al (2011) Investigations of caspr2, an autoantigen of encephalitis and neuromyotonia. Ann Neurol 69:303–311. https://doi.org/10.1002/ana.22297 p g 3. Lai M, Huijbers MGM, Lancaster E et al (2010) Investigation of LGI1 as the antigen in limbic encephalitis previously attributed to potassium channels: a case series. Lancet Neurol 9:776–785. https://doi.org/10.1016/S1474-4422(10)70137-X 11. Gallus M, Roll W, Dik A, Barca C (2023) Translational imag- ing of TSPO reveals pronounced innate inflammation in human and murine CD8 T cell-mediated limbic encephalitis. Sci Adv 9:eabq7595. https://doi.org/10.1126/sciadv.abq7595 p g 4. Mueller C, Langenbruch LM, Rau JMH et al (2021) Determinants of cognition in autoimmune limbic encephalitis—a retrospective cohort study. Hippocampus 31:1092–1103. https://doi.org/10. 1002/hipo.23375 12. Höftberger R, Fink S, Aboul-Enein F et al (2010) Tubulin polym- erization promoting protein (TPPP/p25) as a marker for oligoden- droglial changes in multiple sclerosis. Glia 58:1847–1857. https:// doi.org/10.1002/glia.21054 5. Bien CG, Mirzadjanova Z, Baumgartner C et al (2017) Anti-con- tactin-associated protein-2 encephalitis: relevance of antibody titres, presentation and outcome. Eur J Neurol 24:175–186. https:// doi.org/10.1111/ene.13180 5. Bien CG, Mirzadjanova Z, Baumgartner C et al (2017) Anti-con- tactin-associated protein-2 encephalitis: relevance of antibody titres, presentation and outcome. Eur J Neurol 24:175–186. https:// doi.org/10.1111/ene.13180 13. Kuehn JC, Scheuerle A, Bauer J et al (2020) A 64-year-old patient with a mesiotemporal mass and symptomatic epilepsy. Brain Pathol 30:413–414. https://doi.org/10.1111/bpa.12818 14. Körtvelyessy P, Bauer J, Stoppel CM et al (2015) Complement- associated neuronal loss in a patient with CASPR2 antibody-asso- ciated encephalitis. Neurology(R) Neuroimmunol Neuroinflamm 2:e75. https://doi.org/10.1212/NXI.0000000000000075 6. Zinnhardt B, Müther M, Roll W et al (2020) TSPO imaging- guided characterization of the immunosuppressive myeloid tumor microenvironment in patients with malignant glioma. Neuro Oncol 22:1030–1043. https://doi.org/10.1093/neuonc/noaa023 p g 15. References Kornau H-C, Kreye J, Stumpf A et al (2020) Human cerebrospinal fluid monoclonal LGI1 autoantibodies increase neuronal excitabil- ity. Ann Neurol 87:405–418. https://doi.org/10.1002/ana.25666 g 15. Kornau H-C, Kreye J, Stumpf A et al (2020) Human cerebrospinal fluid monoclonal LGI1 autoantibodies increase neuronal excitabil- ity. Ann Neurol 87:405–418. https://doi.org/10.1002/ana.25666 p g 7. Backhaus P, Roll W, Beuker C et al (2020) Initial experience with [18F]DPA-714 TSPO-PET to image inflammation in primary angiitis of the central nervous system. Eur J Nucl Med Mol Imag- ing 47:2131–2141. https://doi.org/10.1007/s00259-019-04662-4 g p g 8. Gershen LD, Zanotti-Fregonara P, Dustin IH et al (2015) Neuro- inflammation in temporal lobe epilepsy measured using positron 8. Gershen LD, Zanotti-Fregonara P, Dustin IH et al (2015) Neuro- inflammation in temporal lobe epilepsy measured using positron
https://openalex.org/W2605507429	https://apb.tbzmed.ac.ir/PDF/APB_14593_20161228211124	English	null	Quality and Oxidative Properties of Sesame and Olive Oils Incorporated with Flaxseed Oil	Advanced pharmaceutical bulletin	2,017	cc-by	3,981	Quality Properties of Sesame and Olive Oils Incorporated with Flaxseed Oil Fataneh Hashempour-Baltork1, Mohammadali Torbati 1, Sodeif Azadmard-Damirchi2, Geoffrey Peter Savage3 1 Department of Food Science and Technology, Faculty of Nutrition, Tabriz University of Medical Sciences, Tabriz, Iran. 2 Department of Food Science and Technology, Faculty of Agriculture, University of Tabriz, Tabriz, Iran. 3 Food Group, Department of Wine, Food and Molecular Biosciences, Lincoln University, Canterbury, New Zealand. Article History: Received: 28 December 2016 Revised: 29 January 2017 Accepted: 30 January 2017 ePublished: 13 April 2017 Abstract Purpose: Suitable ratio of essential fatty acids has an important role in maintaining good health. There is no pure oil with an ideal fatty acid composition and oxidative stability. The main goal of the present study was to evaluate the physical, chemical and nutritional properties of oil obtained by blending flaxseed oil as a rich source of ω3 fatty acids with sesame and olive oils. Methods: Three different ratios (65:30:5, 60:30:10 and 55:30:15) were prepared using olive, sesame and flaxseed oils. These mixtures were stored at 4°C and 24°C and their quality and physicochemical properties were determined by measuring the fatty acid composition, phenolic compound, peroxide, anisidine values and schaal tests. Keywords: Blending oil Essential fatty acid Flaxseed oil Nutrition Oxidation stability Results: Fatty acid composition indicated that adding 10% and 15% flaxseed oil into blends have suitable ratio of essential fatty acids. The sample which contained 5% flaxseed oil had the highest phenolic content among treatments and these compounds showed a significant decrease during storage. A significant increase (p<0.05) was observed in peroxide values of all samples during storage. Increasing the flaxseed oil content in the blends, lead to an increase of the anisidine value. Conclusion: Blending sesame and olive oils with flaxseed oil produced oil blends with a good balance of essential fatty acids. Although peroxide and anisidine values increased during storage of the oil blends; the blends were of a good quality for home and industrial use. ©2016 The Authors. This is an Open Access article distributed under the terms of the Creative Commons Attribution (CC BY), which permits unrestricted use, distribution, and reproduction in any medium, as long as the original authors and source are cited. No permission is required from the authors or the publishers. Advanced Ph Adv Pharm Bull, 2017, 7(1), 97-101 doi: 10.15171/apb.2017.012 http://apb.tbzmed.ac.ir Corresponding author: Mohammadali Torbati , Fax: +98 41 33379529, Email: torbatima@yahoo.com Research Article Research Article Advanced Pharmaceutical Bulletin Pharmaceutical Bulletin Introduction Oils and fats have important roles in cooking, frying and salad dressings or in food formulations; they make an important contribution to our diet and our health. Omega-3 (ω3) fatty acids such as linolenic acid, and omega-6 (ω6) fatty acids such as linoleic acid, are essential fatty acids; thus, they must be supplied through foods. Today, most of the oils consumed in our daily diet such as corn, sunflower, rice bran and grape seed oils have a high amount of ω6fatty acids and therefore, lead to ω6:ω3 ratio increases ranging from 8-12:1; however, this ratio should be 4:1.1 such as their low amount of ω3 essential fatty acids. Instead, flaxseed oil is a rich source of ω3, tocopherols and other bioactive compounds,2 but in pure form it is very unstable and oxidizes quickly. Sesame oil is a good source of ω6 fatty acids and has a considerable level of sesamin and sesamolin lignans, which have different bioactive and health promoting effects.3 Furthermore, sesame oil has anti-inflammatory activity and antiproliferative effects on cancer cells caused by tocopherol homologues.4,5 Sesame oil in spite of containing 85% unsaturated fatty acids, is one of the most stable vegetable oils to oxidation.6 However, sesame oil, which has positive nutritious and healthy effects, is low in ω3 fatty acids and because of its high price has limited application in the food industry. ω3 and ω6 fatty acids are essential for normal growth and have an important role in the prevention of cancer and cardiovascular diseases as well as the improvement of immune function. The ω6 eicosanoids without ω3 fatty acids are pro-inflammatory and may lead to high blood pressure, cardiovascular diseases and arthritis.1 Therefore, their balance is very important in our daily intake. Olive oil is a good source of ω9 and bioactive compounds such as phenolic compounds and phytosterols7,8 and because of its fatty acid composition and natural antioxidants such as tocopherols and polyphenols, it is more stable to heat treatments.9 Different fats/oils have various chemical and physical properties. Pure vegetable oils can have low functional characterization or nutritional properties. For example, using olive oil or sesame oil alone has some drawbacks Consuming edible oils with a suitable ratio of essential fatty acids, appropriate stability to heat treatment and storage is a very important issue in the food industry. Articleinfo Hashempour-Baltork et al. Materials and Methods Materials Fatty acid composition of edible oils and fats is important from the technological and nutritional points of view. Therefore, one of the main analyses in oils and fats field is fatty acid composition. Results show that fatty acid compositions of flaxseed, sesame and olive oils used in this study are in agreement with previously published data.16 Sesame and olive oils had very low amounts of linolenic acid, but flaxseed was a rich source of linolenic acid (Table 1). The flaxseed and sesame oils were obtained from seeds using a cold press (Screw Press Model 85 mm) and olive oil was purchased from local market (Tabriz, Iran). All chemicals used in this study were of analytical grade and purchased from Sigma Chemical Co (St. Louis, Mo, USA). Total Phenolic Compound Total phenolic content (TPC) was determined using the Folin–Ciocalteau reagent and caffeic acid for calibration curve by spectrophotometer (CECIL, Aquaris 1100, England) at 725 nm according to Capannesi et al.13 TPC was determined by calibration curve which was achieved from defined concentration of caffeic acid absorbance, and the final results were reported as mg caffeic acid/kg of oil. A suitable intake of the essential fatty acids is very important in daily diet. According to the literature, the optimal ratio of ω6:ω3 is reported to range from 1:1 to 4:1.1 The ratio of ω6:ω3 in olive, sesame and flaxseed oils were 9.5:1, 50.1:1 and 0.22:1, respectively (Table 1). This result shows that olive and sesame oils do not have optimal ω6:ω3 ratios. Therefore, consumers should have other ω3 rich oils in their diet for long term usage. Based on the high content of linolenic acid in flaxseed oil, which is about 50 times higher than olive and sesame oils (Table 1), addition of flaxseed oil to sesame and olive oil mixtures caused an effective decrease in ω6:ω3 ratio (Table 2), which is very important from nutritional prospective. Therefore, blending flaxseed oil with other vegetable oils which are low in ω3 fatty acid can change their fatty acid composition and increase ω3 fatty acid content. Samples containing 10% and 15% flaxseed have optimal levels of ω6:ω3 ratio as stated in the literature for better health achievement (Table 2).1 98 \| Advanced Pharmaceutical Bulletin, 2017, 7(1), 97-101 Methods Table 1. Fatty acid composition (%) of pure olive, sesame and flaxseed oils Fatty Acid Composition y p Fatty acid compositions were measured as fatty acid methyl esters by gas chromatograph (GC-1000, DANI, Italy) according to the method described by Savage et al.12 Blending of sesame and olive oils with flaxseed oil leads to significant changes (p<0.05) in the fatty acid composition (Table 2). Linolenic acid was increased by increasing flaxseed oil in the blend. Total Phenolic Compound Blending Process Three oils were prepared using olive: sesame: flaxseed in ratios of 65:30:5, 60:30:10 and 55:30:15 in triplicate. The oil mixtures were stored at 4 and 24°C for 90 days. Experiments were carried out on the production day and every 30 days. The fatty acid profile was determined on the first and 90th days. Fatty acid Flaxseed oil Olive oil Sesame oil C16:0 6.7 13.3 10.7 C18:0 2.5 4 6.5 C18:1 20.3 69.1 41.8 C18:2 12.9 11.4 40.1 C18:3 57.1 1.2 0.8 ω6:ω3 0.22:1 9.5:1 50.1:1 Fatty acid Flaxseed oil Olive oil Sesame oil Stability Tests Peroxide value (PV) was determined using the AOAC methods.14 P-anisidine value (AnV) of the oils was determined by dissolving the oil samples in iso-octane, and the absorbance was measured at 350 nm after 10 minutes by spectrophotometer according to ISO 6885.15 Schaal test Introduction in the oil samples, and peroxide value was determined in triplicate after each 24 hours for 10 days. Unfortunately, no pure oil has both an ideal fatty acid composition and good oxidative stability. Blending is the simplest physical and economical procedure to change fatty acid composition, increase the bioactive components and natural antioxidants and make a new at an affordable price.10,11 Statistical Analysis All parameters were measured in triplicate. Data obtained were analyzed by ANOVA using the SPSS statistical software (Chicago, IL, USA) in factorial experiments in completely randomized design, and the results were reported as mean ± standard deviation (SD) of three amounts. Duncan’s multiple range post hoc test was used to analyze significant differences at the 0.05 level. Flaxseed and products fortified with flaxseed powder and oil is gaining more attention because of its valuable nutritional properties.12 In the present study, flaxseed oil was blended with sesame and olive oils in different ratios to provide optimal essential fatty acids with a suitable stability and high bioactive content. These mixtures were stored at different temperatures to study the feasibility of introducing these functional vegetable oils to the market. Total phenolic compound The amount of phenolic compounds is an important quality factor in oils because of their role in oxidation stability, nutritional and organoleptic qualities.18 Olive oil is distinguished from the other vegetable oils by having a high content of phenolic compounds.19 Sample 65:30:5 had the highest TPC among the treatments, which is related to the higher amount of olive oil in the blend. Vegetable oils generally are kept at two different temperatures; in refrigerator (4°C) or in ambient conditions (24-25°C). Storage of these oil mixtures for up to 90 days at either of these temperatures caused significant changes on fatty acid composition. TPC of the samples showed a significant (p< 0.05) slow decrease during the storage until the 60th day (Table 3). This decrease was caused by oxidation and hydrolytic activities which occur because of the effects of temperature, oxygen and enzymes during storage.20 On the 90th day TPC increased significantly (p< 0.05) in all samples, which may come from the breakdown of the complex phenolic compounds in the oils to simple phenols. These simple phenolic compounds give a more intense colour with the Folin Ciocalteu reagent. Addition of flaxseed oil, which is known as highly- unsaturated and unstable oil but with high nutritional quality, to the mixture of olive and sesame oils, had a positive nutritional effect on the fatty acid profile by increasing of ω3 fatty acid ratio. Mixtures of oils with flaxseed oil had excellent oxidation stability during 90- days of storage. Table 3. Total phenolic compounds (mg caffeic acid/kg of oil) of the three oil blends during 90 days storage at different temperatures. Day 1 30 60 90 Sample* 4°C 25°C 4°C 25°C 4°C 25°C 4°C 25°C 65:30:5 1652 ± 12.5 e 1652 ± 12.58 e 1611 ± 8.0 f 1448 ± 12.5 h 1568 ± 12.5 g 1423 ± 7.02 hi 1915 ± 10.5 c 2093 ± 60.2 a 65:30:5 1402 ± 7.63 i 1402 ± 7.63 i 1351 ± 9.0 j 1201 ± 9.53 k 1318 ± 7.63 j 1098 ± 12.5 m 1893 ± 11.5 c 2006 ± 10.0 b 65:30:5 1210 ± 10.0 k 1210 ± 10.3 k 1160 ± 8.5 l 1048 ± 7.21 n 1043 ± 15.2 n 983 ± 15.2 o 1798 ± 12.5 d 1983 ± 15.2 b All values are the mean of three replicates ± standard deviation of the mean. Different letters represent significant differences (p<0.05). Total phenolic compound For treatments see Table 2. Schaal test The oil samples were placed in a series of 20 ml clear glass bottles incubated in a forced-draft air oven set at 65℃ for 10 days. The oxidation reaction was accelerated 98 \| Advanced Pharmaceutical Bulletin, 2017, 7(1), 97-101 Sesame and olive oils incorporated with flaxseed oil Table 2. Fatty acid composition (%) of the three oil blends on the first and 90th day storage at different temperatures. Fatty acid 65:30:5 60:30:10 55:30:15 Day1 Day 90 Day1 Day 90 Day1 Day 90 25°C 4°C 25°C 25°C 4°C 25°C 25°C 4°C 25°C C16:0 12.1 ± 0.2f 12.2 ± 0.2f 12.5 ± 0.3e 11.60 ± 0.1g 13.9 ± 0.1d 14.0 ± 0.1c 10.6 ± 0.1 h 14.2 ± 0.2 b 14.4 ± 0.4a C18:0 4.0 ± 0.3h 6.2 ± 0.1f 6.50 ± 0.1e 14.1 ± 0.2g 8.2 ± 0.2d 8.4 ± 0.4c 3.8 ± 0.3i 8.9 ± 0.2b 9.0 ± 0.5a C18:1 56.8 ± 0.2b 59.5 ± 0.1a 59.7 ± 0.1a 52.2 ± 0.2f 55.5 ± 0.5d 55.8 ± 0.1c 49.9 ± 0.4g 52.7 ± 0.1f 53 ± 0.1e C18:2 22.2 ± 0.2b 17.5 ± 0.1d 17.8 ± 0.1c 23.9 ± 0.4a 16.2 ± 0.2f 16.4 ± 0.1f 23.9 ± 0.5a 17.0 ± 0.1e 17.4 ± 0.1d C18:3 4.3 ± 0.3f 2.8 ± 0.1i 3.0 ± 0.1h 8.2 ± 0.2b 4.2 ± 0.1g 4.5 ± 0.2e 11.4 ± 0.2a 6.0 ± 0.1d 6.2 ± 0.1c ω6:ω3 5.2:1 c 6.2:1a 5.9:1b 2.9:1f 3.8:1d 3.6:1e 2.1:1h 2.8:1fg 2.8:1g Treatments indicated as olive: sesame: flaxseed, respectively. Different letters represent significant differences (p<0.05). All values are the mean of three replicates ± standard deviation of the mean. Fatty acid composition of oils can change during storage. Results shown in Table 2 confirm that there was a significant change (p< 0.05) during 90 days storage of oil samples. There was a significant decrease (p< 0.05) in 18:2 and 18:3. However, this decrease could not increase the ω6:ω3 ratio beyond optimal levels. It has been previously reported that changes in fatty acid composition of olive oil occurs during storage.17 Total phenolic compound \| 99 Advanced Pharmaceutical Bulletin, 2017, 7(1), 97-101 Stability parameters Day 1 30 60 90 Sample 4°C 25°C 4°C 25°C 4°C 25°C 4°C 25°C 65:30:5 5.23 ± 0.25 k 5.23 ± 0.25 k 5.93 ± 0.20 j 6.77 ± 0.25 hi 6.40 ± 0.17 ij 9.47 ± 0.15 d 6.83 ± 0.40 hi 10.97 ± 0.15 c 65:30:5 4.83 ± 0.30 kl 4.83 ± 0.30 kl 6.10 ± 0.26 j 8.90 ± 0.20 e 7.00 ± 0.26 h 9.83 ± 0.15 d 8.10 ± 0.26 f 13.63 ± 0.15 b 65:30:5 4.63 ± 0.30 l 4.63 ± 0.30 l 6.77 ± 0.20 hi 9.40 ± 0.26 d 7.57 ± 0.20 g 11.10 ± 0.36 c 9.57 ± 0.20 d 16.00 ± 0.36 a All values are the mean of three replicates ± standard deviation of the mean. Different letters represent significant differences (p<0.05). For treatments see Table 2. Lipid oxidation involves the formation of hydroperoxides as primary oxidation products which are unstable and may break down to a variety of volatile and nonvolatile compounds as secondary products. Secondary products are responsible for off flavors and some of them can be toxic. Secondary products are determined by measuring the p-anisidine content.22 Generally, there were no significant increases in AnV until 30 days of storage, but after that it increased in all samples (Table 5). Increase in AnV of oils stored at room temperature was greater than that in the oils kept at refrigerator. Samples containing 15% of flaxseed oil had higher AnV, but it was still lower than for a standard range for common with high stability edible oils.23 Table 5. Anisidine values of oil blends during 90 days storage at different temperatures. Day 1 30 60 90 Sample 4°C 25°C 4°C 25°C 4°C 25°C 4°C 25°C 65:30:5 3.00 ± 0.05 i 3.00 ± 0.05 i 3.00 ± 0.3 i 3.00 ± 0.02 i 3.00 ± 0.02 i 3.50 ± 0.04 g 3.38 ± 0.04 g 3.86 ± 0.05 ef 65:30:5 2.00 ± 0.02 k 2.00 ± 0.02 k 3.00 ± 0.04 i 2.99 ± 0.08 i 3.10 ± 0.01 hi 3.69 ± 0.08 f 4.01 ± 0.06 de 4.25 ± 0.05 c 65:30:5 3.00 ± 0.02 i 3.01 ± 0.02 i 3.00 ± 0.03 i 3.00 ± 0.02 i 3.21 ± 0.03 h 4.03 ± 0.07 d 4.68 ± 0.07 b 6.62 ± 0.04 a All values are the mean of three replicates ± standard deviation of the mean. Stability parameters higher PV than the samples stored in a refrigerator. Oxidation is retarded at low temperature. Peroxides generated by the oxidation of fatty acids can affect the quality of oils and also food containing them. Also, oil blends with a higher amount of flaxseed oil showed a greater increase in PV (Table 4) which comes from the higher amounts of fatty acids in this oil. q y g Oxidation can be promoted by proxidant metals such as iron and copper, temperature, light, sensitizers such as chlorophyll. Olive oil can have very high PV among oils used in this study. Commercially available virgin olive oils can have PV values of up to 15 meqO2/kg oil.21 However, flaxseed oil can also oxidize very fast and can have very high PV because they contain high amounts of polyunsaturated fatty acids. The sample containing 5% flaxseed oil had the highest PV at the early stages of storage but during storage, it was more stable and oxidized less (Table 4). This is related to the high amount of olive oil and low amount of flaxseed oil in the first treatment. Olive oil had high PV but it is stable and does not oxidize as fast as flaxseed oil. In this study, all samples showed a significant increase (p<0.05) in PV, after 3 months storage (Table 4). As expected, oil samples kept at room temperature had Temperature had a large effect on the PV of the samples. Oxidation at low temperature was almost two \| 99 Advanced Pharmaceutical Bulletin, 2017, 7(1), 97-101 Hashempour-Baltork et al. observed that blends containing high amounts of flaxseed oil could be stored at low temperatures for a long period of time. times less than that at high temperature (Table 4). Only the 55:30:15 ratio mixture had a higher PV (16 meqO2/kg) after storage at room temperature. It was Table 4. Peroxide values (meq O2/kg oil) of the three oil blends during 90 days storage at different temperatures. 100 \| Advanced Pharmaceutical Bulletin, 2017, 7(1), 97-101 References 1. Simopoulos AP. Omega-6/Omega-3 essential fatty acid ratio and chronic diseases. Food Rev Int 2004;20(1):77-90. doi: 10.1081/FRI-120028831 Figure 1. Peroxide value (meq O2/kg) of pure olive, sesame and flaxseed oil and their mixtures following an oven test (65°c). For treatments see Table 2. 2. Obranovic M, Skevin D, Kraljic K, Pospisil M, Nederal S, Blekic M, et al. Influence of climate, variety and production process on tocopherols, plastochromanol-8 and pigments in flaxseed oil. Food Technol Biotechnol 2015;53(4):496-504. doi: 10.17113/ftb.53.04.15.4252 Conflict of Interest Conflict of Interest The authors have no conflicts of interest to declare. Stability parameters Different letters represent significant differences (p<0.05). *For treatments see Table 2. Ethical Issues Not applicable. Conflict of Interest The authors have no conflicts of interest to declare. Schaal test nutritional quality of final product. Oil blends with 10 and 15% of flaxseed oil showed the optimal ratio of essential fatty acids but considering all quality parameters, oil blends with 10% flaxseed oil had the highest oxidation stability. This study illustrates that using flaxseed oil in vegetable oil blends can yield an effective level of bioactive compounds, balanced ω6:ω3 ratio and suitable stability. PV results of the oil blends stored at 65°C for 10 days showed that olive oil is more stable than sesame and flaxseed oils. Flaxseed oil had the lowest stability which has also been previously reported.24 Oil blends had a significant differences in PV obtained using the Schaal test (Figure 1). As the amount of flaxseed oil was increased in the oil blends the oxidation stability of the oil blends decreased. Figure 1. Peroxide value (meq O2/kg) of pure olive, sesame and flaxseed oil and their mixtures following an oven test (65°c). For treatments see Table 2. Acknowledgments This research was financially supported by Tabriz University of Medical Sciences. Ethical Issues Not applicable. Ethical Issues Conclusion Incorporating sesame and olive oils with flaxseed oil led to improvement in oxidative stability parameters during storage and also had positive effect on 100 \| Advanced Pharmaceutical Bulletin, 2017, 7(1), 97-101 Sesame and olive oils incorporated with flaxseed oil polyphenols detection in olive oils. Food Chem 2000;71(4):553-62.doi:10.1016/s0308-8146(00) 00211-9 3. Moazzami, AA, Stefanie L, Kamal-Eldin H. Lignan contents in sesame seeds and products. Eur J Lipid Sci Technol 2007; 109(10):1022–7. doi: 10.1002/ejlt.200700057 14. AOAC. Official methods of analysis. Oils & fats. H. William (Ed.). Virginia: Association of Official Analytical Chemists; 2000. 4. Rangkadilok N, Pholphana N, Mahidol C. Variation of sesamin, sesamolin and tocopherols in sesame (Sesamum indicum L.) seeds and oil products in Thailand. Food Chem 2010;122(3):724-30. doi: 10.1016/j.foodchem.2010.03.044 15. ISO 6885, Animal vegetables fats and oils – Determination of anisidine value. https://law.resource.org/pub/et/ibr/et.iso.6885.2012. pdf 5. Williamson KS, Morris JB, Pye QN, Kamat CD, Hensley K. A survey of sesamin and composition of tocopherol variability from seeds of eleven diverse sesame (sesamum indicum l.) genotypes using hplc- pad-ecd. Phytochem Anal 2008;19(4):311-22. doi: 10.1002/pca.1050 16. Codex Alimentarius Commission, Codex Stan 19. Edible Fats and Oils Not Covered by Individual Standards, 1999. http://codexalimentarius.net. 17. Gomez-Alonso S, Mancebo-Campos V, Salvador MD. Evolution of major and minor components and oxidation indices of virgin olive oil during 21 months storage at room temperature.. Food Chem 2007;100(1):36-42. doi: 10.1016/j.foodchem.2005.09.006 6. Abou-Gharbia HA, Shehata AAY, Shahidi F. Effect of processing on oxidative stability and lipid classes of sesame oil. Food Res Int 2000;33(5):331-40. doi: 10.1016/S0963-9969(00)00052-1 7. Azadmard-Damirchi S. Review of the use of phytosterols as a detection tool for adulteration of olive oil with hazelnut oil. Food Addit Contam Part A Chem Anal Control Expo Risk Assess 2010;27(1):1-10. doi: 10.1080/02652030903225773 18. Nergiz C, Unal K. Determination of phenolic acids in virgin olive oil. Food Chem 1991; 39(2):237-40. doi:10.1016/0308-8146(91)90164-j. 19. Hohmann CD, Cramer H, Michalsen A, Kessler C, Steckhan N, Choi K, et al. Effects of high phenolic olive oil on cardiovascular risk factors: A systematic review and meta-analysis. Phytomedicine 2015;22(6):631-40. doi: 10.1016/j.phymed.2015.03.019 8. Morello JR, Motilva MJ, Tovar MJ. Changes in commercial virgin olive oil (cv Arbequina) during storage, with special emphasis on the phenolic fraction. Food Chem 2004;85(3):357-64. doi: 10.1016/j.foodchem.2003.07.012 20. Cinquanta L, Esti M, Lanotte E. Evaluation of phenolic compounds in virgin olive oil during storage. J Am Oil Chem Soc 1997;74(10):1259-64. doi: 10.1007/s11746-997-0054-8 9. Ayton J, Rodney J, Graham M. \| 101 Advanced Pharmaceutical Bulletin, 2017, 7(1), 97-101 Conclusion The effect of storage conditions on extra virgin olive oil quality. Australia: RIRDC; 2012. 10. Obrien RD. Fats and oils: Formulating and processing for application. 2nd ed. Florida: CRC Press; 2004. 21. Codex Alimentarius Commission. Codex Stan 33. Standard for olive oils and olive pomace oils codex standard 33. 1981. http://.codexalimentarius.org (Revision: 2015). 11. Hashempour-Baltork F, Torbati M, Azadmard- Damirchi S, Savage GP.Vegetable oil blending: A review of physicochemical, nutritional and health effects. Trends in Food Sci Technol 2016; 57: 52- 58. doi: 10.1016/j.tifs.2016.09.007 22. Kamal-Eldin A. Lipid oxidation pathways. Champaign: AOCS Press; 2003. 23. Gupta MK. Frying oils. In: Shahidi F, editor. Bailey's industrial oil and fat products. 6th ed. New Jersey: John Wiley & Sons; 2005. 12. Savage GP, Mcneil DL, Dutta PC. Lipid composition and oxidative stability of oils in hazelnuts (Corylus avellana L.) Grown in New Zealand. J Am Oil Chem Soc 1997;74(6):755-9. doi: 10.1007/s11746-997-0214-x 24. Lukaszewicz M, Szopa J, Krasowska A. Susceptibility of lipids from different flax cultivars to peroxidation and its lowering by added antioxidants. Food Chem 2004;88(2):225-31. doi:10.1016/j.foodchem.2003.12.042 13. Capannesi C, Palchetti I, Mascini M, Parenti A. Electrochemical sensor and biosensor for
https://openalex.org/W4283776600	https://hal.science/hal-03716457/document	English	null	Intensive care unit-to-unit capacity transfers are associated with increased mortality: no hasty conclusions in the event of a crisis	Annals of intensive care	2,022	cc-by	1,555	Intensive care unit-to-unit capacity transfers are associated with increased mortality: no hasty conclusions in the event of a crisis Benoit Painvin, Stephan Ehrmann, Arnaud W. Thille, Jean-Marc Tadie To cite this version: Benoit Painvin, Stephan Ehrmann, Arnaud W. Thille, Jean-Marc Tadie. Intensive care unit-to-unit capacity transfers are associated with increased mortality: no hasty conclusions in the event of a crisis. Annals of Intensive Care, 2022, 12 (1), pp.60. 10.1186/s13613-022-01031-7. hal-03716457 Distributed under a Creative Commons Attribution 4.0 International License Dear Editor, shock. In light of these results, one could wonder whether it is safe or not for the patient to undergo a night ICU-to- ICU transfer compared to withholding the interhospital transport for a few hours until the sun rises, since night- shift patient’s discharge has been associated with a higher mortality [4]. Dear Editor, We read with great interest the study of Parenmark et al., a large retrospective study, including 15,588 ICU-to-ICU interhospital transfers in Sweden over a 2-year period [1]. The authors describe three types of interhospital trans- fers: clinical transfer (need for specialised care not avail- able in the admitting hospital), capacity transfer (making room for patients with more urgent need for intensive care when all ICU beds are occupied) and repatriation (return to the home ICU following initial treatment at another unit) the last one being labelled as reference. Their main result indicates an increase mortality within 30 days following discharge from the referring ICU in the subgroups of clinical and capacity transfers, with adjusted odds ratio of 1.17 (95% CI 1.02–1.36) and 1.25 (95% CI 1.06–1.49), respectively. Second, the authors pointed out that Sweden has a low number of ICU bed which could play a role in the higher mortality rate found following interhospital capacity and clinical transfers [5]. Solutions to tackle this higher mortality related to interhospital transfer would be to build up local resources for critical care: increasing ICU beds, recruiting ICU highly trained staff and Intensivist doctors to avoid trans- fers of critically ill patients at nights with severe unstable pathologies (especially during wintertime when respira- tory sepsis and acute respiratory distress occur more fre- quently [6]). As the authors notice, the main result is somewhat surprising as higher mortality has not been reported in recent literature [2, 3]. Reasons could be explained as follow: first, the authors specify that 20% of capacity transfers occurred at night, involving severe critically ill patients with acute lung injury, sepsis, and cardiogenic Furthermore, the authors’ message must be balanced when facing crisis, such as the COVID-19 pandemic. Assuming that interhospital transfers are unsafe and choosing a strategy of implementation of new ICU beds to face surge of critically ill patients could lead to a higher mortality [7]. © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. HAL Id: hal-03716457 https://hal.science/hal-03716457v1 Submitted on 7 Jul 2022 L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. Distributed under a Creative Commons Attribution 4.0 International License Painvin et al. Annals of Intensive Care (2022) 12:60 https://doi.org/10.1186/s13613-022-01031-7 Open Access Dear Editor, During the first months of the COVID-19 crisis, countries planned and organized large-scale inter- hospital transfers either for clinical or capacity reasons and demonstrated that transferred patients did not have a higher mortality rate [2, 3, 8]. However, we agree with the authors and acknowledge that “understanding the impact of ICU-to-ICU transfer on patient outcome is complex and must consider a couple of important aspects” such as identifying appropriate control patients. This comment refers to the article available online at https://doi.org/10.1186/ s13613-022-01003-x. Correspondence: benoit.painvin@chu-rennes.fr; jean-marc.tadie@chu- rennes.fr 1 Service des Maladies Infectieuses et de Réanimation Médicale, Centre Hospitalier Universitaire de Rennes, Hôpital Pontchaillou, 2 rue Henri le Guilloux, 35033 Rennes, France Full list of author information is available at the end of the article Correspondence: benoit.painvin@chu-rennes.fr; jean-marc.tadie@chu- rennes.fr Funding Funding This study did not receive any funding. Funding This study did not receive any funding. Availability of data and materials Not applicable. Availability of data and materials Not applicable. Author details 1 1 Service des Maladies Infectieuses et de Réanimation Médicale, Centre Hospitalier Universitaire de Rennes, Hôpital Pontchaillou, 2 rue Henri le Guil‑ loux, 35033 Rennes, France. 2 Service de Médecine Intensive Et Réanimation, CRICS- Triggersep Research Network, Centre Hospitalier Régional Universi‑ taire de Tours, CIC INSERM 1415Hôpital Bretonneau 2, boulevard Tonnellé, 27044 Tours, France. 3 Centre d’étude des pathologies respiratoires, INSERM U1100, Université de Tours, Tours, France. 4 Service de Médecine Intensive Et Réanimation, Centre Hospitalier Universitaire de Poitiers, 2 rue de la Milétrie, 90577 86000 Poitiers, France. 5 Faculté de Médecine, Université de Rennes 1, Unité INSERM CIC 1414, IFR 140, Rennes, France. Received: 4 May 2022 Accepted: 20 May 2022 Received: 4 May 2022 Accepted: 20 May 2022 Abbreviations 8. Painvin B, Messet H, Rodriguez M, Lebouvier T, Chatellier D, Soulat L, et al. Inter-hospital transport of critically ill patients to manage the intensive care unit surge during the COVID-19 pandemic in France. Ann Intensive Care. 2021;11:54. Acknowledgements None. Acknowledgements None. Declarations Ethics approval and consent to participate Not applicable. Publisher’s Note BP and JMT conceived the letter; BP wrote the manuscript and SE, AWT and JMT revised it. The authors read and approved the final manuscript. Springer Nature remains neutral with regard to jurisdictional claims in pub‑ lished maps and institutional affiliations. Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Page 2 of 2 Page 2 of 2 Painvin et al. Annals of Intensive Care (2022) 12:60 Painvin et al. Annals of Intensive Care (2022) 12:60 Painvin et al. Annals of Intensive Care Finally, as mentioned by the authors, robust prospec- tive studies including before departure, ongoing trans- port and arrival data are needed to determine the timing of the transfer, the safest medical condition allowing for transfer, and whether transport impacts ICU mortality. 5. Bauer J, Brüggmann D, Klingelhöfer D, Maier W, Schwettmann L, Weiss DJ, et al. Access to intensive care in 14 european countries: a spatial analysis of intensive care need and capacity in the light of COVID-19. Intensive Care Med. 2020;46:2026–34. 5. Bauer J, Brüggmann D, Klingelhöfer D, Maier W, Schwettmann L, Weiss DJ, et al. Access to intensive care in 14 european countries: a spatial analysis of intensive care need and capacity in the light of COVID-19. Intensive Care Med. 2020;46:2026–34. 6. Danai PA, Sinha S, Moss M, Haber MJ, Martin GS. Seasonal variation in the epidemiology of sepsis. Crit Care Med. 2007;35:410–5. 7. Taccone FS, Vangoethem N, Depauw R, Wittebole X, Blot K, Vanoyen H, et al. The role of organizational characteristics on the outcome of COVID- 19 patients admitted to the ICU in Belgium. Lancet Reg Health Europe. 2020;2:100019. 7. Taccone FS, Vangoethem N, Depauw R, Wittebole X, Blot K, Vanoyen H, et al. The role of organizational characteristics on the outcome of COVID- 19 patients admitted to the ICU in Belgium. Lancet Reg Health Europe. 2020;2:100019. Competing interests h h d l h Competing interests The authors declare that they have no competing interests. References 1. Parenmark F, Walther SM. Intensive care unit to unit capacity transfers are associated with increased mortality: an observational cohort study on patient transfers in the swedish intensive care register. Ann Intensive Care. 2022;12:31. 1. Parenmark F, Walther SM. Intensive care unit to unit capacity transfers are associated with increased mortality: an observational cohort study on patient transfers in the swedish intensive care register. Ann Intensive Care. 2022;12:31. 2. Sanchez M-A, Vuagnat A, Grimaud O, Leray E, Philippe J-M, Lescure F-X, et al. Impact of ICU transfers on the mortality rate of patients with COVID-19: insights from comprehensive national database in France. Ann Intensive Care. 2021;11:151. 3. Chen E, Longcoy J, McGowan SK, Lange-Maia BS, Avery EF, Lynch EB, et al. interhospital transfer outcomes for critically Ill patients with coronavirus disease 2019 requiring mechanical ventilation. Crit Care Explor. 2021;3: e0559. 3. Chen E, Longcoy J, McGowan SK, Lange-Maia BS, Avery EF, Lynch EB, et al. interhospital transfer outcomes for critically Ill patients with coronavirus disease 2019 requiring mechanical ventilation. Crit Care Explor. 2021;3: e0559. 4. Duke GJ, Green JV, Briedis JH. Night-shift discharge from intensive care unit increases the mortality-risk of ICU survivors. Anaesth Intensive Care. 2004;32:697–701.
https://openalex.org/W4388469508	https://www.rbciamb.com.br/Publicacoes_RBCIAMB/article/download/1634/915	English	null	Bibliometric review of electro-electronic waste (WEEE) in the Web of Science database: groups’ production and main themes	Revista Brasileira de Ciências Ambientais	2,023	cc-by	7,188	R E SUMO Os resíduos de equipamentos elétricos e eletrônicos (REEE) têm sido dos maiores e crescentes resíduos gerados em todo o mundo, tornando- se um dos grandes desafios da humanidade. O objetivo do artigo foi mapear a produção cientifica sobre REEE na última década (2012–2022), adotando como método de pesquisa uma análise bibliométrica com base no levantamento de documentos obtidos da base de dados Web of Science. O total de 278 artigos de pesquisa e revisão foi selecionado para análise utilizando o software Vosviewer e RStudio. Obteve-se, como resultado, um aumento significativo nos números de publicações na última década, com 86% dos artigos publicados entre 2015 e 2022. Além disso, foi possível obter o ranking dos autores mais importantes e revistas mais utilizadas para publicação dos artigos; constatou-se que o continente asiático, europeu e americano foram os que tiveram maior contribuição. Na análise de acoplamento de documentos, combinada com a de palavras-chaves, constataram-se as principais áreas pesquisadas atualmente em relação ao REEE: reciclagem de lixo eletrônico; impactos ambientais; sustentabilidade; economia circular; gestão eficiente lixo eletrônico e tecnologias para reciclagem lixo eletrônico. além disso, as palavras-chave “e-waste” e “polybrominated diphenyl ethers” foram as com maior frequência utilizadas pelos autores para representar a temática. Pode-se concluir que a temática tem se destacado ao longo dos últimos anos com diversas publicações, fornecendo implicações gerenciais e políticas para pesquisadores e profissionais. The waste of electrical and electronic equipment (WEEE) has been one of the largest and growing wastes generated in the world, turning into a great challenge for humanity. The objective of the article was to map the scientific production on WEEE in the last decade (2012– 2022), adopting a bibliometric analysis as a research method based on the survey of documents obtained from the Web of Science database. A total of 278 research and review articles were selected for analysis, with the use of Vosviewer and RStudio software. As a result, there was a significant increase in the number of publications in the last decade, with 86% of articles published between 2015 and 2022. In addition, it was possible to obtain the ranking of the most important authors, and the journals most used for publication of articles; it was found that the Asian, European and American continents had the greatest contribution. Revista Brasileira de Ciências Ambientais Brazilian Journal of Environmental Sciences ISSN 2176-9478 Volume 56, Number 1, March 2021 Revista Brasileira de Ciências Ambientais Brazilian Journal of Environmental Sciences Bibliometric review of electro-electronic waste (WEEE) in the Web of Science database: groups’ production and main themes Revisão bibliométrica de resíduos eletroeletrônicos (REEE) no banco de dados da Web of Science: produção de grupos e principais temas Bibliometric review of electro-electronic waste (WEEE) in the Web of Science database: groups’ production and main themes Revisão bibliométrica de resíduos eletroeletrônicos (REEE) no banco de dados da Web of Science: produção de grupos e principais temas Magdala Gelilarck Bizerra1 , Liliana Andrea Santos2 , Luiz Filipe Alves Cordeiro1 , Aldo Torre 1Instituto de Tecnologia de Pernambuco – Recife (PE), Brazil. 2Universidade Federal Rural de Pernambuco – Recife (PE), Brazil. 3Universidade Federal de Pernambuco – Recife (PE), Brazil. Correspondence address: Magdala Gelilarck Bizerra – Rua Dr. Enéas de Lucena, 265 – Apto. 3501 – Encruzilhada – CEP: 52041-090 – Recife (PE), Brazil. E-mail: magdalagb@hotmail.com Conflicts of interest: the authors declare no conflicts of interest. Methodologyh The data was analyzed through bibliometric indicators, which took into account both the quantitative and qualitative aspects of scientific production, and scientometric indicators, which were used to analyze the quantitative aspects of science. This allowed for the analysis of the scientific productions that were included in the database (Kücher and Feldbauer-Durstmüller, 2019). The bibliometric analysis for this study was carried out gathering infor- mation through the WoS database, and the following filters were applied: period (2012 to 2022), type of document (article review), category (envi- ronmental science) and language (English); on the subject of electronic waste, reverse logistics and environmental impacts. The platform was cho- sen because it provides access to the main citation databases in the world. Southeast Asia and Africa are now some of the largest recipients of e-waste in the world (Lepawsky, 2015). However, this data goes against sustainability; according to a report by the International Telecommu- nication Union, only 0.1% of e-waste is formally recycled in Africa, which is targeting a global collection and recycling rate of 30% by 2023 (Forti et al., 2020). The demarcation of the articles using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (Prisma) method was con- ducted through the systematization of four steps: identification, selec- tion, eligibility and inclusion (Moher et al., 2009). This method is widely used to analyze studies published in different areas of science, thus al- lowing a systematic and integrated analysis of published scientific data. Therefore, the increasing levels of electronic waste, as well as the in- appropriate and unsafe treatment and disposal, represent significant chal- lenges for the environment and human health. Their inadequate man- agement contributes significantly to global warming, which makes the intensification of scientific studies to find a better use for this waste nec- essary, as much as sustaining public policies to properly dispose of WEEE and mitigate the environmental impacts caused by incorrect disposal. A brief description of the steps followed for analysis can be seen in Figure 1. In the literature, several authors reported different reviews on electronic waste that support this research. Bibliometric review of electro-electronic waste (WEEE) in the Web of Science database: groups’ production and main themes Bibliometric review of electro-electronic waste (WEEE) in the Web of Science database: groups’ production and main themes Introductionh In the last 10 years, the research on WEEE has been increasing, as demonstrated in this study. Several peer-reviewed articles have been published in high-impact journals focusing on WEEE and its subtop- ics. However, there are few bibliometric studies aligned with this theme (Zhang et al., 2019). Therefore, the purpose of this work was to map the scientific production on WEEE using bibliometrics. The strong inclusion of electronic equipment in our lives and the speed at which such equipment become technologically outdated has led to an unprecedented increase in the waste of electrical and elec- tronic equipment (WEEE) (Perkins et al., 2014). Emerging technolo- gies (e.g., 5G technology and virtual reality) compound the acceler- ating rate of electronic obsolescence, generating new e-waste streams (Shittu et al., 2021). This article used the Web of Science (WoS) database to retrieve publication data in the form of ranking authors, countries, journals, citations, and search fields as important keywords. The electronics industry produces different types of WEEE, which contain toxic and hazardous substances that significantly threaten the environment and human health (Bressanelli et al., 2019). On the other hand, e-waste contains precious metals and raw mate- rials that, if recycled, could be useful for long-term economic and environmental sustainability. By mapping the domain of knowledge, this bibliometric analysis accounted for the main terms used in scientific production on the sub- ject, identifying research networks, research flow and relevant topics, becoming a basis for updated guidance for future research in this area. The results present a panoramic view of research directions and several questions derived from documents published on the subject. According to the 2020 global e-waste monitoring report (Forti et al., 2020), e-waste production in 2019 was around 53.6 million tons, of which 17.4% were duly collected and recycled, preventing the equiv- alent of up to 15 million tons of carbon dioxide from being released into the environment. The remaining 82.6% did not reach processing units for this type of waste, indicating they were inappropriately dis- posed of, mostly in the ecosystem. This data becomes more alarming when estimates indicate that global electronic waste should reach 74.7 million metric tons by 2030 (Forti et al., 2020), with 80% being gener- ated in developed countries in Europe and North America. R E SUMO In the analysis of document coupling, combined with that of keywords, the main areas connected to WEEE currently researched were found: electronic waste recycling; environmental impacts; sustainability; circular economy; efficient e-waste management and e-waste recycling technologies; in addition, the keywords “e-waste” and “polybrominated diphenyl ethers” were the most frequent words used by the authors to represent the theme. It can be concluded that the theme has stood out over the last few years, with several publications providing managerial and political implications for researchers and professionals. Palavras-chave: análise bibliométrica; cconomia circular; impacto ambiental; lixo eletrônico; sustentabilidade. Keywords: bibliometric analysis; circular economy; environmental Impact; electronic waste; sustainability. 1Instituto de Tecnologia de Pernambuco – Recife (PE), Brazil. 2Universidade Federal Rural de Pernambuco – Recife (PE), Brazil. 3Universidade Federal de Pernambuco – Recife (PE), Brazil. Correspondence address: Magdala Gelilarck Bizerra – Rua Dr. Enéas de Lucena, 265 – Apto. 3501 – Encruzilhada – CEP: 52041-090 – Recife (PE), Brazil. E-mail: magdalagb@hotmail.com Conflicts of interest: the authors declare no conflicts of interest. Funding: none. Received on: 05/22/2023. Accepted on: 09/12/2023. https://doi.org/10.5327/Z2176-94781634ii 342 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 1Instituto de Tecnologia de Pernambuco – Recife (PE), Brazil. 2Universidade Federal Rural de Pernambuco – Recife (PE), Brazil. 3Universidade Federal de Pernambuco – Recife (PE), Brazil. Correspondence address: Magdala Gelilarck Bizerra – Rua Dr. Enéas de Lucena, 265 – Apto. 3501 – Encruzilhada – CEP: 52041-090 – Recife (PE), Brazil. E-mail: magdalagb@hotmail.com Conflicts of interest: the authors declare no conflicts of interest. Funding: none. Received on: 05/22/2023. Accepted on: 09/12/2023. https://doi.org/10.5327/Z2176-94781634 This is an open access article distributed under the terms of the Creative Commons license. Received on: 05/22/2023. Accepted on: 09/12/2023. https://doi.org/10.5327/Z2176-94781634 This is an open access article distributed under the terms of the Creative Commons license. 342 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 Methodologyh Among them are studies on its management (Li et al., 2013; Pérez-Belis et al., 2015; Tsai et al., 2020; Shittu et al., 2021); reverse logistic (Govindan and Soleimani, 2017; Islam and Huda, 2018; Türkeli et al., 2018); the production of WEE (Ismail and Hanafiah, 2020); its recycling (Zhang and Xu, 2016); life cycle evaluation (Ismail and Hanafiah, 2019; Corrêa Nunes et al., 2021), emphasising the challenges and opportunities in a global con- text (Sharma et al., 2010; Ilankoon et al., 2018; Goodship et al., 2019; Gollakota et al., 2020); or a case study of one particular country (de Oliveira Neto et al., 2019). Figure 1 – Systematic flowchart search in the database. Source: adapted from Guedes et al. (2022). Figure 1 – Systematic flowchart search in the database. Source: adapted from Guedes et al. (2022). 343 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 Bizerra, M.G. et al. In the identification stage, a survey was applied to the Web of Science database — Main Collection (CLARIVATE ANALYTICS) platform through “all fields”, with the simultaneous application of the terms (“WEEE” OR “e-waste” OR “electronic waste” AND “re- verse logistic” AND “environmental impact”), in order to try to reach a greater scope. Then, in the selection stage, the initial search was carried out in April 2022, and the following filters were applied: period (2012 to 2022), type of document (article review); catego- ry (environmental science); language (English). Afterwards, in the eligibility stage, the titles and abstracts of the articles were read to confirm their relationship with the addressed theme. Finally, in the inclusion stage, the articles were read in full to leave only those that matched the focus of this study. There was an increase in the number of publications on WEEE and its aspects in the last decade (Figure 2). Within the evaluated pe- riod, most of the articles on the subject were published in the last sev- en years (2015–2022). Of the total sample, 36 articles were published before 2015, while 232 articles were published from 2015 onwards. The peak number of articles was published in 2022 (44 articles). The results establish an equivalence between this study and the ones conducted by Zhang et al. (2019), Singh et al. (2021) and Concari et al. (2022) in which this topic is considered to be of global interest. Main productive authors and citationsh The analysis accounted for a total of 1,024 authors, with a minimum of four published documents (Figure 3). The most outstanding authors in the theme (cluster 1 — red) were: Li jinhui, with 23 publications and 1,791 citations, followed by Zeng, Xianlai, with ten publications and 886 citations; Awasthi, Abhishek Kumar with seven publications; and Song, Quingbin with six publications; the primary authors of these clusters are from Tsinghua University in China. It must be highlighted that the members of Chinese universities together, with a minimum of five published documents, represent the majority of the total number of authors in the sample (80%). In the bibliographic coupling analysis of documents, the links in- dicate the number of shared cited references (Van Eck and Waltman, 2010), that is, when two documents cite the same document, showing the strength of a particular publication compared to other ones (Mu- let-Forteza et al., 2018; Cavalcante et al., 2021). The bibliometric analysis was conducted using the R software (Bibliometrix package). For this stage, the impact factor (IF) of to- tal citations, index H, and the number of articles published by the researcher were considered, which obtain citations greater than or equal to this number, following the methodology described by Almeida et al. (2018). Figure 2 – Annual distribution, citations and average citations per year of published articles on e-waste compiled in the Web of Science database (2012–2022), using RStudio. Finally, the result of the different bibliometric parameters data mapped the bibliometric information of scientific publications in the last decade on the proposed theme. Methodologyh There was a notable growth in the amount of research published from 2015 to 2022. The increase in publications in the sample period can be at- tributed to the international agreements and directives of the United Na- tions (UN) and public policies on WEEE (ONU, 1992, 2000, 2015; Brasil, 2010, 2020), to restrict the use of certain hazardous substances in WEEE, aiming to reduce its environmental impact at the end of its life cycle. As a result, 278 articles filled the selection criteria and were used as the final sample for analysis. Then, the data containing information about the articles were exported in .bib and .csv excel format to be sub- mitted to bibliometric analysis using the Vosviewer software (1.6.17 version) and RStudio. Overall, the average number of citations per document was 56.3, or about 5 citations per year per document. The observed trend showed a decrease in the number of citations in the last two years of the study, but the prevailing phenomenon did not indicate distancing from the scientif- ic community over time. Therefore, the drop in 2021 and 2022 citations does not mean a drop in reader interest, but only a brief gap between publication and citation conversion. The number of publications and ac- ademic involvement in the subject increased between 2015 to 2022. The Vosviewer software was used to map the bibliometric net- works, enabling the quantitative analysis of data regarding the be- havior of scientific production on WEEE in the analyzed period. This method made it possible to conduct the research in a more assertive and uniform way on specific groups (author citation, au- thor co-citation, journal citation, country citation, bibliographic coupling of documents and keyword co-occurrence), confirming the study hypotheses. Bibliometric analysis of co-citation between authors Bibliometric analysis of co-citation between authors A total of 14,623 authors were obtained with a co-citation relation- ship, with a minimum of 20 citations per author, totaling 171 collabo- rating and cooperating authors (Figure 4). In Figure 4, the lines between authors represent their cooper- ation links through citations, while the different colors represent the four collaboration clusters. The main researchers in the net- work were “Song, Qb” (cluster 3 — blue); Li (cluster 2 — green); Leung, Aow (cluster 1 — red) and Awual, Mr (cluster 4 — yellow). Evolution of publications over time Evolution of publications over time From the Web of Science database, a total of 6,637 articles participated in the research on electronic waste and produced, after choosing the filters, a total of 278 articles for the bibliomet- ric analysis. Figure 2 – Annual distribution, citations and average citations per year of published articles on e-waste compiled in the Web of Science database (2012–2022), using RStudio. 344 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 Bibliometric review of electro-electronic waste (WEEE) in the Web of Science database: groups’ production and main themes Research by Zhang et al. (2019) also showed a correlation with this anal- ysis, where out of a total of 6,601 authors, 14 published more than 30 articles. The most productive authors according to the study are the Chinese, further indicating that Chinese researchers have paid more attention to this field. Of these authors, “Song, Qb” has the highest link and total link strength (166/4,046 and 115 citations). Other authors were linked to one of these main authors. It was observed that, although some authors have a relatively low- er link, their articles were more cited and contributed significantly to scientific knowledge on the WEEE theme. The authors were “Zeng, Xl” (127 citations); “Cui, Jr” (90 citations); “Unep” (84 citations) and “Islam”, with 46 citations. It is important to mention that most of these publications focus on the environmental impact caused by the im- proper disposal of WEEE. In recent years, researchers have focused their attention on this topic due to its potential economic and envi- ronmental benefits. Another prominent author was Li Jinhui (104 ci- tations), with publications that focus on the theme of metal pollution and waste recycling. In cluster 2 — green (right side, Figure 3), the most outstanding author was: Xu, Zhenming (Jiao Tong University, China); with ten pub- lications and 1,103 citations; followed by He, Wenzhi (University of Science and Technology, China), who contributed with four publica- tions and 233 citations of the used sample. Main sources and its citationsh The primary sources analysis indicates the predominance of jour- nal publications in environmental science and pollution, environment and sustainability. Figure 3 – Map of network with the ten most influential and productive authors in waste electrical and electronic equipment research. Figure 3 – Map of network with the ten most influential and productive authors in waste electrical and electronic equipment research. Figure 4 – Main authors — co-citation network map. Figure 4 – Main authors — co-citation network map. 345 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 Bizerra, M.G. et al. A total of 51 major journals that published 229 documents on the subject (2012–2022) were analyzed. The main ten journals account for approximately 75.9% of the selected articles in the database, indicating that the topic interests many journals (Figure 5). It could be noticed that the journals that published the most on the subject were not necessarily the most cited. The Journal of Cleaner Production (IF — 11.072) was considered one of the most important journals using the bibliometric method of evaluating scientific journals, based on received citations and articles pub- lished by the journal, its high impact factor and academic prestige. Next are the Journal Waste Management (IF — 8.816) and Environmen- tal Science and Pollution Research (IF — 5.8). In Figure 5, the size of the circles represents the number of publi- cations in a journal; the wider circles imply more journal contributions for the subject. The color of the circles shows the year of publications on electronic waste. The bibliometric analysis shows that the Journal of Cleaner Production and Environmental Science and Pollution is that which plays the most dominant and influential role on the subject, with 58 (25%) of the publications, followed by Resources Conservation and Recycling, Waste Management, Waste Management & Research, Science of the Environment, Environmental International, Environ- mental Pollution and Critical Environmental Reviews with 23 (10%), 23 (10%), 16 (7%), 14 (6.1%), 10 (4.3%), 10 (4.3%), and 9 (3.9%) articles, respectively. Ranking of contribution per countries/regions of the world Figure 6 – Map of citations per country in waste of electrical and electronic equipment research during the period of 2012 to 2022 (sample 66 countries, minimum 3 articles per country, 392 documents — VOSviewer). Figure 7 – Bibliographic coupling network of research data (VOSVIEWER). China, India, Australia, USA and Italy are leaders in research and publications on the subject, with 106 (27%), 33 (8.4%), 28 (7.1%), 29 (7.3%) and 21 (5.3%) articles, respectively. The developing countries contributed to this sample with a signif- icant number of publications. China (106 publications; 27%), Malaysia (12 publications; 3%) and Brazil (6 publications; 1.5%) contributing to 31.6% of the research, while Pakistan (7 publications; 1.8%); Turkey (6 publications; 1.5%); Singapore (5 publications; 1.3%); Bangladesh (6 publications; 1.5%) and Ghana (7 publications; 1.8%) contributed with 7.9% of this research. The other developing countries in the analyzed bib- liometric network contributed with less than five publications (16.6%). Figure 7 – Bibliographic coupling network of research data (VOSVIEWER). The obtained bibliographic couplings and their influential articles are essential to perform the qualitative content analysis and discover the main themes and research directions. Based on the bibliometric results of this study, in the last decade, countries around the world have studied ways to properly dispose of WEEE, especially developed countries (USA, Japan and Russia), the main producers of WEE, and the developing countries (China, India, Brazil) (Forti et al., 2020). Analysis of the most cited articles reveals several groupings indicated by the names of the authors, as shown in Figure 7. The red cluster shows the perspective of studies on “electronic waste recycling”. The most cited article is that by Hahladakis et al. (2018), with an overview of chemical plastic additives and their environmental impacts. This analysis also in- cludes Tansel’s (2017) article on consumer electronics and e-waste. The research conducted by Concari et al. (2022) corroborates this study by confirming the expansion of academic influence in several countries on the subject and the prominence of China in this field, due to its various works on WEEE. The green cluster is characterized by numerous quotes from Kumar et al. (2017), who address an overview of electronic waste, generation, collection, legislation and recycling practices in his article. This set of articles addresses some issues related to waste management using a broader approach. Ranking of contribution per countries/regions of the world Figure 6 ranks the leading nations and their evolution in terms of scientific contribution concerning WEEE during the study. In this analysis, the larger each circle is, the greater the number of documents that the corresponding country has. Furthermore, the thick- er the link between circles, the more they collaborated. The distance be- tween the clusters indicates the strength between them and how much these authors publish in co-authorship. It appears that China and the USA have great cooperative relationships in studies on the subject. The Journal of Cleaner Production collaborated with 30 (13.1%) published documents and 1,691 citations, currently oc- cupying the rank of the journal that published the most articles on the subject. Regarding the Waste Management Journal, with its 23 (10%) published documents and 2,914 citations, it is the most cited journal globally, ranking first in citations, containing 154 link strengths. Figure 6, cluster (5) — lilac shows that China, USA, Pakistan and the Netherlands are in evidence, since together they represent 38.7% of publications. The bibliometric network shows that the Asian continent (Chi- na, India, Malaysia and Japan) had the greatest contribution, with 163 (41.5%) publications, followed by the European continent (Italy, England, Germany and the Netherlands), with 77 (19.6%) publica- tions, and the American continent (USA, Canada and Brazil), with 54 (13.7%). The other continents and countries contributed with 98 (25%) publications, highlighting a worldwide interest in the subject. It is opportune to verify the results of research carried out by Zhang et al. (2019) and Singh et al. (2021), where the most productive jour- nals on the subject were Waste Management (8.54%), Journal of Cleaner Production (5.20%), Environmental Science & Technology (4.43%) and Resources Conservation and Recycling, demonstrating that this field of research attracted several scientific journals globally. Figure 5 – Map of main journals that publish papers concerning the waste of electrical and electronic equipment. Figure 5 – Map of main journals that publish papers concerning the waste of electrical and electronic equipment. 346 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 Bibliometric review of electro-electronic waste (WEEE) in the Web of Science database: groups’ production and main themes Figure 6 – Map of citations per country in waste of electrical and electronic equipment research during the period of 2012 to 2022 (sample 66 countries, minimum 3 articles per country, 392 documents — VOSviewer). Ranking of contribution per countries/regions of the world For example, Islam and Huda (2018) address both the reverse logistics of electronics, the management of this waste and the implications of the circular economy on sustainable development. Bibliometric analysis: bibliographic coupling of documents Bibliometric analysis: bibliographic coupling of documents A total of 278 articles were identified as the most relevant in the WEEE theme in the Web of Science database. The analysis results gen- erated five article clusters, with 15,807 links in total (Figure 7). 347 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 Bizerra, M.G. et al. The bibliometric analysis proves that articles characterized by nu- merous references have greater influence on bibliographic coupling, and that the five analyzed clusters have interconnections with each other and with the subtopics presented in the keywords. The blue cluster shows that the authors worry constantly about the environmental sustainability theme. This topic has been effusively addressed in recent articles linked to subareas of interest. The main articles on the electronic waste management approach are from Kid- dee et al. (2013), and, on environmental pollution and electronics recy- cling, by Awasthi et al. (2016). According to Zhang et al. (2019) and Singh et al. (2021), WEEE research hotspots focused mainly on recycling technologies, environ- mental impacts and WEEE policies, closed loop and reverse logistics; in addition to this study, this analysis found that, over time, the fron- tiers of research on WEEE followed an evolutionary path, including dominant themes such as sustainability and circular economy. The yellow agglomerate is characterized by articles concerned with the recycling of metals from electronic waste. The main arti- cles are by Kaya (2016), on the recovery of metals and non-metals from electronic waste, Zhang and Xu (2016), on the analysis of recy- cling technologies for waste from electronic equipment. The most cited article was Ilankoon et al. (2018), which analyzed electron- ic waste in the international context, a review of the dangers and management strategies. Bibliometric review of electro-electronic waste (WEEE) in the Web of Science database: groups’ production and main themes Bibliometric review of electro-electronic waste (WEEE) in the Web of Science database: groups’ production and main themes On the networks by the co-occurrence of terms located in the ti- tles and abstracts (Figure 10) analysis, it is observed that the synergy between the words occurs around the words “recycling” and “e-waste”, which are interconnected with all the other highlighted keywords. Regarding the statistical analysis of the titles and abstracts of the ar- ticles in the sample, it was verified through the cloud of words (Figure 9) that the words that stood out the most were, in order of frequency: “re- view”, “e-waste”, “polybrominated diphenyl ethers” and “electronic waste”, clearly showing trends in research on electronic waste. Identifying the most recent active keywords in all these WEEE analyses can provide researchers with the most attractive research frontiers and areas of investigation within the explored field. The re- sults demonstrate that the search criteria in the databases are well aligned with the purpose of the research. In the word cloud of the abstracts, the words “e-waste” (793), “waste” (532), “recycling” (454), “management” (316) and “envi- ronmental” (277) are highlighted. Among the keywords that make up the cloud of titles in our sample, the most significant are “re- view” (147), “waste” (95), “recycling” (58), “electronic” (54) and “management” (51). Contribution of authors: BIZERRA, M. G. C.: Conceptualization, Methodology, Data curation, Investigation; Formal Analysis; Data curation; Writing — Original Draft. SANTOS, L. A.: Formal Analysis; Writing — Review & Editing. CORDEIRO, L. F. A.: Writing — Review & Editing. SALES, A. T.: Supervision; Writing — Review & Editing. Keywords analysis In the analyzed documents, a list of predominant words and their frequency in scientific articles were detected (1,795, minimum 5 oc- currences), and those more prominent in the textual analysis are high- lighted in the central area and are more extensive (Figure 8). A smaller cluster (lilac) focuses on environmental concerns with the impacts and treatment of electronic waste in the local and spatial dimension of recycling behavior; the main works are Premalatha et al. (2014), on “The generation, impact and manage- ment of electronic waste”, and Tembhare et al. (2022), on e-waste recycling practices. Figure 8 also shows that the themes with the highest occurrence in the keywords were “e-waste” and “polybrominated diphenyl ethers”, ap- pearing with a 107 and 61 times frequency, respectively, and being the keywords most used as an index for the analyzed articles. The distance between these two main keywords demonstrates the relative strength and similarity of the topics. The circle in the same grouping color sug- gested a similar topic among these publications. The size of the circles represents the occurrence of the keywords; that is, the number of times the keyword was co-selected in publications on WEEE. The qualitative analysis shows that Premalatha et al. (2014), Pérez-Be- lis et al. (2015), Al-Salem et al. (2022) and Tembhare et al. (2022) are sig- nificant, because although they still have few citations in their articles, the strength of intensity and cooperation of their links is revealing, demon- strating the importance of coupling data with other clusters. Figure 8 – Chain of connectivity about the co-occurrence of keywords in articles about waste of electrical and electronic equipment (visualization based on occurrence and periodicity). Figure 8 – Chain of connectivity about the co-occurrence of keywords in articles about waste of electrical and electronic equipment (visualization based on occurrence and periodicity). about the co-occurrence of keywords in articles about waste of electrical and electronic equipment (visualization based o 348 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 Conclusionsh The bibliometric analysis of published articles on electronic waste has grown significantly in the last ten years, revealing various themes and perspectives. Figure 9 – Title keyword cloud (RStudio — bibliometrix package). Figure 10 – Abstract keywords cloud (RStudio — bibliometrix package). Figure 9 – Title keyword cloud (RStudio — bibliometrix package). Of the 278 articles obtained from the Web of Science database, 40% had at least fifty citations. Among these, the five documents with the highest number of citations were Kiddee et al. (2013), Kaya (2016), Zhang and Xu (2016), Kumar et al. (2017) and, Hahladakis et al. (2018). The results show that the Journal of Cleaner Production and En- vironmental Science and Pollution are the most dominant journals in number of publications, with approximately 25% of the total. The Jour- nal of Cleaner Production tends to be more influential than others, with a higher impact factor and h-index. Scientific research on the subject is concentrated in about 66 countries. The nations with the highest vol- ume of scientific contributions are from the Asian continent. China is the most productive country, followed by India. “E-waste” and “poly- brominated diphenyl ethers” were the keywords with the highest fre- quency to represent the central theme of the study. The scientific map- ping indicates that the main research themes on WEEE are recycling, environmental impacts, sustainability, circular economy, efficient man- agement and technologies for recycling electronic waste. Figure 9 – Title keyword cloud (RStudio — bibliometrix package). Figure 10 – Abstract keywords cloud (RStudio — bibliometrix package). This research is not without limitations. First, we considered only peer-reviewed publications and excluded books and other types of documents. Second, we only considered the WoS database in this study. Extracting data from other scientific databases such as Scopus and Google Scholar, academic document types (e.g. book chapters and conference papers), as well as non-English papers may provide further information for future bibliometric analyses. In addition, for a more detailed analysis, a domain review would be desirable. Figure 10 – Abstract keywords cloud (RStudio — bibliometrix package). Contribution of authors: Contribution of authors: References Al-Salem, S.M.; Leeke, G.; El-eskandarany, M.S.; Haute, M.V.; Constatinou, A.; Dewill, R.; Baeyens, J., 2022. On the implementation of the circular economy route for E- waste management: A critical review and an analysis for the case Al-Salem, S.M.; Leeke, G.; El-eskandarany, M.S.; Haute, M.V.; Constatinou, A.; Dewill, R.; Baeyens, J., 2022. On the implementation of the circular economy route for E- waste management: A critical review and an analysis for the case Almeida, I.C.S; Galvão, R.; Resende, L., 2018. Academic rankings and pluralism: The case of Brazil and the new version of Qualis. EconomiA, v. 19, (3), 293-313. https://doi.org/10.1016/j.econ.2018.03.003 349 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 Bizerra, M.G. et al. of the state of Kuwait. Journal of Environmental Management, v. 323, 116-118. https://doi.org/10.1016/j.jenvman.2022.116181 Bibliometric Indicators. Journal of Environmental Analysis and Progress, v. 7, (2), 52-61. https://doi.org/10.24221/jeap.7.2.2022.4509.052-061 Awasthi, A.K.; Zeng, X.; Li, J., 2016. Relationship between e-waste recycling and human health risk in India: a critical review. Environmental Science and Pollution Research, v. 23, p. 11509-11532. https://doi.org/10.1007/s11356-016- 6085-7 Hahladakis, J.N.; Velis, C.A.; Eber, R.; Lacovidou, E.; Purnell, P., 2018. An overview of chemical additives present in plastics: Migration, release, fate and environmental impact during their use, disposal and recycling. Journal of Hazardous Materials, v. 344, 179-199. https://doi.org/10.1016/j. jhazmat.2017.10.014 Brasil, 2000. Implementação de sistema de logística reversa de produtos eletroeletrônicos e seus componentes de uso doméstico. Decreto nº 10.240/2000, de 12 fevereiro de 2020 Accessed October 22, 2022) at:. https:// legis.senado.leg.br Ilankoon, I.; Ghorbani, Y.; Chong, M.; Herath, G.; Moyo, T.; Petersen, J., 2018. E-waste in the international context – A review of trade flows, regulations, hazards, waste management strategies and technologies for value recovery. Waste Management, v. 82, 258-275. https://doi.org/10.1016/j. wasman.2018.10.018 Brasil, 2010. Política nacional de resíduos sólidos. Lei nº 12.305, de 2 de agosto de 2010 Accessed October 23, 2022) at:. http://www.planalto.gov.br. Islam, M.D.T.; Huda, N., 2018. Reverse logistics and closed-loop supply chain of Waste Electrical and Electronic Equipment (WEEE)/E-waste: A comprehensive literature review. Resources, Conservation and Recycling, v. 137, 48-75. https://doi.org/10.1016/j.resconrec.2018.05.026 Bressanelli, G.; Perona, M.; Saccani, N., 2019. Challenges in supply chain redesign for the Circular Economy: A literature review and a multiple case study. International Journal of Production Research, v. 57, (23), 7395-7422. https://doi.org/10.1080/00207543.2018.1542176 Ismail, H.; Hanafiah, M.M., 2019. References An overview of LCA application in WEEE management: Current practices, progress and challenges. Journal of Cleaner Production, v. 232, p. 79-93, https://doi.org/10.1016/j.jclepro.2019.05.329 Cavalcante, W.Q.F.; Coelho, A.; Bairrada, C.M., 2021. Sustainability and tourism marketing: A bibliometric analysis of publications between 1997 and 2020 using vosviewer software. Sustainability, v. 13, (9), 4987. https://doi. org/10.3390/su13094987 Ismail, H.; Hanafiah, M.M., 2020. Uma revisão da geração e gestão sustentável de lixo eletrônico: perspectivas presentes e futuras. Journal of Environmental Management, v. 264, 110495. https://doi.org/10.1016/j.jenvman.2020.110495 Concari, A.; Kok, G.; Martens, P., 2020. Revisão sistemática da literatura sobre conceitos e fatores relacionados ao comportamento pró-ambiental do consumidor em relação ao gerenciamento de resíduos por meio de uma abordagem interdisciplinar. Sustentabilidade, v. 12, (11), 4452. https://doi. org/10.3390/su12114452 Concari, A.; Kok, G.; Martens, P., 2020. Revisão sistemática da literatura sobre conceitos e fatores relacionados ao comportamento pró-ambiental sobre conceitos e fatores relacionados ao comportamento pró-ambiental do consumidor em relação ao gerenciamento de resíduos por meio de uma abordagem interdisciplinar. Sustentabilidade, v. 12, (11), 4452. https://doi. org/10.3390/su12114452 Kaya, M., 2016. Recovery of metals and nonmetals from electronic waste by physical and chemical recycling processes. Waste Management, v. 57, 64-90. https://doi.org/10.1016/j.wasman.2016.08.004 Kiddee, P.; Naidu, R.; Wong, M.H., 2013. Electronic waste management approaches: An overview. Wast Management, v. 33, (5), 1237-1250. https://doi. org/10.1016/j.wasman.2013.01.006 Concari, A.; Kok, G.; Martens, P., 2022. Recycling behaviour: Mapping knowledge domain through bibliometrics and text mining. Journal of Environmental Management, v. 303, 114160. https://doi.org/10.1016/j. jenvman.2021.114160 Kücher, A.; Feldbauer-Durstmüller, B., 2019. Organizational failure and decline – A bibliometric study of the scientific frontend. Journal of Business Research, v. 98, 503-516. https://doi.org/10.1016/j.jbusres.2018.05.017 Corrêa Nunes, I.; Kohlbeck, E.; Hänsch Beuren, F.; Borges Fagundes, A.; Pereira, D., 2021. Life cycle analysis of electronic products for a product- service syste. Journal of Cleaner Production, v. 314, 127926. https://doi. org/10.1016/j.jclepro.2021.127926 Kumar, A.; Holuszko, M.; Espinosa, D.C.R., 2017. E-waste: An overview on generation, collection, legislation and recycling practices. Resources, Conservation and Recycling, v. 122, 32-42. https://doi.org/10.1016/j. resconrec.2017.01.018 Forti, V.; Baldé, C. P.; Kuehr, R.; Bel, G., 2020. The global e-waste monitor 2020: Quantities, flows and the circular economy potential. ITU (Accessed January 01, 2023) at:. https://ewastemonitor.info/wp-content/uploads/2020/11/ GEM_2020_def_july1_low.pdf. Lepawsky, J., 2015. “The changing geography of global trade in electronic discards: Time to rethink the e-waste problem. The Geographical Journal, v. 181, (2), 147-159. https://doi.org/10.1111/geoj.12077 Forti, V.; Baldé, C. P.; Kuehr, R. Bel. G., 2020. The Global E-waste Monitor 2020: quantities, flows and the circular economy potential. References Bonn/Geneva/ Rotterdam: United Nations University /United Nations Institute for Training and Research – co-hosted SCYCLE Programme, International Telecommunication Union & International Solid Waste Association. Forti, V.; Baldé, C. P.; Kuehr, R. Bel. G., 2020. The Global E-waste Monitor 2020: quantities, flows and the circular economy potential. Bonn/Geneva/ Forti, V.; Baldé, C. P.; Kuehr, R. Bel. G., 2020. The Global E-waste Monitor 2020 titi fl d th i l t ti l B /G / Li, J.; Lopez, B.N.; Liu, L.; Zhao, N.; Yu, K.; Zheng, L., 2013.Regional or global WEEE recycling. Where to go? Waste Management, v. 33, 923-934. https://doi. org/10.1016/j.wasman.2012.11.011 l Rotterdam: United Nations University /United Nations Institute for Training and Research – co-hosted SCYCLE Programme, International Telecommunication Union & International Solid Waste Association. Training and Research co hosted SCYCLE Programme, International Telecommunication Union & International Solid Waste Association. Moher, D.; Liberati, A.; Tetzlaff, J.; Altman, D.G., 2009. Reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Medicine, v. 6, (7), 264-269. https://doi.org/10.1371/journal.pmed.1000097 Gollakota, A.R.K.; Gautam, S.; Shu, C.M., 2020. Inconsistências na gestão de lixo eletrônico em países em desenvolvimento – fatos e soluções plausíveis. Journal of Environmental Management, v. 261, 110234. https://doi. org/10.1016/j.jenvman.2020.110234 Mulet-Forteza, C.; Martorell-Cunill, O.; Merigó, JM; Genovart-Balaguer, J.; Mauleon-Mendez, E., 2018. Twenty five years of the journal of travel & tourism marketing: a bibliometric ranking. Journal of Travel & Tourism Marketing, v. 35, (9), 1201-1221. https://doi.org/10.1080/10548408.2018. 1487368 Govindan, K.; Soleimani, H., 2017. Uma revisão da logística reversa e cadeias de suprimentos de circuito fechado: um diário com foco na produção mais limpa. Journal of Cleaner Production, v. 142, (Part 1), 371-384. https://doi. org/10.1016/j.jclepro.2016.03.126 Organização das Nações Unidas (ONU), 1991. CNUNMA – Conferência das Nações Unidas sobre Meio Ambiente e Desenvolvimento. Agenda 21. Washington: ONU (Accessed January 03, 2023) at:. https://nacoesunidas.org Goodship, V.; Stevels, A.; Huisman, J. (Eds.)., 2019. Manual de Resíduos de Equipamentos Elétricos e Eletrônicos. Cambridge: Woodhead Publishing (Accessed December 22, 2023) at:. https://doi.org/10.1016/C2016-0-03853-6 Guedes, F.L.; El-Deir, S.G.; Aragão Júnior, W.R.; Jucá, J.F.T., 2022. Analysis of Scientific Production of Refused Derived Fuel through Scientometric and Goodship, V.; Stevels, A.; Huisman, J. (Eds.)., 2019. Manual de Resíduos de Equipamentos Elétricos e Eletrônicos. Cambridge: Woodhead Publishing (Accessed December 22, 2023) at:. https://doi.org/10.1016/C2016-0-03853-6 (Accessed December 22, 2023) at:. https://doi.org/10.1016/C2016-0-03853-6 Organização das Nações Unidas (ONU), 2015. A ONU e o meio ambiente (Accessed January 03, 2023) at:. https://nacoesunidas.org/acao/meio- ambiente/. References Guedes, F.L.; El-Deir, S.G.; Aragão Júnior, W.R.; Jucá, J.F.T., 2022. Analysis of Scientific Production of Refused Derived Fuel through Scientometric and 350 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478 Bibliometric review of electro-electronic waste (WEEE) in the Web of Science database: groups’ production and main themes Tansel, B., 2017. From electronic consumer products to e-wastes: Global outlook, waste quantities, recycling challenges. Environnent international, v. 98, 35-45. https://doi.org/10.1016/j.envint.2016.10.002 Tansel, B., 2017. From electronic consumer products to e-wastes: Global outlook, waste quantities, recycling challenges. Environnent international, v. 98, 35-45. https://doi.org/10.1016/j.envint.2016.10.002 Organização das Nações Unidas (ONU), 2020. China e Estados Unidos lideram lista de países que mais geram lixo eletrônico. Clima e meio ambiente (Accessed January 03, 2023) at:. https://news.un.org/pt/ story/2020/07/1719142. Tembhare, S.P.; Bhanvase, B.A.; Barai, D.P.; Dhoble, S.J., 2022. E-waste recycling practices: a review on environmental concerns, remediation and technological developments with a focus on printed circuit boards. Environment, Development and Sustainability, v. 24, 1-83. https://doi. org/10.1007/s10668-021-01819-w Pérez-Belis, V.; Bovea, M.D.; Ibáñez-Forés, V., 2015. An in-depth literature review of the waste electrical and electronic equipment context: Trends and evolution. Waste Management & Research, v. 33, (1), 3-29. https://doi. org/10.1177/0734242X14557382 Perkins, D.N.; Drisse, N.B.; Nxele, T.; Sly, P.D., 2014. E-Waste: A Global Hazard. Icahn School of Medicine at Mount Sinai. Annals of Global Health, v. 80, 286-295. https://doi.org/10.1016/j.aogh.2014.10.001 Tsai, F.M; Bui, T.-D.; Tseng, M.-L.; Lim, M.K.; Hu, J., 2020. Municipal solid waste management in a circular economy: A data-driven bibliometric analysis. Journal of Cleaner Production, v. 275, p. 124132. https://doi.org/10.1016/j. jclepro.2020.124132 Premalatha, M.; Abbasi, T.; Abbasi, T.; Abbasi, S., 2014. The generation, impact, and management of e-waste: State of the art. Critical Reviews in Environmental Science and Technology, v. 44, (14), 1577-1678. https://doi.org/ 10.1080/10643389.2013.782171 Türkeli, S.; Kemp, R.; Huang, B.; Bleischwitz, R.; McDowall, W., 2018. Circular economy scientific knowledge in the European Union and China: A bibliometric, network and survey analysis (2006–2016). Journal of Cleaner Production, v. 197, (Part 1), 1244-1261. https://doi.org/10.1016/j. jclepro.2018.06.118 Sharma, M.; Joshi, S.; Kumar, A., 2020. Assessing enablers of e-waste management in circular economy using DEMATEL method: An Indian perspective. Environmental Science and Pollution Research, v. 27, (12), 13325- 13338. https://doi.org/10.1007/s11356-020-07765-w Van Eck, J.N.; Waltman, L., 2010. Software survey: VOSviewer, a computer program for bibliometric mapping. Scientometrics, v. 84, 523-538. https://doi. org/10.1007/s11192-009-0146-3 Shittu, O.S.; Williams, I D.; Shaw, P.J., 2021. Global E-waste management: Can WEEE make a difference? References A review of e-waste trends, legislation, contemporary issues and future challenges. Waste Management, v. 120, 549- 563. https://doi.org/10.1016/j.wasman.2020.10.016 Zhang, L.; Xu, Z., 2016. A review of current progress of recycling technologies for metals from waste electrical and electronic equipment. Journal of Cleaner Production, v. 127, 19-36. https://doi.org/10.1016/j.jclepro.2016.04.004 Singh, S.; Trivedi, B.; Dasgupta, M. S.; Routroy, S., 2021. A bibliometric analysis of circular economy concept in E-waste research during the period 2008–2020. Materials Today: Proceedings, v. 46, (Part 17), 8519-8524. https:// doi.org/10.1016/j.matpr.2021.03.525 Zhang, Y.; Geng, Y.; Zhong, Y.; Dong, H.; Liu, Z., 2019. Uma análise bibliométrica da pesquisa de resíduos eletroeletrônicos. Environmental Science and Pollution Research, v. 26, 21098-21108. https://doi.org/10.1007/ s11356-019-05409-2 351 RBCIAMB \| v.58 \| n.3 \| Sep 2023 \| 342-351 - ISSN 2176-9478
https://openalex.org/W3157079905	https://hts.org.za/index.php/hts/article/download/6324/17526	English	null	The religious phenomenon of Juche ideology as a political tool	HTS teologiese studies	2,021	cc-by	5,600	HTS Teologiese Studies/Theological Studies ISSN: (Online) 2072-8050, (Print) 0259-9422 Page 1 of 7 Original Research Page 1 of 7 Original Research Original Research Page 1 of 7 Dates: Keywords: ideology; Juche; North Korea; religion; political. The religious phenomenon of Juche ideology as a political tool This study aims to determine the motive that led to the establishment of Juche by Kim Il Sung amidst the influence of communism and its transformation into religion in North Korea. North Korea is a communist country dictated by Kim Jong-Un of the Kim dynasty and known for its cruelty. The country underwent several changes from Marxism-Leninism to familism to determine its strength in Juche. This ideology that acts as a religion was influenced and strengthened Kim Jong Il to Kim Jong-Un and built by shifting the concept of marxism- Leninism to construct a new understanding of Juche. It will be demonstrated that this ideology was influenced by Confucianism, Christianity, Nationalism, Chinese Communism, and Russian Communism. In the modern era, imperialism was used as an ideological tool to restrict backwardness. This theory allegedly helped Kim Il-Sung establish a unitary, one- person rule over North Korea. ‘It will be examined whether Juche ideology is a tool the state has used to convince people of their government. Pronouncements, an intentional religion in which the people were to believe that their Ruler (Kim Il Sung) was a supreme human or an ideology that morphed into a religion’. It will be demonstrated that, when they started honoring Kim as their god, no other religion was permitted. Affiliations: 1Graduate School, Faculty of Theology, Real Theological Seminary Batam (STT Real Batam), Batam, Indonesia 2Christian Religious Education, Teacher Training and Education, Christian University of Indonesia, Jakarta, Indonesia Contribution: This research offers readers an understanding of the value of humanity in the binding ideology of Juche. However, the Juche Ideology can serve as a missiological bridge towards mission goals, which require the experience of spiritual, physical, and social liberation. Introduction How to cite this article: Widjaja, F.I., Boiliu, N.I., Simanjuntak, I.F., Gultom, J.M.P. & Simanjuntak, F. 2021, ‘The religious phenomenon of Juche ideology as a political tool’, HTS Teologiese Studies/ Theological Studies 77(4), a6324. https://doi.org/ 10.4102/hts.v77i4.6324 Kim Jong-un, the current leader of North Korea, is the successor of Kim Jong II. ‘Jong-Il pioneered and established the ideological foundation as a means of policy implementation and sustaining power’ (Park 2014:1–14). It shows the concept of ‘familism, socialism and religious politics’ (Armstrong 2005:383–394). The family as a primary social unit is the starting point for emerging political figures, including ‘religious leaders’. Armstrong’s article reacts to Sergei Kurbanov’s research on North Korea’s Juche ideology: primitive communism or traditional thinking? It will be examined whether Juche ideology is a tool the state has used to convince people of their government’s pronouncements, an intentional religion in which the people were to believe that their Ruler (Kim Il-Sung) was a supreme human or an ideology that morphed into a religion (Bandow 2019). Copyright: © 2021. The Authors. Licensee: AOSIS. This work is licensed under the Creative Commons Attribution License. Research method The data collection technique is carried out in two ways: firstly, the literature study method searches for various literature and other credible sources related to the research topic directly in the Juche library. Secondly, by conducting interviews to gather information or perceptions from informants regarding the issue, the author wanted to research whilst travelling in North Korea. The historical and descriptive approach was applied in this study. The author visited the Juche Center located in Pyongyang from 30 October to 02 November 2018. Two persons accompanied the author during the trip: Ms. Kim, a young lady, as an interpreter and an old gentleman (67 years) as a tour guide. They are both interviewed and several individuals met at the centre (principles of Juche Center), and librarians were also interviewed regarding Juche ideology. Widjaja visited Pyong Yang, capital of North Korea, from 29 October to 03 November 2018 to look into the social life context. A lot needs to be understood regarding this country. North Korea is still governed by the legacies left by its founder Kim Il-Sung who died in 1994. Following the Juche philosophy, which is also signified in its meaning ‘Ju’ depicts owner, or lord whilst ‘che’ implies body, essence or existence, and nature. This meaning is similar to the basic philosophical principle usually expressed in Korea, which stated that ‘man is the master of everything and decides everything’ (Kurbanov 2019:296–305). This research tries to describe the Juche ideology and the North Korean society’s pluralistic conditions and examines the missiological paradigm contextualisation in their social culture. This research also interpreted the theological vocation of the church in a pluralistic world. Widjaja sees the pluralistic society of North Korea as a bridge for doing contextualised evangelistic task. Furthermore, when the author and team travelled to North Korea, they saw several banners with a proverb, which stated that ‘Kim Il-Sung lives forever. The citizens see Kim Il-Sung as a god. The recent inception of the former dictator’s worship has become one of the world’s acclaimed religions, with approximately 23 million followers. It is more famous than most renowned faiths globally, such as Judaism, Sikhism, Jainism, Bahaism and Zoroastrianism. However, according to Belke, Juche ideology is the tenth-largest religion that has not been studied from the missiological perspective (Belke 1999:1). Page 2 of 7 Page 2 of 7 Ginkel (Ginkel 1982), when an individual or group of persons has chosen a particular goal and when that goal becomes the only way of giving meaning to life, then, said Goudzwaard, that individual or group is on the way to forming an ideology. He further links idolatry to come together with ideology when the goals set become so crucial that the means of attaining them become idols. characterised by an important spiritual being’. Widjaja (2019:591; cf. Widjaja & Boiliu on a plurality [Widjaja & Boiliu 2019]) stated that pluralism is a condition that encompasses a society that tolerates diversity. Korean culture has tended towards being mono-ideological because it always had a single ideology, Confucianism. Widjaja defined this new religion and the North Korean society as having a natural affinity towards communism. Foreigners had limited access to information apart from those communicated by the early missionaries. In 1884, the first Protestant missionary from the United States of America visited the country. Revivals began on the Korean peninsula in 1906 and climaxed in 1909 when the Pentecostal spiritual movement ‘A million souls for Christ’ occurred (Lamport 2018:431). Copyright: Copyright: © 2021. The Authors. Licensee: AOSIS. This work is licensed under the Creative Commons Attribution License. The citizens were made to respect and accept these teachings. North Korea is always in the global media, and it is often reported as a dangerous, provocative, irrational, poor and totalitarian country. Several dominant metaphors appeared (Dalton 2019) in the coverage. These framed North Korea as a military threat (conflict metaphor), unpredictable, irrational and ruthless (psychopathology metaphor), isolated and secretive (pariah metaphor) and a cruel dystopia (Orwellian metaphor) impoverished (basket-case metaphor). However, it is one of the poorest nations that metaphorically isolate itself from the rest of the world. It is referred to as the Hermit Kingdom (Griffis 2009:66) derived from the Chosun Dynasty. To date, it is known as the ‘Hermit kingdom’, notably when it closed its borders to foreigners and restricted its citizens from travelling abroad since the Korean War in 1950. Juche ideology is a form of idol worship. Juche ideology is a form of idol worship. In this case, Juche ideology has been developed into a world or a personal view of life. As Goudzwaard said in http://www.hts.org.za Original Research Page 2 of 7 Result and discussion In 1994 after Il-Sung’s death, North Korea was hit by severe famine ; the successor Kim Jong-Il opened the United Nations (UN) and Non-Government organization (NGOs) borders assistance. At this point, foreigners realised they had a biased view of North Korean culture and religion. The puzzles concerning Kim Il-Sungism started to be understood (Lee 2018). The world could not develop a holistic picture because previous studies on North Korea did not approach Korean ideology from an ethnic position. We have been able to gain new insights from defectors who gave us a new understanding of Juche theology and its development, and its becoming a religion. However, the complete picture was not detected because ethnic approaches did not wholly analyse the pieces. The previous studies were carried out from the political perspective rather than a missiological concept. However, some defectors aided in understanding the Juche ideology and why it was developed as a North Korean religion. Hwang Jang-Yup, an engineer primarily responsible for crafting Juche, defected from North Korea in 1997 (Kihl & Kim 2015:62). It gave foreigners an insight into Kim II Sungism that led to the factors that influenced Juche theology being identified. Subsequently, other religions are not permitted, except for the worship of Kim Il-Sung. Newbigin (Newbigin 1989:66) reported that ‘throughout history, all religions globally have been developed in the tradition of sustainable, rational arguments. They have all attracted the professor’s authority in the past’. It is suggested in the writings of Newbigin that an ideology can develop into a religion that the present author identifies with the Korean Confucian tradition. The author sees that it is an aspect of the North Korean pluralism reported by Newbigin (Newbigin 1989:1). ‘In Western ideology, pluralism is considered an appropriate characteristic of a secular society that lacks an officially approved pattern of belief or behaviour. Therefore, it is also presumed as a free society because accepted dogmas do not control it; rather, it is mainly http://www.hts.org.za Open Access Open Access Page 3 of 7 Page 3 of 7 Page 3 of 7 Original Research Original Research Kim Il-Sung’s family migrated to China when Kim was aged 5 years because the Japanese police threatened his father because of the anti-Japanese movement’s threats. Kim’s father studied the Lenin Revolution in Russia and told Kim Il-Sung about it, hoping to influence Korea. (Kim 1992:79). Result and discussion Kim Il-Sung witnessed how his father fought the Japanese to regain their independence. Kim’s father taught him the Christian faith values, thereby causing Kim to view the world’s religious disposition. Numerous Christians influenced Kim’s father because the Japanese had taken advantage of them for their worship of God. The father’s patriotism led to the development of Christian social religion’. Although Kim Il-Sung reported in the memoir of his father that he was an atheist (Kim 1992:20). Kim Il-Sung also stated that his mother was only a churchgoer and not a born again Christian (Kim 1992:79). Although Kim Il-Sung had frequently attended a church at a young age, Kim categorically stated that they did not believe in Jesus as their saviour. Christianity influence Kim Il-Sung was born in 1912, 2 years after Japan annexed Korea (Kim 1992:17). His father, Kim Hyung-Jik, gained admission into Soong-Shil Junior High School in Pyongyang (Kim 1992:4), an academy established by missionaries in 1897 to educate and raise Christian leaders in Korea. Irrespective of the fact that Kim Hyung-Jik did not complete his education, Kim was enlightened, and the school broadened his worldview. Kim was fortunate to travel outside Korea with the help of the foreign Christian teachers. Kim Hyung-Jik became a nationalist and started fighting for Chosun’s independence. Confucianism influence Kim Il-Sung was raised in a typical Korean family under the influence of Confucianism. Irrespective of the fact that most of them had survived under Shamanism’s influence (belief in a supernatural being is the earliest form of religious practice). Shamans (Kim 1998:33) are healers or spiritual leaders believed to interact with the spirit world through altered states of consciousness such as trance. Consequently, when they are unconscious, their souls leave their bodies and travel beyond a spirit guide. They carry out various forms of healing, namely physical, psychological and spiritual. Confucianism filled the Shamanism structured teaching gap, and it is the Chosun’s leading religion (Young 2013:51–66). The Confucian philosophy was introduced in Korea 1600 years ago from China. Its originator Confucius lived from 551 BC to 479 BC, and the doctrines were humanistic and philosophical teachings concerning the importance of family and social harmony. It focused on the way of life (Lee 2018). However, it is the reason the political system of the North Korean government resembles the order of the Presbyterian church, thereby providing evidence of the influence of Christianity on the life of Kim. Confucius taught moral and political values to control the people authoritatively. Moreover, patriotism was also included. Besides, some other scholars added different ethics to the original philosophy, emphasising that a well-behaved lifestyle on earth leads to a better life after death. Finally, ancestral worship was included, and it was observed as a religion (Lee 2018). Communism influence in China Kim Il-Sung returned to China at the age of 14 and was trained as an anti-Japanese guerrilla. In 1926 (Kim 1992:96), Kim learned to fight after studying Korean history. During this period, Kim studied Chinese communism and war strategies from militaristic organisations. Koreans allied with the Chinese revolutionary army to fight the Japanese who had invaded eastern China. Whilst in the camp, Kim started reading and studying Communism by Marx and Lenin, which was not included in the training curriculum (Belke 1999:171). In 1392, the Chosun Dynasty (Griffis 2009:73) replaced Buddhism, the Koryo Kingdom’s recognised religion, with Confucianism, which forces people to obey their teachers, fathers and king. It consists of strict rules following maintaining a hierarchically structured society. The king rules with the monarchy and is above the law. Fathers were also the heads of household and their decisions are final. Under the influence of Confucianism, this patriarchal society maintains a strict structured order and does not allow individual family members’ freedom. Women do not have a say in family matters; instead, they obey the men’s rankings in the household. The eldest son is the only person who has authority over the house layout. The father’s obligation and duty is to maintain power in a society influenced by Confucianism (Griffis 2009:73). Kim formulated ‘The Way of the Korean Revolution’ on June 30, 1930, in China. Kim reported that ‘the leaders’ of the movement needed to be amongst the masses (people) to make them understand that they are fighting a political war, against the ‘anti-Japanese struggle’ (Belke 1999:170). Kim’s approach of combining communism and religion led to the Juche being founded, which insisted that the state become autonomous without other countries’ interference. Subsequently, Kim joined and participated in the anti- Japanese guerrilla war in Manchu, China. Kim fought as a commander at Bo-Chun-Bo and Mu-San from 1937 to 1939, respectively. Kim was only 27 years old then, which led to the recognition by the Soviet Union as a ‘general’ that fought against the Japanese (Kim 2003:106). Open Access The influence of Russian communism The Korean peninsula is one of the oldest and most dangerous conflict-prone areas globally (Ponomareva & Rudov 2015: 45–56). According to the biography, Kim stated the war’s happenings in 1940 during the International Communist Party’s Conference held by the Soviet Union. Kim was http://www.hts.org.za Open Access Page 4 of 7 Page 4 of 7 Original Research and constitution (Kementerian Unifikasi Korsel, Institute for Military History dan Agency for KIA Recovery & Identification di Kementerian Pertahanan Korea Selatan 2020). and constitution (Kementerian Unifikasi Korsel, Institute for Military History dan Agency for KIA Recovery & Identification di Kementerian Pertahanan Korea Selatan 2020). guaranteed the support of an organised and trained force to fight against the Japanese troops in Manchu, China. The Soviet Union needed someone to introduce Chosun in communist Russia, a satellite country. Kim was appointed as the commander of the Chosun People’s Revolution and was given the southern camp (Kim 1992:71). Kim was assigned to combat Japanese expansion in the Khabarovsk region. In 1942, during the summer of 1942, China and the Soviet Army formed a joint force and attacked Japan. It was reported that the country was under new communism different from Marx or Lenin’s ideology. This unique development of combining a Marxist of Lenin’s ideology with Confucianism led to Juche ideology, defended by Kim’s dictatorship (Jang-yop 1999:127). Kim was an aggressive communist soldier and was favoured by Stalin. On 25 August 1945, the Soviet Union special forces entered Pyongyang City without resistance from the Japanese Army (Seth 2018:27). Kim was only 33 years old then. Kim gained power in North Korea because of the help of the Soviets and was surrounded by advisers from the Union (Belke 1999:171). On 25 August 1948, Kim was elected as the Prime Minister because of the influence of Marxist communism and the brutal application of Stalin’s Soviet-style policies (Seth 2018:27). With the Soviet Union’s military assistance, Kim invaded South Korea in 1950 over the 38th parallel line drawn at Potsdam, Germany, in 1945. Kim desired to bring the peninsula under communism. The war continued until the United States signed the Armistice agreement in collaboration with the United Nations on 27 July 1953. The principle ‘Juche’ considers the fact that generally humans are the driving force in history (Grzelczyk 2012:33–68) and ‘are the master of their destiny’. Kim Jong-Il joined the Communist Party after graduating from the university in 1964. The influence of Russian communism Kim increasingly fought the struggle for the North Korean leader succession. It led to competitive personal deification in which Stalin was involved by supporting economic aid and rebuilding North Korea after war (Jang-yop 1999:1292). In 1972, Kim Jong-Il conferred the Juche Tower and the new Doctrine on the father’s 60th birthday. Pyongyang stated that Juche is an ideology completely different from Marxism–Leninism. These changes were reflected in the constitution in 1974. The new form disclosed that the people supported their leaders. Formation of Juche ideology The dogma of Marxism–Leninism, without a doubt, was accepted as the law from 1945 to 1952 (Belke 1999:173). Kim focused on rebuilding the country destroyed by a war based on the Russian communist system. Furthermore, Kim broadened the teachings of Stalin’s Communist-policies in the country. Kim Il-Sung was seen as a god. It contradicts the statement made by Kim Il-Sung before joining Marxist communism. ‘Religion is a superstitious belief, irrespective of whether an individual believes in Christianity or Buddhism. After Stalin’s death on 05 March 1953, Russia experienced chaos because of a lack of a strong leader. Stalin’s death also led to specific political challenges in North Korea because they withdrew their support, and the Chinese sought to exercise control over the nation. Kim founded the Juche ideology in its embryo form as a means to justify a series of brutal cleansing to cut off excessive Soviet and Chinese influences and eliminate political rivals (Belke 1999:173). Kim (28 December 1955) stated that North Korea needs to become independent following the Juche ideology. Kim’s speech was titled ‘Eliminating Dogmatism and Formalism and Building Juche in Ideological Work’. The Workers’ Party in North Korea started to develop this idea and new policies began to emerge. It is one of the most powerful ideologies that incite people’s quest for independence and ‘appeals to creativity at the highest level but a creativity that brings captivity and bondage’ (Col 2:8): Historically, religion has been dominated by the ruling class and is used as a tool to deceive, exploit and oppress people. Presently, imperialism has been used as an ideological tool to curb backwardness (Kim 2017:160). Hwang stated that ‘assuming the “Juche” doctrine is summarised in one sentence, it tends to be reported as the ruler (Kim Il-Sung) is the inviolable god who is in absolute control and conducts the affairs of the society because they are perceived as almighty (Jang-yop 1999:147). North Korea has a religion in which Kim Il-Sung is worshipped following the Juche. Juche as a political tool To establish Juche means ‘to be revolutionised and reconstructed in the nation itself’ (Lee 1997). It becomes the basis for the rise of ‘totalitarianism’ (Park 2008), in which the mechanism is to create figures as deified. In principle, ‘politics uses religion or religion uses politics’ (Linz 2004). The point is ‘to create power as widely and as strong as possible’ (Schafer 2004) ‘as “dictator and fascism” (Gregor 2012) where “the principle of fascism is the state as religion” (Armstrong 2016). Juche is used as a political tool (religious politics) in North Korea. Historically, Juche was also rooted in Marxism, because for Khazanov, ‘Marxism–Leninism as a secular religion’ (Khazaniv 2008). Therefore, the theory that religion can be used as a political tool or politically using religion is correct, as is North Korea’s case. Juche did not offer salvation such as religion in general, but Juche found it a political tool of totalitarianism. The indoctrination that has taken place in North Korea through the instrument of Juche ideology and it is a barrier that undermines the introduction of other ideologies and religion. Furthermore, Christianity rejects the concept of absolute human power. On the contrary, Christians are believed to live by imitating Christ, who, although in God’s image, humbled and emptied himself to bring salvation and renewal to human life. So, there is quite a significant difference between Juche ideology and Christianity. Although North Koreans are not aware of this ideology’s influences, they had been taught to observe Juche. They make individual life decisions without acknowledging God’s existence, and they have a humanistic worldview as a result. Many North Koreans are not aware or have never heard of Christianity or that Jesus came to save humanity; however, they believe in Communist materialism and its evolution. North Korea built over 40 000 statues of Kim Il-Sung in the country (Lee 2018). Married couples usually visit the statue to take pictures. It validates their marriage after the ceremony. They often see with flowers to honour their gods. Death of God In 1958, all the churches in North Korea were destroyed and believers eliminated by the communists during the anti- intelligence movement. The remaining ones were underground and remained as universal, invisible churches (Lee 2018). The citizens were shocked when Kim Il-Sung died in 1994; some of them even fainted. Several of them cried for days because it was not the death of an ordinary person (Seth 2018:154). Incredibly, many North Koreans believed and hoped that Kim Il-Sung was an immortal being. From a foreigners’ perspective, it is hard to believe that the people thought that Kim, a human was immortal. After Kim Il-Sung’s death, Kim Jong-Il immediately made a banner that stated that ‘Kim Il-Sung lives forever among them’. Kim Jong-Il led people to honour the father, Kim Il-Sung, under Confucian ancestor worship and spiritual being under the Juche ideology. Juche’s religious practices The exact number of underground believers in North Korea is unknown. However, 16 984 martyrs have been recorded from 1945 to 2006. It shows the existence of underground churches in this nation (Lee 2018). It was discovered that approximately 66% of the martyrs were converted to Christianity before 1945, their parents influenced 20%, 11% reported that they were Christians and 2% stated that they encountered God in China (Lee 2009). Consequently, whenever foreigners visit North Korea and make inquiries on the definition of Juche, only a few people provide accurate answers, according to Juche’s guide book (2008:17). They have been indoctrinated to act and dwell under the Juche ideology in which Kim Il-Sung teaches that ‘Humans are the master of the universe and are in control of their destinies as well as the right to have it developed’ (Juche Guideline 2008:15). ‘From an atheist’s point of view, man’s absolute ability to control his destiny has replaced the belief in God’s existence’. Lee (2018) testified as a defector after studying the Bible and having discovered God and wrote that Juche is a concept adopted from the Bible. It might be agreed on one hand because the Bible attributed a high value on human life. Yet, Christianity teaches that human life is subject to God’s authority. Thus, a human being is God’s appointed steward responsible for keeping God’s commandment and glorifying God (Cho 2002). Official church The communist even closed a church approved by the government in 1958 and reopened it in 1972 to show the world that there is religious freedom similar to South Korean Christianity. Juche culture See to it that no one takes you captive through philosophy and empty deception, according to the tradition of men, according to the elementary principles of the world, rather than according to Christ. The author’s visit to North Korea could not produce an accurate missiological study because of its limited nature. The movement of foreigners and citizens is restricted in this nation. The author could not leave the hotel without an escort and could not visit many cities. People are not aware of each others’ activities because freedom of movement is prohibited. All those who desired to be independent united and cooperated with support from the communist manifesto http://www.hts.org.za http://www.hts.org.za Page 5 of 7 Page 5 of 7 Original Research Therefore, because of this fact, this study is based on the author’s observations of Juche’s religious view. Therefore, because of this fact, this study is based on the author’s observations of Juche’s religious view. Every city centre has a Juche ideology study room and a temple to worship Kim Il-Sung. North Koreans enter this Hall in holiness, and people are prohibited from speaking out loud except to pay respect in the form of worship. People visit this place in holiness, communicate their desires and receive spiritual power. The restriction of movement has been instrumental in the formation of a Juche culture. It is the control of communication through the limiting of movement and access to technology that has entranced Juche culture (Güven 2019). Authors’ contributions Juche’s ideology was influenced by Confucianism, Christianity, Nationalism, Chinese and Russian communism. North Korean liberators modified Russian communism. Subsequently, Kim Jong-Il changed communism into a religion according to the Juche doctrine, making the people believe that Kim was immortal. All authors contributed equally to this research article. Missiological perspective A missiological perspective on Juche opens a broad view of the universal value of human beings. Juche valued humans as the master of the universe and are in control of their destinies as well as the right to have it developed, but his deviation was the abuse of one person. This article also wants to show that http://www.hts.org.za Open Access Open Access Page 6 of 7 Original Research the mission’s purpose is to convey that all people experience liberation spiritually, physically and socially. Toruan to accompany and help the author (Fransiskus Irwan Widjaja) in collecting the information to North Korea and for documentation and partner for discussion during their time in North Korea in 2018. The authors are grateful to Mr Noh I. Boiliu for editing and providing additional supporting articles. The authors are also thankful to Mr Irfan Feriando Simanjuntak for delivering the concept and required description and Mr Joni Manumpak Parulian Gultom for reviewing this article before its publication. The authors also thank Mr Fredy Simanjuntak for taking part in the sentence editing process. In the present context, the church cannot perform its mission because of the socio-political problems. However, the task cannot be understood in the narrow sense of preaching the Gospel to North Korea’s people. This article encourages and voices that everyone has an equal position before God in the Christian mission perspective. The act of suppressing individual rights in North Korea is deviant. Therefore, the idea of ‘individual equality’ in Juche must be upheld (cf. Widjaja, Simanjuntak & Boiliu 2020:189–193). References Armstrong, C., 2016, ‘Political religion’, in P. Corner & J.H. Lim (eds.), The Palgrave handbook of mass dictatorship, pp. 67–80, Foreign Languages Publishing House. Therefore, when the people started to honour Kim as their god, no other religion was permitted. It led to an increase in the persecution of underground churches because of various reasons. Generally, Marxist communism does not allow Christianity because the Christian faith does not support materialism adhered to by the communists. Lenin called it ‘the opium of the masses’, and it tends to destroy the society. Armstrong, C.K., 2005, ‘Familism, socialism, and political religion in North Korea’, Totalitarian Movements and Political Religions 6(3), 383–394. https://doi. org/10.1080/14690760500317743 Belke, T., 1999, Juche: A Christian study of North Korea’s state religion, Living Sacrifice Book Company. Cho, E., 2002, ‘The encounter between the Juche idea and Christianity’, Mission Studies XIX(1), 1–37. https://doi.org/10.1163/157338302X00062 Dalton, B., 2019, ‘Hermit kingdom, nuclear nation … If the media keep calling North Korea names, it will only prolong the conflict’, The Conversation. Ginkel, A.V., 1982, ‘Goudzwaard expose today’s idol’, Perspective (Institute for Christian Studies) 16(2), 1–4. Gregor, A.J., 2012, Totalitarianism and political religion: An intellectual history, Stanford University Press. The author reported that there are similarities between Juche ideology and Confucianism. Initially, they started as philosophies, although later converted to religion for their selfish political reasons. Juche has become a major religious force that has manoeuvered North Korea to replace traditional Marxist communism. ffis, W.E., 2009, Corea, the hermit nation (1882), Cambridge Scholars Publish Grzelczyk, V., 2012, ‘In the name of the father, son, and grandson: Succession patterns and the Kim dynasty’, The Journal of Northeast Asian History 9(2), 33–68. GÜVEN, E., 2019, ‘The Juche System and the DPR Korea Media as Official Mouthpiece of the Kim Family’, Global Media Journal TR Edition 10(19), 194–215. Jang-yop, H., 1999, I saw the truth of the history, Han-Yul. Kementerian Unifikasi Korsel, Institute for Military History dan Agency for KIA Recovery & Identification di Kementerian Pertahanan Korea Selatan, dan B.S.N.K., 2020, Indeologi ‘Juche’ (mandiri secara swasembada), Semenanjung Korea, A to Z. Funding information This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. Kim Il-Sung initially opposed religion because Kim believed it is superstitious to either believe in Jesus or Buddhism. Historically, religion has been dominated by the ruling class and is used as a tool to deceive, exploit and oppress the people. In the modern era, imperialism was used as an ideological tool to restrict backwardness. Disclaimer The views and opinions expressed in this article are those of the authors and do not necessarily reflect the official policy or position of any affiliated agency of the authors. The North Koreans were made to worship Kim with Juche as the political philosophy of a ruling government. However, it was transformed into religion because the people believed that the ruler (Kim Il Sung) is not elected; instead, he is a supreme human. Competing interests The authors declare that they have no financial or personal relationships that may have inappropriately influenced them in writing this article. Data availability Data sharing is not applicable to this article as no new data were created or analysed in this study. Acknowledgements The authors would like to thank Winson Simamora, Nelsong Sembiring, Wellhelem Manuhutu and Berliando Lumban Khazaniv, A.M., 2008, ‘Marxism-Leninism as a secular religion’, in R. Griffin, R. Mallett & J. Tortorice (eds.), The sacred in twentieth century politics: Essays in honour of professor Stanley G. Payne, pp. 119–142, Palgrave Macmillan. http://www.hts.org.za http://www.hts.org.za Open Access http://www.hts.org.za Original Research Page 7 of 7 Newbigin, L., 1989, The Gospel in a pluralist society, Eerdmans Publishing Co. Kihl, Y.W. & Kim, H.N., 2015, North Korea: The politics of regime survival, 2nd edn., Routledge. Newbigin, L., 1989, The Gospel in a pluralist society, Eerdmans Publishing Co. Park, C.H., 2008, ‘The traditional morality of totalitarianism – Analysis of Juche ideology through honoring parents’, in Proceedings of the 52nd annual meeting of the ISSS-2008, pp. 1–11. Kim, I.S., 1992, Kim Il-Sung heads of state, Korea (North), biography: Korean resistance movements, 1905–1945, p. 3447, Foreign Languages Publishing House (English edn.). Park, Y.S., 2014, ‘Policies and ideologies of the Kim Jong-un regime in North Korea: Theoretical implications’, Asian Studies Review 38(1), 1–14. https://doi.org/10.10 80/10357823.2013.868864 Kim, K-U., 2003, Studies of North Korean political history (History of establishment), Sonin. Kim, S., 2017, Authority and emotions: Kim Jong Il and religious imagination in North Korean literature, Harvard. Ponomareva, E. & Rudov, G., 2015, ‘Russia–North Korea: State of affairs and trends’, Journal of Asian Public Policy 9(1), 45–56. https://doi.org/10.1080/17516234.201 5.1122716 Kim, T., 1998, Korean Shamanism: Muism, Jimoondang. Schafer, M., 2004, ‘Luigi Sturzo as a theorist of totalitarianism’, in H. Maier (ed.), Totalitarianism and political religions, I, pp. 21–30, Routledge, Taylor, and Francis. Kurbanov, S.O., 2019, ‘North Korea’s Juche ideology: Indigenous communism or traditional thought?’, Critical Asian Studies 51(2), 296–305. https://doi.org/10.10 80/14672715.2019.1566750 Seth, M.J., 2018, North Korea: A history, Red Globe Press. Lamport, M.A., 2018, ‘North Korea’, in Encyclopedia of Christianity in the global south, 2nd edn., p. 1122, Rowman & Littlefield, London. Widjaja, F.I., 2019, ‘Pluralitas dan Tantangan Misi: Kerangka Konseptual untuk Pendidikan Agama Kristen dalam Masyarakat Majemuk’, Pendidikan Agama Kristen Regula Fidei 4(1), 591. https://doi.org/10.33541/jrfvol1iss1pp115 Lee, G., 1997, ‘The Political Philosophy of Juche’, Stanford Journal of East Asian Affairs 3(1), 105–112. Widjaja, F.I. & Boiliu, N.I., 2019, Misi dan Pluralitas Keyakinan di Indonesia, Andi Offset. Lee, P., 2009, Toward a missiological understanding of the persecuted church in North Korea, Fuller Theological Seminary. Widjaja, F.I., Simanjuntak, F. & Boiliu, N.I., 2020, ‘Repositioning mission in postmodern culture’, 414 (Iceshe 2019), pp. 189–193. https://doi.org/10.2991/assehr.k. 200311.038 Lee, P., 2018, Peeling the onion formation of North Korean Juche as a religion. Linz, J.J., 2004, ‘The religious use of politics and/or the political use of religion: Ersatz ideology versus Ersatz religion’, in H. Maier (ed.), Totalitarian movements and political religions, I, pp. 102–119, Routledge, Taylor, and Francis. Young, C., 2013, ‘Into the sunset: Ch’ŏndogyo in North Korea, 1945–1950’, Journal of Korean Religions 4(2), 51–66. https://doi.org/jkr.2013.0010 http://www.hts.org.za Open Access
https://openalex.org/W1980774504	https://vtechworks.lib.vt.edu/bitstream/10919/48972/1/journal_pone_0005797.pdf	English	null	Genomic Characterization of Methanomicrobiales Reveals Three Classes of Methanogens	PloS one	2,009	cc-by	8,079	Abstract Background: Methanomicrobiales is the least studied order of methanogens. While these organisms appear to be more closely related to the Methanosarcinales in ribosomal-based phylogenetic analyses, they are metabolically more similar to Class I methanogens. Methodology/Principal Findings: In order to improve our understanding of this lineage, we have completely sequenced the genomes of two members of this order, Methanocorpusculum labreanum Z and Methanoculleus marisnigri JR1, and compared them with the genome of a third, Methanospirillum hungatei JF-1. Similar to Class I methanogens, Methanomicrobiales use a partial reductive citric acid cycle for 2-oxoglutarate biosynthesis, and they have the Eha energy-converting hydrogenase. In common with Methanosarcinales, Methanomicrobiales possess the Ech hydrogenase and at least some of them may couple formylmethanofuran formation and heterodisulfide reduction to transmembrane ion gradients. Uniquely, M. labreanum and M. hungatei contain hydrogenases similar to the Pyrococcus furiosus Mbh hydrogenase, and all three Methanomicrobiales have anti-sigma factor and anti-anti-sigma factor regulatory proteins not found in other methanogens. Phylogenetic analysis based on seven core proteins of methanogenesis and cofactor biosynthesis places the Methanomicrobiales equidistant from Class I methanogens and Methanosarcinales. Conclusions/Significance: Our results indicate that Methanomicrobiales, rather than being similar to Class I methanogens or Methanomicrobiales, share some features of both and have some unique properties. We find that there are three distinct classes of methanogens: the Class I methanogens, the Methanomicrobiales (Class II), and the Methanosarcinales (Class III). Citation: Anderson I, Ulrich LE, Lupa B, Susanti D, Porat I, et al. (2009) Genomic Characterization of Methanomicrobiales Reveals Three Classes of Methanogens. PLoS ONE 4(6): e5797. doi:10.1371/journal.pone.0005797 Editor: Niyaz Ahmed, University of Hyderabad, India Received March 30, 2009; Accepted May 7, 2009; Published June 4, 2009 Received March 30, 2009; Accepted May 7, 2009; Published June 4, 2009 Copyright: 2009 Anderson et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Copyright: 2009 Anderson et al. This is an open-access article distributed under the terms of the Creative Commons unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This work was performed under the auspices of the US Department of Energy’s Office of Science, Biological and Environmental Research Program, and by the University of California, Lawrence Berkeley National Laboratory under Contract No. Abstract DE-AC02-05CH11231, Lawrence Livermore National Laboratory under Contract No. DE-AC52-07NA27344, and Los Alamos National Laboratory under Contract No. DE-AC02-06NA25396. L. E. U. was supported by grant number GM72285 from the National Institutes of Health. B. L., I. P., M. S.-L., and W. B. W. were supported by DOE contract number DE-FG02-97ER20269. D. S., L. D., and B. M. were supported by NASA Astrobiology, Exobiology, and Evolutionary Biology grant NNG05GP24G. M. L. was supported by the Department of Energy under contract DE-AC05-000R22725. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Department of Energy Office of Biological and Environmental Research: http://www.er.doe.gov/ober/ National Institutes of Health: http://www.nih.gov National Aeronautics and Space Administration http://www.nasa.gov Competing Interests: The authors have declared that no competing interests exist. * E-mail: IJAnderson@lbl.gov ological and Environmental Sciences Directorate, Oak Ridge National Laboratory, Oak Ridge, Tennessee, United States of America ¤ Current address: Biological and Environmental Sciences Directorate, Oak Ridge National Laboratory, Oak Ridge, Tennessee, United Genomic Characterization of Methanomicrobiales Reveals Three Classes of Methanogens Iain Anderson1*, Luke E. Ulrich2, Boguslaw Lupa3, Dwi Susanti4,7, Iris Porat3¤, Sean D. Hooper1, Athanasios Lykidis1, Magdalena Sieprawska-Lupa3, Lakshmi Dharmarajan4,7, Eugene Goltsman1, Alla Lapidus1, Elizabeth Saunders8, Cliff Han8, Miriam Land9, Susan Lucas1, Biswarup Mukhopadhyay4,5,6,7, William B. Whitman3, Carl Woese10, James Bristow1, Nikos Kyrpides1 1 Joint Genome Institute, Walnut Creek, California, United States of America, 2 Joint Institute for Computational Sciences, University of Tennessee – Oak Ridge National Laboratory, Oak Ridge, Tennessee, United States of America, 3 Department of Microbiology, University of Georgia, Athens, Georgia, United States of America, 4 Virginia Bioinformatics Institute, Bioinformatics and Computational Biology Graduate Program, Virginia Polytechnic Institute and State University, Blacksburg, Virginia, United States of America, 5 Department of Biochemistry, Bioinformatics and Computational Biology Graduate Program, Virginia Polytechnic Institute and State University, Blacksburg, Virginia, United States of America, 6 Department of Biological Sciences, Bioinformatics and Computational Biology Graduate Program, Virginia Polytechnic Institute and State University, Blacksburg, Virginia, United States of America, 7 Department of Genetics, Bioinformatics and Computational Biology Graduate Program, Virginia Polytechnic Institute and State University, Blacksburg, Virginia, United States of America, 8 Joint Genome Institute, Los Alamos National Laboratory, Bioscience Division, Los Alamos, New Mexico, United States of America, 9 Bioscience Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee, United States of America, 10 Department of Microbiology, University of Illinois, Urbana, Illinois, United States of America Genomes of Methanomicrobiales Genomes of Methanomicrobiales thermophiles, and thermoacidophiles. Recently a third kingdom, Thaumarchaeota, has been proposed that includes mesophilic organisms previously classified as Crenarchaeota [3]. This is the first publication to describe genomes from the order Methanomicrobiales. We report here the genome of the marine methanogen Methanoculleus marisnigri JR1 [11] and that of Methanocorpusculum labreanum Z, a methanogen isolated from tar pit sediments [12]. We include comparisons of these two with the genome sequence of Methanospirillum hungatei JF-1, a spiral-shaped methanogen isolated from sewage sludge [13]. We also present a comparative analysis of Methanomicrobiales genomes with those of Methanosarcinales and Class I methanogens. Methanogens play a major role in the global carbon cycle [4] by carrying out the final steps in the anaerobic degradation of organic material. In the process, they are estimated to produce close to 400 million metric tons of methane per year. Much of the methane is converted back to carbon dioxide by methanotrophs, but some is released to the atmosphere where it is a potent greenhouse gas. As a result of human activities, the concentration of methane in the atmosphere has almost tripled in the last 200 years [5]. Methanogens are currently classified in five orders: Methano- bacteriales, Methanococcales, Methanopyrales, Methanomicro- biales, and Methanosarcinales. General features The genomes of M. labreanum and M. marisnigri consist of one chromosome and no plasmids (Table 1), and the same is true for M. hungatei. The size of the M. hungatei genome is substantially larger than those of the other two. M. marisnigri has only one ribosomal RNA (rRNA) operon, while M. labreanum has three and M. hungatei has four. In two of the M. hungatei rRNA operons, there are two copies of the 5S rRNA. It has been recognized that the methanogens can be divided into two major groups based on phylogenetic analysis [6,7]. The first group contains the orders Methanobacteriales, Methanococ- cales, and Methanopyrales, and has been named Class I methanogens by Bapteste et al. [7]. The second group, the Class II methanogens, includes Methanomicrobiales and Methanosarci- nales. However, the Methanomicrobiales are physiologically more similar to the Class I methanogens than to the Methanosarcinales, growing on H2/CO2 or formate, while members of the Methanosarcinales can produce methane from acetate, methanol, methylamines, and other C-1 compounds. Recently Thauer et al. [8] have argued that methanogens can be divided into two groups based on the presence or lack of cytochromes, with Methano- sarcinales alone possessing cytochromes. The Methanomicrobiales thus belong to the phylogenetic group of Class II methanogens, but to the physiological group of methanogens without cyto- chromes. Introduction kingdoms, the Crenarchaeota and the Euryarchaeota [2]. The Crenarchaeota consist mainly of thermophiles and thermoacido- philes while the Euryarchaeota contains a wider variety of organisms including the methanogens, the extreme halophiles, The Archaea were discovered to form a distinct domain in 1977 [1] and subsequently were found to be comprised of two major PLoS ONE \| www.plosone.org June 2009 \| Volume 4 \| Issue 6 \| e5797 PLoS ONE \| www.plosone.org 1 June 2009 \| Volume 4 \| Issue 6 \| e5797 Methanogenesis Frh Mvh Eha Ehb Ech Mbh Class I methanogens All All all except Msp all except Mka Methanosarcinales Mac, Mba, Mmz Mba, Mmz Methanomicrobiales All Mmar All All Mlab, Mhun Frh: F420-reducing hydrogenase; Mvh: F420-non-reducing hydrogenase; Eha: energy-converting hydrogenase A; Ehb: energy-converting hydrogenase B; Ech: energy- converting hydrogenase; Mbh: membrane-bound hydrogenase; Msp: Methanosphaera stadtmanae; Mka: Methanopyrus kandleri; Mac: Methanosarcina acetivorans; Mba: Methanosarcina barkeri; Mmz: Methanosarcina mazei; Mmar: Methanoculleus marisnigri; Mlab: Methanocorpusculum labreanum; Mhun: Methanospirillum hungatei. doi:10.1371/journal.pone.0005797.t002 Frh Mvh Eha Ehb Ech Mbh Class I methanogens All All all except Msp all except Mka Methanosarcinales Mac, Mba, Mmz Mba, Mmz Methanomicrobiales All Mmar All All Mlab, Mhun Frh: F420-reducing hydrogenase; Mvh: F420-non-reducing hydrogenase; Eha: energy-converting hydrogenase A; Ehb: energy-converting hydrogenase B; Ech: energy- converting hydrogenase; Mbh: membrane-bound hydrogenase; Msp: Methanosphaera stadtmanae; Mka: Methanopyrus kandleri; Mac: Methanosarcina acetivorans; Mba: Methanosarcina barkeri; Mmz: Methanosarcina mazei; Mmar: Methanoculleus marisnigri; Mlab: Methanocorpusculum labreanum; Mhun: Methanospirillum hungatei. doi:10.1371/journal.pone.0005797.t002 Frh: F420-reducing hydrogenase; Mvh: F420-non-reducing hydrogenase; Eha: energy-converting hydrogenase A; Ehb: energy-converting hydrogenase B; Ech: energy- converting hydrogenase; Mbh: membrane-bound hydrogenase; Msp: Methanosphaera stadtmanae; Mka: Methanopyrus kandleri; Mac: Methanosarcina acetivorans; Mba: Methanosarcina barkeri; Mmz: Methanosarcina mazei; Mmar: Methanoculleus marisnigri; Mlab: Methanocorpusculum labreanum; Mhun: Methanospirillum hungatei. doi:10.1371/journal.pone.0005797.t002 F420-dependent secondary alcohol dehydrogenase from Methano- culleus thermophilus, but M. labreanum and M. hungatei do not (no BLAST hit with cutoff of 10210). found only in Class I methanogens. The other two genomes lack genes assigned to COG4074 and thus are unlikely to have this enzyme. The enzyme functions under conditions of nickel limitation (reviewed in [18]), so this suggests that M. labreanum can tolerate lower environmental nickel concentrations. When this gene is found in a Class I methanogen genome, it is often accompanied by one or two paralogs of unknown function belonging to COG4007. However, M. labreanum lacks these paralogs. In Class I methanogens, F420-non-reducing hydrogenase provides electrons to heterodisulfide reductase, and its D subunit interfaces with heterodisulfide reductase [17]. In all three Methanomicrobiales the gene for the D subunit of the hydroge- nase (Mlab_0242, Memar_0622, Mhun_1839) is adjacent to the genes for heterodisulfide reductase, but phylogenetic analysis of hydrogenase alpha subunits (not shown) reveals that only M. marisnigri possesses the F420-non-reducing hydrogenase (Memar_1007–1008) (Table 2). Apparently, M. labreanum and M. hungatei use a different source of electrons for their heterodisulfide reductase. Methanogenesis Based on the lack of F420-nonreducing hydrogenase and the fact that the Eha hydrogenase is located adjacent to formylmethanofuran dehydrogenase (Fmd) (see below), we pro- pose that, in at least some Methanomicrobiales, Fmd and heterodisulfide reductase are linked to transmembrane proton or sodium ion transport (Figure 1) rather than flavin-based electron bifurcation as proposed by Thauer et al. [8]. Methanogenesis As expected, the three Methanomicrobiales have all of the genes required for methanogenesis from hydrogen and carbon dioxide. All three species are capable of utilizing formate, and they have formate transporters as well as cytosolic formate dehydrogenases that probably reduce coenzyme F420. No homologs were found to C-1 compound:corrinoid methyltransferases, corrinoid proteins, and methylcobalamin:Coenzyme M methyltransferases involved in methanogenesis from methanol and methylamines (no BLAST hit to Methanosarcina acetivorans proteins with 1025 cutoff value). Some methanogens, including M. marisnigri, can utilize secondary alcohols as electron donors for methanogenesis [10], whereas M. hungatei JF-1 can not [14], and M. labreanum has not been tested. Alcohol dehydrogenases that oxidize secondary alcohols and use the electrons to reduce coenzyme F420 have been characterized [15] and the structure has been determined for one enzyme [16]. M. marisnigri has a gene (Memar_0783) that is closely related to the Members of the order Methanomicrobiales have few known unique properties. However their membrane lipid composition is distinctive, and they are unique in possessing aminopentanetetrols in their lipids (reviewed in [9]). In addition to growth on H2/CO2 or formate, some are capable of using secondary alcohols as electron donors [10]. Methanomicrobiales have been detected in marine environments, in landfills and wastewater reactors, and as symbionts of ciliates (reviewed in [9]). Table 1. General genome statistics. M. labreanum M. marisnigri M. hungatei Genome size (bp) 1,804,962 2,478,101 3,544,738 G+C content (bp) 902,600 (50.0%) 1,537,981 (62.1%) 1,600,415 (45.1%) Number of genes 1828 2559 3305 RNA genes 63 (3.4%) 53 (2.1%) 66 (2.0%) Protein-coding genes 1765 (96.6%) 2506 (97.9%) 3239 (98.0%) Pseudogenes 26 (1.4%) 17 (0.7%) 99 (3.0%) Genes in ortholog clusters 1676 (91.7%) 2294 (89.6%) 3031 (91.7%) Genes assigned to COGs 1358 (74.3%) 1832 (71.6%) 2314 (70.0%) Genes assigned to Pfam domains 1335 (73.0%) 1790 (69.9%) 2326 (70.4%) Genes with signal peptides 406 (22.2%) 620 (24.2%) 771 (23.3%) Genes with transmembrane helices 368 (20.1%) 595 (23.3%) 762 (23.1%) Fusion genes 73 (4.0%) 104 (4.1%) 171 (5.2%) Transposable elements 0 3 76 CRISPR-associated genes 8 1 21 CRISPR repeat arrays 1 0 6 doi:10.1371/journal.pone.0005797.t001 Table 1. General genome statistics. Table 1. General genome statistics. June 2009 \| Volume 4 \| Issue 6 \| e5797 PLoS ONE \| www.plosone.org 2 Genomes of Methanomicrobiales Table 2. Hydrogenases in methanogen genomes. Membrane-Bound Hydrogenases Methanogens have several families of membrane-bound hydrogenases that participate in various processes including methanogenesis and biosynthesis (reviewed in [19]). These hydrogenases are encoded by a core of conserved genes that includes from six to more than 20 subunits. The three Methanomicrobiales genomes encode two to three membrane- bound hydrogenases (Table 2). All three possess the genes for a membrane-bound hydrogenase similar to that encoded by the Methanothermobacter thermautotrophicus eha operon (Memar_1172– 1185, Mlab_0561–0573, Mhun_2094–2106). Their genes for the enzyme subunits are in the same order as those in the M. M. labreanum has a hydrogen-forming methylene-tetrahydro- methanopterin dehydrogenase (COG4074), an enzyme previously Figure 1. Proposed pathway for methanogenesis in Methanomicrobiales. Methanomicrobiales are predicted to couple formylmethanofuran formation and CoM-CoB heterodisulfide reduction to ion gradients. Fd: ferredoxin; MF: methanofuran; H4MPT: tetrahydromethanopterin. doi:10.1371/journal.pone.0005797.g001 Figure 1. Proposed pathway for methanogenesis in Methanomicrobiales. Methanomicrobiales are predicted to couple formylmethanofuran formation and CoM-CoB heterodisulfide reduction to ion gradients. Fd: ferredoxin; MF: methanofuran; H4MPT: tetrahydromethanopterin. doi:10.1371/journal.pone.0005797.g001 June 2009 \| Volume 4 \| Issue 6 \| e5797 PLoS ONE \| www.plosone.org 3 Genomes of Methanomicrobiales Figure 2. Alternate pathways for synthesis of 2-oxoglutarate from oxaloacetate. Class I methanogens and Methanomicrobiales use a partial reductive citric acid cycle while Methanosarcinales use a partial oxidative citric acid cycle. doi:10.1371/journal.pone.0005797.g002 thermautotrophicus eha operon. However, some of the smaller subunits have diverged so extensively that homology can not be detected, and subunits A and M are absent. Adjacent to the hydrogenase operon are genes for the subunits of formylmetha- nofuran dehydrogenase, suggesting that the Eha hydrogenase may reduce the ferredoxin used by this enzyme (Figure 1). All three genomes also have a six-subunit membrane-bound hydrogenase operon similar to Ech hydrogenase (Mlab_1619– 1624, Memar_0359–0364, Mhun_1741–1746), which has multi- ple functions in Methanosarcina barkeri [20]. M. labreanum and M. hungatei, but not M. marisnigri, also have an operon very similar to the mbh operon of Pyrococcus furiosus. Since this hydrogenase is found in the two Methanomicrobiales genomes that lack F420- nonreducing hydrogenase, the Mbh hydrogenase may be involved in heterodisulfide reduction (Figure 1). M. hungatei has another operon similar to membrane-bound hydrogenases (Mhun_1817– 1822). Homologous operons are absent from the other two Methanomicrobiales, but they are found in two Methanosarci- nales, Methanosarcina acetivorans and Methanosarcina mazei. Metabolism and Transport The Embden-Meyerhof pathway is present in many methan- ogens, where it is thought to play a role in the metabolism of stored glycogen. Although M. hungatei and M. marisnigri have putative glycogen phosphorylases (Mhun_1203, Memar_1262, Memar_ 2480), and M. marisnigri has a putative glycogen branching enzyme (Memar_1265), none of the three Methanomicrobiales has an identifiable glycogen synthase. M. marisnigri and M. hungatei appear to encode a complete glycolysis pathway. This suggests that they may be able to utilize glucose from the environment (although they lack identifiable sugar transporters) or that they have a novel glycogen synthase. M. hungatei was previously reported to lack phosphofructokinase activity [21], but the genome contains two putative phosphofructokinase genes (Mhun_0556 and Mhun_ 1465). The Embden-Meyerhof pathway appears to be absent from M. labreanum as it lacks both phosphofructokinase and pyruvate kinase. A gluconeogenesis pathway is present in all three, as it is necessary for biosynthesis of pentoses and hexoses. in M. jannaschii with similarity to conserved region 4 of bacterial sigma factors [23], and the Methanomicrobiales have homologs of three of these proteins (MJ0173, MJ0272, and MJ1243). However, the SpoIIAB anti-sigma factor binds to three separate regions of sigma F [24] corresponding to conserved regions 2, 3, and 4, and regions 2 and 3 are not present in the archaeal proteins. Therefore the targets of these archaeal anti-sigma factors can not be determined from the genome sequence. in M. jannaschii with similarity to conserved region 4 of bacterial sigma factors [23], and the Methanomicrobiales have homologs of three of these proteins (MJ0173, MJ0272, and MJ1243). However, the SpoIIAB anti-sigma factor binds to three separate regions of sigma F [24] corresponding to conserved regions 2, 3, and 4, and regions 2 and 3 are not present in the archaeal proteins. Therefore the targets of these archaeal anti-sigma factors can not be determined from the genome sequence. Membrane-Bound Hydrogenases However, the hydrogenase large subunits of these operons appear to lack the cysteine residues necessary for binding to the nickel-iron center, so these operons may not encode hydrogenases. Figure 2. Alternate pathways for synthesis of 2-oxoglutarate from oxaloacetate. Class I methanogens and Methanomicrobiales use a partial reductive citric acid cycle while Methanosarcinales use a partial oxidative citric acid cycle. doi:10.1371/journal.pone.0005797.g002 Phylogenetic Analysis of Enzymes for Methanogenesis and Cofactor Biosynthesis Bapteste et al. [7] determined the relationships among the various groups of methanogens by generating phylogenetic trees for enzymes of methanogenesis and cofactor biosynthesis. Their analysis found that methanogens could be divided into two groups: Class I and Class II methanogens. We present here an updated analysis that includes additional sequenced genomes. Further- more, the protein-coding genes that we used in the analysis (see Materials and Methods) are present in only one copy per genome. Inclusion of the additional genomes reveals that, surprisingly, Methanomicrobiales are equally distant from Class I methanogens and from the Methanosarcinales (Figure 3). Therefore there appear to be three distinct classes of methanogens: the Class I methanogens, the Methanomicrobiales (that we have termed Class II methanogens), and the Methanosarcinales (that we have termed Class III methanogens). The pathway for 2-oxoglutarate production differs significantly between the Methanosarcinales and the Class I methanogens. Methanosarcinales generate 2-oxoglutarate through a partial oxidative TCA cycle with isocitrate as an intermediate, while Class I methanogens use a partial reductive TCA cycle with succinate as an intermediate (Figure 2). Methanomicrobiales appear to use the partial reductive TCA cycle, similar to the Class I methanogens, as they have genes for all of the necessary enzymes and they lack genes for citrate synthase and isocitrate dehydro- genase. They possess genes encoding the two subunits of the predicted archaeal aconitase [22], but this enzymatic activity has not been verified experimentally. PLoS ONE \| www.plosone.org Sigma Factor Regulators Now that several sequenced genomes from the order Metha- nomicrobiales are available, it is possible to carry out comparative genomic analyses between this order and the other methanogens. We used a protein clustering method to identify and cluster related proteins from 15 species representing Class I, II, and III methanogens (see Materials and Methods for the list of organisms). We then searched for the signature clusters, i.e. clusters of homologous proteins that are present in all members of a phylogenetic group and absent from other groups. Of particular interest are the exclusive signature clusters, those whose member Both M. labreanum and M. marisnigri contain an anti-anti-sigma factor (Memar_02467, Mlab_1451), an anti-sigma factor (Memar_2469, Mlab_1452), and a serine phosphatase (Memar_2468, Mlab_1450) that are similar to the SpoIIAA/ SpoIIAB/SpoIIE components of the Bacillus subtilis sporulation pathway. Moreover, these SpoII-type proteins are also found in M. hungatei, but not outside the order Methanomicrobiales. This finding is intriguing given that no bona fide sigma factors have been identified in Archaea. Kyrpides and Ouzounis identified proteins PLoS ONE \| www.plosone.org June 2009 \| Volume 4 \| Issue 6 \| e5797 4 Genomes of Methanomicrobiales Figure 3. Phylogenetic tree of methanogens based on seven core enzymes of methanogenesis and cofactor biosynthesis. See Materials and Methods for a list of the proteins and organisms included. Protein sequences were concatenated and aligned with Clustal W. The tree was generated with MrBayes 3.1.2 and viewed with TreeView. doi:10.1371/journal.pone.0005797.g003 Figure 3. Phylogenetic tree of methanogens based on seven core enzymes of methanogenesis and cofactor biosynthesis. See Materials and Methods for a list of the proteins and organisms included. Protein sequences were concatenated and aligned with Clustal W. The tree was generated with MrBayes 3.1.2 and viewed with TreeView. doi:10.1371/journal.pone.0005797.g003 three classes, as well as those shared by Methanomicrobiales and Methanosarcinales. proteins are found in all sequenced genomes from only one class. We also identified shared signature clusters (present in only two classes) and common signature clusters (present in all three). Class I. The Class I methanogen exclusive signature clusters include two LSU ribosomal proteins (L34E and L14E) and three enzymes of coenzyme M (CoM) biosynthesis (phosphosulfolactate synthase, phosphosulfolactate phosphatase, and sulfolactate dehydrogenase). This suggests that other methanogens possess either unrelated genes for these enzymes or a different pathway for CoM biosynthesis. Also present in only Class I methanogens is 2- phosphoglycerate kinase, an enzyme used in the synthesis of cyclic 2,3-diphosphoglycerate, which is thought to be a thermoprotectant. Its presence in mesophilic Class I methanogens suggests that it carries out a different function in these organisms. The second enzyme of the pathway, cyclic 2,3-diphosphoglycerate synthetase, is found only in a subset of Class I methanogens and is not part of the signature. We found 413 common signature clusters (Figure 4, Supple- mentary Table S1). These proteins are involved primarily in core information processing and essential metabolic activities (i.e. transcription, translation, methanogenesis, etc.). We found 62 exclusive signature clusters for Methanomicrobiales, 24 for Class I methanogens, and 48 for Methanosarcinales. Given the relatively close phylogenetic relationship between Methanomicrobiales and Methanosarcinales in ribosomal RNA and ribosomal protein- based trees, it is surprising that they share only 33 clusters to the exclusion of the Class I methanogens. While this is more than either class shares with the Class I methanogens, it represents but a very small proportion of the genome. PLoS ONE \| www.plosone.org Methanomicrobiales (Class II). Of the 62 exclusive signature clusters for Methanomicrobiales, 26 are hypothetical proteins, reflecting the fact that this order has been less studied. A serine/threonine kinase and a serine phosphatase, both of which regulate sigma factors in bacteria (see the Sigma Factor Regulators section above), are part of the Methanomicrobiales signature. In addition to a full-length heterodisulfide reductase subunit A (HdrA), Methanomicrobiales also contain a homolog that is truncated at both the N- and C-terminus. Similarly, the A and G subunits of their tetrahydromethanopterin S-methyltransferase are fused. A separate A subunit was found, but no other G subunit is present. Methanomicrobiales (Class II). Of the 62 exclusive signature clusters for Methanomicrobiales, 26 are hypothetical proteins, reflecting the fact that this order has been less studied. A serine/threonine kinase and a serine phosphatase, both of which regulate sigma factors in bacteria (see the Sigma Factor Regulators section above), are part of the Methanomicrobiales signature. In addition to a full-length heterodisulfide reductase subunit A (HdrA), Methanomicrobiales also contain a homolog that is truncated at both the N- and C-terminus. Similarly, the A and G subunits of their tetrahydromethanopterin S-methyltransferase are fused. A separate A subunit was found, but no other G subunit is present. Missing from Class I. There are 33 clusters shared by Methanomicrobiales and Methanosarcinales that are absent from Class I methanogens. Among these are the DNA mismatch repair proteins MutL and MutS. MutH, however, is not present in any methanogen. This suggests that, if Class I methanogens have methyl-directed mismatch repair, they use a different system. Class I methanogens also lack DNA gyrase subunits A and B. This is unexpected as several Class I methanogens were found to be sensitive to coumarins that target bacterial DNA gyrase [29]. Furthermore, DNA gyrase is the only protein known to introduce negative supercoils into DNA, and these are required for many cellular processes including transcription and DNA replication [30]. Another enzyme missing from Class I methanogens is 5- amino-4-imidazolecarboxamide ribonucleotide (AICAR) transformylase in the pathway for de novo purine synthesis. Since most Class I methanogens are autotrophs, they must have this capability provided by a protein unrelated to the known enzyme. Another shared cluster is the one containing Ech hydrogenase subunit A. Although M. acetivorans lacks Ech, it does have the F420H2 dehydrogenase subunit L and a subunit of multisubunit sodium/proton antiporters, both of which cluster with EchA. In the following sections we describe some of the signature proteins associated with each of the PLoS ONE \| www.p June 2009 \| Volume 4 \| Issue 6 \| e5797 PLoS ONE \| www.plosone.org 5 Genomes of Methanomicrobiales Figure 4. Venn diagram of signature clusters. The clusters were generated using a spectral clustering procedure (see Materials and Methods section for details). Signature protein clusters were identified as clusters for which a member protein was present in every analyzed species from one or more classes of methanogens. The number of exclusive, shared, and common signature clusters associated with each methanogen class are shown. The functions of characterized proteins belonging to exclusive signature clusters and to clusters shared between the Methanomicrobiales and the Methanosarcinales are also noted. doi:10.1371/journal.pone.0005797.g004 Figure 4. Venn diagram of signature clusters. The clusters were generated using a spectral clustering procedure (see Materials and Methods section for details). Signature protein clusters were identified as clusters for which a member protein was present in every analyzed species from one or more classes of methanogens. The number of exclusive, shared, and common signature clusters associated with each methanogen class are shown. The functions of characterized proteins belonging to exclusive signature clusters and to clusters shared between the Methanomicrobiales and the Methanosarcinales are also noted. doi:10.1371/journal.pone.0005797.g004 All Class I methanogens also have a homolog of seryl- tRNA(Sec) selenium transferase, used for the synthesis of selenocysteine in bacteria. However, this gene is likely to have a different function in archaea because not all of the Class I methanogens use selenocysteine [25], and those that do utilize a different pathway for selenocysteine synthesis, one that is shared with eukaryotes [26,27]. Experimental testing of this protein found that it did not catalyze selenocysteine formation [28]. cell membranes. Likewise, they encode two proteins involved in DNA compaction: the non-histone chromosomal protein MC1 and a unique variant (,200 amino acids longer) of the ScpB subunit of the condensin complex. That these two chromosome condensation proteins are found only in Methanosarcinales may be related to the larger genome size of some members. The other two components of the condensin complex, ScpA and Smc, are present in most methanogens, including Methanosarcinales. In addition to these DNA condensation proteins, all methanogens have at least one histone gene. Most also have a copy of the gene encoding the Alba protein, but among Methanosarcinales it is present only in Methanosaeta thermophila. PLoS ONE \| www.plosone.org Phylogenetics The sequencing of the genomes of M. labreanum and M. marisnigri reported in this paper, combined with the previously sequenced genome of M. hungatei, has enabled further characterization of the order Methanomicrobiales and clarification of its relationship to other methanogens. Our analyses including these species reveal that the order Methanomicrobiales is clearly distinct from other methanogens. The phylogenetic tree built for seven core methanogenesis and cofactor biosynthesis enzymes reveals three discrete groups of methanogens: the Class I methanogens, the Methanomicrobiales (termed here Class II), and the Methano- sarcinales (termed here Class III). This classification differs significantly from the previous study by Bapteste et al. [7] that divided the methanogens into two major groups. In that earlier study, the order Methanosarcinales was represented by only species from the genus Methanosarcina, whereas our study also included two genomes from other genera. Likewise, their analysis included only one representative of the Methanomicrobiales, while we included four species from this order. Because our study encompassed more species and greater diversity, our results may be a more accurate representation of the relationships among these groups. A relatively close relationship was previously seen between Methanosarcinales and Methanomicrobiales in 16S rRNA trees [3,31] and ribosomal protein trees [3,7]. In contrast, Methanomicrobiales are equally distant from Class I methanogens and Methanosarcinales in the tree built in this study from core methanogenesis proteins. All Methanomicrobiales also contain genes for the Ech hydrogenase that has been characterized in M. barkeri. Ech hydrogenase is involved in reduction of ferredoxin for the first step of methanogenesis from H2/CO2, in the reduction of ferredoxin for biosynthesis, and in the formation of H2 from ferredoxin during aceticlastic methanogenesis [20]. Since the Ech hydroge- nase is found in all three Methanomicrobiales, it is likely that its function is common to all three, e.g. the reduction of ferredoxin for 2-oxoglutarate synthesis. Another putative membrane-bound hydrogenase (Pmh) is found only in M. hungatei where it may perform a function that is unique to this organism, such as producing ferredoxin for acetyl-CoA decarbonylase/synthase, an enzyme that is absent from the other two. Experimental evidence is needed to determine the functions of these hydrogenases. Nevertheless, their distribution within the Methanomicrobiales is clearly distinct from that in the Class I methanogens and the Methanosarcinales (Table 2), supporting the functional and evolutionary uniqueness of this group. Materials and Methods The protein clustering results reported also suggest a significant distance between Methanomicrobiales and all other methanogens. They share only 6 signature clusters with Class I methanogens and 33 with Methanosarcinales. In addition, the number of exclusive signature clusters for the Methanomicrobiales is of the same magnitude as the signatures for the other two groups. The complement of membrane-bound hydrogenases also shows the uniqueness of Methanomicrobiales. They all have the Eha hydrogenase similar to Class I methanogens and the Ech hydrogenase found in Methanosarcinales, while some of them have hydrogenases similar to Mbh from P. furiosus and a putative membrane-bound hydrogenase from Methanosarcinales. Genomes of Methanomicrobiales Genomes of Methanomicrobiales Lastly, Methanomicrobiales and Methanosarcinales have one form of glucose-6-phosphate isomerase (COG2140), while most Class I methanogens use another (COG0166). Since a glucose-6- phosphate isomerase could not be identified in Methanopyrus kandleri or in Methanobacteriales, there is probably a third form of this enzyme. synthesis and heterodisulfide reduction to membrane ion gradi- ents, even though they lack cytochromes and methanophenazine that are present in Methanosarcinales. Hydrogenases Methanomicrobiales encode from two to four membrane-bound hydrogenases. In all three genomes (M. labreanum, M. marisnigri, and M. hungatei), the genes for Eha hydrogenase are found adjacent to genes for formylmethanofuran dehydrogenase (Fmd), suggesting that the Eha hydrogenase may reduce a low potential ferredoxin that is required for the reduction of CO2 to formylmethanofuran. In contrast, in the Class I methanogen Methanococcus maripaludis, the eha and fmd operons are not linked, and Eha hydrogenase presumably plays a role in carbon assimilation similar to Ehb and not methanogenesis [8,32,33]. Phylogenetics Their distribution and other features of the operons suggest that their roles in energy conservation differ in Class I methanogens, Methanosarcinales, and Methanomicrobiales. Methanosarcinales (Class III). Among the exclusive signature proteins found in Methanosarcinales are subunits A, K, and N of reduced coenzyme F420 (F420H2) dehydrogenase. Since only Methanosarcinales can use methyl compounds as a substrate for methanogenesis, it is not surprising that this enzyme, used for growth on methyl compounds, is not found in the other methanogens. All methanogens have the archaeal bisphosphoglycerate-independent phosphoglycerate mutase, but Methanosarcinales also have a bacterial version. Similarly, Methanosarcinales use the bacterial adenylate kinase while other methanogens have the archaeal enzyme. Methanosarcinales exclusive signature proteins include phage shock protein A, a protein that functions in the repair of damaged Methanosarcinales (Class III). Among the exclusive signature proteins found in Methanosarcinales are subunits A, K, and N of reduced coenzyme F420 (F420H2) dehydrogenase. Since only Methanosarcinales can use methyl compounds as a substrate for methanogenesis, it is not surprising that this enzyme, used for growth on methyl compounds, is not found in the other methanogens. All methanogens have the archaeal bisphosphoglycerate-independent phosphoglycerate mutase, but Methanosarcinales also have a bacterial version. Similarly, Methanosarcinales use the bacterial adenylate kinase while other methanogens have the archaeal enzyme. Methanosarcinales exclusive signature proteins include phage shock protein A, a protein that functions in the repair of damaged PLoS ONE \| www.plosone.org June 2009 \| Volume 4 \| Issue 6 \| e5797 6 Supporting Information Table S1 Signature clusters of methanogens. List of signature clusters including exclusive clusters that are present in one class only, shared clusters that are present in two classes, and common clusters that are present in all three classes. p Found at: doi:10.1371/journal.pone.0005797.s001 (0.11 MB TXT) Genome Analysis Automatic genome annotation was performed at Oak Ridge National Laboratory. Genes were identified using a combination of Critica [38] and Glimmer [39]. In addition, predicted coding regions (CDSs) were manually curated using JGI’s Gene-PRIMP Quality Assurance pipeline (http://tunis.jgi-psf.org/geneprimp) (Pati et al., in preparation). Comparative genome analysis was performed within the Integrated Microbial Genomes (IMG) system [40]. CRISPR repeats were identified with the CRISPR Recognition Tool [41]. Author Contributions Conceived and designed the experiments: CW JB NCK. Performed the experiments: SDH AL EG AL ES CSH MLL SL. Analyzed the data: IA LEU BL DS IP LD BM WW NCK. Contributed reagents/materials/ analysis tools: MSL. Wrote the paper: IA LEU BL DS IP BM WW NCK. For protein clustering, methanogens were included from all three groups: six Class I methanogens, four Methanomicrobiales, and five Methanosarcinales. Class I included Methanocaldococcus Accession Numbers The genome sequences of Methanoculleus marisnigri JR1, Metha- nocorpusculum labreanum Z, and Methanospirillum hungatei JF-1 can be accessed in GenBank (CP000562, CP000559, and CP000254, respectively). The Genomes OnLine Database accession numbers are Gc00512, Gc00506, and Gc00350, respectively. A phylogenetic tree was constructed using the concatenated sequences of seven core proteins found in all methanogens and involved in methanogenesis and cofactor biosynthesis. The genes included are F420-dependent methylenetetrahydromethanopterin dehydrogenase (mtd, COG1927), tetrahydromethanopterin:coen- zyme M methyltransferase subunits B (mtrB, COG4062), C (mtrC, COG4061), D (mtrD, COG4060) and E (mtrE, COG4059), FO synthase subunit 1 (cofG), and sulfopyruvate decarboxylase alpha subunit (comD). Protein sequences were downloaded from IMG [40]. The concatenated amino acid sequences were aligned with Clustal W [42], and the tree was generated with MrBayes 3.1.2 [43] with 1,000,000 generations sampled every 100 generations. The first 250,000 generations were discarded as burn-in. The tree was visualized with TreeView [44]. Genome Sequencing and Assembly Genome Sequencing and Assembly The genome of M. labreanum Z was sequenced at the Joint Genome Institute (JGI) using a combination of Sanger shotgun sequencing and 454 sequencing-by-synthesis technology. All general aspects of library construction and sequencing performed at the JGI can be found at http://www.jgi.doe.gov/sequencing/ protocols/prots_production.html. Draft assemblies were based on 26,432 Sanger shotgun and 390,106 pyrosequencing reads. The combined reads provided 346 coverage of the genome. The Newbler assembly software (www.454.com) and the Paracel Genome Assembler (Paracel, Pasadena, CA) were used for fragment assembly, and the Consed finishing package (www. phrap.org) was used for quality assessment and editing. All mis- assemblies were corrected and all gaps between contigs were closed by custom primer walk using subclones or PCR products as templates. A total of 196 additional reactions were run to close gaps and to raise the quality of the finished sequence. The genome of M. marisnigri JR1 was sequenced at the Joint Genome Institute (JGI) using a combination of 3 kb, 7 kb and 36 kb (fosmid) DNA libraries. Draft assemblies were based on 29,769 total reads. The three libraries combined provided 116 coverage of the genome. The Phred/Phrap/Consed software package (www.phrap.com) was used for sequence assembly and quality assessment [35–37]. All mis-assemblies were corrected and all gaps between contigs were closed by custom primer walk using subclones or PCR products as templates. A total of 702 primer walk reactions, PCR end reads and 3 mini-libraries were required to close gaps and to raise the quality of the finished sequence. The second eigenvalue of the transition matrix is a measure of how easily a graph (i.e. a cluster) can be partitioned. A cutoff value of 0.8 was applied; if the second eigenvalue exceeds 0.8, the cluster is further partitioned. This approach provides a relatively flexible partitioning that can reveal protein similarities despite sequence differences due to phylogenetic distance. Signature protein clusters were identified as clusters for which a member protein was present in every analyzed species from one (or more) class of methanogens. Those clusters were binned into groups: exclusive signature clusters found in all members of only one class, shared signature clusters found in all members of a specified pair of classes, and common clusters found in all three classes. The resultant cluster distribution was visualized as a Venn diagram. 4. Ferry JG (1997) Methane: small molecule, big impact. Science 278: 1413– 1414. 5. Etheridge DM, Steele LP, Francey RJ, Langenfelds RL (1998) Atmospheric methane between 1000 A.D. and present: evidence of anthropogenic emissions and climatic variability. J Geophys Res 103: 15979–15993. 6. Fox GE, Magrum LJ, Balch WE, Wolfe RS, Woese CR (1977) Classification of methanogenic bacteria by 16S ribosomal RNA characterization. Proc Natl Acad Sci U S A 74: 4537–4541. DNA Preparation M. marisnigri strain JR1 was obtained from the ATCC (ATCC 35101). It was cultured at room temperature in modified McC medium [34] that contained 0.1 M NaCl, 3 g/L of sodium bicarbonate, 2 g/L of Trypticase (replacing yeast extract), and 0.17 g/L of Na2S?9H2O. M. labreanum strain Z was obtained from the ATCC (ATCC 43576). It was cultured at 37uC in MS-OCM Base Medium with 2.5 g/L NaCl, 5 mM sodium acetate, 50 mM sodium formate, and 2.5% (v/v) of rumen fluid. , ( ) For DNA isolation, cells were suspended in TE buffer (10 mM Tris, 1 mM EDTA, pH 8.0). Sodium dodecyl sulfate was added to a final concentration of 0.5% and proteinase K was added to make 100 micrograms/ml, then the solution was incubated at 37uC for 1 hour. After adding NaCl to 0.5 M concentration, the solution was approximately 0.9 ml. Next, 0.5 ml chloroform:isopropyl alcohol (24:1) was added. The solution was mixed and then centrifuged at 13,0006g for 10 minutes. The aqueous phase was transferred to a new tube, combined with 0.5 ml phenol:chloroform:isoamyl alcohol (25:24:1), mixed, and centrifuged at 13,0006g for 10 minutes. The aqueous phase was collected, combined with 0.6 ml isopropanol, incubated at room temperature for 30 minutes, and then centrifuged at 13,0006g for 5 minutes. The pellet was washed with 70% ethanol, resuspended in TE+RNAse (100 micrograms/ ml), and incubated at 37uC for 20 minutes. Methanomicrobiales share some capabilities with Class I methanogens to the exclusion of Methanosarcinales. Both groups are capable of using only H2/CO2 or formate for methanogenesis. The genomes show that they also share the pathway for 2- oxoglutarate synthesis. Both use a partial reductive TCA cycle, while Methanosarcinales use a partial oxidative TCA cycle. This could reflect the observations that Methanomicrobiales efficiently use low concentrations of H2, while the Methanosarcinales dominate in environments in which acetate is plentiful. The partial oxidative TCA cycle results in the loss of one carbon as CO2, therefore the use of the reductive cycle by Methanomicro- biales and Class I methanogens would be predicted to preserve more fixed carbon. On the other hand, we propose that, similar to Methanosarcinales, Methanomicrobiales link formylmethanofuran P June 2009 \| Volume 4 \| Issue 6 \| e5797 PLoS ONE \| www.plosone.org 7 p y g 2. Woese CR, Kandler O, Wheelis ML (1990) Towards a natural system of organisms: proposal for the domains Archaea, Bacteria, and Eucarya. Proc Natl Acad Sci U S A 87: 4576–4579. 1. Woese CR, Fox GE (1977) Phylogenetic structure of the prokaryotic domain: the primary kingdoms. Proc Natl Acad Sci U S A 74: 5088–5090. 3. Brochier-Armanet C, Boussau B, Gribaldo S, Forterre P (2008) Mesophilic crenarchaeota: proposal for a third archaeal phylum, the Thaumarchaeota. Nat Rev Microbiol 6: 245–252. 5. Etheridge DM, Steele LP, Francey RJ, Langenfelds RL (1998) Atmospheric methane between 1000 A.D. and present: evidence of anthropogenic emissions and climatic variability. J Geophys Res 103: 15979–15993. 6. Fox GE, Magrum LJ, Balch WE, Wolfe RS, Woese CR (1977) Classification of methanogenic bacteria by 16S ribosomal RNA characterization. Proc Natl Acad Sci U S A 74: 4537–4541. Genomes of Methanomicrobiales Genomes of Methanomicrobiales jannaschii, Methanothermobacter thermautotrophicus, Methanopyrus kandleri, Methanococcus maripaludis S2, Methanobrevibacter smithii, and Methano- sphaera stadtmanae. Methanomicrobiales included Methanocorpusculum labreanum, Methanoculleus marisnigri, Methanospirillum hungatei, and Candidatus Methanoregula boonei. Methanosarcinales included Methanosarcina acetivorans, Methanosarcina mazei, Methanosarcina barkeri, Methanococcoides burtonii, and Methanosaeta thermophila. Protein sequences for these organisms were downloaded from IMG [40]. We applied a spectral clustering procedure [45,46] to cluster similar proteins based on the topology of their similarity graph, rather than using a fixed threshold value for sequence similarity. The proteins are represented as nodes in a connected undirected graph with edges that carry weights based on node-to-node similarity according to the protein identity. The clustering procedure is analogous to a random walk of a particle moving on this graph from one node to another. In each node the particle moves to another node based on the probabilities corresponding to the weights of the edges. The amount of time the particle spends in a given subgraph will determine whether this is indeed a cluster of its own or not. 4. Ferry JG (1997) Methane: small molecule, big impact. Science 278: 1413– 1414. 1. Woese CR, Fox GE (1977) Phylogenetic structure of the prokaryotic domain: the primary kingdoms. Proc Natl Acad Sci U S A 74: 5088–5090. 2. Woese CR, Kandler O, Wheelis ML (1990) Towards a natural system of organisms: proposal for the domains Archaea, Bacteria, and Eucarya. Proc Natl Acad Sci U S A 87: 4576–4579. 3. Brochier-Armanet C, Boussau B, Gribaldo S, Forterre P (2008) Mesophilic crenarchaeota: proposal for a third archaeal phylum, the Thaumarchaeota. Nat Rev Microbiol 6: 245–252. Genomes of Methanomicrobiales J y 14. Widdel F, Rouvie`re PE, Wolfe RS (1988) Classification of secondary alcohol- utilizing methanogens including a new thermophilic isolate. Arch Microbiol 150: 477–481. p J 33. Lupa B, Hendrickson EL, Leigh JA, Whitman WB (2008) Formate-dependent H2 production by the mesophilic methanogen Methanococcus maripaludis. Appl Env Microbiol 74: 6584–6590. 15. Bleicher K, Winter J (1991) Purification and properties of F420- and NADP+- dependent alcohol dehydrogenases of Methanogenium liminatans and Methanobacter- ium palustre, specific for secondary alcohols. Eur J Biochem 200: 43–51. 34. Whitman WB, Shieh J-S, Sohn S-H, Caras DS, Premachandran U (1986) Isolation and characterization of 22 mesophilic methanococci. Syst Appl Microbiol 7: 235–240. 35. Ewing B, Green P (1998) Base-calling of automated sequencer traces using phred. II. Error probabilities. Genome Res 8: 186–194. p y 16. Aufhammer SW, Warkentin E, Berk H, Shima S, Thauer RK, et al. (2004) Coenzyme binding in F420-dependent secondary alcohol dehydrogenase, a member of the bacterial luciferase family. Structure 12: 361–370. 36. Ewing B, Hillier L, Wendl MC, Green P (1998) Base-calling of automated sequencer traces using phred. I. Accuracy assessment. Genome Res 8: 175–185. 17. Stojanowic A, Mander GJ, Duin EC, Hedderich R (2003) Physiological role of the F420-non-reducing hydrogenase (Mvh) from Methanothermobacter marburgensis. Arch Microbiol 180: 194–203. 37. Gordon D, Abajian C, Green P (1998) Consed: a graphical tool for sequence finishing. Genome Res 8: 195–202. g 38. Badger JH, Olsen GJ (1999) CRITICA: coding region identification tool invoking comparative analysis. Mol Biol Evol 16: 512–524. 18. Shima S, Thauer RK (2007) A third type of hydrogenase catalyzing H2 activation. Chem Rec 7: 37–46. 19. Hedderich R, Forzi L (2005) Energy-converting [NiFe] hydrogenases: more than just H2 activation. J Mol Microbiol Biotechnol 10: 92–104. 39. Delcher AL, Harmon D, Kasif S, White O, Salzberg SL (1999) Improved microbial gene identification with GLIMMER. Nucleic Acids Res 27: 4636–4641. 20. Meuer J, Kuettner HC, Zhang JK, Hedderich R, Metcalf WW (2002) Genetic analysis of the archaeon Methanosarcina barkeri Fusaro reveals a central role for Ech hydrogenase and ferredoxin in methanogenesis and carbon fixation. Proc Natl Acad Sci U S A 99: 5632–5637. 40. Markowitz VM, Szeto E, Palaniappan K, Grechkin Y, Chu K, et al. (2008) The integrated microbial genomes (IMG) system in 2007: data content and analysis tool extensions. Nucleic Acids Res 36: D528–D533. 21. Verhees CH, Tuininga JE, Kengen SWM, Stams AJM, van der Oost J, et al. Genomes of Methanomicrobiales 7. Bapteste E´, Brochier C, Boucher Y (2005) Higher-level classification of the Archaea: evolution of methanogenesis and methanogens. Archaea 1: 353–363. 26. Yuan J, Palioura S, Salazar JC, Su D, O’Donoghue P, et al. (2006) RNA- dependent conversion of phosphoserine forms selenocysteine in eukaryotes and archaea. Proc Natl Acad Sci U S A 103: 18923–18927. 8. Thauer RK, Kaster A-K, Seedorf H, Buckel W, Hedderich R (2008) Methanogenic archaea: ecologically relevant differences in energy conservation. Nat Rev Microbiol 6: 579–591. 27. Xu X-M, Carlson BA, Mix H, Zhang Y, Saira K, et al. (2007) Biosynthesis of selenocysteine on its tRNA in eukaryotes. PLoS Biol 5: e4. selenocysteine on its tRNA in eukaryotes. PLoS Biol 5: e4. y y 28. Kaiser JT, Gromadski K, Rother M, Engelhardt H, Rodnina MV, et al. (2005) 9. Garcia J-L, Ollivier B, Whitman WB (2006) The order Methanomicrobiales. Prokaryotes 3: 208–230. 28. Kaiser JT, Gromadski K, Rother M, Engelhardt H, Rodnina MV, et al. (2005) Structural and functional investigation of a putative archaeal selenocysteine synthase. Biochemistry 44: 13315–13327. 10. Zellner G, Winter J (1987) Secondary alcohols as hydrogen donors for CO2- reduction by methanogens. FEMS Microbiol Lett 44: 323–328. y y 29. Sioud M, Possot O, Elie C, Sibold L, Forterre P (1989) Coumarin and quinolone action in archaebacteria: evidence for the presence of a DNA gyrase-like enzyme. J Bacteriol 170: 946–953. 11. Romesser JA, Wolfe RS, Mayer F, Spiess E, Walther-Mauruschat A (1979) Methanogenium, a new genus of marine methanogenic bacteria, and character- ization of Methanogenium cariaci sp. nov. and Methanogenium marisnigri sp. nov. Arch Microbiol 121: 147–153. 30. No¨llman M, Crisona NJ, Arimondo PB (2007) Thirty years of Escherichia coli DNA gyrase: From in vivo function to single-molecule mechanism. Biochimie 89: 490–499. 12. Zhao Y, Boone DR, Mah RA, Boone JE, Xun L (1989) Isolation and characterization of Methanocorpusculum labreanum sp. nov. from the LaBrea tar pits. Int J Syst Bacteriol 39: 10–13. 31. Schleper C, Jurgens G, Jonuscheit M (2005) Genomic studies of uncultivated archaea. Nat Rev Microbiol 3: 479–488. 13. Ferry JG, Smith PH, Wolfe RS (1974) Methanospirillum, a new genus of methanogenic bacteria, and characterization of Methanospirillum hungatii sp. nov. Int J Syst Bacteriol 24: 465–469. 32. Porat I, Kim W, Hendrickson EL, Xia Q, Zhang Y, et al. (2006) Disruption of the operon encoding Ehb hydrogenase limits anabolic CO2 assimilation in the archaeon Methanococcus maripaludis. J Bacteriol 188: 1373–1380. References PLoS ONE \| www.plosone.org June 2009 \| Volume 4 \| Issue 6 \| e5797 8 Genomes of Methanomicrobiales (2001) ADP-dependent phosphofructokinases in mesophilic and thermophilic methanogenic archaea. J Bacteriol 183: 7145–7153. 41. Bland C, Ramsey TL, Sabree F, Lowe M, Brown K, et al. (2007) CRISPR Recognition Tool (CRT): a tool for automatic detection of clustered regularly interspaced palindromic repeats. BMC Bioinformatics 8: 209. 22. Makarova KS, Koonin EV (2003) Filling a gap in the central metabolism of archaea: prediction of a novel aconitase by comparative-genomic analysis. FEMS Microbiol Lett 227: 17–23. 42. Thompson JD, Higgins DG, Gibson TJ (1994) CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res 22: 4673–4680. 23. Kyrpides NC, Ouzounis CA (1997) Bacterial sigma 70 transcription factor DNA-binding domains in the archaeon Methanococcus jannaschii. J Mol Evol 45: 706–707. 43. Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572–1574. 24. Decatur AL, Losick R (1996) Three sites of contact between the Bacillus subtilis transcription factor sigma F and its antisigma factor SpoIIAB. Genes Dev 10: 2348–2358. 44. Page RDM (1996) TREEVIEW: an application to display phylogenetic trees on personal computers. Comput Appl Biosci 12: 357–358. 44. Page RDM (1996) TREEVIEW: an application to displ 45. Paccanaro A, Casbon JA, Saqi MAS (2006) Spectral clustering of protein sequences. Nucleic Acids Res 34: 1571–1580. 25. Zhang Y, Turanov AA, Hatfield DL, Gladyshev VN (2008) In silico identification of genes involved in selenium metabolism: evidence for a third selenium utilization trait. BMC Genomics 9: 251. q 46. Brewer ML (2007) Development of a spectral clustering method for the analysis of molecular data sets. J Chem Inf Model 47: 1727–1733. PLoS ONE \| www.plosone.org June 2009 \| Volume 4 \| Issue 6 \| e5797 9
https://openalex.org/W3044509874	http://enviro.vgtu.lt/index.php/enviro/2020/paper/download/630/525	English	null	AN INNOVATIVE TOOL FOR THE EVALUATION OF NOx EMISSIONS FROM ROAD TRAFFIC	Environmental engineering	2,020	cc-by	5,428	Introduction Transport-related emissions show a significant growth over the past decade and this trend is expected to continue on the long term. In this respect, decarbonising the transport sector by eliminating fossil fuel could be a very effective policy on the long term. Meanwhile, smaller scale interventions in the urban area are already possible by implementing investments in clean infrastructure (Zhang et al., 2018) or traffic management. Urban areas are acknowledged to have the highest share in generating pollution related to transport and to face the most negative direct impact (Covrig et al., 2016; Roșu et al., 2018). Traffic emissions are included in the background air pollution. However, their impact is very important (Alam et al., 2017; Banica et al., 2017; Condurat, 2016; Tongwane et al., 2015). p ( g ) Specifically, the impact is more significant in areas where there is a large component of heavy traffic (Laña et al., 2016) or where car traffic level ranges from moderate to heavy (Matz et al., 2018). Taking into consideration this specific context, pollutant emissions related to mobility are strictly correlated with the traffic dynamics such as queue, spillback and speed variation (Afotey et al., 2013; Barth & Boriboonsomsin, 2008; Zhao et al., 2017). Sustainable Urban Mobility Plans (SUMPs) (EC, 2013; ELTIS, 2014) are strategic documents for transport planning. They promote more sustainable mobility systems and they analyse the impact of any trends or policies for urban mobility in which the estimation of pollutants is an important project indicator. In some cases, urban regeneration models have been developed (García-Fuentes & de Torre, 2017; Kepaptsoglou et al., 2015) with the main aim to reduce air pollution related to urban mobility (Pérez et al., 2019). There are studies using vehicle speed data for the direct estimation for the calculation of air pollutant emissions in the urban area (Zhao et al., 2017; Mitran & Ilie, 2014; Chang & Lin, 2018; Toșa et al., 2015, Jiang et al., 2017; Gori et al., 2014; Li et al., 2016). Many different road traffic emission models depending on speed analysis have been developed at different level of aggregation in space and time. According to Wang et al. (2018), the main classification includes a) static emission models (aggregated models, average-speed models, traffic situation models) and b) dynamic emission models (regression-based models, modal models, instantaneous models). 11th International Conference “Environmental Engineering” Vilnius Gediminas Technical University Lithuania, 21–22 May 2020 Section: Smart Cities, Roads and Railways http://enviro.vgtu.lt 11th International Conference “Environmental Engineering” Vilnius Gediminas Technical University Lithuania, 21–22 May 2020 Section: Smart Cities, Roads and Railways http://enviro.vgtu.lt eISSN 2029-7092 / eISBN 978-609-476-232-1 Article ID: enviro.2020.630 https://doi.org/10.3846/enviro.2020.630 Article ID: enviro.2020.630 https://doi.org/10.3846/enviro.2020.630 Article ID: enviro.2020.630 https://doi.org/10.3846/enviro.2020.630 An Innovative Tool for the Evaluation of Nox Emissions from Road Traffic Rozalia Melania Boitor 1, Rodica Dorina Cadar 2, Petru Daniel Maran 3, Marco Petrelli 4* 1–3Transport Systems Research Group, Technical University of Cluj-Napoca, Cluj-Napoca, Romania 4Department of Engineering, Roma Tre University, Roma, Italia Received 4 February 2020; accepted 31 March 2020 Rozalia Melania Boitor 1, Rodica Dorina Cadar 2, Petru Daniel Maran 3, Marco Petrelli 4* 1–3Transport Systems Research Group, Technical University of Cluj-Napoca, Cluj-Napoca, Romania 4Department of Engineering, Roma Tre University, Roma, Italia Received 4 February 2020; accepted 31 March 2020 Abstract. In the last years, there was great interest in the development of tools for an effective evaluation of road transport pollutant-related emissions, especially in the urban areas. This paper represents an innovative approach for identifying criticalities about pollutant emissions associated with road traffic and for defining effective policies in order to decrease pollutant emissions. The proposed tool concerns the development of an emission indicator, a proxy measure, which is useful for the assessment of emission problems, based on the use of GPS (Global Positioning System) instantaneous vehicle speed data. The tool can be considered an innovative and adequate solution in many cases in which the development of a valid and robust traffic simulation model, especially DTA (dynamic traffic assignment) is not available in the medium- and short-term horizon. The methodological process concerns the monitoring of road traffic conditions using GPS data from probe vehicles in combination with the use of GIS (Geographic Information System) for the estimation of an emission indicator. The tool has been tested on a real case study in the city of Cluj in Romania for the NOx emissions. The results show the utility of the tool in supporting policy and decision making, due to its ease of application and consistency, especially in defining critical areas. Keywords: NOx emissions, spatial analysis, GPS instantaneous vehicle speed, COPERT model, urban road traffic. Corresponding author. E-mail: marco.petrelli@uniroma3.it Introduction The input data for all these is the traffic assignment model, classified according to temporal dimension: DTA, semi-dynamic traffic assignment and static traffic assignment (STA) models. Corresponding author. E-mail: marco.petrelli@uniroma3.it Copyright © 2020 The Author(s). Published by VGTU Press R. M. Boitor et al. An innovative tool for the evaluation of Nox emissions from road traffic R. M. Boitor et al. An innovative tool for the evaluation of Nox emissions from road traffic Significant advances have been made about the accuracy of the emission models depending on the results of traffic assignment. However, these models need a large amount of data for calibration and validation and in many cases, it is very difficult to implement a satisfactory representation of the car traffic demand. Significant advances have been made about the accuracy of the emission models depending on the results of traffic assignment. However, these models need a large amount of data for calibration and validation and in many cases, it is very difficult to implement a satisfactory representation of the car traffic demand. The aim of this research is to provide a simple but effective tool for the analysis of emission zones and for the evaluation of their impact in order to tackle the problem of reducing air pollutants related to road transport. The proposed methodology develops an emission indicator, a proxy measure, based on GPS instantaneous speed data. This does not represent a tool for estimating pollutant emissions, which implies the computation of emissions based on the knowledge of the level of traffic, but rather an indicator of these emissions. This indicator leads to the evaluation of the impact of different policies, as well as the adoption of control strategies in a simpler but nevertheless robust approach. The approach is based on using data from GPS (Kong et al., 2018) for monitoring road traffic situations, in combination with the use of GIS (Liu et al., 2017). GIS tools permit the definition and representation of the spatial distribution of many variables, the pollutant emission in this case (Cai et al., 2015; Masood et al., 2017; Tenailleau et al., 2016). This paper is a refinement of Boitor et al. (2019) innovative approach of identifying the criticalities of pollutant emissions related to road traffic. It can be considered an adequate solution especially when the development of valid DTA model is not available. It would also be useful for the evaluation of different management actions. Such an approach can be used for any specific pollutant emissions and the case study application deals with the NOx emissions. The numerical application is applied in the case study of the city of Cluj-Napoca, Romania. The area is divided into three different speed-emission zones: Hot emission area, Medium emission area, Cold emission area. R. M. Boitor et al. An innovative tool for the evaluation of Nox emissions from road traffic These zones This paper is a refinement of Boitor et al. (2019) innovative approach of identifying the criticalities of pollutant emissions related to road traffic. It can be considered an adequate solution especially when the development of valid DTA model is not available. It would also be useful for the evaluation of different management actions. Such an approach can be used for any specific pollutant emissions and the case study application deals with the NOx emissions. emissions related to road traffic. It can be considered an adequate solution especially when the development of valid DTA model is not available. It would also be useful for the evaluation of different management actions. Such an approach can be used for any specific pollutant emissions and the case study application deals with the NOx emissions. The numerical application is applied in the case study of the city of Cluj-Napoca, Romania. The area is divided into three different speed-emission zones: Hot emission area, Medium emission area, Cold emission area. These zones were graphically represented using only the NOx indicator values. The numerical application is applied in the case study of the city of Cluj-Napoca, Romania. The area is divided into three different speed-emission zones: Hot emission area, Medium emission area, Cold emission area. These zones were graphically represented using only the NOx indicator values. The paper is structured as follows: Section 1 describes the methodology of the proposed tool, Section 2 presents the case study application while Section 3 shows the results of the application. Finally, conclusions are reported. Copyright © 2020 The Author(s). Published by VGTU Press This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. 1. Methodology The speed – emissions relationship is relevant for both instantaneous as well as average speed (Hu et al., 2018; Li et al. 2019). Instantaneous speed, along with instantaneous emissions are more accurate than average emissions (Ryu et al., 2015) for the representation of real-world situations for speeds below 20 km/h. Speed-emissions correlation is sensitive to varying real-world driving conditions in the urban context (Pathak et al., 2017), where the speed profile is a crucial parameter for fuel consumption and emission estimation, along with other factors such as time, distance, acceleration and deceleration (Zhao et al., 2016). General process of the new tool for the evaluation of NOx emissions from road traffic is reported in Figure 1. Figure 1. General process of the new tool for the evaluation of NOx emissions from road traffic GPS Raw Data GPS Data Filtering Speed filtering algorithm Zero speed (ZS) Moving (S>0) Congestion (ZC) Parking (ZP) GIS Data Analysis Data validation Heat maps of speed data and points' distribution 3D speed distribution Global Moran I and Hot Spot analysis of speed data NOx Emission Indicator Computation of NOx emission indicators Heat maps of NOxemission indicators 3D of NOx emission indicators Hot Spot of NOx indicators in the urban area and on the road network Outcome 3 Levels of Indicators Medium emission area Hot emission area Cold emission area Low Speed - Congestion (LSC) Low Speed - local condition (LSLC) 3 Levels of Indicators Figure 1. General process of the new tool for the evaluation of NOx emissions from road traffic Figure 1. General process of the new tool for the evaluation of NOx emissions from road traffic 2 2 R. M. Boitor et al. An innovative tool for the evaluation of Nox emissions from road traffic Step 1 refers to GPS data elaboration and filtering for the numerical application after the map matching and the extraction of the GPS data for the specific area. GPS data filtering for the numerical application includes the elimination of the parking related points (ZP) which have zero speed (ZS) while maintaining the other points with zero speed, which are related to congestion (ZC). Zero speed congestion points are part of the moving process of vehicles, while parking points are excluded from the analysis. 1. Methodology The process of filtering is also considering the moving points for ensuring that GPS instantaneous data can be considered a reliable proxy measure of the average speed of the traffic flow. For avoiding the overestimation of emission indicator, the low speed points are filtered, excluding the points representing the absence of interference with other vehicles, that generally do not produce reaction with changes in acceleration and deceleration. This is the typical condition of the local roads and they are generally considered low speed - local condition (LSLC) points. Low speed – congestion (LSC) points are part of the moving process of vehicles, while LSLC points are excluded from the analysis. This step allows the use of instantaneous speed data as satisfactory indicators of traffic density, thus considering congestion conditions. y g g Step 2 concerns the GIS component of the application, starting from the creation and analysis of the speed database for the citywide area. Simple visualization of the GPS points and speed weighted points in heat maps are used to check their distribution. Furthermore, the large dataset of GPS points is aggregated using a hexagonal tessellated grid (10,000 square meters hexagons). Specifically, this step involves the analysis of speed distributions and its relationships with different levels of traffic congestion. Spatial distributions are created for identifying some general patterns in the area, which can be validated with spatial statistics indicators, such as Global Moran I and Hot Spot analysis (Getis-Ord Gi). Step 3 deals with the calculation and visualisation of the NOx emissions indicator. The computation of the emission indicator is made using COPERT IV model (Ntziachristos et al., 2009). COPERT is considered the standard emission inventory system and it is used to estimate emissions from road transport in many countries. The hot emissions equations are considered for the calculation of the emission factor for NOx pollutant. The visualization of NOx emissions indicator is conducted using GIS tools for the emission indicator zoning process. The final output of the tool consists of a map showing the emission zones. The analysis of hot spots and cold spots is carried out using Getis-Ord Gi statistic, by calculating a Z-score and a p-value for each feature in the dataset. The three types of emission areas – Cold, Medium, and Hot emission areas – are validated with geo-statistical analysis. For additional details about the spatial analysis and spatial statistics see Boitor et al. (2019). 2. Case study: Cluj-Napoca, Romania The proposed tool is applied in the municipality of Cluj-Napoca, Romania. Cluj-Napoca is a mid-sized city with a population of more than 300,000 inhabitants and the transport system relies mostly on roads for both interurban and intra-urban mobility. In the urban road network, the speed limit is 50 km/h. The database consists of 1,551,135 GPS points collected in the city of Cluj-Napoca, second-by-second between 09.02.2015 – 19.02.2018, using passenger car probe vehicles monitored with Track GPS application. The database contains information regarding time, latitude, longitude, speed and course for every point. 376,699 zero speed (ZS) points are divided into zero parking speeds (ZP) and zero congestion speed (ZC). The result was that 10% of the recorded points are ZP and they are excluded. A similar study (Zhao et al., 2016) presents that 87% of the recordings indicated a moving car and 13% of the recordings indicated that the car had stopped moving or the car had stopped. Furthermore, from the new dataset excluding ZP points, 20% of the points are LSLC points and they are excluded from the analysis. y The spatial distribution of the GPS points in Figure 2a shows a good coverage of the city, considering the road network, the number of observations as well as the spread locations. The speed weighted points’ distribution in Figure 2b shows a very good overlap with the arterial collectors and the main urban streets and it reflects traffic patterns and road traffic congestion areas. The spatial distribution of speed weighted points was further carried out to determine the spatial autocorrelation of speeds in the data set. The Global Moran analysis reported a value of 0.177713 for Global Moran’s Index, a Z-score of 52.4 and a p-value of 0.00000001. The results indicated a strong spatial autocorrelation of the speed values. This enabled the hot spot analysis (Getis-Ord Gi) for more complex pattern analysis. Map in Figure 2c present a more practical perspective of the speed analysis. A honeycomb network is used to aggregate the points for the analysis and some pattern determination. The honeycomb is classified using speed specific percentiles (v25, v50, v75, v85) and the posted speed (similar to v95). 2. Case study: Cluj-Napoca, Romania The speed patterns, resulting from the map can be characterized by three main aspects:  the presence of high-speed corridors (arterial or collector street) in red, located outside the built-up area, where secondary streets give way to main traffic roads;  the presence of high-speed corridors (arterial or collector street) in red, located outside the built-up area, where secondary streets give way to main traffic roads;  the presence of high-speed clusters (red hexagons) located within the built-up area, on high-speed streets/collectors with the right of way and near signalized intersection;  the presence of high-speed clusters (red hexagons) located within the built-up area, on high-speed streets/collectors with the right of way and near signalized intersection;  the presence of low-speed hexagons on the main road corridors close to signalized intersections or roundabouts. 3 R. M. Boitor et al. An innovative tool for the evaluation of Nox emissions from road traffic Figure 2. Spatial distribution of the GPS points in the dataset: a) density of points; b) density of speed weighted points; c) honeycomb speed distribution Figure 2. Spatial distribution of the GPS points in the dataset: a) density of points; b) density of speed weighted points; c) honeycomb speed distribution The numerical application is carried out by using only passenger cars since the percentage of heavy traffic vehicles in the general flow in Cluj-Napoca is very low, largely under 10% based on numerous traffic censuses in different points in the city. Therefore, the study includes only the speed data of the passenger cars. The NOx evaluation in COPERT model is made for the passenger cars disaggregated by type of vehicle and legislation standard. Considering the data from the National Institute for Statistics (2018), at the national level there are 55% of gasoline passenger cars and 45% diesel passenger cars. The average age of the passenger car fleet in the city is related to the emission standard Euro 4. 3. Results of the application/spatial analysis of data The results of the application of the tool in Cluj-Napoca presented in this section are focused on the analysis of the results about the emissions indicator. According to the methodology, GIS-based methods are used, such as heat maps, spatial distributions, and hotspots. GIS representation for emission indicator visualization is conducted using the honeycomb network and different ranges for the classification of the cells. Figure 3a presents the distribution of emissions indicator weighted points. It shows that the high concentration of emissions is found in the central area of the city, while the sparse density covering the urban areas suggest that the points are spread all over the urban built area. High concentration of NOx indicator values is also found in congested intersections and on the road that are not modernized. Therefore, the details about the state of the pavement should be considered for further and more detailed analysis. The Global Moran analysis of the NOx emission factor reports a value of 0.091235 for Global Moran’s Index, a Z-score of 26.9 and a p-value of 0.00000001. The results indicate a strong spatial autocorrelation of the NOx emission indicator values. Figure 3b shows the results of the hot-spot analysis. First, it underlines that, unlike the spatial distribution of speed data, the NOx emission indicator presents high values, correlated to the high NOx emission case, in many places throughout the city, not only limited to the central part of the city, where spread data show the existence of the traffic congestion case. The high emission area distribution is due to the presence of the two worst cases: slow speed for congested traffic and high-speed vehicles with increasing fuel consumption. The classification of the road transport 4 R. M. Boitor et al. An innovative tool for the evaluation of Nox emissions from road traffic system including all its components for the evaluation of the ecological status of these areas was also proposed by Andrei and Condurat (2018). Taking into account this approach, the emission zones in terms of NOx pollutant are a good indicator for identifying the need of intervention. Specifically, the use of Getis- Ord Gi statistic underlines the presence of three different areas according to the point of view of pollutant emissions. 3. Results of the application/spatial analysis of data This adjustment of the results, by filtering and clustering the points, enables a more homogeneous and clear distribution of the emission areas in the urban area model, which is more consistent with real traffic conditions (Figure 4b). Therefore, mapping the areas based on the values of the NOx emission indicator (Figure 4a), it is possible to observe that the proxy indicator allows to obtain results similar to the ones derived from the classic emission models. Figure 3. Computation of NOx emission zones: a) density of NOx emission indicator/heat map; b) hot-spot results of emission indicator Figure 3. Computation of NOx emission zones: a) density of NOx emission indicator/heat map; b) hot-spot results of emission indicator Figure 4. Computation of NOx emission zones: a) NOx emission indicator classification in honeycomb network; b) NOx emission areas classification Figure 4. Computation of NOx emission zones: a) NOx emission indicator classification in honeycomb network; b) NOx emission areas classification Conclusions This section reports the description of the methodology proposed and the results of the numerical application of this to the real case study of the city of Cluj-Napoca. The basis of the method is related to the use of new technologies that allow the collection of an extensive amount of data referring to real vehicle travel history. These data, as also described in the literature, are used for developing tools involving the evaluation of the impact of various policies for decreasing pollutants emissions, especially in urban areas. Therefore, the aim of this study is to propose a simple, but effective tool for the analysis of emission zones and for the evaluation of the impact of management policies. The proposed methodology develops an emission indicator, a proxy measure, based on GPS instantaneous speed data. This does not represent a tool for estimating pollutant emissions, which implies the computation of emissions based on the knowledge of the level of traffic, but rather an indicator of these emissions. For this purpose, the evaluation and visualization of NOx emissions indicator based on GPS speed data from several probe vehicles is conducted in four steps. The first step of the tool refers to raw data processing, namely map 5 R. M. Boitor et al. An innovative tool for the evaluation of Nox emissions from road traffic matching and extraction of specific area. Secondly, GPS data filtering is conducted in order to identify ZP points associated with parking and LSLC points, which are excluded from the emission indicator analysis. Then, the validation is carried out by means of GIS tools. At last, the emission indicator is obtained and validated. About the application carried out in Cluj-Napoca, it is possible to observe that the emission zones were classified based on the values of the NOx emission indicator. A share of 11% Hot emission areas, 44% Medium emission areas and 45% Cold emission areas resulted. Moreover, Hot emission areas are easy to spot and further analyse. NOx instantaneous emissions have been found to be more reliable for capturing congestion impact. The integration of emission data with GIS is an effective method to perform complex environmental analysis by means of pollutant spatial distribution. Conclusions The results show the utility of the tool for policy and planning process of road traffic management, due to its ease of application and consistency especially for the definition of critical areas and prioritising of the improvements with mobility projects. The input data can be dynamically updated, which implies the repetition of the numerical application. p pp Further efforts have to be done to better refine the method. The efforts should include the following aspects: (1) the study is focused on a small fleet of probe vehicles. More complete analyses, on a wider and more robust sample of probe vehicles need to be conducted; (2) the study provides a methodology that the use of the speed data related to single vehicles. The activities carried out are mainly focused on the spatial analysis of the data from GPS points but there is the need of additional refinements in order to study the temporal changes of speed vehicles as well, for defining congested and free-flow conditions in a more robust way; (3) further development will deal with the validation of the model using two possible sets of data for comparison: the real-world data concerning pollutants emission and concentration from sensors and the emissions estimated using a detailed traffic model based on DTA. Another important step for the overall validation of the tool will test the capacity to evaluate various strategies to reduce of emissions such as the congestion mitigation actions or speed management techniques. References Afotey, B. N., Sattler, M. L., Mattingly, S. P., & Chen, V. C. (2013). Statistical model for estimating carbon dioxide emissions from a light-duty gasoline vehicle. Journal of Environmental Protection, 4, 8–15. https://doi.org/10.4236/jep.2013.48A1002 Alam, M. S., Duffy, P., Hyde, B., & McNabola, A. (2017). Improvement in the estimation and back-extrapolation of CO2 emissions from the Irish road transport sector using a bottom-up data modelling approach. Transportation Research Part D: Transport and Environment, 56, 18–32. https://doi.org/10.1016/j.trd.2017.07.011 Andrei, R., & Condurat, M. (2018). Evaluation of ecological impact of roads function of their technical condition and the intensity of traffic flow (II) Practical approach. Case studies and technical recommendations. Buletinul Institutului Politehnic din lasi. Sectia Constructii, Arhitectura, 64, 109–123. http://www.bipcons.ce.tuiasi.ro/Archive/652.pdf Banica, A., Bobric, E. D., Cazacu, M. M., Timofte, A., Gurlui, S., & Breaban, I. G. (2017). Integrated assessment of exposure to traffic-related air pollution in Iasi City, Romania. Environmental Engineering and Management Journal, 16, 2147–2163. https://doi.org/10.30638/eemj.2017.222 Barth, M., & Boriboonsomsin, K. (2008). Real-world carbon dioxide impacts of traffic congestion. Transportation Research Record: Journal of the Transportation Research Board, 2058, 163–171. https://doi.org/10.3141/2058-20 Boitor, R. M., Cadar, R. D., Măran, P. D., Mannini, L., & Petrelli, M. (2019). A new tool for the evaluation of CO2 emissions from road traffic: a case study in Cluj-Napoca, Romania. Environmental Engineering & Management Journal, 18, 2083–2093. https://doi.org/10.30638/eemj.2019.198 Cai, M., Zou, J., Xie, J., & Ma, X. (2015). Road traffic noise mapping in Guangzhou using GIS and GPS. Applied Acoustics, 87, 94–102. https://doi.org/10.1016/j.apacoust.2014.06.005 Chang, C. T., & Lin, T. P. (2018). Estimation of carbon dioxide emissions generated by building and traffic in Taichung city. Sustainability, 10, 112–130. https://doi.org/10.3390/su10010112 Condurat, M. (2016). Chains of causality associated with the environmental impact of road transport system. Journal of Sustainable Architecture and Civil Engineering, 14, 20–30. https://doi.org/10.5755/j01.sace.14.1.14658 Covrig, I., Oroian, I., Odagiu, A., Holonec, L., & Oroian, E. (2016). A. hippocastanum L. and T. cordata mill. as biomonitoring plants for air pollution in urban areas. A case study: City of Cluj-Napoca. Environmental Engineering & Management Journal (EEMJ), 15(5), 995-1002. http://eemj.eu/index.php/EEMJ/article/view/2919 EC. (2013). COM (2013) 913 final. Together towards competitive and resource efficient urban mobility, A concept for sustainable urban mobility plans, Annex 1. https://ec.europa.eu/transport/sites/transport/files/themes/urban/doc/ump/com%282013%29913-annex_en.pdf ELTIS. (2014). Guidelines. Developing and implementing a sustainable urban mobility plan. http://www.eltis.org/sites/default/files/sump_guidelines_en.pdf ELTIS. (2014). Guidelines. Developing and implementing a sustainable urban mobility plan. References http://www.eltis.org/sites/default/files/sump_guidelines_en.pdf ( ) p g p g y p http://www.eltis.org/sites/default/files/sump_guidelines_en.pdf García-Fuentes M., & de Torre C. (2017). Towards smarter and more sustainable regenerative cities: the REMOURBAN model. Entrepreneurship and Sustainability Issues, 4, 328–338. https://doi.org/10.9770/jesi.2017.4.3S(8) Gori, S., La Spada, S., Mannini, L., & Nigro, M. (2014). Emission dynamic meso-simulation model to evaluate traffic congested urban networks. IET Intelligent Transport Systems, 9, 333–342. https://doi.org/10.1049/iet-its.2013.00 Grote, M., Williams, I., Preston, J., & Kemp, S. (2018). A practical model for predicting road traffic carbon dioxide emissions using Inductive Loop Detector data. Transportation Research Part D: Transport and Environment, 63, 809–825. https://doi.org/10.1016/j.trd.2018.06.026 6 R. M. Boitor et al. An innovative tool for the evaluation of Nox emissions from road traffic Hu, X., Xu, D., & Wan, Q. (2018). Short-term trend forecast of different traffic pollutants in Minnesota based on spot velocity conversion. International Journal of Environmental Research and Public Health, 15(9), 1925. Jiang, Z., Chen, X. M., & Ouyang, Y. (2017). Traffic state and emission estimation for urban expressways based on heterogeneous data. Transportation Research Part D: Transport and Environment, 53, 440–453. https://doi.org/10.1016/j.trd.2017.04.042 Kepaptsoglou, K., Karlaftis, M. G., Gkotsis, I., Vlahogianni, E., & Stathopoulos A. (2015). Urban regeneration in historic downtown areas: an ex-ante evaluation of traffic impacts in Athens, Greece. International Journal of Sustainable Transportation, 9, 478– 489. https://doi.org/10.1080/15568318.2013.811331 Kong, X., Li, M., Ma, K., Tian, K., Wang, M., Ning, Z., Xia, F. (2018). Big trajectory data: A survey of applications and services. IEEE Access, 6, 58295–58306. https://doi.org/10.1109/ACCESS.2018.2873779 Laña, I., Del Ser, J., Padró, A., Vélez, M., & Casanova-Mateo, C. (2016). The role of local urban traffic and meteorological conditions in air pollution: A data-based case study in Madrid, Spain. Atmospheric Environment, 145, 424–438. https://doi.org/10.1016/j.atmosenv.2016.09.052 g j Li, M., Yu, L., Zhai, Z., He, W., & Song, G. (2016). Development of emission factors for an urban road network based on speed distributions. Journal of Transportation Engineering, 142, 1–9. https://doi.org/10.1061/(ASCE)TE.1943-5436.0000858 Liu, Y., Yan, X., Wang, Y., Yang, Z., & Wu, J. (2017). Grid mapping for spatial pattern analyses of recurrent urban traffic congestion based on taxi GPS sensing data. Sustainability, 9, 533–548. https://doi.org/10.3390/su9040533 Masood, A., Kafeel, A., & Shamshad, A. (2017). Urban roadside monitoring, modeling and mapping of air pollution. Applied Journal of Environmental Engineering Science, 3, 179–194. Matz, C. J., Stieb, D. M., Egyed, M., Brion, O., & Johnson, M. (2018). References Evaluation of daily time spent in transportation and traffic- influenced microenvironments by urban Canadians. Air Quality, Atmosphere & Health, 11, 209–220. https://doi.org/10.1007/s11869-017-0532-6 Mitran, G., & Ilie, S. (2014). Method for inventorying CO emissions from road traffic in urban areas through transport modeling. Environmental Engineering and Management Journal, 13(8), 1945–1956. National Institute for Statistics – ROMANIA. (2018). Transport equipment, vehicles registered and road traffi http://www.insse.ro/cms/ro/content/mijloace-de-transport-vehicule-înmatriculate-și-accidente-de-circulație-rutie National Institute for Statistics ROMANIA. (2018). Transport equipment, vehicles registered and road traffic accidents. http://www.insse.ro/cms/ro/content/mijloace-de-transport-vehicule-înmatriculate-și-accidente-de-circulație-rutieră Ntziachristos, L., Gkatzoflias, D., Kouridis, C., & Samaras Z. (2009). COPERT: A European road transport emission inventory Ntziachristos, L., Gkatzoflias, D., Kouridis, C., & Samaras Z. (2009). COPERT: A European road transport emissio model (pp. 491–504). Springer. https://doi.org/10.1007/978-3-540-88351-7_37 Pathak, S. K., Singh, Y., Sood, V., & Channiwala, S. A. (2017). On-road vehicle driving and energy requirements and impact on unregulated exhaust emissions under urban driving conditions. SAE International Journal of Engines, 10, 1866–1879. https://doi.org/10.4271/2017-01-1013 Pérez, J., de Andrés, J. M., Borge, R., de la Paz, D., Lumbreras, J., & Rodríguez, E. (2019). Vehicle fleet characterization study in the city of Madrid and its application as a support tool in urban transport and air quality policy development. Transport Policy, 74, 114–126. Ryu, B. Y., Jung, H. J., & Bae, S. H. (2015). Development of a corrected average speed model for calculating carbon dioxide emissions per link unit on urban roads. Transportation Research Part D: Transport and Environment, 34, 245–254. https://doi.org/10.1016/j.trd.2014.10.012 Roșu, L., Istrate, M., & Bănică, A. (2018). Passenger car dependency and consequent air pollutants emissions in Iasi metropolitan area (romania). Environmental Engineering and Management Journal, 17(4), 865–875. http://eemj.eu/index.php/EEMJ/article/view/3551 Tenailleau, Q. M., Bernard, N., Pujol, S., Parmentier, A. L., Boilleaut, M., Houot, H., & Mauny, F. (2016). Do outdoor environmental noise and atmospheric NO2 levels spatially overlap in urban areas. Environmental Pollution, 214, 767–775. https://doi.org/10.1016/j.envpol.2016.04.082 Tongwane, M., Piketh, S., Stevens, L., & Ramotubei T. (2015). Greenhouse gas emissions from road transport in South Africa and Lesotho between 2000 and 2009. Transportation Research Part D: Transport and Environment, 37, 1–13. https://doi.org/10.1016/j.trd.2015.02.017 Toşa, C., Antov, D., Köllő, G., Rõuk, H., & Rannala M. (2015). A methodology for modelling traffic related emissions in suburban areas. Transport, 30, 80–87. https://doi.org/10.3846/16484142.2013.819034 Wang, Y., Szeto, W. Y., Han, K., & Friesz, T. L. (2018). Dynamic traffic assignment: A review of the methodological advances for environmentally sustainable road transportation applications. Transportation Research Part B: Methodological, 111, 370– 394. R. M. Boitor et al. An innovative tool for the evaluation of Nox emissions from road traffic References Zhang, L., Long, R., Chen, H., & Geng, J. (2018). A review of China’s road traffic carbon emissions. Journal of Cleaner Production, 207, 569–581. https://doi.org/10.1016/j.jclepro.2018.10.003 Zhao, X., Carling, K., & Håkansson, J. (2016). Residential planning, driver mobility and CO2 emission. Working papers in transport, tourism, information technology and microdata analysis, 5. Dalarna University. https://www.diva-portal.org/smash/get/diva2:1067701/FULLTEXT01.pdf Zhao, X., Carling, K., & Håkansson, J. (2017). Residential planning, driver mobility and CO2 emission: a microscopic look at Borlänge in Sweden. European Planning Studies, 25(9), 1597–1614. 7 7
https://openalex.org/W3081233215	https://www.epj-conferences.org/articles/epjconf/pdf/2020/14/epjconf_eosam2020_07002.pdf	English	null	Hybrid plasmonic/photonic crystals for optical detection of bacterial contaminants	EPJ web of conferences	2,020	cc-by	1,003	Hybrid plasmonic/photonic crystals for optical detection of bacterial contaminants Giuseppe M. Paternò1, Liliana Moscardi1,2, Stefano Donini1, Aaron M. Ross2, Silvia M. Pietralunga1,3, Nicholas Dalla Vedova1, Simone Normani1, Ilka Kriegel4, Guglielmo Lanzani1,2 and Francesco Scotognella Giuseppe M. Paternò1, Liliana Moscardi1,2, Stefano Donini1, Aaron M. Ross2, Silvia M. Pietralunga1,3 Vedova1, Simone Normani1, Ilka Kriegel4, Guglielmo Lanzani1,2 and Francesco Scotognella 1Center for Nano Science and Technology, Istituto Italiano di Tecnologia (IIT), Via Pascoli 10, 20133, Milano, Italy 2Physics Department, Politecnico di Milano, Piazza L. da Vinci 32, 20133 Milano, Italy 3Institute for Photonics and Nanotechnologies (IFN), Consiglio Nazionale delle Ricerche (CNR), Piazza L. da Vinci 32, 20133 Milano, Italy 4Department of Nanochemistry, Istituto Italiano di Tecnologia (IIT), via Morego, 30, 16163 Genova, Italy nt of Nanochemistry, Istituto Italiano di Tecnologia (IIT), via Morego, 30, 16163 Genova, Italy Abstract. Here, we show that a hybrid plasmonic/photonic crystal consisting of a thin layer of bioactive plasmonic material (i.e. silver) deposited on top a 1D PhC can detect one of the most common bacterial contaminant, namely Escherichia coli. We speculate that the change in the plasmon charge density brought about by metal/bacterium interaction results in a variation of the plasmon resonance that, in turns, translates in a shift of the photonic structural color. References 1. Hong, W., Hu, X., Zhao, B., Zhang, F. & Zhang, D. 22, 5043–5047 (2010). 2. Kanai, T., Lee, D., Shum, H. C., Shah, R. K. & Weitz, D. A. Adv. Mater. 22, 4998–5002 (2010). 2. Kanai, T., Lee, D., Shum, H. C., Shah, R. K. & Weitz, D. A. Adv. Mater. 22, 4998–5002 (2010). 3. Fudouzi, H. & Sawada, T. Langmuir 22, 1365– 1368 (2006). 3. Fudouzi, H. & Sawada, T. Langmuir 22, 1365– 1368 (2006). 2 Results Here, we show that a novel hybrid plasmonic−photonic device consisting of a thin layer of silver deposited on top of a solution-processed BS is responsive to bacterial contamination. In particular, we assessed the validity of our approach by detecting one of the most hazardous Gram-negative bacterial contaminants in food and water, Escherichia coli. We integrated the bio-responsive silver layer on top of a 1D PhC consisting of alternating layers of SiO2 and TiO2 nanoparticles (5 bilayers). In this context, the plasmonic metal can be seen as defective layer of the photonic crystal that affects the optical response of the PhC via its free carrier density.> Thus, the main goal is to modify the dielectric properties at the PhC/metal interface and, thus, the overall optical read-out by exploiting the possible change in the silver complex dielectric function caused by Ag/bacteria interaction. To localise strongly the plasmonic response at the interface of the photonic structure, we selected the minimum Ag thickness attainable with our deposition apparatus. Interestingly, while interaction with the culture medium (pristine) does not alter appreciably the photonic read-out (Fig. 1), when the hybrid structure is interfaced with the bacterial colonies we noticed a 40 nm blue-shift of the photonic band-gap. EPJ Web of Conferences 238, 07002 (2020) EPJ Web of Conferences 238, 07002 (2020) EPJ Web of Conferences 238, 07002 (2020) EOSAM 2020 https://doi.org/10.1051/epjconf/202023807002 1 Introduction Photonic crystals (PhCs) have been largely employed as detection/sensing devices in recent years, since the photonic stop-band can be tuned by applying a number of external stimuli, such as chemical1, thermal2 and mechanical triggers3. In this context, we have recently proposed porous 1D photonic structures exhibiting electro-optical tunability, due to the incorporation of optoelectronically-active plasmonic nanoparticles in the photonic structures.4–6 Fig. 1. Transmission spectrum of the pristine and E.coli- contaminated Ag/DBR. The inset shows the visual change in the reflection properties of the sample. 400 500 600 700 800 0.0 0.2 0.4 0.6 0.8 1.0 E. coli Ag/DBR pristine Ag/DBR E.coli Transmittance Wavelength (nm) Pristine 400 500 600 700 800 0.0 0.2 0.4 0.6 0.8 1.0 E. coli Ag/DBR pristine Ag/DBR E.coli Transmittance Wavelength (nm) Pristine p Here, we show that a hybrid plasmonic/photonic crystal consisting of a thin layer of bioactive plasmonic material (i.e. silver) deposited on top a 1D PhC can detect one of the most common bacterial contaminant, namely Escherichia coli.7 We speculate that the change in the plasmon charge density brought about by metal/bacterium interaction results in a variation of the plasmon resonance which, in turns, translates in a shift of the photonic structural color. Fig. 1. Transmission spectrum of the pristine and E.coli- contaminated Ag/DBR. The inset shows the visual change in the reflection properties of the sample. * Corresponding author: giuseppe.paterno@iit.it © The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (http://creativecommons.org/licenses/by/4.0/). 4. Aluicio-Sarduy, E. et al. Beilstein J. Nanotechnol. 7, 1404–1410 (2016). 5. Paternò, G. M. et al. Sci. Rep. 8, 3517 (2018). 6. Paternò, G. M., Moscardi, L., Kriegel, I., Scotognella, F. & Lanzani, G. J. Photonics Energy 8, 1 (2018). 7. Paternò, G. M. et al. J. Phys. Chem. Lett. 10, 4980–4986 (2019). 8. Drude, P. Ann. Phys. 312, 687–692 (1902). * Corresponding author: giuseppe.paterno@iit.it © The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (http://creativecommons.org/licenses/by/4.0/). © The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (http://creativecommons.org/licenses/by/4.0/). EPJ Web of Conferences 238, 07002 (2020) EPJ Web of Conferences 238, 07002 (2020) EOSAM 2020 https://doi.org/10.1051/epjconf/202023807002 4. Aluicio-Sarduy, E. et al. Beilstein J. Nanotechnol. 7, 1404–1410 (2016). 5. Paternò, G. M. et al. Sci. Rep. 8, 3517 (2018). 6. Paternò, G. M., Moscardi, L., Kriegel, I., Scotognella, F. & Lanzani, G. J. Photonics Energy 8, 1 (2018). 7. Paternò, G. M. et al. J. Phys. Chem. Lett. 10, 4980–4986 (2019). 8. Drude, P. Ann. Phys. 312, 687–692 (1902). 6. Paternò, G. M., Moscardi, L., Kriegel, I., Scotognella, F. & Lanzani, G. J. Photonics Energy 8, 1 (2018). 7. Paternò, G. M. et al. J. Phys. Chem. Lett. 10, 4980–4986 (2019). 8. Drude, P. Ann. Phys. 312, 687–692 (1902). 2 2
W4391822025.txt	https://journals.wlb-stuttgart.de/ojs/index.php/wfr/article/download/10234/10113	de		Rezension von: Ulbrich, Claudia, Leibherrschaft am Oberrhein im Spätmittelalter	Württembergisch Franken	2,024	cc-by	1,128	204 Neue Bücher Gründungen zusammenfassend. Wie Tübinger ist und Familie anzuschließen der bis 1634 überlebende Zweig des Hauses an die welche Schicksale diese Grafen als Herren von Lichteneck im Breisgau (1356—1634) erlebten, klärt W. Setzler. Für uns ist diese Linie durch den Tod des Grafen Konrad auf der Waldenburger Fasnacht 1570 von Interesse (vgL WFr. 1957); durch seine Schwester Agathe stammt das Haus Hohenlohe-Waldenburg von diesen Grafen und Pfalzgrafen ab. Der aufschlußreiche Band erschließt und Tatsachen neue bereichert unsere Kenntnis. Wu Claudia Ulbrich: Leibherrschaft am Oberrhein im Spätmittelalter (= Veröffentlichungen des Max-Planck-Instituts für Geschichte, 58). Göttingen: Vandenhoeck und Ruprecht 1979. 327 S., 2 Ktn. Die Geschichte der Leibherrschaft im deutschen Südwesten dürfte schon bisher als relativ gut erforscht gelten. Wir erinnern hier an die Schriften Theodor Knapps über Württemberg, die neueren über Forschungen des leider früh verstorbenen Zürcher Rechtshistorikers Walter Müller die st. gallischen Gotteshausleute oder die auf Oberschwaben (Allgäu) bezogenen Arbeiten des Saarbrücker Sozialhistorikers Peter Blickle. Aus der Schule des letzteren stammt die vorliegende Dissertation, mit der das bisher erforscht wird und mit der sich - nur wenig untersuchte Oberrheingebiet das darf gleich eingangs festgehalten werden - die Verf. der Reihe der hier genannten Vorgänger würdig anschließt. Die Arbeit besticht schon hinsichtlich ihrer Anlage. Ulbrich hat sich nicht darauf beschränkt, einzelne Herrschaften isoliert zu beschreiben. Sie erforscht vielmehr eine Reihe benachbarter Herrschaften unterschiedlicher Größe und Verfassungs- bzw. Besitzerstruktur in einem größeren Gebiet, und die zwar geistlichen Herrschaften des Klosters St. Blasien und der Deutschordenskommende Beuggen, die städtischen Landgebiete von Basel, Solothurn und Freiburg/Brsg. sowie die Fürstentümer Baden und Bistum Basel. Durch diese unterschiedlichen rechtlichen und setzen und aufwendige, aber lohnende Methode gelingt es, die Ausformungen und Funktionen derLeibeigenschaft mit den verschiedenen wirtschaftlichen Zuständen der jeweiligen Herrschaften in Beziehung damit weitgehend zu befriedigend zu erklären. Sie kann auch die durch Ab- und Zuwanderung von Leibeigenen entstehenden Probleme und die Folgen solcher Wanderungsbewegungen für das Verhältnis benachbarter Herrschaften erhellen Kampf der Leibeigenen um die rechtliche Fixierung ihrer - Fragen, die neben dem Abgaben und Dienste zu den interessantesten der Leibeigenschaft gehören. Die Geschichte der Leibeigenschaft, dies wird auch in der vorliegenden Arbeit immer wieder deutlich, ist durch eine eigenartige Umkehr in der Funktion gekennzeichnet. Im Verlauf dieser Funktionsänderung wandelte sich die Leibeigenschaft als ursprünglich typische Erscheinungsform der Personalherrschaft zum Mittel der Territorialherrschaft. Die an die Person anknüpfende Leibherrschaft bot den Leibherren den vor allem im Spätmittelalter angesichts einer höheren Mobilität der Bevölkerung wichtigen Vorteil, Herrschaftsrechte auch nach Wegzug der Grundholden bzw. ihrer Kinder in fremde Gebiete ausüben zu können. einem Schwierigkeiten ergaben sich zunächst nur aus fremder Herrschaften, der sogenannten der Eheschließung mit leibeigenen Frauen ungenoßsamen Ehe. Um zu verhindern, daß die Kinder dem Leibherrn der Frau zufielen und damit dem Leibherrn des Mannes verlorengingen, wurden Genoßsameverträge zwischen den Herrschaften geschlossen. zum Verbot ten Später ging man der ungenoßsamen Ehe über. Im 14. Jahrhundert mehrten sich die Schwierigkei- bei der Rückforderung von Leibeigenen, die in Städte oder andere Herrschaften gezogen waren. ihren Die Leibherren suchten sich gegen den Verlust zunächst dadurch zu schützen, daß sie Leibeigenen Eide, Verschreibungen und Bürgschaften abverlangten, hoben aber seit dem Ende des 14. Jahrhunderts zunehmend dieFreizügigkeit ihrer Eigenleute ganz auf. Dieser Tendenz zur Einschränkung der Freizügigkeit entsprach der Kampf gegen Eigenleute fremder Herrschaften. So setzte Basel Grafschaft Rheinfelden) im 16. Jahrhundert mit Verträgen (1527 Solothurn, 1534 durch, daß Zuziehende künftig Basler Leibeigene werden sollten (»des Bann, des Mann«), Parallel zu dieser »Territorialisierung« vollzog sich eine Nivellie- 205 Neue Bücher rung der Pflichten von Leibeigenen und freien Untertanen. Schon Frondienste von jedem, der im Landgebiet »gesessen« war. 1411 verlangte etwa Basel 1514 wurde die Refspflicht (Kriegsfolge) auf alle Untertanen ausgedehnt. Die Leibeigenschaft ging so in einer allgemeinen Untertänigkeit weitgehend auf. Nur die Eheverbote und Freizügigkeitsbeschränkungen, auch die Möglichkeit, sich durch eine Manumissionsgebühr loszukaufen, erinnerten noch an die Besonderheit dieses Statusverhältnisses. Die weitgehende Gleichstellung von Rechten und Pflichten führte schon im 15. Jahrhundert zu Abgrenzungsschwierigkeiten zwischen Grund-, Gerichts- und Leibherrschaft. Insgesamt bewertet Ulbrich die Rolle höher, als dies in jüngster Zeit der Leibherrschaft für Eigenleute und Herrschaften etwa Lütge und K. S. Bader getan haben. Zu Recht weist sie darauf hin, daß etwa die Ablieferung des Besthaupts in einer kleinbäuerlichen Wirtschaft eine ganz erhebliche Belastung darstellte. Aber nicht nur die wirtschaftliche, auch die rechtliche Bedeutung wird hier aufgewertet, freilich schaft wird Leibherren als Mittel für »zahlreiche in einer sehr differenzierten Weise. Die Leibherr- wirtschaftliche Leibherrschaft auf einer vernünftigen Mittellinie und und angesehen. Damit scheint sich das Urteil über politische Zielsetzungen« der die geschichtliche Bedeutung der einzupendeln. Nachdem die aus politischen ideologischen Gründen im 19. Jahrhundert vorherrschende Dämonisierung durch die jüngere Forschung weitestgehend abgebaut werden konnte, ist in den letzten Jahren die wirtschaftliche, soziale und rechtliche Bedeutung der Leibherrschaft manchmal vielleicht schon zu gering eingestuft worden. Ulbrich s Verdienst ist es, hier korrigierend einzugreifen. Sie hat durch eine glückliche Verbindung von Detailuntersuchung und vergleichender Betrachtung ein ebenso dichtes wie differenziertes und lebensnahes Bild von den Möglichkeiten und Grenzen entworfen, die sich für die Herrschaften des Oberrheingebiets aus dem Institut der Leibherrschaft ergaben. R. J. W. Sankt Elisabeth. Fürstin Dienerin Heilige. Aufsätze, Dokumentation, Katalog. Hrsg, von der Philipps-Universität Marburg i. Verbindung mit dem Hessischen Landesamt für geschichtliche Landeskunde. Sigmaringen: Thorbecke 1981. XIV, 570 S., 3 Ktn. Wir leben im Zeitalter der großen historischen Ausstellungen, die sich besonderer Beliebtheit beim Publikum erfreuen. Diese Ausstellungen erfordern in der Regel gründliche jahrelange deren Ertrag im Katalog für die künftige Forschung festgehalten wird. Ein gelungenes Beispiel bietet die Marburger Elisabethausstellung, die zur 750. Vorarbeiten, besonders Wiederkehr ihres Todes (1231) veranstaltet wurde. 17 Autoren behandeln in ihren Aufsätzen Themen im Umkreis der Heiligen, von der zeitgenössischen Überlieferung und der Stellung der Frau im Ordenswesen bis zu Kunstwerken und Reliquien. Der reich illustrierte Katalog (von S. 315 an) bringt in 8 Abschnitten mit ausführlichen Texten Belege zu Leben und Nachleben der ungarischen Königstochter, die bereits als Kind an den Thüringer Landgrafenhof gebracht wurde. Leider fehlt eine Abhandlung und Begründung für Elisabeths Ahnentafel; die auf S. 330 abgedruckte Stammtafel des Hauses Andechs wirft einige Fragen auf: So fehlt unter den Kindern Bertholds VI. und der Agnes v. Groitzsch Kunigunde, die Gemahlin des Grafen Eberhard v. Eberstein (deren Tochter das Kloster Gnadental gründete, vgl. ZGO 1975). War Hedwig, die Gemahlin Bertolds V., wirklich eine Wittelsbacherin? Der schöne und reichhaltige Band bietet vielfache Anregung und wird eine Grundlage der Elisabethfor- schung bleiben. Wu Amedeo Molnär: Die Waldenser. Geschichte und europäisches Ausmaß einer Ketzerbewegung. Göttingen: Vandenhoeck und Ruprecht 1980. 456 S. Verfasser behandelt Waldes und die Armen von Lyon mit Der Glaubensbrüdern, aber er läßt die eigentlichen »Ketzer«, die ihren lombardischen Katharer und Albigenser, beiseite. Er untersucht das Weiterleben der Waldenser Frömmigkeitsbewegung im Unter- grund, ihre Einwirkung auf Böhmen und die Hussiten und ihre Beziehungen zur Schweizer Reformation. Literatur und Botschaft, d.h. Theologie der Waldenser, werden knapp und
https://openalex.org/W2965817255	https://www.nature.com/articles/s41598-019-47882-2.pdf	English	null	Simultaneous quantification of alpha-aminoadipic semialdehyde, piperideine-6-carboxylate, pipecolic acid and alpha-aminoadipic acid in pyridoxine-dependent epilepsy	Scientific reports	2,019	cc-by	7,651	Simultaneous quantification of alpha-aminoadipic semialdehyde, piperideine-6-carboxylate, pipecolic acid and alpha- aminoadipic acid in pyridoxine- dependent epilepsy Received: 2 November 2018 Accepted: 2 July 2019 Published: xx xx xxxx Jiao Xue1, Junjuan Wang2, Pan Gong1, Minhang Wu2, Wenshuang Yang3, Shiju Jiang3, Ye Wu Yuwu Jiang1, Yuehua Zhang1, Tatiana Yuzyuk4,5, Hong Li6 & Zhixian Yang1 Jiao Xue1, Junjuan Wang2, Pan Gong1, Minhang Wu2, Wenshuang Yang3, Shiju Jiang3, Ye Wu1, Yuwu Jiang1, Yuehua Zhang1, Tatiana Yuzyuk4,5, Hong Li6 & Zhixian Yang1 The measurements of lysine metabolites provide valuable information for the rapid diagnosis of pyridoxine-dependent epilepsy (PDE). Here, we aimed to develop a sensitive method to simultaneously quantify multiple lysine metabolites in PDE, including α-aminoadipic semialdehyde (a-AASA), piperideine-6-carboxylate (P6C), pipecolic acid (PA) and α-aminoadipic acid (α-AAA) in plasma, serum, dried blood spots (DBS), urine and dried urine spots (DUS). Fifteen patients with molecularly confirmed PDE were detected using liquid chromatography-mass spectrometry (LC-MS/MS) method. Compared to the control groups, the concentrations of a-AASA, P6C and the sum of a-AASA and P6C (AASA-P6C) in all types of samples from PDE patients were markedly elevated. The PA and a-AAA concentrations ranges overlapped partially between PDE patients and control groups. The concentrations of all the analytes in plasma and serum, as well as in urine and DUS were highly correlated. Our study provided more options for the diverse sample collection in the biochemical tests according to practical requirements. With treatment modality of newly triple therapy investigated, biomarker study might play important role not only on diagnosis but also on treatment monitoring and fine tuning the diet. The persistently elevated analytes with good correlation between plasma and DBS, as well as urine and DUS made neonatal screening using DBS and DUS possible. Pyridoxine-dependent epilepsy (PDE; OMIM 266100), a rare autosomal recessive disorder, is caused by muta- tions in the gene coding for aldehyde dehydrogenase 7 A1 (ALDH7A1), also known as antiquitin1. In mam- mals, antiquitin involves in lysine catabolism where it metabolizes α-aminoadipic semialdehyde (α-AASA) to α-aminoadipic acid (α-AAA)2. With antiquitin dysfunction, metabolic intermediates of lysine catabolism path- way including α-AASA, piperideine-6-carboxylate (P6C) and pipecolic acid (PA) accumulate, which are mark- edly elevated in urine, plasma and cerebrospinal fluid, and can be used as diagnostic markers of PDE. Moreover, the potential toxicity of these accumulating compounds might directly cause cellular injury and further contrib- ute to the neurodevelopmental impairments in PDE3. www.nature.com/scientificreports www.nature.com/scientificreports www.nature.com/scientificreports Material and Methods Ethics statement. This study was approved by the Biomedical Research Ethical Committee of Peking University First Hospital, and written informed consents were obtained from the legal guardians (parents) of the children. All experiments were performed in accordance with relevant guidelines and regulations. All data were analyzed anonymously. Patient and control samples preparation. A total of five types of samples from 15 patients with PDE were collected, including plasma (n = 15), serum (n = 14), DBS (n = 15), urine (n = 15) and DUS (n = 15). All samples were freshly collected before meal and taking pyridoxine. Then, blood and urine were spotted onto filter paper card and allowed to dry at room temperature for about four hours to prepare DBS and DUS. Blood aliquots were also used to prepare plasma and serum samples. No serum sample was collected from patient 14. The clinical features and genotypes of 7 out of the 15 PDE patients (patient 3, 5, 6, 8, 9, 11, 15) have been reported previously14,15. All patients were on daily pyridoxine supplements (30–360 mg/d) without specific diet restric- tion at the time of testing. The psychomotor development was assessed according to clinical judgment and the restandardization of the Adaptive Scale of Infant and Children16.f p Unaffected controls samples were collected from the patients with genetic generalized epilepsy (excluding PDE genetically), tic disorders or simple upper respiratory infection. Considering that all our PDE patients were older than 1 year at the time of the testing, we selected the children aged 1–13 years old as controls. Only samples with normal findings on routine biochemical tests and/or metabolic screening were included. The control ranges for plasma (n = 28), serum (n = 25), DBS (n = 25), urine (n = 25) and DUS (n = 25) were established. Because the P6C concentration of one control plasma sample was unusually much higher than the others (about 15 times) and there was not enough sample volume for the retest, we did not include it into statistics.h g p The specimens were then kept on dry ice during shipment and then stored at −80 °C until analysis. The determination of the lysine metabolites using LC-MS/MS. Reagents. The following reagents were purchased: hydrochloic acid, 3Nin n-butanol (Regis Technologies, Inc.), methanol and acetonitrile (meker, us), formicacid, a-AAA, PA, 5-sulfosalicylic acid dihydrate Amberlyst® 15 dry resin and allysine-ethylene acetal (AEA) (Sigma), d9-PA and d3-a-AAA (CDN Isotopes). Simultaneous quantification of alpha-aminoadipic semialdehyde, piperideine-6-carboxylate, pipecolic acid and alpha- aminoadipic acid in pyridoxine- dependent epilepsy The combination of a lysine-restricted diet with pyridoxine and arginine supplements, known as triple therapy, decreased accumulation of PDE biomarkers and improved cognitive and motor development in majority of reported patients4,5. Plasma levels of the sum of AASA and P6C (AASA-P6C) and PA directly correlated with plasma lysine5. Therefore, the measurements of a-AASA, P6C and 1Department of Pediatrics, Peking University First Hospital, Beijing, China. 2Zhejiang Biosan Biochemical Technologies Co., Ltd, Zhejiang, China. 3Department of Clinical Laboratory, Peking University First Hospital, Beijing, China. 4Department of Pathology, University of Utah, Salt Lake City, UT, USA. 5ARUP Laboratories, ARUP Institute for Clinical and Experimental Pathology, Salt Lake City, UT, USA. 6Department of Human Genetics, Emory University, School of Medicine, America, Atlanta, USA. Correspondence and requests for materials should be addressed to Z.Y. (email: zhixian.yang@163.com) Scientific Reports \| (2019) 9:11371 \| https://doi.org/10.1038/s41598-019-47882-2 www.nature.com/scientificreports/ PA provide valuable information for the rapid diagnosis and monitoring the treatment of patients with PDE, as well as prognosis prediction. Recently, Wempe et al.6 reported a novel biomarker 6-oxo-pipecolate (6-oxo-PIP) for PDE that was stable at room temperature, while the sensitivity and specificity of it still need further studies.h p y pi y The measurement of a-AASA, P6C and PA was still limited to few laboratories world widely. Several studies on the determination of plasma and urinary a-AASA and/or P6C by liquid chromatography-mass spectrometry (LC-MS/MS) method, and determination of PA by gas chromatography-mass spectrometer (GC-MS) or LC-MS/ MS were previously published1,7–9. Moreover, detection of a-AASA and P6C in dried blood spots (DBS) from newborn patients with PDE using LC-MS/MS or a novel method HILIC-ESI-MS were reported10,11. However, these analytes in different types of samples (such as plasma, urine and DBS), or even different analytes in the same sample, were usually need to be measured by different methods with various specimen preparation and turned around time. In a 2016 publication, Pena et al.12 described a protocol that used LC-MS/MS to simultaneously detect the PA, P6C as well as saccharopine, glutamic acid and pyridoxal-5′-phosphate in mouse plasma samples. Thus far, only one study on simultaneous detection of AASA-P6C and PA in plasma and urine from PDE patients by LC-MS/MS were reported13. Simultaneous quantification of alpha-aminoadipic semialdehyde, piperideine-6-carboxylate, pipecolic acid and alpha- aminoadipic acid in pyridoxine- dependent epilepsy Here, we modified the method described previously13, and made it suitable for the simultaneous quantification of multiple metabolites including a-AASA, P6C, AASA-P6C, PA and α-AAA in plasma, serum and DBS as well as urine and dried urine samples (DUS), which could be clinically used for early diagnosis and monitoring patients with PDE. Material and Methods Three DBS (ID 3 mm) was mixed with 150 μL of 50% methanol containing 10.0 μM of d3-a-AAA and d9-PA as the internal standard. DUS specimens were prepared as the urine after reconstituted by water. Plasma, urine and DBS specimens. An aliquot of 20 μL of urine was mixed with 120 μL of acetonitrile containing 15.0 μM of d3-a-AAA and d9-PA as the internal standards. An aliquot of 50 μL of plasma was mixed with 220 μL of acetonitrile containing 7.5 μM of d3-a-AAA and d9-PA as the internal standard. Three DBS (ID 3 mm) was mixed with 150 μL of 50% methanol containing 10.0 μM of d3-a-AAA and d9-PA as the internal standard. DUS specimens were prepared as the urine after reconstituted by water. p p pt y The mixture was vortexed for 2 min, allowed to stand for 3 min, and then centrifuged at 20000 g at 4 °C for 10 min. An aliquot of supernatant was transferred to a clean tube and dried by nitrogen flow in room temperature. The residue was derivatized with 100 μl of 3 N HCl in n-butanol (v/v) at 65 °C and 600 rpm for 30 min dried as described above and reconstituted in 100 μL water/methanol (70:30) containing 0.1% of formic acid. Detection methods. Separation was achieved on an ACQUITY BEH-C18 column (2.1 × 50 mm, 1.7 m) at a tem- perature of 40 °C using a linear gradient of mobile phase A (0.1% of formic acid in water) and mobile phase B (0.1% of formic acid in methanol) as follows: 0 min, 10% B (0.4 mL/min); 1.5 min, 45% B (0.3 mL/min); 2.5 min, 75% B (0.3 mL/min); 3 min, 95% B (0.3 mL/min); 4–5 min, 10% B (0.3 mL/min). The mass spectrometer was operated in positive ion mode on a Waters Xevo TQD MS/MS with a 1.2 kV capillary voltage. The source and desolvation gas temperature was 150 °C and 550 °C, respectively. The data were acquired in with multiple reaction monitoring (MRM) mode. The cone and collision energy for the detection of a-AASA, P6C, a-AAA, d3-a-AAA, PA and d9-PA were shown in Table S1. The examples of extract ion chromatogram (EIC) of a-AASA, P6C, PA, a-AAA, d9-PA and d3-AAA in control and PDE patient in DBS, plasma and urine samples were shown in Figs 1, S1–2, respectively. LOD/LOQ. Material and Methods The limit of detection (LOD) was determined by analyzing plasma, DBS, and urine lowest con- trol samples by diluting concentrations progressively (n = 6) until a minimum signal-to-noise ratio (S/N) of 3 was achieved. The limit of quantification (LOQ) for all analytes was determined as LOD until a minimum signal-to-noise ratio (S/N) of 10 was achieved. Usually, LOQ is defined as the lowest concentration of analyte, at which the measured concentration is within ±10–30% of the target concentration with the precision (CV) of <10–20%. recision. Three replicates of each QC level for plasma, DBS and urine were used to calculate the intra-assay recision. The inter-assay precision is evaluated by analyzing the samples over five days. Stability test for plasma, urine, DBS and DUS. Whole blood and urine samples were collected from five healthy adult volunteers in EDTA tubes, and then pooled according to the sample types. QC plasma samples were pre- pared by spiking normal plasma with the following concentrations of added AASA-P6C: 4.8, 6.4, 16 μmol/L, and separately PA and AAA: 2, 4, 16 μmol/L. Urine QC samples were prepared by spiking normal urine with the following concentrations of added AASA-P6C: 15, 25, 65 μmol/L, and separately PA and AAA: 5, 20, 80 μmol/L. DBS samples were prepared by spiking whole blood with the following concentrations of added AASA-P6C: 4, 12, 24 μmol/L, and separately PA and AAA: 5, 15, 30 μmol/L. DUS samples were prepared by spiking urine matrix with the following concentrations of added AASA-P6C: 4, 20, 60 μmol/L, and separately PA and AAA: 5, 25, 75 μmol/L. DBS and DUS samples were spotted onto filter paper card (Protein Saver™ 903® Card, Whatman Inc, Piscayaway, NJ), allowing to dry at room temperature for about four hours and then stored in sealed plastic bags. The stability of AASA-P6C, a-AAA and PA in plasma, urine, DBS and DUS were tested at room temperature over 7 days, and at 4 °C over 30 days. Two PDE patients’ urine, plasma, DBS and DUS samples stored in −80 °C over 8 months were also tested to get the stability of analytes. Statistical analysis. All statistical analysis were completed using SPSS 16.0. The Shapiro-Wilk test was used to test whether variables were normally distributed. A Student’s t-test (2-tailed) or Mann-Whitney U test was used to test differences in a-AASA, P6C, AASA-P6C, PA and a-AAA concentrations between patients and control group. Material and Methods The association of the metabolites between different samples was analyzed by calculating the correlation coefficient, and R2 > 0.7 was considered to be a strong correlation. The significance level was set at 0.05 and all tests to assess P values were two-sided. Material and Methods AASA-P6C synthesis. Lacking commercially available standards, the AASA-P6C reference material (a mix of a-AASA and P6C) was synthesized from AEA using Amberlyst® 15 bead according to published procedures1,8,13. The efficiency of the conversion of AEA to AASA-P6C and residue traces of AEA were confirmed as the published procedure8. According to these results, the final concentration of AASA-P6C in reference material was calculated based on the assumption of 100% of synthesis efficiency. Reproducibility of AASA-P6C synthesis was evaluated by the peak areas ratio of a-AASA and P6C for the same amounts of synthesized material from the different batches on three different days. We used 1:3 ratio to approximate a-AASA and P6C concentrations based on assumption that the ionization efficiency of a-AASA and P6C were close as the published procedure8,13,17. Control samples and calibration. Six non-zero calibrators were prepared at concentrations of 2–400 umol/L for AASA-P6C, 0.5–100 umol/L for PA and 0.5–100 umol/L for a-AAA in the buffer (2.5% BSA and 0.8% NaCl), normal urine or 30% acetonitrile in water for the use with plasma, DBS or urine, respectively.h Three plasma quality controls (QC) were prepared by spiking normal plasma with AASA-P6C, a-AAA and PA standards at low (5 µmol/L), medium (20 µmol/L) and high (100 umol/L) concentrations. To prepare three DBS QC, normal whole blood was spiked with the same levels of AASA-P6C, a-AAA and PA, and spotted onto filter paper card (Protein Saver™ 903® Card, Whatman Inc, Piscayaway, NJ). After drying at room temperature overnight, DBS were stored in sealed plastic bags. Three QCs were prepared in urine at low (5 mol/L AASA-P6C, a-AAA and PA), medium (20 mol/L AASA-P6C and AAA, 50 mol/L PA) and high (200 mol/L AASA-P6C and a-AAA, 100 mol/L PA) concentrations. The three DUS QCs were prepared as the same concentrations as in urine QCs by soaking filter paper card and drying at room temperature overnight. All types of QCs were stored at −80 °C. Scientific Reports \| (2019) 9:11371 \| https://doi.org/10.1038/s41598-019-47882-2 www.nature.com/scientificreports/ Plasma, urine and DBS specimens. An aliquot of 20 μL of urine was mixed with 120 μL of acetonitrile containing 15.0 μM of d3-a-AAA and d9-PA as the internal standards. An aliquot of 50 μL of plasma was mixed with 220 μL of acetonitrile containing 7.5 μM of d3-a-AAA and d9-PA as the internal standard. Results l i Analytical performance of the method. The intra-assay precision of AASA-P6C, a-AAA, and PA in plasma, DBS and urine were within 10%. The inter-assay precision of AASA-P6C, a-AAA, and PA in urine were 1.9–6.2%, 2.6–8.7% and 2.7–5.9% respectively. The inter-assay precision of AASA-P6C, a-AAA, and PA in plasma were 4.1–5.9%, 7.0–9.8% and 4.0–10.3% respectively. The inter-assay precision of AASA-P6C, a-AAA, and PA in DBS were 3.4–5.8%, 2.8–4.2% and 4.4–5.7% respectively. The results for LOQ and LOD were shown in Table S2. Stability of plasma, urine, DBS and DUS. As the Fig. 2 shown, when the plasma and urine samples were stored either at room temperature for 24 hours or 4 °C for 2 days, the low, medium and high level of AASA-P6C dropped to below 60%, 70% and 80% of the initial value respectively. When DBS and DUS were stored at room temperature after 7 days, the low, medium and high AASA-P6C concentration were above 70% of the initial value. When DBS and DUS were stored at 4 °C after 32 days, the low, middle and high AASA-P6C concentration were above 85% of the initial value. The concentrations of PA and α-AAA in the plasma, urine, DBS and DUS were above 80% and 85% of the initial value at room temperature and 4 °C respectively. The plasma, urine, DBS and DUS samples of two PDE patients were stored at −80 °C for 8 months, and the concentrations of AASA-P6C, PA and α-AAA in all samples are above 85% of the initial value. In conclusion, DBS and DUS samples were more stable than plasma and urine at room temperature and 4 °C, allowing for more reliable clinical screening. Scientific Reports \| (2019) 9:11371 \| https://doi.org/10.1038/s41598-019-47882-2 www.nature.com/scientificreports/ www.nature.com/scientificreports/ Figure 1. The classic extract ion chromatogram of a-AASA, P6C, PA, a-AAA, d9-PA and d3-AAA in control and PDE patient in DBS samples. Figure 1. The classic extract ion chromatogram of a-AASA, P6C, PA, a-AAA, d9-PA and d3-AAA in contro and PDE patient in DBS samples. General information of PDE patients. A total of 15 PDE patients were included in this study (1.3–8.6 years old) (Table 1). All patients were confirmed the diagnosis through mutation analysis of ALDH7A1 and being treated with oral pyridoxine at the time of testing, at doses of 30–360 mg/d. Results l i The PA (R2 = 0.958, p < 0.001), a-AASA (R2 = 0.733, p < 0.001) and AASA-P6C (R2 = 0.826, p < 0.001) concentrations were highly correlated between DBS and plasma samples, but no obvious correlation was found in P6C (R2 = 0.039, p = 0.241) and a-AAA (R2 = 0.419, p = 0.005) concentra- tions (Fig. S3). Compared the metabolites concentrations in DBS and plasma, the P6C, AASA-P6C and PA levels were higher in plasma (p < 0.001), and a-AAA level was higher in DBS (p < 0.001). anges overlapped partially between PDE patients and control groups in all specimens (Fig. 3). And as for every DE patient, PA and a-AAA could be moderate elevated or within normal range. ranges overlapped partially between PDE patients and control groups in all specimens (Fig. 3). And as for every PDE patient, PA and a-AAA could be moderate elevated or within normal range. Comparison the metabolites concentrations between different samples in PDE patients. The concentration of a-AASA (R2 = 0.958, p < 0.001), P6C (R2 = 0.968, p < 0.001), AASA-P6C (R2 = 0.981, p < 0.001), PA (R2 = 0.919, p < 0.001) and a-AAA (R2 = 0.940, p < 0.001) was positively correlated between plasma and serum. Similarly, the concentration of the a-AASA (R2 = 0.987, p < 0.001), P6C (R2 = 0.961, p < 0.001), AASA-P6C (R2 = 0.988, p < 0.001), PA (R2 = 0.929, p < 0.001) and a-AAA (R2 = 0.987, p < 0.001) in urine and DUS was positively correlated also. The PA (R2 = 0.958, p < 0.001), a-AASA (R2 = 0.733, p < 0.001) and AASA-P6C (R2 = 0.826, p < 0.001) concentrations were highly correlated between DBS and plasma samples, but no obvious correlation was found in P6C (R2 = 0.039, p = 0.241) and a-AAA (R2 = 0.419, p = 0.005) concentra- tions (Fig. S3). Compared the metabolites concentrations in DBS and plasma, the P6C, AASA-P6C and PA levels were higher in plasma (p < 0.001), and a-AAA level was higher in DBS (p < 0.001). Comparison the metabolites concentrations between different samples in PDE patients. Results l i Neurodevelopmental evaluation showed 10 out of 15 patients had expressive speech delay, most with mild delay (7/10), two with moderate delay and one with severe delay; Only 2 showed motor delay, all others were normal on motor development. Concentrations of PDE biomarkers in PDE patients and unaffected controls. Control samples were collected from the patients with genetic generalized epilepsy (excluding PDE genetically), tic disorders or simple upper respiratory infection (1–13 yrs). The mean and ranges of the five metabolites concentrations in each type of control samples were shown in the Table 2. The a-AASA concentrations in plasma and serum samples were too low to determine here, which were recorded as 0.00 µmol/L.hif The concentrations of biomarkers varied significantly among different PDE patients, and the detailed data were shown in Supplementary Dataset. Compared to the control groups, the a-AASA, P6C and AASA-P6C in all types of samples were markedly elevated in the PDE patients (p < 0.001). In PDE patients, the mean concentra- tion of PA was elevated in plasma, serum and DBS (p < 0.001), and mean concentrations of a-AAA was elevated in plasma, serum and DBS (p < 0.001), as well as in urine (p < 0.05). However, the PA and a-AAA concentration Scientific Reports \| (2019) 9:11371 \| https://doi.org/10.1038/s41598-019-47882-2 www.nature.com/scientificreports/ Figure 2. The stability of AASA-P6C in plasma, urine, DBS and DUS at room tempreture and 4 °C, respectively. igure 2. The stability of AASA-P6C in plasma, urine, DBS and DUS at room tempreture and 4 °C, respectively ranges overlapped partially between PDE patients and control groups in all specimens (Fig. 3). And as for every PDE patient, PA and a-AAA could be moderate elevated or within normal range. Comparison the metabolites concentrations between different samples in PDE patients. The concentration of a-AASA (R2 = 0.958, p < 0.001), P6C (R2 = 0.968, p < 0.001), AASA-P6C (R2 = 0.981, p < 0.001), PA (R2 = 0.919, p < 0.001) and a-AAA (R2 = 0.940, p < 0.001) was positively correlated between plasma and serum. Similarly, the concentration of the a-AASA (R2 = 0.987, p < 0.001), P6C (R2 = 0.961, p < 0.001), AASA-P6C (R2 = 0.988, p < 0.001), PA (R2 = 0.929, p < 0.001) and a-AAA (R2 = 0.987, p < 0.001) in urine and DUS was positively correlated also. Results l i The concentration of a-AASA (R2 = 0.958, p < 0.001), P6C (R2 = 0.968, p < 0.001), AASA-P6C (R2 = 0.981, p < 0.001), PA (R2 = 0.919, p < 0.001) and a-AAA (R2 = 0.940, p < 0.001) was positively correlated between plasma and serum. Similarly, the concentration of the a-AASA (R2 = 0.987, p < 0.001), P6C (R2 = 0.961, p < 0.001), AASA-P6C (R2 = 0.988, p < 0.001), PA (R2 = 0.929, p < 0.001) and a-AAA (R2 = 0.987, p < 0.001) in urine and DUS was positively correlated also. The PA (R2 = 0.958, p < 0.001), a-AASA (R2 = 0.733, p < 0.001) and AASA-P6C (R2 = 0.826, p < 0.001) concentrations were highly correlated between DBS and plasma samples, but no obvious correlation was found in P6C (R2 = 0.039, p = 0.241) and a-AAA (R2 = 0.419, p = 0.005) concentra- tions (Fig. S3). Compared the metabolites concentrations in DBS and plasma, the P6C, AASA-P6C and PA levels were higher in plasma (p < 0.001), and a-AAA level was higher in DBS (p < 0.001). Results l i Table 1. Summary of clinical and molecular findings of the fifteen PDE patients. Samples α-AASA P6C α-AASA-P6C PA α-AAA PDE patients/ (1.3–8.6y), n = 15 Plasma (µmol/L) 4.89 (0.59–14.57) 5.99 (2.95–11.58) 10.87 (4.40–24.17) 4.45 (1.49–8.12) 4.05 (2.19–6.69) Serum (µmol/L) 3.86 (0.28–11.36) 5.93 (3.25–10.17) 9.79 (3.73–20.65) 4.72 (1.63–8.29) 4.49 (2.55–6.95) DBS (µmol/L) 3.80 (1.17–9.49) 0.64 (0.38–1.56) 4.44 (2.30–10.04) 3.38 (1.38–5.81) 5.64 (2.66–9.45) Urine (µmol/mmolCr) 31.4 (1.81–94.35) 10.53 (1.41–27.66) 41.95 (3.22–122.00) 0.13 (0.02–0.38) 27.99 (12.25–74.89) DUS (µmol/mmolCr) 30.05 (2.07–86.32) 10.44 (2.69–28.55) 40.49 (4.76–114.87) 0.13 (0.01–0.37) 28.54 (10.64–75.47) Controls/(1–13y) Samples/mean ages (ranges) α-AASA P6C α-AASA-P6C PA α-AAA Plasma (µmol/L)/n = 28 0.00 0.40 (0.25–0.58) 0.40 (0.25–0.58) 1.50 (0.38–7.61) 2.22 (1.04–4.72) Serum (µmol/L)/n = 25 0.00 0.35 (0.34–0.378) 0.35 (0.34–0.38) 0.91 (0.33–3.52) 2.00 (0.97–4.23) DBS (µmol/L)/n = 25 0.02 (0.00–0.08) 0.17 (0.15–0.23) 0.19 (0.15–0.26) 0.81 (0.17–3.37) 1.60 (0.91–2.90) Urine (µmol/mmolCr)/n = 25 0.12 (0.00–0.53) 0.43 (0.10–1.10) 0.55 (0.18–1.11) 0.17 (0.00–0.94) 15.92 (2.62–51.12) DUS (µmol/mmolCr)/n = 25 0.10 (0.00–0.67) 0.41 (0.10–1.13) 0.51 (0.18–1.13) 0.31 (0.01–1.19) 19.72 (2.51–53.70) Table 2. The ranges of the metabolites concentrations in PDE patients and control groups. Table 2. The ranges of the metabolites concentrations in PDE patients and control groups. Results l i Scientific Reports \| (2019) 9:11371 \| https://doi.org/10.1038/s41598-019-47882-2 www.nature.com/scientificreports/ No./ Sex Seizure onset age Age at test Pyridoxine dose Duration of pyridoxine treatment ALDH7A1 mutations (NM_001182.4) Development Xue et al., 201514 Xue et al., 201615 Language Motor 1/F 3m 4y3m 360 mg/d 3y10m c.1061 A > G (p.Y354C); Deletion of exon 8–13 Moderate Normal / / 2/M 3.5m 8y7m 200 mg/d 4y c.1553 G > C (p.R518X); c.1061 A > G (p.Y354C) Normal Normal / / 3/M 23d 3y4m 90 mg/d 3y c.1279 G > C (p.E427Q); c.1279 G > C (p.E427Q) Mild Normal Mild/1y4m Mild/1y11m 4/F 6m 4y5m 180 mg/d 4m c.1547 A > G (p.Y516 C); c.212 C > T (p.P71L) Normal Normal / / 5/F 2d 3y 150 mg/d 3y c.1008 + 1 G > A (IVS11 + 1 G > A); c.796 C > T (p.R266X) Mild Normal Mild/1y Moderate/1y6m 6/M 1d 5y6m 180 mg/d 3y9m IVS17-1_7delCCACTAG + c.1566_1567delTA; c.871 + 5 G > A (IVS9 + 5 G > A) Moderate Normal Severe/3y5m Severe/4y 7/F 1y1m 5y4m 180 mg/d 1y4m c.1279 G > C (p.E427Q); c.986 G > A (p.R329K) Mild Normal / / 8/F 8d 6y1m 180 mg/d 5y2m c.965 C > T (p.A322V); c.952 G > C (p.A318P) Normal Normal Mild/4y Mild/4y7m 9/M 2m 5y8m 180 mg/d 5y1m c.410 G > A (p.G137E); c.1008 + 1 G > A (IVS11 + 1 G > A) Normal Mild Severe/3y7m Severe/4y2m 10/F 1d 4y5m 150 mg/d 4y c.1415 + 1 G > T (IVS15 + 1 G > T); c.871 + 5 G > A (IVS9 + 5 G > A) Mild Normal / / 11/M 1d 3y10m 120 mg/d 3y c.1531 G > A (p.D511N); c.1008 + 1 G > A (IVS11 + 1 G > A) Severe Normal Severe/1y9m Severe/2y3m 12/F 5m 5y3m 240 mg/d 3y5m c.1061 A > G (p.Y354C); c.1008 + 1 G > A (IVS11 + 1 G > A) Mild Normal / / 13/M 1m 4y9m 90 mg/d 2m c.1547 A > G (p.Y516 C); c.1061 A > G (p.Y354C); Mild Mild / / 14/F 1.5m 1y4m 180 mg/d 11m c.1547 A > G (p.Y516 C); c.1547 A > G (p.Y516 C) Normal Normal / / 15/M 8d 2y6m 30 mg/d 2y c.1547 A > G (p.Y516 C); c.1072 C > T (p.R358X) Mild Normal / Mild/1y Table 1. Summary of clinical and molecular findings of the fifteen PDE patients. Discussion Early recognition and diagnosis of PDE were highly desirable as prompt treatment could maximize benefits18,19. The availability of biomarker testing would facilitate early diagnosis of this treatable condition. This might be also potentially applicable to newborn screening. Currently, most researches on biochemical detection were focused on one or several analytes, and restricted to the limited number of PDE patients8,10,17. Here, we simultaneously detected a-AASA, P6C, AASA-P6C, PA and a-AAA from five different types of samples, including plasma, serum, DBS, urine and DUS, freshly collected from 15 patients with PDE, which was the most comprehensive study in a relative larger cohort at present. The reference range of each metabolite in PDE patients and control groups were determined, providing possible medical reference ranges for the biochemical screening. The successful detection of the biomarkers in DBS and DUS might promote the establishment of neonatal screening of PDE.h g p g The elevated concentration ranges of a-AASA, P6C and AASA-P6C in plasma and urine were basically in agreement with previously published in PDE patients, which confirmed the reliability of our test1,7,8,13,20. In our study, though the mean concentration of PA was elevated in blood specimens in PDE patients, the PA concentra- tion ranges partially overlapped between PDE patients and control groups in all the types of samples. Previous reports suggested that PA levels might be more responsive to pyridoxine treatment and could be normalized after many months to years successful treatment with pyridoxine or with age8,18,21. Moreover, it could not be ruled out that in some of our cases, PA levels were normal from the onset as reported previously22. In addition, elevated concentrations of PA were also encountered in other conditions, such as generalized peroxisomal dysfunction, Scientific Reports \| (2019) 9:11371 \| https://doi.org/10.1038/s41598-019-47882-2 www.nature.com/scientificreports/ hyperlysinemia, defects of proline metabolism, chronic liver dysfunction, or even in patients without ent cause, giving it a low specificity19,23. The current research of biochemical detection in PDE patients mainly concentrated on plasm a-AASA, P6C and PA, seldom on the a-AAA, the direct downstream metabolite of a-AASA. At 20 Figure 3. Box-plot of a-AASA, P6C, AASA-P6C, PA and a-AAA in plasma, serum, DBS, urine and D unaffected controls and confirmed PDE patients. Figure 3. Box-plot of a-AASA, P6C, AASA-P6C, PA and a-AAA in plasma, serum, DBS, urine and DUS from unaffected controls and confirmed PDE patients. www.nature.com/scientificreports/ system with human ALDH7A1 mutation cloned into. Our study is the first to detect α-AAA levels in samples of PDE patients. Different from those reported previously, in our testing, the a-AAA concentration could be within normal range or even elevated in both the blood and urine specimens for each PDE patient. Based on the limited research at present, we could not judge whether this difference was related to the fact that more inevitable influencing factors existed in human samples than in vitro cell culture. The possible explanations for our results were as follows: The degradation of a-AAA is catalyzed by a pyridoxal-5′-phosphate (PLP) depend- ent transaminase (a-AAA transaminase)24,27. In PDE patients, the chemical condensation of the accumulating P6C and PLP lead to a secondary deficiency of PLP, which hampers the degradation of a-AAA to a-ketoadipic acid. Meanwhile, the antiquitin dysfunction reduce the synthesis of a-AAA from a-AASA. It might be the com- bination effects of these two aspects led to the final result together, which was influenced by different dosage of pyridoxine supplementation and different α-AASA dehydrogenase activity caused by various ALDH7A1 muta- tions. In addition, the wide range of a-AAA concentrations in both PDE patients and control groups indicated that it might be susceptible to other factors such as dietary lysine intake and sampling time. However, due to lacking uniform baseline of pyridoxine dosage and lysine restrictions, the explanations above were needed to be further confirmed in the future. Our results showed that a-AAA was not specific enough to be a biomarker of PDE. However, considering the better stability of a-AAA and PA, we suggested that a-AAA and PA could be simultaneously quantified with a-AASA and P6C as an auxiliary index. The individual only with high a-AAA and/or PA level should receive a further detection to avoid missed diagnosis caused by degradation of other biomarkers, though that rarely happens.h The concentrations of a-AASA, P6C, AASA-P6C, PA and a-AAA in plasma, serum, DBS as well as in urine and DUS, varied considerably in our patients. The difference between patients could be several times or even up to dozens of times. This remarkably wide range of metabolites levels had also been reported previously1,7. We had no clear explanation for this wide range. It might reflect different levels of a-AASA dehydrogenase residual activity caused by different ALDH7A1 mutations or dietary protein (L-lysine) intake. www.nature.com/scientificreports/ Few investiga- tions of the relationship between metabolites concentrations and either pyridoxine dose or neurodevelopmen- tal phenotype in PDE patients have been reported. Sadilkova et al.8 reported that in five patients with PDE, the one receiving the highest daily dose of pyridoxine had the lowest α-AASA. However, in our study, patient 5 and patient 12 usually had the highest metabolites levels, who had very different onset age (2 days and 5 months, respectively) and had been receiving relative higher daily dose of pyridoxine (150 mg/d and 240 mg/d respectively) for more than 3 years in both; while the lowest metabolites levels were usually present in patient 3 and patient 13, who had been receiving the lower daily dose of pyridoxine (90 mg/d in both) for 3 years and 2 months respectively. Furthermore, all these four patients had mild language and/or motor development delay, without obvious difference in degree. So, our study indicated that no clear correlations existed between the metabolites levels and age (recommend for uses in patients more than 1 year old), pyridoxine dosage or psych- omotor development, as well as and onset age and duration of pyridoxine treatment. Recently, triple therapy on PDE patients showed improved biomarkers and phychomotor development, and the direct correlation between lysine levels and PDE biomarkers had been reported5. We had not concurrently evaluated the dietary intake because of no specific diet restriction of our patients at the time of sample collected, and did not clear if various lysine intake attributed to the variation.f y As shown in Table 1, most patients had normal motor development but different degrees of language barriers, which was consistent with the literature reports that verbal skills were more impaired than nonverbal skills18,19,28. The clinical features of 7 of the 15 patients had been described previously14,15. During the 2-year follow-up peri- ods from the first study14, the psychomotor development of these patients improved with age, and the gaps with normal children were gradually narrowing. This was encouraging as early initiation of treatment was expected to continuously improve long-term prognosis for PDE patients. However, it was a pity that although triple therapy had been reported to be highly efficient in decreased accumulation of PDE biomarkers and improved develop- ment in the majority of PDE patients4,5, it had been not yet available in China at present and none of our patients were started on it. Discussion hyperlysinemia, defects of proline metabolism, chronic liver dysfunction, or even in patients without any appar- ent cause, giving it a low specificity19,23. Th t h f bi h i l d t ti i PDE ti t i l t t d l d i hyperlysinemia, defects of proline metabolism, chronic liver dysfunction, or even in patients without any appar- ent cause, giving it a low specificity19,23.h i The current research of biochemical detection in PDE patients mainly concentrated on plasma and urine a-AASA, P6C and PA, seldom on the a-AAA, the direct downstream metabolite of a-AASA. At 2003, Baxter reviewed that the a-AAA level was normal in patients with pyridoxine-dependent seizures without genetic diagnosis24. Recently, Crowther et al.25 demonstrated that α-AAA was decreased in fibroblasts from PDE patients compared to controls, and similarly, Coughlin et al.26 reported that α-AAA production was signif- icantly decreased and correlateed well with α-AASA dehydrogenase activity in an E.coli based expression Scientific Reports \| (2019) 9:11371 \| https://doi.org/10.1038/s41598-019-47882-2 www.nature.com/scientificreports/ References 1. Mills, P. B. et al. Mutations in antiquitin in individuals with pyridoxine dependent seizures. Nat Med. 12, 307 309 (2006). 2. Tsai, C. H. & Henderson, L. M. Degradation of O-phosphohydroxylysine in rat liver. Purification and properties of 2-aminoadip semialdehyde dehydrogenase. J Biol Chem. 249(18), 5790–5792 (1974). 2. Tsai, C. H. & Henderson, L. M. Degradation of O-phosphohydroxylysine in rat liver. Purification and properties of 2-amino semialdehyde dehydrogenase. J Biol Chem. 249(18), 5790–5792 (1974). y y g ( ) ( ) 3. Jansen, L. A. et al. Glial localization of antiquitin: implications for pyridoxine-dependent epilepsy. Ann Neurol. 75(1), 22–32. hl d l l h h d l d d l d q p py p p p y 4. Coughlin, C. R. 2nd. et al. Triple therapy with pyridoxine, arginine supplementation and dietary lysine restriction in pyridoxine dependent epilepsy: Neurodevelopmental outcome. Mol Genet Metab. 116(1-2), 35–43 (2015).f 4. Coughlin, C. R. 2nd. et al. Triple therapy with pyridoxine, arginine supplementation and dietary lysin dependent epilepsy: Neurodevelopmental outcome. Mol Genet Metab. 116(1-2), 35–43 (2015).f 5. Yuzyuk, T. et al. Effect of dietary lysine restriction and arginine supplementation in two patients with pyridoxine-dependent epilepsy. Mol Genet Metab. 118(3), 167–172 (2016).i y 6. Wempe, M. F. et al. Identification of a novel biomarker for pyridoxine-dependent epilepsy: Implications for newborn screening. J Inherit Metab Dis. 42(3), 565–574 (2019). 7. Bok, L. A. et al. Pyridoxine-dependent seizures in Dutch patients: diagnosis by elevated urinary alpha-aminoadipic semialdehyde levels. Arch Dis Child. 92(8), 687–689 (2007). 8. Sadilkova, K., Gospe, S. M. Jr. & Hahn, S. H. Simultaneous determination of alpha-aminoadipic semialdehyde, piperideine-6- carboxylate and pipecolic acid by LC-MS/MS for pyridoxine-dependent seizures and folinic acid-responsive seizures. J Neurosci Methods. 184(1), 136–141 (2009). ( ) ( ) 9. Plecko, B. et al. Biochemical and Molecular Characterization of 18 Patients With Pyridoxine-Dependent Epilepsy and Mutations o the Antiquitin (ALDH7A1) Gene. Hum Mutat. 28(1), 19–26 (2007). q 0. Jung, S., Tran, N. T., Gospe, S. M. Jr. & Hahn, S. H. Preliminary investigation of the use of newborn dried blood spots for screening pyridoxine-dependent epilepsy by LC-MS/MS. Mol Genet Metab. 110(3), 237–240 (2013).i py p p p y y 1. Mathew, E. M. et al. Biomarker Profiling for Pyridoxine Dependent Epilepsy in Dried Blood Spots by HILIC-ESI-MS. Int J Ana Chem. 2018, 2583215 (2018). 2. Pena, I. A. et al. www.nature.com/scientificreports/ doubt that they were the best choices when the samples need to be transported to other laboratories for testing, or be stored for future use.i In conclusion, a modified method was used to simultaneously quantify the a-AASA, P6C, AASA-P6C, PA and a-AAA in different samples including plasma, serum, DBS, urine and DUS. No definite correlations was found between the metabolites levels and age, pyridoxine dosage or psychomotor development in PDE patients. The reference ranges of the metabolites concentration in each sample were determined for PDE patients and con- trol groups respectively, which provided more options for the diverse sample collection in the biochemical tests according to practical requirements. With treatment modality of newly triple therapy investigated, biomarker study might play important roles not only on diagnosis but also on treatment monitoring and fine toning the diet. Reliable biomarker study also helped the interpretation of variants with unknown significance when genetic test is not definitive. The persistently elevated analytes with good correlation between plasma and DBS, as well as urine and DUS made neonatal screening using DBS and DUS possible. g g p However, we acknowledged that there were several limitations in our study. First, healthy control samples should be investigated to establish a reference range, and individuals with more wide age ranges should be included into the control group. Second, more PDE patients samples should be studied for validation, especially the samples prior to pyridoxine treatment. And patients and controls under 1 year old should be included to be further analyzed. Third, other factors, such as genotypes, should also be considered and analyzed in the future. Forth, series follow up measurement might provide the information if other factors affected the level of biomark- ers, such as dietary intake. References Simultaneous detection of lysine metabolites by a single LC-MS/MS method: monitoring lysine degradation in mouse plasma. Springerplus. 5, 172 (2016).i 3. Yuzyuk, T. et al. A novel method for simultaneous quantification of alpha-aminoadipic semialdehyde/piperideine-6-carboxylate and pipecolic acid in plasma and urine. J Chromatogr B Analyt Technol Biomed Life Sci. 1017–1018, 145–52 (2016). p p p g y f 14. Xue, J. et al. A cohort study of pyridoxine-dependent epilepsy and high prevalence of splice site IVS11 + 1G > A mutation in Chinese patients. Epilepsy Res. 118, 1–4 (2015). p p y 5. Xue, J. et al. Clinical and genetic characteristics and detection of urinary pipecolic acid in pyridoxine dependent epilepsy. Zhonghua Er Ke Za Zhi. 54(8), 592–596 (2016).h 16. Zhang, Z. et al. The Restandardization of the Adaptive Scale of Infant and Children. Chinese. Journal of Clinical Psychology. 3(1), 6–11 (1995). 7. Ferrer-López, I. et al. Determination of urinary alpha-aminoadipic semialdehyde by LC-MS/MS in patients with congenita metabolic diseases. J Chromatogr B Analyt Technol Biomed Life Sci. 944, 141–143 (2014). 18. Basura, G. J., Hagland, S. P., Wiltse, A. M. & Gospe, S. M. Jr. Clinical features and the management of pyridoxine-dependent and pyridoxine-responsive seizures: review of 63 North American cases submitted to a patient registry. Eur J Pediatr. 168(6), 697–704 (2009). ( ) 9. Stockler, S. et al. Pyridoxine dependent epilepsy and antiquitin deficiency: clinical and molecular characteristics and recommendations for diagnosis, treatment and follow-up. Mol Genet Metab. 104(1-2), 48–60 (2011).f 20. Salomons, G. S. et al. An intriguing “silent” mutation and a founder effect in antiquitin (ALDH7A1). Ann Neurol. 62(4), 414–418 (2007). 21. Bennett, C. L. et al. Prevalence of ALDH7A1 mutations in 18 North American pyridoxine-dependent seizure (PDS) patients. Epilepsia. 50(5), 1167–1175 (2009). p p 2. Mercimek-Mahmutoglu, S., Donner, E. J. & Siriwardena, K. Normal plasma pipecolic acid level in pyridoxine dependent epilepsy due to ALDH7A1 mutations. Mol Genet Metab. 110(1–2), 197 (2013). 23. Peduto, A. et al. Hyperpipecolic acidaemia: a diagnostic tool for peroxisomal disorders. Mol Genet Metab. 82(3), 224–230 (2004). 24 B P P id i d d i li i l d bi h i l d Bi hi Bi h A 1647(1 2) 36 41 . Peduto, A. et al. Hyperpipecolic acidaemia: a diagnostic tool for p d d d l l d b h 23. Peduto, A. et al. Hyperpipecolic acidaemia: a diagnostic tool for peroxisomal disorders. Mol Genet Metab. www.nature.com/scientificreports/ Plasma and urine were the most commonly used specimens for biomarkers determination in PDE patients. There was significant positive correlation between the metabolites concentrations in plasma and serum. When testing, one could choose plasma or serum, depending on the actual situation at that time, for example, the type of samples the other upcoming tests required. The significant positive correlation between the analytes levels in urine and DUS provided more options for choosing urine specimen. Our results here showed the lysine metabo- lites were more stable in DBS and DUS, than in other types of samples. Additionally, DBS and DUS were easier to be prepared with a small amount of blood and urine, and were ideal alternative samples compared to the others, particularly for newborn. Therefore, the successfully simultaneously detection of the biomarkers in DBS and DUS here suggested possibility of neonatal screening for PDE. This feasibility of newborn screening for PDE had ever been suggested by Jung et al.10 and Mathew et al.11, after implementing the detection of a-AASA and P6C in neo- natal DBS using LC-MS/MS or a novel method HILIC-ESI-MS respectively. Recently, Wempe et al.6 reported a novel biomarker 6-oxo-PIP that might be used in newborn screening of PDE. While further studies were needed to establish the sensitivity and specificity of 6-oxo-PIP for PDE. Our data showed that among the metabolites with specificity in PDE (a-AASA, P6C and AASA-P6C), the correlation between plasma and DBS was signifi- cantly positive for a-AASA and AASA-P6C, but not for P6C. The level of P6C in DBS was much lower than that in plasma, though the concentration was also high enough to recognize PDE. This indicated that P6C might not be so stable in DBS. Considering the difficulty in precise quantification of a-AASA or P6C respectively due to the spontaneous equilibrium between them, it seemed that AASA-P6C might be the most reliable indicator in screen- ing. However, limited to that all the specimens here were post-treatment, larger difference of the metabolites concentrations was expected between naive patient and control, which require further pilot study before applying to newborn screening. In addition, for the advantages of DBS and DUS in transportation and preservation, no Scientific Reports \| (2019) 9:11371 \| https://doi.org/10.1038/s41598-019-47882-2 www.nature.com/scientificreports/ Acknowledgements g We thank the patients and their families for participating. This work was supported by National Science Foundation of China (No. 81771393), Beijing Municipal Science & Technology Commission (No. Z171100001017125). None of these authors have any conflict of interest to disclose. Author Contributions Conceived and designed the experiments: Zhixian Yang, Jiao Xue. Performed the experiments: Jiao Xue, Junjuan Wang, Pan Gong, Minhang Wu, Wenshuang Yang, Shiju Jiang. Analyzed the data: Zhixian Yang, Jiao Xue, Junjuan Wang, Tatiana Yuzyuk, Hong Li. Contributed reagents/materials/analysis tools: Ye Wu, Yuwu Jiang, Yuehua Zhang, Tatiana Yuzyuk, Hong Li. Wrote the paper: Jiao Xue. References 82(3), 224–230 (2004). 24. Baxter, P. Pyridoxine-dependent seizures: a clinical and biochemical conundrum. Biochim Biophys Acta. 1647(1–2), 36–41 (2003). 25. Crowther, L. M. et al. New insights into human lysine degradation pathways with relevance to pyridoxine-dependent epilepsy due to antiquitin deficiency. J Inherit Metab Dis. [Epub ahead of print] (2019).h her, L. M. et al. New insights into human lysine degradation pathwa 25. Crowther, L. M. et al. New insights into human lysine degradation pathways with releva to antiquitin deficiency. J Inherit Metab Dis. [Epub ahead of print] (2019).h i 26. Coughlin, C. R. 2nd. et al. The genotypic spectrum of ALDH7A1 mutations resulting in pyridoxine dependent epilepsy: A common epileptic encephalopathy. J Inherit Metab Dis. 42(2), 353–361 (2019).l h epileptic encephalopathy. J Inherit Metab Dis. 42(2), 353–361 (20 p p p p y ( ) ( ) 27. Plecko, B. et al. Pipecolic acid elevation in plasma and cerebrospinal fluid of two patients with pyridoxine-dependent epilepsy Neurol. 48(1), 121–125 (2000). 28. Gospe, S. M. Jr. Neonatal vitamin-responsive epileptic encephalopathies. Chang Gung Med J. 33(1), 1–12 (2010). 28. Gospe, S. M. Jr. Neonatal vitamin-responsive epileptic encephalopathies. Chang Gung Med J. 33(1), 1–12 (2010). Scientific Reports \| (2019) 9:11371 \| https://doi.org/10.1038/s41598-019-47882-2 www.nature.com/scientificreports/ Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-019-47882-2. Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-019-47882-2 Competing Interests: The authors declare no competing interests. Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre- ative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not per- mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2019 Scientific Reports \| (2019) 9:11371 \| https://doi.org/10.1038/s41598-019-47882-2 10
https://openalex.org/W2970589605	https://hal.archives-ouvertes.fr/hal-02490536/file/2020_P%C3%A9ron%20et%20al.%20Anim%20Biotelem%20--%20flight%20height%20error.pdf	English	null	The challenges of estimating the distribution of flight heights from telemetry or altimetry data	bioRxiv (Cold Spring Harbor Laboratory)	2,019	cc-by	11,986	Abstract Background: Global positioning systems (GPS) and altimeters are increasingly used to monitor vertical space use by aerial species, a key aspect of their ecological niche, that we need to know to manage our own use of the air‑ space, and to protect those species. However, there are various sources of error in flight height data (“height” above ground, as opposed to “altitude” above a reference like the sea level). First the altitude is measured with a vertical error from the devices themselves. Then there is error in the ground elevation below the tracked animals, which translates into error in flight height computed as the difference between altitude and ground elevation. Finally, there is error in the horizontal position of the animals, which translates into error in the predicted ground elevation below the animals. We used controlled field trials, simulations, and the reanalysis of raptor case studies with state-space models to illustrate the effect of improper error management. Results: Errors of a magnitude of 20 m appear in benign conditions for barometric altimeters and GPS vertical positioning (expected to be larger in more challenging context). These errors distort the shape of the distribution of flight heights, inflate the variance in flight height, bias behavioural state assignments, correlations with environmental covariates, and airspace management recommendations. Improper data filters such as removing all negative flight height records introduce several biases in the remaining dataset, and preclude the opportunity to leverage unam‑ biguous errors to help with model fitting. Analyses that ignore the variance around the mean flight height, e.g., those based on linear models of flight height, and those that ignore the variance inflation caused by telemetry errors, lead to incorrect inferences. Conclusion: The state-space modelling framework, now in widespread use by ecologists and increasingly often automatically implemented within on-board GPS data processing algorithms, makes it possible to fit flight models directly to the output of GPS devices, with minimal data pre-selection, and to analyse the full distribution of flight heights, not just the mean. In addition to basic research about aerial niches, behaviour quantification, and environ‑ mental interactions, we highlight the applied relevance of our recommendations for airspace management and the conservation of aerial wildlife. Keywords: Flight height, 3D, Movement variance, State-space model, Telemetry error, Soaring The challenges of estimating the distribution of flight heights from telemetry or altimetry data Guillaume Péron1* , Justin M. Calabrese2,3, Olivier Duriez4, Christen H. Fleming2,3, Ruth García‑Jiménez5, Alison Johnston6,7, Sergio A. Lambertucci8, Kamran Safi9 and Emily L. C. Shepard10 Animal Biotelemetry Animal Biotelemetry Péron et al. Anim Biotelemetry (2020) 8:5 https://doi.org/10.1186/s40317-020-00194-z Background Describing the distribution of animals in environmental space is fundamental to understanding their resource requirements, cognitive processes, energetic strate- gies, and ecological characteristics. The distribution Correspondence: guillaume.peron@univ‑lyon1.fr 1 CNRS, Laboratoire de Biométrie et Biologie Evolutive UMR5558, Univ Lyon, Université Lyon 1, 69622 Villeurbanne, France Full list of author information is available at the end of the article Review of the sources of error in flight height data from GPS and altimetershl Throughout we refer to flight height h, which is the dis- tance to the ground below the bird, different from flight altitude z. The flight altitude denotes the distance to a reference altitude, often the ellipsoid, i.e., a geometri- cally perfect (but simplistic) model of the sea level, as documented by the World Geodetic System (WGS84 or EPSG:4326). Alternatively, some GPS devices may pro- vide the altitude relative to the empirical sea level, as measured at a reference point over a reference period. For example, in France the “NGF-IGN 1969” norm means that altitude is measured relative to the mean sea level in the port of Marseille between 1884 and 1896. Alternatively again, some GPS devices may measure the altitude relative to the geoid, which is a model of the sea level if it was only influenced by the local gravitational field and the rotation of the Earth, i.e., without the effect of landmasses and wind [30]. There are databases and simple formulae to convert from one system of reference to another, but this nevertheless represents a first poten- tial source of error in flight height data. However, monitoring vertical airspace use by wildlife remains challenging. Ground-based surveys are limited in their field of vision and time window. Airborne moni- toring (e.g., from glider planes) is logistically challeng- ing and constrained by weather conditions. Radar-based methodologies are not usually specific enough to assign records to species (but see [13, 14]). Animal-borne track- ing methodologies such as global positioning systems (GPS) and altimeters have therefore become popular to monitor flying species [15]. They record data even when the animals are out of sight for ground-based observers, over extensive, potentially uninterrupted periods of time, and with no uncertainty about which species or individu- als are being monitored. For example, we can record rap- tors soaring over the high sea at night [16]. However, the data that GPS and altimeters record are not error-free [17–20]. Usually, a few unambiguously erroneous posi- tions are recorded beyond unpassable barriers like the ground [10, 21–25], making the occurrence of errors particularly more obvious in flight height data than other movement tracking data. © The Author(s) 2020. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativeco mmons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/ zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Page 2 of 13 Péron et al. Anim Biotelemetry (2020) 8:5 of animals in horizontal space has dominated ecologi- cal studies [1], however the vertical dimension is also important for flying animals, and for that matter also div- ing and tree-climbing animals [2–5]. For example, flight height data document the vertical niche and community ecology of aerial foragers [6, 7]. Flight height data quan- tify the flight strategies and associated energy allocation tactics [8, 9], and their relationships with environmental factors (e.g., [10]). Lastly, from an applied perspective, we need an accurate, error-free description of the distribu- tion of birds and other animals in the aerosphere to avoid collisions with man-made structures and aircraft, in the current context of increasing human encroachment into the airspace [11, 12]. [10], and complement them with novel data from con- trolled field trials and from simulations, in order to illustrate the stakes of proper error-handling in vertical airspace use data. Review of the sources of error in flight height data from GPS and altimetershl l Flight height above the ground is computed as h = z −zDEM x, y , where zDEM x, y is the ground altitude predicted by a digital elevation model (DEM) at the recorded horizontal position x, y , in the same system of reference as z. Errors in h can then be caused by errors in any of the three com- ponents: z , zDEM , or x, y (Fig. 1). Importantly, depending on the application, researchers might want to study z not h [8, 9]. In the list below, only the first and second sources of error influence z. The other three influence h but not z. Most of the research into ways to deal with sampling errors in positioning data has focused on horizontal animal movement [20, 26–28]. There is very little guid- ance for ecologists about the challenges specific to ver- tical space use data [29]. Many practitioners consider that vertical movement data need to be “filtered” before analysis, i.e., they discard some records before proceed- ing with the analysis. They may discard records that are too far from preceding ones (as often done for horizontal data [27]), too far beyond impassable barriers [24, 25], or obtained from an unreliable configuration of the GPS sat- ellite network [29]. Instead of discarding the more erro- neous records, researchers have also sometimes chosen to reset them to plausible values [21, 23]. However, when applied improperly, such filters can have undesirable con- sequences. We start by reviewing the sources of error in GPS and altimeter flight height data. Next, we reanalyse case studies into the flight height of three raptor species 1. Error in z when z is given by a GPS In other words, altimeters can be more accurate than GPS to monitor flight height, but only over short periods of time when the weather can be considered con- stant and the altimeter is calibrated for that weather. One should ideally regularly re-calibrate the altimeters using direct observations of flight height and accurate meas- ures of PREF and T. Unfortunately, field calibrations are rarely feasible in practice (but see [37, 38]). The conse- quence is that altimeters are often miscalibrated. The degree of miscalibration depends mostly on the weather. This generates temporal autocorrelation in the error time series. Over a restricted time period, the error patterns are thus more akin to a bias (a systematic over- or under- estimation of flight height) than to an error in the statisti- cal sense of a zero-mean, identically and independently distributed random process. Importantly, altimeter data still allow one to compute the derivative of flight height, i.e., climb rate, because the amount of bias can be con- sidered constant over short periods of time. In the fol- lowing (cf. “The magnitude of vertical errors in GPS and altimeters” in “Results”), we will directly compare the errors from GPS and altimeters using controlled field experiments. [33]. The σUERE is however reputedly larger in the verti- cal axis than the horizontal axes [19, 34], meaning that manufacturer-provided σUERE should be considered con- servative for vertical applications and should be used with appropriate caution. The vertical position dilution of precision factor (VDOP) quantifies the effect of changes in the size and spatial configuration of the available satellite network on the precision of GPS records [31, 32] (Additional file 1: Fig. S1). The more satellites are available and the more evenly spread apart they are, the more reliable the posi- tioning is. Some GPS manufacturers do provide a VDOP value for each record, but many only provide a more generic DOP value. When σUERE and VDOP are known, the error-gen- erating process can then be approximated by a Gauss- ian process with time-varying standard deviation σz(t) = VDOP(t) · σUERE (Eq. 6.45 in [32]). Therefore, the DOP is not a direct index of precision. The spread of the error distribution increases with the DOP, but the error on any given record is stochastic. 1. Error in z when z is given by a GPS If recorded by a GPS, z is affected by the “user equiva- lent range error” (UERE) and the “vertical dilution of pre- cision” (VDOP) [31, 32].hff The UERE stems from diffusion and diffraction in the atmosphere, reflection from obstacles, and receiver noise [31, 32]. The acronym UERE usually directly refers to the root mean squared error, but here we will use the nota- tion σUERE instead. σUERE is usually in the order of a few meters and considered constant over time for a given device. Some GPS manufacturers specify the horizontal σUERE , or alternatively it can be estimated from the data Péron et al. Anim Biotelemetry (2020) 8:5 Page 3 of 13 Péron et al. Anim Biotelemetry Fig. 1 Illustration of the difference between true and recorded flight height. a True flight height above ground (htrue), and true elevation above ellipsoid (ztrue). b Adding the five sources of error, with circled numbers referring to headings in “Review of the sources of error”. DEM stands for digital elevation model. c Two tracks with the same amount of error. The bird of track 1 is flying high so all the recorded flight height data remain positive despite the errors. The bird of track 2 is flying low, so some of the recorded data fall below the digital elevation model Fig. 1 Illustration of the difference between true and recorded flight height. a True flight height above ground (htrue), and true elevation above ellipsoid (ztrue). b Adding the five sources of error, with circled numbers referring to headings in “Review of the sources of error”. DEM stands for digital elevation model. c Two tracks with the same amount of error. The bird of track 1 is flying high so all the recorded flight height data remain positive despite the errors. The bird of track 2 is flying low, so some of the recorded data fall below the digital elevation model and PREF is the air pressure at an elevation of reference (both pressures in mbar or in Pascal). However, this for- mula only holds when the atmosphere is at equilibrium. Changes in temperature, pressure, and air composition, i.e., the weather, alter the link between z and P. These influences are difficult to control fully because one would need to measure the weather variables both where the bird is, and at the reference elevation immediately below the bird. 2. Error in z when z is given by an altimeter 1. Error in z when z is given by a GPS The DOP is therefore not intended to be used as a data filter (e.g., discard any data with DOP above a given threshold), but instead it should be used to model the error-generating process. Simulations of flight tracks l d fl h l We simulated flight tracks that followed Ornstein– Uhlenbeck processes [43]. This is a well-studied class of continuous-time stochastic models, which is not spe- cific to vertical movement or even to movement [43]. In the case of vertical movement, the parameters of an Ornstein–Uhlenbeck process represent the mean flight height, the variance in flight height, and the autocor- relation time. The mean flight height varied from 10 to 800 m (drawn from a uniform distribution). The variance in flight height varied from 10 to 750 m2 (6 values within this range). The autocorrelation time varied between 0.1 and 1.5 arbitrary time units (uniform distribution). We transformed the raw Ornstein–Uhlenbeck simulations using an atanh link as described by Péron et al. [10] to enforce positive flight height. Because these are simula- tions, we then knew both the true flight height and the recorded flight height, which is the true flight height plus an independent and identically distributed zero-mean Gaussian error. 2. Error in z when z is given by an altimeter If recorded using an altimeter, z is computed from the barometric pressure, using the formula z = c · T · log (PREF/P) [35, 36]. c is a calibration con- stant that mostly depends on the composition of the air (e.g., percentage of vapour) and on the gravitational field. T is the air temperature in Kelvin, P is the air pressure, 3. GPS horizontal error Page 4 of 13 Page 4 of 13 Péron et al. Anim Biotelemetry (2020) 8:5 Péron et al. Anim Biotelemetry (2020) 8:5 which we used to recalibrate the altimeter post hoc. Lastly, the drone carried a separate, on-board, altimeter. In a second, separate experiment, we attached two “Gipsy 5” GPS units (Technosmart) to an ultra-light air- craft, with a vertical distance of 1.8 m between the two units. We then flew the aircraft near Radolfzell while the two units simultaneously tracked its flight height, col- lecting one record per second for a total of 11.5 h over 5 days, flying between 0 and 243 m above ground. which we used to recalibrate the altimeter post hoc. Lastly, the drone carried a separate, on-board, altimeter. x, y is also affected by a user equivalent range error and a dilution of precision (Fig. 1). The horizontal error in x, y can thus also be described as a Gauss- ian process with time-varying standard deviation: σxy(t) = 1/ √ 2 · HDOP(t) · σUERE . Note that we use here a horizontal dilution of precision factor, HDOP. An often-overlooked consequence of errors in the horizontal position is that they introduce flaws in the link to spatially explicit environmental covariates [39, 40]. In particular, the ground elevation zDEM is extracted from a location x, y that is slightly different from the true location [24]. If the terrain is very rough, then the ground elevation at the recorded location x, y may be significantly differ- ent from the ground elevation below the actual location of the bird. In the following (cf. “Horizontal errors can cause vertical errors” in “Results”), we will use simula- tions to quantify the influence of horizontal errors. In a second, separate experiment, we attached two “Gipsy 5” GPS units (Technosmart) to an ultra-light air- craft, with a vertical distance of 1.8 m between the two units. 4. Interpolation error in zDEM 4. Interpolation error in zDEM zDEM is interpolated from discrete ground elevation measurements [41, 42]. The ground elevation is measured at a few select locations, but it is interpolated between them. The result of the interpolation is then rasterized at a set resolution, and the result is the DEM. This process can be quite imprecise [41, 42]. At a cliff, for example, the ground elevation may drop by several hundred meters within a single pixel of the DEM. 2. Error in z when z is given by an altimeter We then flew the aircraft near Radolfzell while the two units simultaneously tracked its flight height, col- lecting one record per second for a total of 11.5 h over 5 days, flying between 0 and 243 m above ground. l Third, we compared the vertical positions recorded by four different units from three different manufacturers: Technosmart (AxyTrek and Gipsy 5), Microwave (GPS- GSM 20-70), and Ornitela (GPS-GSM Ornitrack 85). We (RG and OD) carried these units to 21 known geodesic points, of which the altitude was precisely documented by the French National Geographic Institute. The units recorded their position once every minute for a total of 894, 934, 560, and 563 data points, keeping only the unit location combinations that yielded more than 25 fixes. We computed the bias and root mean squared error of the vertical measurement by comparing these data to the actual, known altitudes of the geodesic points. Importantly, the manufacturers do not use the same ref- erence to compute the altitude: microwave uses the geoid (WGS 84 EGM-96 norm), whereas the others use the mean sea level (assumed to correspond to the local refer- ence, meaning the NGF-IGN 1969 norm, but see below). We expressed all altitudes in the same norm before com- puting biases and errors, and accounted for sampling effort (number of fixes) and location when comparing the performance of different units. 5. Errors in DEM base data The original measurements from which DEMs are interpolated are not necessarily error-free either. These errors are assumed small relative to the other sources, however, there is, to our knowledge, not much informa- tion available about the base datasets from which DEM are interpolated and their precision. Results The magnitude of vertical errors in GPS and altimeters During the first controlled field trial (with the drone), DOP values between 1.2 and 1.6 indicated that the con- figuration of the satellite network was reliable through- out. Nevertheless, 6.7% of the GPS flight height records were below the rooftop height, i.e., obviously errone- ous. For the altimeter, with default settings, 10% of the records were below the rooftop height. The default set- tings of the altimeter therefore did not correspond to the atmospheric conditions during the experiment. The standard deviation of the difference between the recali- brated altimetry and the GPS data was 22 m, between the recalibrated altimetry and default-setting altimetry it was 14 m, and between the recalibrated altimetry and the on- board drone altimeter it was 19 m. This means that, with default settings, the altimeters had approximately the same precision as the GPS.i For the condors, we selected the period between 11:00 and 15:00, which concentrates condor activity and there- fore flight time, and discarded other records. For the vul- tures, we selected the period between 09:00 and 16:00. For the eagles, we selected the period between 08:00 and 17:00 and, because a lot of time is spent motionless in this species even during their core activity period, we further removed all the records that were less than 15 m from the previous record. We acknowledge the arbitrary nature of this data selection and emphasize that it is not necessary or even recommended to apply such filters before analysis. We, however, stress that in the context of the present study, the case studies perform an illustra- tive function, meaning that we use them to highlight the effect of improper error-handling, at least during the par- ticular time periods that we selected for analysis because we consider them relevant for biological inference, and that the same analytical procedures can indiscriminately be applied to other time frames. During the second controlled field trial (with two GPS units attached to the same aircraft), in 35% of cases, the lower unit was erroneously recorded above the higher unit. The standard deviation of the difference between the heights recorded by the two units was 7.1 m. The highest of the two units recorded 3% of negative flight heights. The lowest unit recorded 13% of negative flight heights. Results During the third controlled field trial (with GPS units carried to a geodesic point of precisely known alti- tude), the mean absolute bias of the vertical measure- ment was 27 m on average across units and locations. The root mean squared error ranged from 14 to 42 m depending on the unit, with a small effect of location. However, the within-session standard deviation ranged only to 28 m, suggesting that a bias in the sea level ref- erence point (probably incorrectly assumed to follow Simulations of synthetic landscapesh above ground (assuming no behavioural adjustment in the presence of wind turbines). Collision risk estimated from GPS tracks is increasingly used to make recommen- dations about the choice of locations for new turbines, or to schedule the operation of existing ones. We expected that the estimated collision risk would depend on flight parameters (mean flight height, variance in flight height), on the magnitude of errors, and on error-handling. For example, a large variance in flight height might lead to a high collision risk even if the mean flight height is beyond the collision zone. Improperly handled errors may lead to positions being erroneously recorded in the collision zone when the birds actually flew outside of it, and vice versa. The same type of thinking could be applied to other types of collision risk, e.g., antennas, utility lines, buildings with bay windows, except that the collision zone would be at a different height. The objective was to quantify the influence of horizon- tal errors. We generated synthetic landscapes of varying complexity and roughness (Additional file 1: Fig. S2). We then transposed the flight track of a lesser kestrel Falco naumanni over these synthetic landscapes. The individ- ual originally flew over extremely flat terrain (the Crau steppe in France). The data (P. Pilard and OD, unpub- lished) were collected every 3 min using a Gipsy 5 GPS unit from Technosmart, and processed through the state- space model of Péron et al. [10] to account for real sam- pling errors before use. We then added simulated random telemetry noise of controlled standard deviation. Materials and methods Controlled field trials To quantify the magnitude of the vertical error in altime- ters and GPS devices, we conducted three controlled trial experiments. First, we attached an “Ornitrack 25” GPS–altimeter unit (Ornitela) to a drone. We then flew the drone above the rooftop of the Max-Planck institute in Radolfzell, Germany, at heights ranging from 0 (drone landed on the rooftop) to 90 m. We conducted 6 flight sessions over 2 days, each lasting between 15 and 140 min, collecting one record every 10 min for a total of 30 records, flying between 0 and 100 m above the rooftop. We also moni- tored the air pressure and temperature on the rooftop, Page 5 of 13 Péron et al. Anim Biotelemetry (2020) 8:5 Péron et al. Anim Biotelemetry Raptor case studies We reanalysed the data from Péron et al. [10], where the field procedure, data selection, and data analysis proce- dures are described in full. Briefly, we studied three spe- cies of large soaring raptors: Andean condors Vultur gryphus (five juveniles, 1692 individual days of monitor- ing, 15 min interval), Griffon vultures Gyps fulvus (eight adults, 2697 individual days, 1–5 min interval), and Golden eagles Aquila chrysaetos (six adults, 3103 individ- ual days, 6–10 min interval). After applying the analyti- cal procedure, for each data point, we could compare the corrected position, an estimate of the true position, to the recorded position, which was affected by the sources of errors we listed under “Background”. Collision risk In several instances, we will illustrate the potential effect of improper data-handling on management recommen- dations by estimating the risk of collision with wind tur- bines as the proportion of records between 60 and 180 m Péron et al. Anim Biotelemetry (2020) 8:5 Page 6 of 13 Page 6 of 13 the French norm) inflated the RMSE. The average bias ranged between − 17 and + 12 m depending on the unit, after correcting for significant location effect, but with- out effect of altitude. Overall, this means that different brands of GPS devices yield different rate of error in their altitude measurements, which can impair the compari- son of datasets collected by different devices. We thereby recommend accommodating the device-specificity of the error-generating process at the data analysis step, and also that the devices record their VDOP at each record (cf. “Statistical solution” below). Further investigation or communication with manufacturers should decipher whether this stems from different fix acquisition proce- dures (e.g., satellite detection) or different post-process- ing algorithms, and should also make clear which sea level reference point different manufacturers are using. heights, i.e., the variance in the true flight height was consistently lower than the variance in the recorded flight height (Fig. 2). In the raptor case studies, we obtained the same result, with the caveat that we did not have access to the true flight height, but we could instead use the corrected flight heights (Fig. 2). l Indeed, if the vertical movement and error processes were independent, the total variance in flight height would exactly correspond to the sum of the movement and sampling variances (e.g., [44]; see also [45] and references therein). When the vertical movement and error processes are not fully independent, the total ver- tical variance is still larger than the vertical movement variance. In other words, the total vertical variance is a biased estimate of the vertical movement variance that confounds telemetry errors with rapid movements. The animals would appear more vertically mobile and with a more spread-out distribution in the aerosphere than they actually are. This type of issue is potentially quite widespread in other areas of movement ecology that pertain to horizontal movement as well, e.g., in behavioural assignment exercises that use movement variances (daily displacements, turning angles, etc.) to determine the behavioural state of animals. Collision risk f These controlled field trials, along with other similar reports [22, 34], highlight that even in benign conditions, GPS and altimeter data are sufficiently error-prone to tamper with ecological inference in many cases (range of the standard deviation of the error: 4–50 m). The issue is only suspected to be more acute in operational con- ditions when the DOP is larger, the terrain rougher, the weather more variable, and there are more obstacles to signal diffusion than in controlled field trials. Further- more, the rate of error depended on the brand of the unit and on the location, which can be of importance when comparing across studies. G G G G G 50 100 150 200 250 300 350 50 100 150 200 250 300 350 Bias in the observed variance of flight height in 3 raptor species SD of flight height (corrected) SD of flight height (uncorrected) G Species condor eagle vulture Fig. 2 Comparison between the standard deviation of the recorded flight height (y-axis) and of the corrected flight height (x-axis), assumed to represent the true flight height, in three species of large soaring raptors. Each point stands for one bird over its entire monitoring period. The state-space model that we used to correct the flight heights, and in particular its robustness to variation in sampling resolution across populations, is explained in Péron et al. [10]. The diagonal line shows where the points should be if the recorded flight heights were error-free Horizontal errors can cause vertical errors G G G G G 50 100 150 200 250 300 350 50 100 150 200 250 300 350 Bias in the observed variance of flight height in 3 raptor species SD of flight height (uncorrected) G Species condor eagle vulture Bias in the observed variance of flight height in 3 raptor species In the synthetic landscape simulations, the frequency of negative flight height records increased with the standard deviation of both the horizontal and vertical telemetry error (Additional file 1: Fig. S2a), and with the landscape roughness and complexity (Additional file 1: Fig. S2b). However, the various sources of errors acted in a mul- tiplicative way, so that even when the telemetry noise was small (SD of 1 m), the error in h could be large (SD of 20 m; Additional file 1: Fig. S2c; darkest grey curve). Perhaps unexpectedly, when the horizontal error was large, the error in the height above ground tended to be independent of the vertical error in the GPS (on average across all simulations; Additional file 1: Fig. S2c; lightest grey curve). This means that the effect of the horizontal error in the GPS can supersede the effect of the vertical error, if the terrain is rough. Even in the absence of any vertical error, the horizontal error was indeed routinely sufficient to cause 10–20% of the data points to be below ground (Additional file 1: Fig. S2a). SD of flight height (corrected) SD of flight height (corrected) SD of flight height (corrected) Fig. 2 Comparison between the standard deviation of the recorded flight height (y-axis) and of the corrected flight height (x-axis), assumed to represent the true flight height, in three species of large soaring raptors. Each point stands for one bird over its entire monitoring period. The state-space model that we used to correct the flight heights, and in particular its robustness to variation in sampling resolution across populations, is explained in Péron et al. [10]. The diagonal line shows where the points should be if the recorded flight heights were error-free Errors inflate the recorded variance in flight height In the simulations of flight tracks, errors in h inflated the variance in the distribution of recorded flight Page 7 of 13 Péron et al. Anim Biotelemetry (2020) 8:5 Péron et al. Horizontal errors can cause vertical errors Anim Biotelemetry Negative flight height records provide useful information In this section we focus on negative records, i.e., unrealistically low records, but the same logic can be applied to unrealistically high records. Negative flight height records are more likely to occur when animals are near the ground, either perched or flying. If we remove the negative records [29], perching and low flight are under-sampled in the final dataset [21]. To illustrate this point, we used a GPS-tracked flight path from a migrating juvenile osprey (Pandion haliae- tus) as it crossed the sea between the Italian mainland and Corsica [16]. During a portion of that sea cross- ing, its Ornitela GPS unit recorded flight heights that oscillated between − 2 and − 7 m below the sea level (Additional file 1: Fig. S3, inset). The amplitude of the oscillation suggested that the bird followed the swell of the waves. The complete sequence (Additional file 1: Fig. S3) depicts a progressive loss of altitude as the bird glided towards firm ground, and a period of active flap- ping flight (as per the accelerometry record) very low above the waves once the bird had lost all of its accu- mulated potential energy before reaching firm ground. These negative flight height records documented a crit- ical time period. First, the risk of having to make a sea landing were clearly much greater in the few minutes when the osprey was flying low over the waves, com- pared to the rest of the sea crossing when the bird was often soaring high [16]. In addition, when flying low, the bird had no other choice than to flap and therefore expend energy; whereas when higher above the sea, the bird had the option to soar and therefore spare energy. It is critical that negative flight height records like these are maintained, even if, instead of a fully interpretable high-resolution sequence like in this example, there are just a few isolated negative flight height records in a low-resolution dataset. Negative flight height records provide useful information In addition, if we only kept the records with positive flight height, we would obtain a biased sample of the distribution of flight height. Horizontal errors can cause vertical errors Both in simulations and in the raptor case studies, discarding negative flight height records led to the overestimation of the mean flight height in the remaining dataset, the underesti- mation of the variance in flight height, the introduc- tion of a right skew in the distribution of flight height, and the overestimation of the collision risk (Fig. 3). The latter result was because negative records mostly occurred when the bird was flying below the colli- sion zone, and thus removing negative records led to under-sample safe periods of time. Note that this par- ticular result pertains to the wind turbine application Fig. 3 Removing the negative recorded flight heights introduces biases in the distribution of the remaining flight heights. Left group of panels: in simulations, where the true flight height is known. Right group of panels: in the raptor case studies, where the corrected flight height is assumed to represent the true flight height. In all panels, the x-axis features the variance in the true (or corrected) flight height. The y-axis features the percentage bias in a mean flight height; b collision risk (proportion of time spent between 60 and 180 m above ground); c variance in flight height; and d skewness of the distribution of flight height. A percentage bias of + 10% means that the focal quantity is 10% larger after we remove the negative records Fig. 3 Removing the negative recorded flight heights introduces biases in the distribution of the remaining flight heights. Left group of panels: in simulations, where the true flight height is known. Right group of panels: in the raptor case studies, where the corrected flight height is assumed to represent the true flight height. In all panels, the x-axis features the variance in the true (or corrected) flight height. The y-axis features the percentage bias in a mean flight height; b collision risk (proportion of time spent between 60 and 180 m above ground); c variance in flight height; and d skewness of the distribution of flight height. A percentage bias of + 10% means that the focal quantity is 10% larger after we remove the negative records Fig. 3 Removing the negative recorded flight heights introduces biases in the distribution of the remaining flight heights. Left group of panels: in simulations, where the true flight height is known. The mean flight height is not sufficient to describe the distribution of flight heights Flight height datasets are often reduced to a single sum- mary metric, the mean flight height and its variation with environmental and individual covariates [29, 46–49]. This decision is mostly based on the ease of implementing spreadsheets, linear models, moving averages, or spline models. In this section, we instead call for approaches that describe the full distribution of flight heights in the aerosphere, not only the mean flight height. To jus- tify this call, we again focus on collision risk estimation. Indeed, if the variance in flight height is large enough, a proportion of time may be spent in the collision zone even if the mean flight height is outside the collision zone. In simulations, the proportion of time spent in the collision zone indeed depended on both the mean and the variance in flight height (Fig. 4a, b). In the rap- tor datasets, the estimated probability of flying in the collision zone did not decrease much for the individuals whose mean flight height was estimated above the col- lision zone (Fig. 4c). Similarly, the individuals that had an estimated mean flight height well below the collision zone were predicted to spend about 20% of their time in the collision zone (Fig. 4c). We strongly recommend that collision risk forecasts should not be based on the fixed effects of linear models, but instead on the full distribu- tion of flight heights—a recommendation that will likely hold for all studies into vertical airspace use. The simulations nicely complemented the raptor case studies by (1) eliminating any debate about whether the corrected flight heights in the raptor case studies were trustworthy or not (in the simulations, the true flight heights are exactly known) and (2) increasing the range of flight behaviours, since the raptors tended to exhibit lower percentage of time near the ground (in part because we purposely tried to exclude time spent perched) and different distributions of the sampling error. The amount of bias appeared highly dependent on the underlying flight behaviour and error distribu- tion, and therefore not easy to predict and account for without appropriate error-handling methodology. g gy Additionally, there are many other major conse- quences of discarding negative flight heights. One is the disruption of the expected balance of positive and negative errors in the remaining data. The mean flight height is not sufficient to describe the distribution of flight heights Negative flight height records only arise when the error is negative, and so removing them introduces a bias towards posi- tive errors, thereby disrupting the shape of the distri- bution of errors in the remaining data. Yet, we need the full range of errors to fit the statistical solution that we support (cf. “Discussion”). Another, unrelated con- sequence is the disruption of the sampling schedule of the remaining data. Many movement analyses are critically sensitive to the sampling schedule, and there- fore their outcome will not be the same after remov- ing the negative records. Lastly, and perhaps most importantly, negative flight height records can help fit the models that separate the error and movement processes, because they are unambiguously erroneous and can be informed as such in the model-fitting pro- cedure (cf. “Statistical solution” in “Discussion”). Some authors have applied less stringent filters, such as removing only the most negative flight height records and removing an equal amount of extremely positive flight height records. While the effect on the remain- ing distribution, and on the balance of negative and positive errors is supposedly weaker than if removing all of the negative records, we warn that the remaining records are still affected by the same error process that generated the records that were deemed too errone- ous to keep, thus the issues from the previous section (“Errors inflate the recorded variance in flight height”) still need to be addressed. In addition, these extremely erroneous records are potentially the most informative regarding the shape of the error distribution (cf. “Sta- tistical solution” in “Discussion”). Horizontal errors can cause vertical errors Right group of panels: in the raptor case studies, where the corrected flight height is assumed to represent the true flight height. In all panels, the x-axis features the variance in the true (or corrected) flight height. The y-axis features the percentage bias in a mean flight height; b collision risk (proportion of time spent between 60 and 180 m above ground); c variance in flight height; and d skewness of the distribution of flight height. A percentage bias of + 10% means that the focal quantity is 10% larger after we remove the negative records Péron et al. Anim Biotelemetry (2020) 8:5 Page 8 of 13 The mean flight height is not sufficient to describe the distribution of flight heights case only; in other types of collision risk, e.g., build- ings and utility lines, the collision zone starts closer to the ground. Statistical solutions c Same as (b) but using real datasets collected from three raptor species, where the corrected flight height is assumed to represent the true flight height. Each symbol stands for an individual over its entire monitoring period True height at me t Recorded height at me t Observaon error Movement process True height at me t+1 Recorded height at me t+1 Movement process Observaon error Fig. 5 Schematic overview of the principles of a state-space model as applied to the correction of sampling errors in flight height data. The movement (or state) process accounts for the distribution of true flight heights. The observation process introduces sampling errors of various origins (cf. “Review of the sources of error” in “Background”) and yields the recorded flight heights. It also accounts for the sampling schedule. By fitting this model to recorded flight height time series, we can retrospectively compute the corrected flight height, an estimate of the true flight height of the parameters of a state-space model are separately estimable, a phenomenon called “weak identifiability” that occurs when the sampling variance largely exceeds the process variance. An example of weak identifiability is when the difference between two classes of individuals are larger than the differences within the classes [55]. In addition, there are large statistical correlations between variance parameters in a movement model [52], making it extra difficult to accurately separate movements and errors in sparse datasets. In that context, unambiguously erroneous records, such as negative flight heights, repre- sent an additional source of information [20]. They can help separate the process and sampling variances [10] and solve issues of weak identifiability. True height at me t Recorded height at me t Observaon error Movement process True height at me t+1 Recorded height at me t+1 Movement process Observaon error Fig. 5 Schematic overview of the principles of a state-space model as applied to the correction of sampling errors in flight height data. The movement (or state) process accounts for the distribution of true flight heights. The observation process introduces sampling errors of various origins (cf. “Review of the sources of error” in “Background”) and yields the recorded flight heights. It also accounts for the sampling schedule. Statistical solutions Our results illustrate how the improper treatment of vertical errors in telemetry data can flaw the inference about the use of the aerosphere by flying animals. To avoid these issues, the state-space model framework [50] (Fig. 5) has a structure that is naturally aligned with the challenges of sampling errors in vertical space-use data. A state-space model is a stochastic model describing the changes over time in a state variable (here, the true flight height), when that variable is imperfectly observed (here, the recorded flight height). There is a “state pro- cess”, separated from an “observation process” (Fig. 5). State-space models are routinely used to correct for posi- tioning errors in satellite-tracking data (chap. 6 in [32]), including in wildlife studies [20, 26, 33, 51–53]. Impor- tantly, these applications are not to be confused with another application of state-space models to movement data, when the focal state variable is a “behavioural state” whose Markovian transitions drive changes in movement rates [8, 9, 54]. Indeed, when the objective is, like in this study, to correct for positioning errors, the state variable is the position itself. Péron et al. Anim Biotelemetry (2020) 8:5 Page 9 of 13 Péron et al. Anim Biotelemetry (2 Fig. 4 The variance in flight height influences the percentage of time spent in the collision zone of a wind farm (grey area, between 60 and 180 m). a Four simulated tracks (where the true flight height is known) with the same mean flight height (200 m) but different variances (10, 50, 100, and 250 m2). b More extensive simulations. Each point corresponds to one simulated track with a different mean flight height. c Same as (b) but using real datasets collected from three raptor species, where the corrected flight height is assumed to represent the true flight height. Each symbol stands for an individual over its entire monitoring period Fig. 4 The variance in flight height influences the percentage of time spent in the collision zone of a wind farm (grey area, between 60 and 180 m). a Four simulated tracks (where the true flight height is known) with the same mean flight height (200 m) but different variances (10, 50, 100, and 250 m2). b More extensive simulations. Each point corresponds to one simulated track with a different mean flight height. Statistical solutions By fitting this model to recorded flight height time series, we can retrospectively compute the corrected flight height, an estimate of the true flight height i As a perspective, we stress that there are also ways to obtain unambiguously correct records. These records could in theory perform a role similar to that of unam- biguously erroneous records. For example, sometimes the position of the animals can be confirmed, e.g., at a docu- mented feeding site, a nest, or by an incidental ground- based sighting. Those records can be matched to the GPS track, yielding an exact measure of the local error. Animal-borne devices may also include a transponder designed to signal passage near strategically placed emit- ters (e.g., [56]). This type of validation data is routinely used in other applications of the GPS technology [32]. Lastly, the state-space framework is naturally conducive to the joint analysis of multiple sources of error-prone In studies of flight height, the movement model can be set up such that the state variable always stays above zero. Then, if the recorded flight height is − 7 m, the model “knows” that the error was at least 7 m [22], as for example was the case in the osprey example (Additional file 1: Fig. S3). Actually, the presence of unambiguously erroneous records makes flight height studies better- suited to apply state-space models than many studies into horizontal space use by animals. Indeed, even when in theory the model is estimable, sometimes only a subset Péron et al. Anim Biotelemetry (2020) 8:5 Péron et al. Anim Biotelemetry (2020) 8:5 Page 10 of 13 Kalman filter. This algorithm is fast, but requires all the model processes to be Gaussian or approximately Gaussian (no truncation or constraint, no excess extreme values, no excess kurtosis or skew). data (e.g., [57]). In flight height studies, it is therefore possible to jointly analyse GPS and altimeter data, or multiple GPS streams coming from the same animal. This double-data approach is expected to help with statistical covariance issues, but cannot be expected to fully resolve all identifiability issues [58], which only error-free valida- tion data can do. • The TMB package for R [60] approximates the like- lihood of the state-space model using the automatic differentiation algorithm with Laplace approxima- tion. Statistical solutions That approach makes computing times shorter than the next option, while still allowing for flexible modelling such as non-Gaussian errors [26], custom link functions [10], or multiple data streams. We should eventually stress that several wildlife GPS manufacturers already use a state-space model as part of the on-board data pre-processing algorithm, i.e., the released data have already been corrected by a propri- etary state-space algorithm which may furthermore rely on proprietary validation data (Ornitela staff, pers. comm.). From our experience, in wildlife applications, these pre-processing algorithms are only applied during “bursts” of high-frequency data acquisition, not when the users request a more traditional low-frequency data acquisition schedule. Importantly, the data may not be pre-processed across bursts. The error from the first location of a burst is then carried over the entire burst sequence. Flight height tracks affected by this issue would exhibit a staircase-shaped profile. Overall, this type of data pre-processing trades a lower error variance against a larger error autocorrelation. Additional state- space modelling of the released pre-processed data can deal with this type of error autocorrelation, but the mod- els need to be custom-made, i.e., are not routinely imple- mented in software. Perhaps more worryingly, some commercially available GPS units apparently simply trun- cate the recorded height at zero above sea level (pers. obs.). We call for a more open approach to these data manipulations, including making the raw, unprocessed GPS records available, in addition to any pre-processed data, and with a formal description of the pre-processing algorithm. Indeed manufacturers may not be aware of the specificties of vertical animal movements. Vertical move- ments are faster and less temporally autocorrelated than horizontal movements, and they depend on specific envi- ronmental covariates [10], making it necessary that end users obtain the unprocessed flight height data to param- eterize the most ecologically relevant models.i • The Monte Carlo Markov Chain Bayesian framework [61–63] generates parameter distributions that itera- tively converge towards the solution. This option is the most flexible in terms of nonlinearities and non- Gaussian features, such as truncated distributions [20], but the computing time can be prohibitive for large datasets. • The crawl [33] and ctmm [59] packages for R com- pute the likelihood of the state-space model using a Acknowledgements A. Scharf and H. Wehner (MPI) flew the drone and assisted with data collec‑ tion. M. Quetting flew the ultra-light aircraft and E. Lempidakis assisted with data collection. We warmly thank everyone involved in the collection and management of the raptor tracking data, and in particular C. Itty for the eagle data, and the organizations Asters and Vautours des Baronnies for the help with field experiments. Abbreviations h: Flight height above ground; z: Flight altitude (relative to the same reference as the DEM, e.g., the ellipsoid); DEM: Digital elevation model; UERE: User equivalent range error; DOP: Dilution of precision; SD: Standard deviation. Availability of data and materials The data have been uploaded to MoveBank (http://www.movebank.org). Availability of data and materials The data have been uploaded to MoveBank (http://www.movebank.org). Data requirements and data quality checkshi Anim Biotelemetry to correlate with movement velocity or movement behav- iour can also be incorporated using linear links with the movement model parameters (autocorrelation time, dif- fusion rate). about the effect of errors on the estimation of aerial niche overlaps and the quantification of behaviours seem particularly relevant in this context. In conclusion, the issue of properly handling errors in flight height data is key to any aeroecology study. We strongly advise against ad-hoc “data quality” filters, and against statistical tools that only document variation in the mean flight height instead of the full distribution of flight height. Our proposed statistical framework based on state-space models and the analysis of the full distri- bution of flight heights requires interdisciplinary work between experts in flight behaviour and experts in data analysis, and the emergence of interface specialists, but the insights and the applied decisions based on those insights are expected to be more reliable. Supplementary information Supplementary information accompanies this paper at https://doi. org/10.1186/s40317-020-00194-z. Supplementary information accom org/10.1186/s40317-020-00194-z. Supplementary information accompanies this paper at https://doi. org/10.1186/s40317-020-00194-z. Additional file 1. Additional figures S1–S3. Additional file 1. Additional figures S1–S3. Authors’ contributions All authors conceived the project and revised the manuscript. GP performed the analyses and simulations and wrote the first draft of the manuscript. OD, RG, SAL, KS, and ES procured the data. All authors read and approved the final manuscript. Regarding applied consequences, we focused on dem- onstrating how improper methods would imperfectly quantify the time spent by GPS-tracked raptors in the rotor-swept zone of wind turbines (Fig. 3b). There are many other human–wildlife conflicts for the use of the aerosphere, for example bird strikes near airports and disturbance of wildlife by drones and other recreational aircraft. Regarding bird strikes, GPS-based predic- tive models of bird flight height (e.g., Péron et al. [10]) might help plan ahead the operation of airports. The state-space class of model that we advocate is actually already used, in real time, to exploit bird activity data from radar monitors and generate a warning system for airport managers [65]. Regarding recreational aircraft and drones, analysing bird-borne GPS tracks may help reveal the effect of the disturbance, which is expected to increase in frequency as drones in particular become more popular [66]. The recommendations we made Funding SAL thanks PICT-BID 0725/2014 and Eppley Foundation for financial support. JMC and CHF were supported by NSF ABI 1458748. The griffon vulture project was supported by ANR-07-BLAN-0201. RGJ was supported by pre-doctoral Grant (FPI/BES-2016-077510). Availability of data and materials The data have been uploaded to MoveBank (http://www.movebank.org). Conclusion Improper error-handling methodologies yield a flawed picture of aerial niches. For example, discarding negative flight height records artificially truncates the observed distribution of flight heights (Fig. 3), and focusing on the mean flight height alone (for example when using linear models) does not fully describe the aerial niche (Fig. 4). While these observations are quite intuitive, bad prac- tices remain common enough that it was important to stress these issues and illustrate them thoroughly. On the other hand, not addressing the occurrence of errors at all would artificially spread-out the observed distribution of flight heights (Fig. 2), leading for example to increased observed vertical overlap between species and individu- als, which can modify the inference about community processes. Improper error-handling procedures would also tamper with the quantification of behaviour and flight strategies, by increasing or decreasing the observed vertical velocity, and interfere with behavioural state assignments. Lastly, errors may covary with environ- mental covariates. GPS positioning accuracy decreases with terrain roughness [19]. Thereby, selectively discard- ing records based on the number of available satellites or the dilution of precision would lead to biased sampling of terrain roughness. Wind speed decreases near the ground [64]. Discarding negative flight height records (that predominantly occur near the ground) would lead to misrepresent the relationship to wind speed. Data requirements and data quality checkshi The state-space model-fitting procedure simply require the h or z time series along with the times- tamps [10, 26, 33]. The interval between records needs to be sufficiently short that the effect of the temporal autocorrelation is visible, which in practice for raptors means an interval below 1 h and ideally below 30 min [10]. The observation error must not largely exceed the movement variance, otherwise the state-space model is likely to become unidentifiable. In practice, research- ers may therefore find the following rough data quality checks useful: check that the median interval duration is < 1 h (ideally < 30 min), that the number of fixes per day is > 4, and that the proportion of records with negative flight height is < 50%. In addition, if there are very short intervals (< 1 min) we recommend incorporating into the movement model some temporal autocorrelation in the vertical velocity, in addition to temporal autocorrelation in the vertical position. If available, the VDOP or other metrics of triangulation reliability can predict the observation error in the state- space model, using a log-linear link between the stand- ard deviation of the observation error and the VDOP, which should help with model fitting. Similarly, in case the researchers know for sure that there was no error on some of the records, they can fix the error parameter to zero for these records, which should also help with model fitting. On the other hand, the information that some recorded values are impossible is coded up using adequate link functions [10] and would thereby automat- ically inform the model about the minimum magnitude of the error on the involved records. Lastly, for a better fit to the data, environmental covariates that are expected We also acknowledge that the fitting of state-space models to space use data still requires relatively rare sta- tistical skills. Nevertheless, there are already several free, open-source computing environments to fit state-space models to vertical (and horizontal) movement data, and thereby estimate the most likely movement track as a by- product of the estimated parameters, similarly to how the individual values would be computed in a generalized mixed model with individual random effects: • The crawl [33] and ctmm [59] packages for R com- pute the likelihood of the state-space model using a Page 11 of 13 Page 11 of 13 Péron et al. Anim Biotelemetry (2020) 8:5 Péron et al. References 1. Nathan R, Getz WMWM, Revilla E, Holyoak M, Kadmon R, Saltz D, et al. A movement ecology paradigm for unifying organismal movement research. Proc Natl Acad Sci USA. 2008;105:19052–9. 1. Nathan R, Getz WMWM, Revilla E, Holyoak M, Kadmon R, Saltz D, et al. A movement ecology paradigm for unifying organismal movement research. Proc Natl Acad Sci USA. 2008;105:19052–9. 24. Katzner TE, Brandes D, Miller T, Lanzone M, Maisonneuve C, Tremblay JA, et al. Topography drives migratory flight altitude of golden eagles: implications for on-shore wind energy development. J Appl Ecol. 2012;49:1178–86. https://doi.org/10.1111/j.1365-2664.2012.02185.x. 2. Weimerskirch H, Le Corre M, Ropert-Coudert Y, Kato A, Marsac F. The three-dimensional flight of red-footed boobies: adaptations to foraging in a tropical environment? Proc Biol Sci. 2005;272:53–61. 2. Weimerskirch H, Le Corre M, Ropert-Coudert Y, Kato A, Marsac F. The three-dimensional flight of red-footed boobies: adaptations to foraging in a tropical environment? Proc Biol Sci. 2005;272:53–61. 25. Krone O, Treu G. Movement patterns of white-tailed sea eagles near wind turbines. J Wildl Manage. 2018;82:1367–75. https://doi.org/10.1002/ jwmg.21488. 3. Liechti F, Bauer S, Dhanjal-Adams KL, Emmenegger T, Zehtindjiev P, Hahn S. Miniaturized multi-sensor loggers provide new insight into year-round flight behaviour of small trans-Sahara avian migrants. Mov Ecol. 2018;6:19 https://doi.org/10.1186/s40462-018-0137-1. 26. Albertsen CM, Whoriskey K, Yurkowski D, Nielsen A, Flemming JM. Fast fitting of non-Gaussian state-space models to animal movement data via Template Model Builder. Ecology. 2015;96:2598–604. https://doi. org/10.1890/14-2101.1. 4. Bishop CM, Spivey RJ, Hawkes LA, Batbayar N, Chua B, Frappell PB, et al. The roller coaster flight strategy of bar-headed geese conserves energy during Himalayan migrations. Science. 2015;347:250–4. 4. Bishop CM, Spivey RJ, Hawkes LA, Batbayar N, Chua B, Frappell PB, et al. The roller coaster flight strategy of bar-headed geese conserves energy during Himalayan migrations. Science. 2015;347:250–4. 27. Freitas C, Lydersen C, Fedak MA, Kovacs KM. A simple new algorithm to filter marine mammal Argos locations. Mar Mamm Sci. 2008;24:315–25. https://doi.org/10.1111/j.1748-7692.2007.00180.x. g y g 5. Kunz TH, Gauthreaux SA, Hristov NI, Horn JW, Jones G, Kalko EKV, et al. Aeroecology: probing and modeling the aerosphere. Integr Comp Biol. 2007;48:1–11. https://doi.org/10.1093/icb/icn037. 5. Kunz TH, Gauthreaux SA, Hristov NI, Horn JW, Jones G, Kalko EKV, et al. Aeroecology: probing and modeling the aerosphere. Integr Comp Biol. 2007;48:1–11. https://doi.org/10.1093/icb/icn037. 28. Fleming CH, Fagan WF, Mueller T, Olson KA, Leimgruber P, Calabrese JM. References Estimating where and how animals travel: an optimal framework for path reconstruction from autocorrelated tracking data. Ecology. 2016;97:576– 82. https://doi.org/10.1890/15-1607.1. p g 6. Arlettaz R, Perrin N, Hausser J. Trophic resource partitioning and competi‑ tion between the two sibling bat species Myotis myotis and Myotis blythii. J Anim Ecol. 1997;66:897. 29. Poessel SA, Duerr AE, Hall JC, Braham MA, Katzner TE. Improving estimation of flight altitude in wildlife telemetry studies. J Appl Ecol. 2018;55:2064–70. https://doi.org/10.1111/1365-2664.13135. 7. Siemers BM, Schnitzler H-U. Echolocation signals reflect niche differentia‑ tion in five sympatric congeneric bat species. Nature. 2004;429:657–61. 7. Siemers BM, Schnitzler H-U. Echolocation signals reflect niche differentia‑ tion in five sympatric congeneric bat species. Nature. 2004;429:657–61. 8. Murgatroyd M, Photopoulou T, Underhill LG, Bouten W, Amar A. Where eagles soar: fine-resolution tracking reveals the spatiotemporal use of dif‑ ferential soaring modes in a large raptor. Ecol Evol. 2018;8:6788–99. https ://doi.org/10.1002/ece3.4189. g 30. Fowler CM. The solid earth: an introduction to global geophysics. Cam‑ bridge: Cambridge University Press; 2005. 31. Parkinson BW, Spilker JJJ, editors. Global positioning system: theory and applications, vol. 1. Washington, D.C.: American Institute of Aeronautics and Astronautics; 1996. ISBN 1-56347-106-X. 9. Pirotta E, Edwards EWJ, New L, Thompson PM. Central place foragers and moving stimuli: a hidden-state model to discriminate the processes affecting movement. J Anim Ecol. 2018;87:1116–25. 32. Sanz Subirana J, Juan Zornoza JM, Hernandez-Pajares M. GNSS data processing volume I: fundamentals and algorithms. Noordwijk: ESA Com‑ munications; 2013. 10. Péron G, Fleming CH, Duriez O, Fluhr J, Itty C, Lambertucci S, et al. The energy landscape predicts flight height and wind turbine collision hazard in three species of large soaring raptor. J Appl Ecol. 2017;54:1895–906. https://doi.org/10.1111/1365-2664.12909. 33. Johnson DS, London JM, Lea M-AMA, Durban JW. Continuous-time correlated random walk model for animal telemetry data. Ecology. 2008;89:1208–15. https://doi.org/10.1890/07-1032.1. 11. Lambertucci SA, Shepard ELC, Wilson RP. Human–wildlife conflicts in a crowded airspace. Science. 2015;348:502–4. http://www.academia. edu/12206414/Human-wildlife_conflicts_in_a_crowded_airspace. Accessed 19 July 2018. g 34. Bouten W, Baaij EW, Shamoun-Baranes J, Camphuysen KCJ. A flexible GPS tracking system for studying bird behaviour at multiple scales. J Ornithol. 2013;154:571–80. https://doi.org/10.1007/s10336-012-0908-1. 35. Monaldo FM, Goldhirsh J, Walsh EJ. Altimeter height measurement error introduced by the presence of variable cloud and rain attenuation. J Geophys Res. 1986;91:2345. https://doi.org/10.1029/JC091iC02p02345. 12. Davy CM, Ford AT, Fraser KC. Aeroconservation for the fragmented skies. Conserv Lett. 2017;10:773–80. https://doi.org/10.1111/conl.12347. 13. Consent for publication Not applicable. 14. Zaugg S, Saporta G, van Loon E, Schmaljohann H, Liechti F. Automatic identification of bird targets with radar via patterns produced by wing flapping. J R Soc Interface. 2008;5:1041–53. https://doi.org/10.1098/ rsif.2007.1349. The authors declare that they have no competing interests. 15. López-López P. Individual-based tracking systems in ornithology: welcome to the era of Big Data. Ardeola. 2016;63:103–36. https://doi. org/10.13157/arla.63.1.2016.rp5. Received: 22 October 2019 Accepted: 6 February 2020 22. Ross-Smith VH, Thaxter CB, Masden EA, Shamoun-Baranes J, Burton NHK, Wright LJ, et al. Modelling flight heights of lesser black-backed gulls and great skuas from GPS: a Bayesian approach. J Appl Ecol. 2016;53:1676–85. 23. Weimerskirch H, Bishop C, Jeanniard-du-Dot T, Prudor A, Sachs G. Frigate birds track atmospheric conditions over months-long transoceanic flights. Science. 2016;353:74–8. Consent for publication Not applicable. Competing interests The authors declare that they have no competing interests. Consent for publication Not applicable. Author details 16. Duriez O, Péron G, Gremillet D, Sforzi A, Monti F. Migrating ospreys use thermal uplift over the open sea. Biol Lett. 2018;14:20180687. https://doi. org/10.1098/rsbl.2018.0687. 1 CNRS, Laboratoire de Biométrie et Biologie Evolutive UMR5558, Univ Lyon, Université Lyon 1, 69622 Villeurbanne, France. 2 Smithsonian Conserva‑ tion Biology Institute, National Zoological Park, Front Royal, VA 22630, USA. 3 Department of Biology, University of Maryland, College Park, MD 20742, USA. 4 Centre d’Ecologie Fonctionnelle et Evolutive, UMR 5175, CNRS- Université de Montpellier-EPHE-Université Paul Valery, 1919 Route de Mende, 34293 Montpellier cedex 5, France. 5 Department of Animal Science, Faculty of Life Sciences and Engineering, University of Lleida, 25198 Lleida, Spain. 6 Cornell Lab of Ornithology, Cornell University, Ithaca, NY, USA. 7 Department of Zoology, Conservation Science Group, University of Cambridge, Cambridge, UK. 8 Grupo de Investigaciones en Biología de la Conservación, Laboratorio Ecotono, INIBIOMA (CONICET-Universidad Nacional del Comahue), Quintral 1250, 8400 Bariloche, Argentina. 9 Max Planck Institut für Ornithologie, Am Obstberg 1, 78315 Radolfzell, Germany. 10 Department of Biosciences, College of Science, Swansea University, Swansea SA2 8PP, UK. g 17. Frair JL, Nielsen SE, Merrill EH, Lele SR, Boyce MS, Munro RHM, et al. Removing GPS collar bias in habitat selection studies. J Appl Ecol. 2004;41:201–12. https://doi.org/10.1111/j.0021-8901.2004.00902.x. 18. Jerde CL, Visscher DR. GPS measurement error influences on move‑ ment model parameterization. Ecol Appl. 2005;15:806–10. https://doi. org/10.1890/04-0895. 19. D’Eon RG, Serrouya R, Smith G, Kochanny CO. GPS radiotelemetry error and bias in mountainous terrain. Wildl Soc Bull. 2002;30:430–9. 20. Brost BM, Hooten MB, Hanks EM, Small RJ. Animal movement constraints improve resource selection inference in the presence of telemetry error. Ecology. 2015;96:2590–7. gy 21. Roeleke M, Teige T, Hoffmeister U, Klingler F, Voigt CC. Aerial-hawking bats adjust their use of space to the lunar cycle. Mov Ecol. 2018;6:11. https:// doi.org/10.1186/s40462-018-0131-7. Received: 22 October 2019 Accepted: 6 February 2020 Ethics approval and consent to participatef Permissions to trap and tag griffon vultures and lesser kestrels were given by the Centre de Recherche sur la Biologie des Populations d’Oiseaux, Museum National d’Histoire Naturelle, Paris, to O. Duriez (Personal Program PP961) and to P. Pilard (PP311). Permissions to trap and tag osprey were given by Instituto Nazionale per la Protezione e la Ricerca Ambientale (ISPRA) under the authorization of Law 157/1992 [Art. 4 (1) and Art. 7 (5)] and under authoriza‑ tion no. 2502 05.05.2016 and no. 4254 27.03.2018 issued by Regione Toscana. Permissions to trap and tag condors were given by the Dirección de Fauna de Río Negro (DFRN), the Administración de Parques Nacionales (APN) from Argentina and the owners and managers of local farms. All condor procedures were approved by the DFRN, RN132.730-DF, and APN, 14/2011. Page 12 of 13 Page 12 of 13 Péron et al. Anim Biotelemetry (2020) 8:5 Péron et al. Anim Biotelemetry (2020) 8:5 Péron et al. Anim Biotelemetry References Dokter AM, Åkesson S, Beekhuis H, Bouten W, Buurma L, van Gasteren H, et al. Twilight ascents by common swifts, Apus apus, at dawn and dusk: acquisition of orientation cues? Anim Behav. 2013;85:545–52. 36. Crocker C, Jackson G. The use of altimeters in height measurement. 2018. http://www.hills-database.co.uk/altim.html. Page 13 of 13 Péron et al. Anim Biotelemetry (2020) 8:5 Péron et al. Anim Biotelemetry (2020) 8:5 53. Patterson TA, Thomas L, Wilcox C, Ovaskainen O, Matthiopoulos J. State-space models of individual animal movement. Trends Ecol Evol. 2008;23:87–94. 37. Borkenhagen K, Corman A-M, Garthe S. Estimating flight heights of seabirds using optical rangefinders and GPS data loggers: a methodo‑ logical comparison. Mar Biol. 2018;165:17. https://doi.org/10.1007/s0022 7-017-3273-z. 54. Gurarie E, Bracis C, Delgado M, Meckley TD, Kojola I, Wagner CM. What is the animal doing? Tools for exploring behavioral structure in animal movements. J Anim Ecol. 2016;85:69–84. https://doi.org/10.1111/1365- 2656.12379. 38. Shepard ELC, Williamson C, Windsor SP. Fine-scale flight strategies of gulls in urban airflows indicate risk and reward in city living. Philos Trans R Soc Lond B Biol Sci. 2016;371:76–93. 55. Garrett ES, Zeger SL. Latent class model diagnosis. Biometrics. 2000;56:1055–67. https://doi.org/10.1111/j.0006-341X.2000.01055.x. 39. Hays GC, Åkesson S, Godley BJ, Luschi P, Santidrian P. The implications of location accuracy for the interpretation of satellite-tracking data. Anim Behav. 2001;61:1035–40. 56. Rebke M, Coulson T, Becker PH, Vaupel JW. Reproductive improve‑ ment and senescence in a long-lived bird. Proc Natl Acad Sci USA. 2010;107:7841–6. 40. Bradshaw CJA, Sims DW, Hays GC. Measurement error causes scale- dependent threshold erosion of biological signals in animal movement data. Ecol Appl. 2007;17:628–38. https://doi.org/10.1890/06-0964. 57. Péron G, Nicolai CA, Koons DN. Demographic response to perturbations: the role of compensatory density dependence in a North American duck under variable harvest regulations and changing habitat. J Anim Ecol. 2012;81:960–9. Ahttp://apps.isiknowledge.com/full_record.do?produ ct=UA&search_mode=GeneralSearch&qid=9&SID=U24vljJMo88xoTW SEgn&page=1&doc=8. Accessed 7 Aug 2013. 41. Januchowski SR, Pressey RL, VanDerWal J, Edwards A. Character‑ izing errors in digital elevation models and estimating the financial costs of accuracy. Int J Geogr Inf Sci. 2010;24:1327–47. https://doi. org/10.1080/13658811003591680. 42. Gorokhovich Y, Voustianiouk A. Accuracy assessment of the processed SRTM-based elevation data by CGIAR using field data from USA and Thai‑ land and its relation to the terrain characteristics. Remote Sens Environ. 2006;104:409–15. g g g 58. Besbeas P, Morgan BJT. Variance estimation for integrated population models. Adv Stat Anal. 2017;101:439–60. https://doi.org/10.1007/s1018 2-017-0304-5. 43. Dunn JE, Gipson PS. References Analysis of radio telemetry data in studies of home range. Biometrics. 1977;33:85–101. 59. Calabrese JM, Fleming CH, Gurarie E. ctmm: an r package for analyzing animal relocation data as a continuous-time stochastic process. Methods Ecol Evol. 2016;7:1124–32. https://doi.org/10.1111/2041-210X.12559. 44. Auger-Méthé M, Field C, Albertsen CM, Derocher AE, Lewis MA, Jonsen ID, et al. State-space models’ dirty little secrets: even simple linear Gaussian models can have estimation problems. Sci Rep. 2016;6:26677. 60. Kristensen K, Nielsen A, Berg CW, Skaug H, Bell B. TMB: automatic differen‑ tiation and laplace approximation. J Stat Softw. 2014;2014:1–21. 61. Spiegelhalter D, Thomas A, Best N, Way R. WinBUGS user manual. Compo‑ nents. 2003;2:1–60. 45. Gould WR, Nichols JD. Estimation of temporal variability of survival in ani‑ mal populations. Ecology. 1998;79:2531–8. https://doi.org/10.1890/0012- 9658(1998)079%5b2531:eotvos%5d2.0.co;2. 62. Csilléry K, Blum MGB, Gaggiotti OE, François O. Approximate Bayesian computation (ABC) in practice. Trends Ecol Evol. 2010;25:410–8. 46. Walter WD, Fischer JW, Humphrey JS, Daughtery TS, Milleson MP, Tillman EA, et al. Using three-dimensional flight patterns at airfields to identify hotspots for avian–aircraft collisions. Appl Geogr. 2012;35:53–9. 63. Plummer M. JAGS: a program for analysis of Bayesian graphical models using Gibbs sampling. In: Proceedings of the 3rd international workshop on distributed statistical computing. 2003;20–22. http://citeseerx.ist.psu. edu/viewdoc/summary?doi=10.1.1.13.3406. Accessed 24 Aug 2018. 47. Balotari-Chiebao F, Brommer JE, Saurola P, Ijäs A, Laaksonen T. Assessing space use by pre-breeding white-tailed eagles in the context of wind- energy development in Finland. Landsc Urban Plan. 2018;177:251–8. 64. Sachs G. Minimum shear wind strength required for dynamic soaring of albatrosses. Ibis. 2005. https://doi.org/10.1111/j.1474-919x.2004.00295.x. 48. Cleasby IR, Wakefield ED, Bearhop S, Bodey TW, Votier SC, Hamer KC. Three-dimensional tracking of a wide-ranging marine predator: flight heights and vulnerability to offshore wind farms. J Appl Ecol. 2015;52:1474–82. https://doi.org/10.1111/1365-2664.12529. 65. Bruder JA. Bird hazard detection with airport surveillance radar. Radar Syst. (RADAR 97). IEE; 1997. p. 160–3. https://digital-library.theiet.org/ content/conferences/10.1049/cp_19971652. Accessed 14 June 2019. 49. Tikkanen H, Rytkönen S, Karlin O-P, Ollila T, Pakanen V-M, Tuohimaa H, et al. Modelling golden eagle habitat selection and flight activity in their home ranges for safer wind farm planning. Environ Impact Assess Rev. 2018;71:120–31. 66. Rebolo-Ifrán N, Graña Grilli M, Lambertucci SA. Drones as a threat to wildlife: YouTube complements science in providing evidence about their effect. Environ Conserv. 2019;1–6. https://www.cambridge.org/core/ product/identifier/S0376892919000080/type/journal_article. 50. de Valpine P, Hastings A. Fitting population models incorporating process noise and observation error. Ecol Monogr. 2002;72:57–76. https://doi. org/10.1890/0012-9615(2002)072%5b0057:fpmipn%5d2.0.co;2/full. Publisher’s Note S Springer Nature remains neutral with regard to jurisdictional claims in pub‑ lished maps and institutional affiliations. 51. Buderman FE, Hooten MB, Ivan JS, Shenk TM. A functional model for characterizing long-distance movement behaviour. Methods Ecol Evol. 2015;7:264–73. https://doi.org/10.1111/2041-210X.12465. 52. Fleming CH, Sheldon D, Gurarie E, Fagan WF, LaPoint S, Calabrese JM. Kálmán filters for continuous-time movement models. Ecol Inform. 2017;40:8–21. • fast, convenient online submission • thorough peer review by experienced researchers in your ﬁeld • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year • At BMC, research is always in progress. Learn more biomedcentral.com/submissions Ready to submit your research ? Choose BMC and benefit from: • fast, convenient online submission • thorough peer review by experienced researchers in your ﬁeld • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year • At BMC, research is always in progress. Learn more biomedcentral.com/submissions Ready to submit your research ? Choose BMC and benefit from: • fast, convenient online submission • thorough peer review by experienced researchers in your ﬁeld • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year • At BMC, research is always in progress. Learn more biomedcentral.com/submissions Ready to submit your research ? Choose BMC and benefit from:
https://openalex.org/W3022184904	https://europepmc.org/articles/pmc7199946?pdf=render	English	null	Characterizing changes in soil microbiome abundance and diversity due to different cover crop techniques	PloS one	2,020	cc-by	13,782	PLOS ONE PLOS ONE RESEARCH ARTICLE Characterizing changes in soil microbiome abundance and diversity due to different cover crop techniques Charlotte H. Wang1,2☯, Linda Wu2☯, Zengyan WangID3, Magdy S. Alabady4, Daniel Parson1,5, Zainab Molumo1, Sarah C. FankhauserID1* 1 Oxford College of Emory University, Oxford, Georgia, United States of America, 2 Emory University, Atlanta, Georgia, United States of America, 3 Computer Sciences Department of University of Georgia, Athens, Georgia, United States of America, 4 Department of Plant Biology, University of Georgia, Athens, Georgia, United States of America, 5 Oxford College Organic Farm, Oxford, Georgia, United States of America a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 ☯These authors contributed equally to this work. * sarah.fankhauser@emory.edu OPEN ACCESS OPEN ACCESS Citation: Wang CH, Wu L, Wang Z, Alabady MS, Parson D, Molumo Z, et al. (2020) Characterizing changes in soil microbiome abundance and diversity due to different cover crop techniques. PLoS ONE 15(5): e0232453. https://doi.org/ 10.1371/journal.pone.0232453 Editor: Primo Proietti, Universita degli Studi di Perugia, ITALY Received: July 15, 2019 Accepted: April 15, 2020 Published: May 5, 2020 Soil-based microorganisms assume a direct and crucial role in the promotion of soil health, quality and fertility, all factors known to contribute heavily to the quality and yield of agricul- tural products. Cover cropping, used in both traditional and organic farming, is a particularly efficient and environmentally favorable tool for manipulating microbiome composition in agricultural soils and has had clear benefits for soil quality and crop output. Several long- term investigations have evaluated the influence of multi-mix (multiple species) cover crop treatments on soil health and microbial diversity. The present study investigated the short- term effects of a seven species multi-mix cover crop treatment on soil nutrient content and microbial diversity, compared to a single-mix cover crop treatment and control. Analysis of 16S sequencing data of isolated soil DNA revealed that the single-mix cover crop treatment decreased overall microbial abundance and diversity, whereas the control and multi-mix treatments altered the overall microbial composition in similar fluctuating trends. Further- more, we observed significant changes in specific bacteria belonging to the phyla Acidobac- teria, Actinobacteria, Planctomycetes, Proteobacteria and Verrucombicrobia for all treatments, but only the single-mix significantly decreased in abundance of the selected bacteria over time. Our findings indicate that the control and multi-mix treatments are better at maintaining overall microbial composition and diversity compared to the single-mix. Fur- ther study is required to elucidate the specific difference between the treatment effect of the multi-mix treatment and the control, given that their microbial composition changes over time were similar but they diverge into two populations of unique bacterial types by the end of this short-term study. OPEN ACCESS Citation: Wang CH, Wu L, Wang Z, Alabady MS, Parson D, Molumo Z, et al. (2020) Characterizing changes in soil microbiome abundance and diversity due to different cover crop techniques. PLoS ONE 15(5): e0232453. https://doi.org/ 10.1371/journal.pone.0232453 Editor: Primo Proietti, Universita degli Studi di Perugia, ITALY Received: July 15, 2019 Accepted: April 15, 2020 Published: May 5, 2020 Citation: Wang CH, Wu L, Wang Z, Alabady MS, Parson D, Molumo Z, et al. Introduction Soil microbial communities are essential to the fertility of soil and success of agricultural crops, yet a substantial portion of microbial life within the soil remains uncultivated and under-explored [1]. The composition and diversity of these soil microbial communities are important indicators of the soil fertility. Within agricultural soils, microbes play essential roles in supporting plant development through processes such as nitrogen fixation, growth hor- mone synthesis and nutrient recycling of decomposed plant matter [2,3]. An array of farming techniques has been developed with the purpose of manipulating the microbial composition within soil in an effort to promote soil health and enhance crop quality. An approach that has demonstrated much promise in recent years is cover cropping, which is the practice of plant- ing specific species of crop(s) to maturity then, without harvesting, mowing the cover crop back into the soil. Utilized in both conventional and organic agriculture, this technique recy- cles nutrients back into the soil and has been found to contribute to both pest management and the development of a beneficial biota [4]. Soils treated with cover crops have been found to have greater complexity and diversity within their microbial communities than tilled soils. Even against other sustainable techniques, including the use of organic amendments and mini- mum-tillage, cover cropping is comparatively more beneficial for promoting soil viability. Till- age, for example, while considered highly advantageous for soil microbiome maintenance when used conservatively, has been shown to decrease overall microbial diversity over time, whereas cover cropping has been found to consistently preserve and enhance heterogeneity [5]. Funding: We are grateful to Oxford College of Emory University for financial support. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. Competing interests: The authors have declared that no competing interests exist. A variety of classes of cover crops exist, each serving a different and specific function. Legume-based cover crops, in particular, have been found to substantially increase soil nitro- gen concentration by supporting the growth of nitrogen-fixing bacteria. This particular group of cover crops has also been shown to improve overall soil quality and enhance microbial diversity in soil [6]. OPEN ACCESS (2020) Characterizing changes in soil microbiome abundance and diversity due to different cover crop techniques. PLoS ONE 15(5): e0232453. https://doi.org/ 10.1371/journal.pone.0232453 Received: July 15, 2019 Accepted: April 15, 2020 Published: May 5, 2020 Received: July 15, 2019 Accepted: April 15, 2020 Published: May 5, 2020 Received: July 15, 2019 Accepted: April 15, 2020 Published: May 5, 2020 Peer Review History: PLOS recognizes the benefits of transparency in the peer review process; therefore, we enable the publication of all of the content of peer review and author responses alongside final, published articles. The editorial history of this article is available here: https://doi.org/10.1371/journal.pone.0232453 Copyright: © 2020 Wang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information files. Raw sequencing data is uploaded to the NCBI BioProject repository and accessible 1 / 22 PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 PLOS ONE Cover cropping and soil microbiome via the following URL: https://www.ncbi.nlm.nih. gov/bioproject/PRJNA626000/. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 Introduction Of the several cover cropping practices utilized in farming, we compared the effects of single and multispecies, or multi-mix, cover crop mixtures on soil fertility and microbial abundance and diversity. It has previously been found that multi-mixtures of grass and legumes improve crop yield, microbial diversity, soil moisture, and inorganic nitrogen lev- els in soil when paralleled with single, double-mix, and control treatments [7]. These differ- ences may be attributed to the apparent benefits of mixing different species of cover crops with differing actions; effective combinations of these cover crops can work to optimize each cover crops’ effects on the soil. A three-year long-term study evaluating the effects of single versus multi-mix cover cropping on soybean crop quality and output observed greater crop yield in multi-mix plots than single-mix plots; the treatments’ effects on microbial diversity were not assessed [6]. While other long-term studies have investigated the effects of multi-mix cover crop treatments on microbial composition and health, there is a lack of literature that explores the effects of short-term cover cropping (one year) on microbial life [7]. The present study sought to evaluate the effects of short-term cover cropping on soil micro- bial diversity and fertility. The study also aimed to develop a more comprehensive understand- ing of the impacts of combining different organic summer cover crops (multi-mix of legumes and grass as well as single-mix of buckwheat) as it relates to altering soil microbial community composition and nutrient availability, both of which directly influence crop production. We hypothesized that an increased cover crop diversity would lead to a more complex soil rhizo- sphere. We found that soil microbial composition and overall diversity were most altered in plots treated with the control and multi-mix cover crop treatments. While our study is based in a specific agricultural setting, our results demonstrate how specific plant communities can change the microbial ecology of the soil over a short time period. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 2 / 22 PLOS ONE Cover cropping and soil microbiome Experimental timeline and plot preparation The investigation was held on the Oxford College Organic Farm in Oxford, Georgia (33.616247, -83.867004), and the soil has been identified as the Ultisol. It was certified organic in 2015 according to the USDA National Organic Program standards. A 26 x 6 m area of the Oxford Organic Farm was created in Summer 2016 by pilling neighboring soil on top of the target plot and planting Crimson Clover and grain rye in Early Fall. The cover crops were grown for three seasons until the summer of 2017, the experimental plot was designated to this study (Fig 1A). The plot was divided into 9 separate beds that each contained a planting strip (Fig 1B). The 9 subplots provided triplicates of the three treatment levels: a control, single-mix and multi-mix. Control had no cover crop treatment, single-mix consisted of the cover crop: organic non-legume buckwheat species, whereas the multi-mix was a combination of seven Fig 1. A. Timeline tracking the past plot history and key experimental time points. B. Outline of the experimental plots and the independent treatment groups. https://doi.org/10.1371/journal.pone.0232453.g001 Fig 1. A. Timeline tracking the past plot history and key experimental time points. B. Outline of the experimental plots and the independent treatment groups. https://doi.org/10.1371/journal.pone.0232453.g001 PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 3 / 22 PLOS ONE Cover cropping and soil microbiome summer organic cover crops: Japanese Millet, Sorghum-Sundan Grass, Pearl Millet, Soybeans, Sunnhemp, Cowpea, and Buckwheat [8]. The cover crops that were used in this study were selected in consultation with Daniel Parson, the Oxford College Organic Farmer, and based on conventional organic farming practices and economic efficiency. At the start of the study on January 6th 2017, the experimental plot was mowed in preparation for sampling. The plot was tilled, and the cover crop seeds were planted, and the initial soil analysis was conducted. A month later, the cover crops were mowed and allowed to decompose into the soil. Within 20 days, the experimental plots were fertilized and tilled. Then the cash crop, broccoli, was trans- planted into the experimental plot at approximately 20 broccoli per subplot. After four months of growth and maintenance, the broccoli crop was lost due to an unexpected freeze. Data col- lection was initiated in June 2017 and continued until January 2018. Throughout this time period, monthly average temperatures and rainfalls were recorded and compared to previous year records (S1 Fig). Soil nutrient composition Soil tests at UGA-AESL (University of Georgia-Agricultural & Environmental Services Lab) were conducted before the cover crops are planted and before the cash crops are harvested to analyze the organic matter and nutrient availability within the soil. Additionally, cation exchange capacity was measured, which indicated the ability of organic matter to hold onto cations as well as soil fertility and nutrient retention capacity [10]. Active organic carbon was estimated using 5g of soil for each sample using KMnO4 [11]. Experimental timeline and plot preparation A decrease in average temperature was observed compared to 2015– 2016 historical month average temperatures. The summer of 2017 had an increase in total rainfall, but the winter of 2017 was drier. Sample collection Microbial data collection was conducted at three key analysis time points–immediately follow- ing the planting of the cover crops (Time point 1), immediately before the mowing of the cover crops (Time point 2), and before the harvest of the broccoli crops (Time point 3) (Fig 1A). The samples were collected in triplicates for each subplot with a total of 27 samples for each time point, with the exception of time point 1 where only 1 sample per subplot was taken. There was only 1 sample per subplot taken at time point 1 since the soil had just been evenly tilled across each subplot and no further manipulation had been undertaken; thus the extra replicates of the plots were deemed unnecessary. However, we do recognize that this set up led to sample imbalances that should be taken into consideration when interpreting statistical methods used in this manuscript. Soil samples for time points 2 and 3 were collected from the rhizospheres of the plants, which were seeded initially to avoid the edges of the borders of the plot. Previous research has shown that proximity to field edges can positively impact soil nutri- ent content [9]. No vegetative growth was present in time point one. By the end of the experi- ment, there were a total of 63 soil samples used for microbial analysis. Approximately 500uL of soil was collected for each sample at using a 16 mm diameter soil probe at 100- mm depth below the soil surface. Soil samples were frozen at -80˚C for DNA extraction and further analysis. DNA extraction and 16s rDNA sequencing The pool was sequenced on the Miseq PE300 run following illumine standard operating procedures and recommendations [13]. Reads were de-multiplexed using the Illumina bcl2fastq. Adapters of the de-multiplexed data were trimmed by TrimGalore, the sequences were merged, and primers were removed by QIIME2 [14]. The following quality control parameters were used: reads had over 90% bases with the quality score above 30. Representative read picking was performed with the QIIME2 DADA2 denoising algorithm [15] into amplicon sequence variants (ASVs). ASVs were then assigned taxonomy using Greengenes [16], SILVA reference database (Version 132) [17]. The molecular weight of the DNA was determined by running the libraries on the Frag- ment Analyzer instrument (Agilent) using the high sensitivity NGS kit. The NGS fragment analysis confirmed the targeted size for the V3-V4 region of the 16S amplicon. The libraries were then normalized to have equal molarity and were pooled together at equal volumes. The concentration of the pool was assessed with Qubit as well as quantitative PCR to have the most accurate reading possible. After this final quality check, the pool was ready for sequencing. Sequencing and post sequence analysis. The pool was sequenced on the Miseq PE300 run following illumine standard operating procedures and recommendations [13]. Reads were de-multiplexed using the Illumina bcl2fastq. Adapters of the de-multiplexed data were trimmed by TrimGalore, the sequences were merged, and primers were removed by QIIME2 [14]. The following quality control parameters were used: reads had over 90% bases with the quality score above 30. Representative read picking was performed with the QIIME2 DADA2 denoising algorithm [15] into amplicon sequence variants (ASVs). ASVs were then assigned taxonomy using Greengenes [16], SILVA reference database (Version 132) [17]. DNA extraction and 16s rDNA sequencing DNA was isolated from each collected soil sample using the QIAGEN DNeasyaˆ PowerSoilaˆ Kit and protocol per manufacturer’s instructions. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 4 / 22 PLOS ONE Cover cropping and soil microbiome Upon isolation and purification, the samples were sequenced at the Georgia Genomics Facility at University of Georgia where Illumina sequencing was performed using universal primers specific for the V3-V4 regions of the 16s rDNA gene. DNA quality were assessed with a plate fluorometer, then a 16s Library was assembled. After normalizing DNA samples to 5ng/μL, 5 μL of each sample was used for primary PCR. The V3-V4 16S primers (S-D-Bact-0341-b-S-17 CCTACGGGNGGCWGCAG and Reverse: S-D-Bact-0785-a-A-21 GACTACHVGGGTATCTAATCC [12]) were diluted to 2μM. The first PCR reaction mix consisted of 2.5μL of each primer, 12.5uL of KAPA HiFi HotStart ReadyMix and 2.5uL of PCR-grade water [13]. The amplification reaction profile was: 95˚C - 3min; 15 cycles of 95˚C- 30 ec, 56˚C- 30sec, and 72˚C- 30sec; and 72˚C- 4min. A post-PCR cleanup was conducted using 0.8X AMPure beads. The purified PCR product was resuspended in 50uL of elution buffer. Next, 5 uL of the first PCR product for each sample was used in the second PCR amplification. The reaction mix consisted of 5ul first PCR product, 5ul of each i5 and i7 Unique Illumina indexing primers, 25 uL of KAPA HiFi HotStart ReadyMix, and 10 uL of PCR-grade water. The amplification reac- tion profile was: 95˚C - 3min; 12 cycles of 95˚C- 30sec, 56˚C- 30sec, and 72˚C- 30sec; and 72˚C- 4min. The second PCR products were purified using 1X AMPure beads and the cleaned products were eluted in 25uL of elution buffer [13], which constitutes the final 16S sequencing libraries. The library concentrations were measured using the plate fluorometer method. The molecular weight of the DNA was determined by running the libraries on the Frag- ment Analyzer instrument (Agilent) using the high sensitivity NGS kit. The NGS fragment analysis confirmed the targeted size for the V3-V4 region of the 16S amplicon. The libraries were then normalized to have equal molarity and were pooled together at equal volumes. The concentration of the pool was assessed with Qubit as well as quantitative PCR to have the most accurate reading possible. After this final quality check, the pool was ready for sequencing. Sequencing and post sequence analysis. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 Soil composition and fertility Organic carbon content. During the six-month field trials, soil samples were collected at three distinct time points from the three different treatment plots (single-mix, multi-mix, and control) to analyze the soil organic carbon content using KMnO4 extraction (Fig 2). A signifi- cant decrease in active carbon concentration was observed in all three treatments between time point 1 and time point 2 (2 sample t-test, p<0.05). Between time point 2 and time point 3, the active carbon concentration significantly increased for all three treatments, indicating a restoration of carbon in the soil. Across treatments, the differences in active carbon concentra- tion at respective time points were not significant (Fig 2). Cation exchange capacity. Cation exchange capacity (CEC) was used to measure the total exchangeable cations which may be held in soil at specific pH levels. There was no significant difference between the cation exchange capacity in between the treatments at time point 3, but there was a significant increase in CEC in all three sample treatments at time point 3 in com- parison to the CEC at time point 1 (two sample t-test, p<0.05) (Fig 3). Statistical analysis We used the analysis tools of the QIIME2 pipeline to perform the various statistical analyses that are mentioned in this section. Statistical significance analysis of overall microbial composition was based on adjustments to Aitchison’s norm. The Aitchison norm is a statistical analysis method used to normalize compositional data such as microbial/relative abundances. Comparing changes in relative abundances of taxa within a microbial community can present issues when looking at each unique taxon as comprising a part of the whole. Increases in abundance of a specific type of microorganism do not necessarily amount to a decrease in abundance of another or others but can be interpreted as such without first controlling for compositionality so that changes in abundance of one taxonomical unit is not correlated with any observed changes within another [18]. The Aitchison norm utilizes the isometric log ratio (ilr) transformation, which tracks changes in abundance over time and quantifies these changes as log ratio coordinates, or balances [19]. The resulting processed data can then be analyzed using multivariate PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 5 / 22 PLOS ONE Cover cropping and soil microbiome comparative methods such as ANOVA and ANOSIM. This analysis was conducted on six sample categories: 3 time points each of which has three treatments (control, single mix and multi-mix) and 3 treatments each of which has three time points (time point 1, 2, and 3). This analysis identifies whether there is significant difference between the samples in each of the categories. The “group_significance.py” function in QIIME was also used to compare ASV abundance and to identify the statistical level of significance among samples and categories. With this function, we used two different statistical tests: ANOVA and Kruskal-Wallis (KW). The signif- icant ASVs from both tests were cross-referenced to yield a final list of significant ASVs across different groupings (p<0.01). The count of ASV indicated the abundance of each ASV, and it was normalized across samples to account for the difference in the depth of sequencing between samples, thus making it a good candidate for cross-sample comparisons. The normal- ized mean relative abundance for each sample was used for significance analysis. Kruskal-Wal- lis test was used to determine significant differences in average phylum abundance of ASVs that significantly changed over time for each treatment. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 Overall microbial composition The number of reads left after each step in the full amplicon workflow was recorded (Table 1). The full amplicon workflow included input raw reads, filtering based on quality score, denois- ing sequences, merging of paired-end reads and chimera filtering. The average distinct num- ber of ASVs indicated the average total microbial life present, or the total microbial abundance in the sample. The significance of the changes in total microbial life was determined based on ANOVA and ANOSIM based on Aitchison’s distances. In both the control and multi-mix treatments there was a decrease in total microbial life between time point 1 and time point 2, and an increase from time point 2 to time point 3, whereas single-mix decreased in total microbial life over time. At each respective time points, time point 2 and time point 3 showed the greatest difference between the respective treatments (Table 2A). Microbial abundance. The average number of distinct amplicon sequences variances (ASVs) represented microbial abundance of each treatment group at respective time points. From distinct ASV numbers, we observed an increasing trend, although insignificant, in microbial abundance for the control and multi-mix treatments between time point 1 and time point 2, and a decreasing trend from time point 2 to time point 3, whereas the single mix PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 6 / 22 PLOS ONE Cover cropping and soil microbiome Fig 2. Percent change in concentrations of active carbon in each subplot between time points. Bars represent the average of 3 biological replicates. An overall 6–7% increase in active carbon content in MM and SM treatments respectively, as well as an overall 10% decrease in the control group were observed between the 1st and 2nd time point. Fig 2. Percent change in concentrations of active carbon in each subplot between time points. Bars represent the average of 3 biological replicates. An overall 6–7% increase in active carbon content in MM and SM treatments respectively, as well as an overall 10% decrease in the control group were observed between the 1st and 2nd time point. https://doi.org/10.1371/journal.pone.0232453.g002 https://doi.org/10.1371/journal.pone.0232453.g002 (nonsignificant) consistently decreased in microbial abundance over time (Table 1). However, according to statistical tests based on Aitchison’s distances of the treatment groups, none of the changes of a treatment over time was significant (p<0.01), while we do recognize that the p value for the changes in the control over time is considerable low. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 Overall microbial composition Samples taken from time point 1 were used as a standard to compare the effect of cover crops on soil cation exchange capacity at time point 3. A significant increase in CEC was observed in all sample treatments, indicated by the asterisk (2 sample T-test, p<0.05). Multi-mix and single-mix had similar CEC values. There was no significant difference between the treatments. Fig 3. Cation exchange capacity based on routine soil tests. Samples taken from time point 1 were used as a standard to compare the effect of cover crops on soil cation exchange capacity at time point 3. A significant increase in CEC was observed in all sample treatments, indicated by the asterisk (2 sample T-test, p<0.05). Multi-mix and single-mix had similar CEC values. There was no significant difference between the treatments. Fig 3. Cation exchange capacity based on routine soil tests. Samples taken from time point 1 were used as a standard to compare the effect of cover crops on soil cation exchange capacity at time point 3. A significant increase in CEC was observed in all sample treatments, indicated by the asterisk (2 sample T-test, p<0.05). Multi-mix and single-mix had similar CEC values. There was no significant difference between the treatments. https://doi.org/10.1371/journal.pone.0232453.g003 https://doi.org/10.1371/journal.pone.0232453.g003 https://doi.org/10.1371/journal.pone.0232453.g003 measure of general diversity, including richness and evenness [21]. Looking at the diversity of each treatment across the three time points, the Shannon diversity index was only significantly different for the control treatment between all three time points (Based on Kruskal-Wallis (p = 0.0037). Pairwise tests between the time points and Shannon indices showed that the microbial diversity for the control treatment increased between time points 1 and 2 (p = 0.014) and significantly decreased between time points 2 and 3 (p = 0.007), but the overall decrease was not significant between time points 1 and 3 (p = 0.079). The overall trend for microbial diversity in the control indicated a great fluctuation in the microbial diversity (Table 1). Mean- while, the single-mix microbial diversity did not significantly change over time (Kruskal-Wal- lis, p = 0.017). Pairwise tests showed that the single-mix microbial diversity decreased from time point 1 to time point 2 (p = 0.052), decreased from time point 2 to time point 3 (p = 0.102). Overall microbial composition The low p-value of the control (anosim_aitchison, p = 0.112; anova_aitchison, p = 0.072), compared to the higher p- values in the multi-mix and sinlg-mix treatments, indicated that our cover-cropping technique may be responsible for maintaining overall microbial abundance of the soil (Table 2A). With a different analysis parameter, looking at the difference between treatments at individ- ual time points, ANOVA and ANOSIM tests inferred a significant difference between treat- ments at time point 2 and time point 3 (Table 2B). At time point 1, the soil in every plot was untreated, and the p values at time point 1 were consistent in showing no significant difference between the soil microbial abundance. However, at time points 2 and 3, treatments started to significantly differ from each other at the same time point due to treatment effects (Table 2B). From distinct ASV numbers, we observed that there was indeed a significant difference between microbial abundance between the three treatments at time points 2 and 3, with the single-mix having the highest abundance at time point 2, and lowest abundance at time point 3 (Table 1). Microbial diversity. Diversity measurements were used to compare the samples indepen- dent of their phylogenetic composition [20]. The Shannon diversity index (H) was used as a PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 7 / 22 PLOS ONE Cover cropping and soil microbiome Fig 3. Cation exchange capacity based on routine soil tests. Samples taken from time point 1 were used as a standard to compare the effect of cover crops on soil cation exchange capacity at time point 3. A significant increase in CEC was observed in all sample treatments, indicated by the asterisk (2 sample T-test, p<0.05). Multi-mix and single-mix had similar CEC values. There was no significant difference between the treatments. https://doi.org/10.1371/journal.pone.0232453.g003 Fig 3. Cation exchange capacity based on routine soil tests. Samples taken from time point 1 were used as a standard to compare the effect of cover crops on soil cation exchange capacity at time point 3. A significant increase in CEC was observed in all sample treatments, indicated by the asterisk (2 sample T-test, p<0.05). Multi-mix and single-mix had similar CEC values. There was no significant difference between the treatments. https://doi.org/10.1371/journal.pone.0232453.g003 Fig 3. Cation exchange capacity based on routine soil tests. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 Overall microbial composition The overall decrease in diversity index from time point 1 to time point 3 (p = 0.013) inferred that the single-mix treatment had a decreasing trend in microbial diversity over time (Table 1). The trend of microbial diversity change in the multi-mix treatment was similar to measure of general diversity, including richness and evenness [21]. Looking at the diversity of each treatment across the three time points, the Shannon diversity index was only significantly different for the control treatment between all three time points (Based on Kruskal-Wallis (p = 0.0037). Pairwise tests between the time points and Shannon indices showed that the microbial diversity for the control treatment increased between time points 1 and 2 (p = 0.014) and significantly decreased between time points 2 and 3 (p = 0.007), but the overall decrease was not significant between time points 1 and 3 (p = 0.079). The overall trend for microbial diversity in the control indicated a great fluctuation in the microbial diversity (Table 1). Mean- while, the single-mix microbial diversity did not significantly change over time (Kruskal-Wal- lis, p = 0.017). Pairwise tests showed that the single-mix microbial diversity decreased from time point 1 to time point 2 (p = 0.052), decreased from time point 2 to time point 3 (p = 0.102). The overall decrease in diversity index from time point 1 to time point 3 (p = 0.013) inferred that the single-mix treatment had a decreasing trend in microbial diversity over time (Table 1). The trend of microbial diversity change in the multi-mix treatment was similar to PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 8 / 22 PLOS ONE Cover cropping and soil microbiome Table 1. Summary statistics for each step of the full amplicon workflow and diversity index of bacterial communities. Overall microbial composition The drop in the number of reads during the merging step indicated paired-end sequences that could not merge, either because the sequences were too short and did not overlap, or because the ends did not align. Total ASVs included many repeated ASVs for one species, but the distinct ASV represented the count of different ASVs across all species. Shannon index was used to measure general diversity. The control and single-mix treatments altered in microbial diversity over time, whereas the multi-mix did not significantly alter microbial diversity over time. Significantly different from other time points of the same treatment based on ANOSIM (p<0.01) After the basic adapter and quality-based trimming, the number of reads left after each step in the full amplicon workflow was recorded. The denoising step included dereplication and sample inference. The drop in the number of reads during the merging step indicated paired-end sequences that could not merge, either because the sequences were too short and did not overlap, or because the ends did not align. Total ASVs included many repeated ASVs for one species, but the distinct ASV represented the count of different ASVs across all species. Shannon index was used to measure general diversity. The control and single-mix treatments altered in microbial diversity over time, whereas the multi-mix did not significantly alter microbial diversity over time. Significantly different from other time points of the same treatment based on ANOSIM (p<0.01) https://doi.org/10.1371/journal.pone.0232453.t001 https://doi.org/10.1371/journal.pone.0232453.t001 the control—increasing between time points 1 and 2 (p = 0.079) and decreasing between time points 2 and 3(p = 0.270). The multi-mix microbial diversity, however, was not significantly different across time with respect to our alpha value of 0.01 (Kruskal-Wallis, p = 0.08). The overall change in microbial abundance between time point 1 and time point 3 also showed a decreasing trend such as the control, though the change was not significant (p = 0.052). How- ever, the p-values were still considerably low, thus we report this p-value for consideration of the potential biological difference in microbial diversity. The trends in changes in microbial diversity were similar to the changes in microbial abundance (Table 1). For the analysis of different treatments within individual time points, the difference between the diversity index of respective treatments was not significant (Kruskal-Wallis, p>0.01). Overall microbial composition However, the p values of 0.051 at time point 1, 0.10 at time point 2, and 0.070 at time point 3 demonstrated that there still may be biological differences in microbial diversity between the treatments at different time points, even at time point 1, with identical untreated soil. At time point 3, the single-mix also appeared to have the lowest Shannon diversity index (Table 1). This low final microbial diversity for single-mix was consistent with the overall trend that the single-mix decreases microbial diversity. Nevertheless, we do recognize that the Shannon diversity index is very sensitive to rare groups, making the analysis on microbial diversity with room for other potential inferences [21]. Overall microbial composition Time Point 1 Time Point 2 Time Point 3 C SM MM C SM MM C SM MM Input Reads 223909 +/- 16680 190305 +/- 5650 200699 +/- 64896 109408 +/- 45341 164517 +/- 67912 109856 +/- 58548 219351 +/- 121581 106797 +/- 51037 154522 +/- 55159 Filtered Reads 112520 +/- 11608 97432 +/- 4456 104411 +/- 33776 50533 +/- 23198 81510 +/- 35674 51302 +/- 28567 107605 +/- 58804 46983 +/- 27857 74309 +/- 27056 Denoised Reads 105244 +/- 11286 90931 +/- 4259 98519 +/- 33723 46997 +/- 21840 76499 +/- 34254 47590 +/- 27284 102213 +/- 56998 43450 +/- 26490 69467 +/- 26035 Merged Reads 83875 +/- 10130 72336 +/- 4124 81248 +/- 32897 37781 +/- 17771 62323 +/- 29205 37504 +/- 22643 86704 +/- 51071 34349 +/- 21881 56419 +/- 22797 Non-chimeric Reads (Total ASVs) 80797 +/- 9568 70287 +/- 4128 78443 +/- 31193 36572 +/- 17185 60124 +/- 27884 36393 +/- 21951 83328 +/- 48407 33369 +/- 21128 54828 +/- 22095 Distinct ASVs 2376 +/- 224 2183 +/- 155 2019 +/- 216 1069 +/- 483 1625 +/- 624 1154 +/- 591 1947 +/- 684 1019 +/- 558 1523 +/- 420 Shannon 10.24 +/- 0.14 10.12 +/- 0.11 9.93 +/- 0.10 9.37 +/- 0.38 9.58 +/- 0.45 9.53 +/- 0.39 9.81 +/- 0.35 9.20 +/- 0.54 9.56 +/- 0.26 After the basic adapter and quality-based trimming, the number of reads left after each step in the full amplicon workflow was recorded. The denoising step included dereplication and sample inference. The drop in the number of reads during the merging step indicated paired-end sequences that could not merge, either because the sequences were too short and did not overlap, or because the ends did not align. Total ASVs included many repeated ASVs for one species, but the distinct ASV represented the count of different ASVs across all species. Shannon index was used to measure general diversity. The control and single-mix treatments altered in microbial diversity over time, whereas the multi-mix did not significantly alter microbial diversity over time. Significantly different from other time points of the same treatment based on ANOSIM (p<0.01) https //doi org/10 1371/jo rnal pone 0232453 t001 After the basic adapter and quality-based trimming, the number of reads left after each step in the full amplicon workflow was recorded. The denoising step included dereplication and sample inference. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 Analysis of each treatment: Difference across time points Further analysis of each treatment was based on the phylum classification of the ASV identi- fied taxon rather than using more specific classification as the unit. Phylum analysis was uti- lized since ASV taxonomy assignment was inconsistent in terms of the levels of classification due to variations in reads and matches. For example, some ASVs were classified to the family level while some were classified to the genus species level of the taxonomy. In order to reduce PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 9 / 22 PLOS ONE Cover cropping and soil microbiome bias in the analysis of taxonomic assignments, the phylum level classification was most consis- tent across all mapped taxa, thus it was the best unit for further analysis such as changes in sig- nificantly altered bacteria over time for each treatment. bias in the analysis of taxonomic assignments, the phylum level classification was most consis- tent across all mapped taxa, thus it was the best unit for further analysis such as changes in sig- nificantly altered bacteria over time for each treatment. Significant differences in ASV counts were determined across the three time points for each treatment; this analysis was based on cross-references of Kruskal-Wallis (p<0.01) and ANOVA (P<0.01) tests. The control treatment plot showed the greatest number of changes, with a total of 485 ASV aligned bacterial types that differed significantly in their abundances throughout the three time points. Meanwhile, 213 bacterial types were identified to be signifi- cantly different across time points for single-mix, and 138 bacterial types for multi-mix (Table 2). Further analyses of these changes are presented below. Significant changes in phyla over time in control treatment. Out of the 485 bacteria types that were significantly different in relative abundance in the control treatment, 85% of the bacterial types were classified into 14 phyla, and the remaining bacteria were unknown classifications (Acidobacteria, Actinobacteria, Armatimonadetes, Bacteroidetes, Chlorobi, Chloroflexi, Cyanobacteria, Elusimicrobia, Firmicutes, Gemmatimonadetes, Nitrospirae, Planctomycetes, Proteobacteria, Verrucombicrobia) (Table 3). Specifically, Planctomycetes and Proteobacteria were the two major phyla that comprised 45% of bacteria that significantly differed in the control treatment. The average abundance of significantly changing bacteria within each phylum, measured by ASV counts, was determined for each time point (Fig 4A). https://doi.org/10.1371/journal.pone.0232453.t002 PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 Analysis of each treatment: Difference across time points The average relative abundance of significantly changing bacteria in 11 out of the 14 phyla sig- nificantly changed from time point 1 to time point 3 (Kruskal-Wallis p<0.01), as time point 1 had the highest average abundances. This change in average abundance correlated with the overall decreasing and increasing trend in the number of distinct ASVs over time for the con- trol treatment (Table 1). Average counts of ASVs within the phyla of Chlorobi, Elusimicrobia, and Nitrospirae did not change significantly throughout time. Table 2. Reported P-values of ANOVA and ANOSIM tests based on aitchison distances of microbial abundance. A. Reported P-value anosim_aitchison anova_aitchison Control 0.112 0.072 Single-Mix 0.604 0.495 Multi-Mix 0.625 0.274 B. Reported P-value anosim_aitchison anova_aitchison TP1 0.218 0.165 TP2 0.001 0.001 TP3 0.007 0.02 Table 2. Reported P-values of ANOVA and ANOSIM tests based on aitchison distances of microbial abundance. The ANOSIM and ANOVA methods were both used to track any significant changes between each treatment group with regard to species abundance and composition over three different time points. The data were normalized using the Aitchison norm statistical analysis method. An alpha level of 0.01 was used for analysis. the Aitchison norm statistical analysis method. An alpha level of 0.01 was used for analysis. A. Against this alpha level, no significant changes in abundance and/or composition were observed across any of the time points for all treatments in both the ANOSIM and ANOVA analyses. B For the first time point, there were no significant changes in species composition or abundance between the A. Against this alpha level, no significant changes in abundance and/or composition were observed across any of the time points for all treatments in both the ANOSIM and ANOVA analyses. B. For the first time point, there were no significant changes in species composition or abundance between the treatment groups. However, for both time points 2 and 3, significant changes in the microbial composition between treatment groups were supported by both the ANOSIM and ANOVA results which generated p-values below the given alpha-level value of 0.01. An exception was that at time point 3 only the ANOSIM generated a p-value lower than 0.01, but the p-value from ANOVA was also very small (p<0.05). https://doi.org/10.1371/journal.pone.0232453.t002 10 / 22 PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 PLOS ONE Cover cropping and soil microbiome Table 3. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 Analysis of each treatment: Difference across time points Number of bacterial types that significantly differ in relative abundance between all three time points. Total Acidobacteria Actinobacteria Armatimonadetes Bacteroidetes Chlorobi Chloroflexi Cyanobacteria Elusimicrobia Firmicutes Gemmatimonadetes Nitrospirae Planctomycetes Proteobacteria Verrucomicrobia Control 485 23 35 9 44 1 17 2 2 19 5 1 79 138 38 Single-Mix 213 9 10 4 27 1 3 1 1 3 3 0 39 66 16 Multi-Mix 138 8 7 3 4 0 11 0 1 17 4 1 27 30 9 Significant difference was determined based on cross references from Kruskal-Wallis (p<0.01) and ANOVA (P<0.01) tests on average ASV counts of different bacterial types. The number of bacterial types that significantly differed in relative abundance based on ASV counts for the control was more than twice the number of bacterial types that significantly differed in relative abundance for the single and multi-mixed treatments Table 3. Number of bacterial types that significantly differ in relative abundance between all three time points. Total Acidobacteria Actinobacteria Armatimonadetes Bacteroidetes Chlorobi Chloroflexi Cyanobacteria Elusimicrobia Firmicutes Gemmatimonadetes Nitrospirae Planctomycetes Proteobacteria Verrucomicrobia Control 485 23 35 9 44 1 17 2 2 19 5 1 79 138 38 Single-Mix 213 9 10 4 27 1 3 1 1 3 3 0 39 66 16 Multi-Mix 138 8 7 3 4 0 11 0 1 17 4 1 27 30 9 Significant difference was determined based on cross references from Kruskal-Wallis (p<0.01) and ANOVA (P<0.01) tests on average ASV counts of different bacterial types. The number of bacterial types that significantly differed in relative abundance based on ASV counts for the control was more than twice the number of bacterial types that significantly differed in relative abundance for the single and multi-mixed treatments. https://doi.org/10.1371/journal.pone.0232453.t003 Table 3. Number of bacterial types that significantly differ in relative abundance between all three time points. Total Acidobacteria Actinobacteria Armatimonadetes Bacteroidetes Chlorobi Chloroflexi Cyanobacteria Elusimicrobia Firmicutes Gemmatimonadetes Nitrospirae Planctomycetes Proteobacteria Verrucomicrobia Control 485 23 35 9 44 1 17 2 2 19 5 1 79 138 38 Single-Mix 213 9 10 4 27 1 3 1 1 3 3 0 39 66 16 Multi-Mix 138 8 7 3 4 0 11 0 1 17 4 1 27 30 9 Significant difference was determined based on cross references from Kruskal-Wallis (p<0.01) and ANOVA (P<0.01) tests on average ASV counts of different bacterial types. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 Analysis of each treatment: Difference across time points The number of bacterial types that significantly differed in relative abundance based on ASV counts for the control was more than twice the number of bacterial types that significantly differed in relative abundance for the single and multi-mixed treatments. ber of bacterial types that significantly differ in relative abundance between all three time points. Acidobacteria Actinobacteria Armatimonadetes Bacteroidetes Chlorobi Chloroflexi Cyanobacteria Elusimicrobia Firmicutes Gemmatimonadetes Nitrospirae Planctomycetes Proteobacteria Verrucomicrobia 23 35 9 44 1 17 2 2 19 5 1 79 138 38 9 10 4 27 1 3 1 1 3 3 0 39 66 16 8 7 3 4 0 11 0 1 17 4 1 27 30 9 rence was determined based on cross references from Kruskal-Wallis (p<0.01) and ANOVA (P<0.01) tests on average ASV counts of different bacterial types. The number of hat significantly differed in relative abundance based on ASV counts for the control was more than twice the number of bacterial types that significantly differed in relative he single and multi-mixed treatments 11 / 22 PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 PLOS ONE Cover cropping and soil microbiome al pone 0232453 May 5 2020 PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 12 / 22 PLOS ONE Cover cropping and soil microbiome Fig 4. Change in average relative abundance of significantly different ASVs within distinct phyla classifications at all three time points for the three different treatments. A. For the control treatment, the change in average relative abundance of significantly different bacteria taxa in their relative phyla classifications generally decreased throughout consecutive time points. Out of the 14 classified phyla, 11 demonstrated a significant change in average phyla relative abundance from time point 1 to time point 3. B. For the single-mix treatment, the change in average relative abundance of significantly different bacteria taxa in their relative phyla classifications generally decreased throughout consecutive time points. Out of the 8 classified phyla, 4 demonstrated a significant decrease in average phyla relative abundance between time point 1 and time point 3. C. For the multi-mix treatment, the change in average relative abundance of significantly different bacteria taxa in their relative phyla classifications generally increased throughout consecutive time points. Out of the 10 classified phyla, 5 demonstrated a significant change in average phyla relative abundance from time point 1 to time point 3. https://doi.org/10.1371/journal.pone.0232453.g004 Significant changes in phyla over time in single-mix treatment. Analysis of each treatment: Difference across time points Of the 213 bacteria types that were significantly different in relative abundance across time in the single-mix treat- ment, 86% were classified into 13 different phyla (Acidobacteria, Actinobacteria, Armatimona- detes, Bacteroidetes, Chlorobi, Chloroflexi, Cyanobacteria, Elusimicrobia, Firmicutes, Gemmatimonadetes, Planctomycetes, Proteobacteria, Verrucombicrobia), with the remaining taxa having unknown classifications (Table 3). Unlike the control treatment, the average abun- dance for significantly changing bacteria within 5 phyla significantly decreased from time point 1 to time point 3 (Kruskal-Wallis, p<0.01), similar to the decreasing trend of overall microbial abundance and diversity for the single-mix treatment (Fig 4B, Table 1). Significant changes in phyla over time in multi-mix treatment. Of the 138 bacterial types that were significantly different in relative abundance (ASV counts) across time in the multi-mix treatment, 88% were classified into 12 phyla (Acidobacteria, Actinobacteria, Arma- timonadetes, Bacteroidetes, Chloroflexi, Elusimicrobia, Firmicutes, Gemmatimonadetes, Nitrospirae, Planctomycetes, Proteobacteria, Verrucombicrobia) (Table 3). The average rela- tive abundance for 6 significantly changing bacterial taxa classified phyla significantly changed between time point 1 and time point 3 (Kruskal-Wallis p<0.01), as time point 3 had the great- est average abundances (Fig 4C). The average relative abundance of OTUs within the phyla of Actinobacteria, Armatimonadetes, Chloroflexi, Elusimicrobia, Gemmatimonadetes, and Nitrospirae did not change significantly throughout time, however, these phyla also had fewer significantly changing bacteria taxa in each of these phyla. Notably, the total ASV count changed from time point 1 to time point 3 for the multi-mix treatment in a similar trend as the control, which is consistent with the rest of the total micro- bial composition analysis (Table 1). Overall, the microbial composition trends for all three treatments over time were consistent with the phylum grouped analysis of specific bacteria taxa what significantly changed over time. Notably, the most predominant changing bacteria phyla classification such as Bacteroi- detes, Planctomycetes, Proteobacteria, and Verrucombicrobia indicated significant changes in average ASV abundance. These phyla are important to consider in discerning the trends and effects of our cover cropping treatments in the future. Analysis of unique bacterial types across various time points Of all bacterial types analyzed, 13.7% had common bacterial type identifier codes, indicating the same type of bacteria, while unique bacterial types were identified for each treatment group. https://doi.org/10.1371/journal.pone.0232453.g005 https://doi.org/10.1371/journal.pone.0232453.g005 the filtering criteria) and did not allow for a filtering analysis based on average ASV counts as some samples may have a high ASV count in only one of the three technical replicates, which was still filtered out in our unique bacteria analysis in order to eliminate all potential uncer- tainties or sequencing errors. The taxa filtering step eliminated all bacterial taxa that had more than 1/3 samples with zero ASV counts. After the rigorous filtering step, only 9.6% of bacteria from time point 2, 10.2% of bacteria from time point 3 met the filtering criteria and were included in the comparison for unique taxa at respective time points (rowSums(n = = 0)>3). However, due to the fewer replicates at time point 1, the filtering was not as selective for time point 1 and 54% of all bacterial types were selected for analysis of unique bacterial taxa (row- Sums(n = = 0)>1). Distinct identifier codes assigned during taxonomy assignment were used as a unit of measurement of taxa. Thus, the resulted list of unique bacteria may have similar taxon classifications, but were unique in their ASV mapping. At time point 2, both multi-mix and control decreased in the percentage of unique bacterial taxa, while the single mix increased in its proportion of taxa that were unique. At time point 3, the reverse occurred, while the single-mix had only 7 taxa that were unique. For the unique bacterial types in the single-mix, none of them were able to be classified beyond the family level. This change in unique taxa distribution from time point 1 to time point 3 confer with the trend that the single-mix decreased in overall microbial composition, while the control and multi-mix increased in unique bacterial taxa by time point 3, and the divergence between the bacterial taxa at time point 3 was analyzed further to draw out differences between the effect of control and multi-mix treatment at the bacterial genus level. There were 236 bacterial types unique to the control treatment at time point 3, notably, some examples include: Actinomadura, Methylocapsa, Telmatonacter, Candidatus Xiphinema- tobacter, Actinoallomurus. Analysis of unique bacterial types across various time points At time point 1, there were 1735 bacterial types from the control, 1424 from the single-mix and 1236 bacterial types from the multi-mix that had more than 1/3 of the replicates with nonzero ASV counts. Of all bacterial types analyzed, 16.5% had common bacterial type identifier codes, indicating the same type of bacteria, while unique bacterial types were identified for each treatment group. B. At time point 2, there were 386 bacterial types from the control, 488 from the single- mix and 245 bacterial types from the multi-mix that had more than 1/3 of the replicates with nonzero ASV counts. Of all bacterial types analyzed, 15.1% had common bacterial type identifier codes, indicating the same type of bacteria, while unique bacterial types were identified for each treatment group. C. At time point 3, there were 622 bacterial types from the control, 211 from the single-mix and 521 bacterial types from the multi-mix that had more than 1/3 of the replicates with nonzero ASV counts. Of all bacterial types analyzed, 13.7% had common bacterial type identifier codes, indicating the same type of bacteria, while unique bacterial types were identified for each treatment group. Fig 5. Change in unique bacterial types for each treatment at different time points. All identified bacterial taxa were filtered by their respective ASV counts for individual samples. A. At time point 1, there were 1735 bacterial types from the control, 1424 from the single-mix and 1236 bacterial types from the multi-mix that had more than 1/3 of the replicates with nonzero ASV counts. Of all bacterial types analyzed, 16.5% had common bacterial type identifier codes, indicating the same type of bacteria, while unique bacterial types were identified for each treatment group. B. At time point 2, there were 386 bacterial types from the control, 488 from the single- mix and 245 bacterial types from the multi-mix that had more than 1/3 of the replicates with nonzero ASV counts. Of all bacterial types analyzed, 15.1% had common bacterial type identifier codes, indicating the same type of bacteria, while unique bacterial types were identified for each treatment group. C. At time point 3, there were 622 bacterial types from the control, 211 from the single-mix and 521 bacterial types from the multi-mix that had more than 1/3 of the replicates with nonzero ASV counts. Analysis of unique bacterial types across various time points While analyzing overall microbial composition changes of respective treatments, we recog- nized that the overall trend was that single-mix had a decrease in overall microbial composi- tion whereas control and multi-mix treatments alter the microbiome in similar trends. In order to elucidate the difference between the control and multi-mix treatment, even their dif- ference with the single-mix treatment, we analyzed populations of bacteria that were unique to each treatment group at the respective time points (Fig 5). In order to find the truly unique bacterial species that were present in one treatment but not the other, we were very strict in PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 13 / 22 PLOS ONE Cover cropping and soil microbiome Fig 5. Change in unique bacterial types for each treatment at different time points. All identified bacterial taxa were filtered by their respective ASV counts for Fig 5. Change in unique bacterial types for each treatment at different time points. All identified bacterial taxa were filtered by their respective ASV counts for individual samples. A. At time point 1, there were 1735 bacterial types from the control, 1424 from the single-mix and 1236 bacterial types from the multi-mix that had more than 1/3 of the replicates with nonzero ASV counts. Of all bacterial types analyzed, 16.5% had common bacterial type identifier codes, indicating the same type of bacteria, while unique bacterial types were identified for each treatment group. B. At time point 2, there were 386 bacterial types from the control, 488 from the single- mix and 245 bacterial types from the multi-mix that had more than 1/3 of the replicates with nonzero ASV counts. Of all bacterial types analyzed, 15.1% had common bacterial type identifier codes, indicating the same type of bacteria, while unique bacterial types were identified for each treatment group. C. At time point 3, there were 622 bacterial types from the control, 211 from the single-mix and 521 bacterial types from the multi-mix that had more than 1/3 of the replicates with nonzero ASV counts. Of all bacterial types analyzed, 13.7% had common bacterial type identifier codes, indicating the same type of bacteria, while unique bacterial types were identified for each treatment group. https://doi org/10 1371/journal pone 0232453 g005 Fig 5. Change in unique bacterial types for each treatment at different time points. All identified bacterial taxa were filtered by their respective ASV counts for individual samples. A. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 https://doi.org/10.1371/journal.pone.0232453.g005 Soil chemical composition Based on our KMnO4 extraction, we found that that the active carbon content of the soil didn’t significantly differ across treatments but rather changed across time (Fig 2). Between time points 1 and 2, the carbon content decreased, but following time point 2, and the planting of the broccoli crop, the carbon content increased. The consistency between treatments is sim- ilar to what was observed in a long-term study looking at the effects of legume, non-legume cover crops, and control treatments on soil organic carbon content [22]. Similarly, that study showed few differences between the control, non-legume cover crop, and low-legume cover crop treatment. Rather, carbon levels are likely to shift during the cycling of carbon through the cover crop and broccoli growth periods. Based on previous reports, organic carbon levels tend to decrease throughout the cultivation period due to the disturbances to the soil, which exposes the organic matter and allows for its decomposition [23]. This observation may explain why our carbon content decreased from time point 1 to time point 2 (Fig 2). In addi- tion, broccoli is a C3 plant, and previous studies on legume-based cropping have shown that C3 plants provide significant soil organic carbon increases through plant residues [24, 25]. This could explain the carbon content increase between time points 2 and 3 when the broccoli growth took place (Fig 2). Similar to our carbon data, our results on Cation Exchange Capacity (CEC) show that they did not differ based on the cover crop treatment, but increased significantly over time (Fig 3). CEC values are important determinants of soil chemical quality, particularly the retention of major nutrient cations Ca, Mg and K, and immobilization of potentially toxic cations Al and Mn. It can be indicators of soil health as well as soil capacity to absorb nutrients, pesticides, and other chemicals [26]. Our data suggest that cover crop treatments do not significantly affect the soil chemical absorption capacity. Together, based on our results, we did not find an association between the chemical compo- sition and the cover crop treatments. Instead, we found that time played a larger role in influ- encing the active carbon content and CEC of the soil. Previous literature has shown that soil organic carbon content can directly affect the soil microbial community composition [27] and organic carbon content has been found to be a limiting factor in soil bacterial development [28]. Discussion Our investigation aimed to understand the impact of various organic cover crop strategies on the microbial composition within the soil as well as the resulting impact of soil fertility. We hypothesized that the higher cover crop diversity would result in a more complex soil rhizo- sphere, diverse microbial community composition, and higher crop production. Overall, our results supported our hypothesis because the multi-mix treatment had greater diversity relative to the single-mix conditions at the end of the field study. Our preliminary analysis also sug- gests that the multi-mix treatment promoted the abundance of certain bacteria that could potentially provide benefits for plant growth. Thus, our results show that multi-mix is likely a more favorable option compared to single-mix for farmers if the focus is to maintain microbial diversity. Analysis of unique bacterial types across various time points Meanwhile, there were 147 bacterial types unique to the multi-mix treatment, but all the genus level classifications of the unique bacterial types in the multi-mix were the same as the control, despite having different identifier codes. Thus, different bacterial types were unique to the control and multi-mix treatments due to their differences in PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 14 / 22 PLOS ONE Cover cropping and soil microbiome amplicon mapping, but they had similar taxonomical classifications. More inferences based on the unique bacterial genera identified in each treatment helped elucidate the difference between the multi-mix and control treatments. amplicon mapping, but they had similar taxonomical classifications. More inferences based on the unique bacterial genera identified in each treatment helped elucidate the difference between the multi-mix and control treatments. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 Diversity measure Based on the Shannon diversity index, the single-mix treatment demonstrated consistent decreasing microbial diversity over time while the control and multi-mix treatment groups showed a similar trend of decreasing then increasing in diversity over time (Table 1). In fact, by time point 3, single-mix showed the lowest diversity index, suggesting that this treatment likely contributed to decreasing the microbial diversity of the soil over time (Table 1). Com- pared to single-mix, multi-mix and control mix decreased between time point 1 and 2, and increased between time point 2 and 3. In the past, other studies have also observed a temporary decrease in microbial diversity when plant diversity is at its peak, and this supports the same observation in our results [30]. Subsequently, the wide variety of plants in the control and multi-mix treatments likely contributed to the increasing diversity between time points 2 and 3 due to the influence of decomposing plant litter on the activity and community structure of soil microbial communities [31]. This effect has been shown across many types of ecosystems including grasslands, croplands, and natural forests [32, 33]. All together, these results show that limitations on plant diversity can negatively impact microbial diversity as well. Thus, in a practical sense, when determining which treatment may be better for increasing microbial diversity over time, we conclude that multi-mix is better than single-mix for the purpose of increasing potential beneficial microbial diversity. Further analysis remains to be conducted on the specific groups of microbes that are unique to the treatments. It is also worth noting that the Shannon index is very sensitive to rare groups, meaning that while some groups may exist in very low abundance, they still represent part of the overall diversity in the samples [21]. However, when we look at the ANOSIM and ANOVA analysis of Aitchison distances of overall microbial composition, we find that at time point 2 and 3, the treatments differ significantly in microbial abundance. Notably, single-mix showed the highest abundance in time point 2 and the lowest abundance in time point 3 (Table 1). This suggests that within the process of decomposing the cover crops and growing the broccoli crop, the microbial abundance of multi-mix and control treatments increased more than that of single- mix. Diversity measure This may be due to the increased diversity of plant litter that the soil was exposed during the process of decomposition, thus allowing a greater abundance of microbial growth in the treatments with higher plant diversity. This observation is consistent with other studies that observed an increase in microbial biomass with increased plant diversity [31]. Overall, our data, along with the literature, support the conclusion that the increased diver- sity of cover crops promotes greater microbial diversity of the soil. And together with the soil chemical analysis, this change in microbial diversity is independent of carbon content and CEC. This suggests that other environmental factors beyond carbon content and CEC, such as pH, may be a larger determining factor in the soil microbial diversity [34]. Soil chemical composition Interestingly, although we observed an increase in the organic carbon content in time point 3 across all treatments, it did not correlate to consistent changes in microbial diversity. Based on our data, there are other factors beyond carbon content and CEC that contribute to PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 15 / 22 PLOS ONE Cover cropping and soil microbiome the changes in microbial diversity. While the short timeframe of study could explain the lack of association between treatment and chemical composition, others have reported differences across treatments within as few as 60 days [29]. This suggests that even if we extended the timeframe for our study, we may not see substantial changes in CEC and carbon content across the different treatments. Thus, our data demonstrate that the cover crop treatments tested do not substantially impact the carbon content and CEC, and that other measurements, such as microbial diversity, may be more valuable in examining differences in soil profiles. Major phyla and differences across treatments Within each treatment, we identified bacterial types with significantly changing abundance (ASV counts) and classified them into their respective phyla and compared the changes at the PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 16 / 22 PLOS ONE Cover cropping and soil microbiome phyla level across the time points. Notably, 485 bacterial types significantly changed in relative abundance in the control treatment, while only 213 and 138 bacterial types significantly changed in relative abundance in the single-mix and multi-mix, respectively. In the control treatment, the soil was more exposed to environmental and weather changes due to the lack of substantial vegetation covering the soil which may have contributed to the higher fluctuations of microbial abundances [35, 36]. While the multi-mix treatment showed fewer changing ASVs compared to the control, it had more compared to the single-mix. Additionally, while we observed significant decreases in the phyla-level abundances of certain bacterial types over time in the single-mix treatment, the multi-mix showed a decrease then an increase in the abundance of some phyla-level bacterial types (Fig 4B and 4C). Analysis of the significantly changing bacterial types within the multi- mix treatment suggests that the multi-mix cover cropping process may play a role in promot- ing the growth of bacteria that could potentially provide benefits like rhizoremediation and moisture retention [37, 38, 39, 40]. Within all of the treatments, there were significant fluctuations in certain types of bacteria with known beneficial properties (Fig 4). For example, abundance of Proteobacteria, which includes a wide variety of both pathogens and nitrogen-fixing bacteria, increased in the multi- mix and control treatments between time points 2 and 3 [37]. Bacteroidetes, another phylum that significantly changed in abundance across time in all treatments, are known to specialize in the organic matter degradation and is considered a biological indicator of agricultural soil usage [38]. Planctomycetes also altered significantly across time among all treatments. It is known for anaerobic ammonium oxidation and it is associated with low nitrogen fertilizers [39]. Even though some phyla were observed to decrease in abundance in treatments, some unique genera increased in abundance in the multi-mix and control treatments by time point 3, suggesting that these treatments could promote the growth of microbes to that could poten- tially facilitate agricultural development (Fig 4). PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 Unique bacterial types Concurrent with our other data, we observed that the number of unique bacteria for single- mix decreased over time. In comparison, the control and multi-mix treatment first decreased then increased in unique bacterial taxa by time point 3 (Fig 5). Based on our analysis of unique bacteria types at time point 3 in the treatments, we found many potentially beneficial types existing in both control and multi-mix. Many bacterial types have been previously isolated from soil samples and identified for their potential beneficial roles. For example, Actinoma- dura is a known producer of antiviral and antibiotic compounds, Methylocapsa oxidizes atmo- spheric methane aerobically and assimilates carbon from both methane and carbon dioxide, and Telmatonacter is also a potential beneficial bacterium found in many soil communities [40, 41, 42]. Candidatus Xiphinematobacter serves important functions in the degradation of complex organic compounds like cellulose and starch and Actinoallomurus were suggested as plant-growth-promoting agents in the past [43, 44]. However, the limits of detection confined our ability to identify the bacterial types much beyond the genus level, thus, we still cannot exclude the possibility that the multi-mix treatment could contribute to the increase in poten- tially harmful bacteria. We now have a repository list of bacteria to analyze in the future as our understanding of the microbiome improves. By then, we may develop a clearer picture of the positive and/or negative roles that these bacterial types can play in the agricultural process. Our data are even more promising given that we have observed significant changes in rela- tively small plots. Taking into account the edge effect on changing soil nutrient conditions, we considered the possibility that the changes in microbial diversity could have been influenced PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 17 / 22 PLOS ONE Cover cropping and soil microbiome by the changing soil conditions near the edge by neighboring vegetation. In several studies in edge effect, field edges have been associated with increased organic C content [8, 45], however, in our samples, we did not find an increase in the total ASV abundance in our treatments that would likely be associated with this effect [27]. While we did observe an increase in organic C content over time, this could be influenced by the edge effect or the cycling of cover crop and broccoli growth. Unique bacterial types Although we are unable to determine the predominant influence for the phe- nomenon and it remains a possibility that the edge effect impacted carbon and other nutrient levels, ultimately these soil fertility changes were independent of the microbial diversity changes that we observed. Given our observations, and considering the limitations to our analyses, our results suggest that the multi-mix and control treatment selects for potentially agriculturally beneficial bacte- ria while also maintaining overall microbial composition compared to a single-mix treatment. Our data contributes to the growing body of knowledge of how specific agricultural processes can significantly alter the soil microbiome, which may have potential impacts for the health and success of crops. Furthermore, our results provide a practical farming application viable to both conventional and organic farmers. Future investigations include analyzing the specific roles of the bacteria and the effect of these selected bacteria on metabolic processes such as nitrogen fixation. Supporting information S1 Fig. Average monthly temperature (A) and average total rainfall (B) in comparison to data from past two years. A decrease in average temperature was observed. The summer of 2017 had an increase in total rainfall, but the winter of 2017 was drier. (TIF) (TIF) S1 Table. Cover crop characteristics and statistics. (TIF) S2 Table. Aitchison distance tables of normalized distances between different time points of all the treatments. (TIF) S3 Table. Aitchison distance tables of normalized distances between different treatments within each individual time point. (TIF) S1 File. Analysis of ASV abundance of all identified bacterial types in the control treat- ment. ANOVA (S1 File -> anova_ocs.txt) ad Kruskal-Wallis (S1 File -> kw_ocs.txt) tests were performed for the control treatment to find bacterial types that significantly differed in ASV counts across time points. The list of significantly changing bacterial types was included (S1 File -> control_sig.csv) as well as summary statistics (average and standard deviation of abun- dance) of significantly changing bacteria categorized by their phyla classifications (S1 File -> control_significant_taxa_stats.csv). (ZIP) S2 File. The alpha rarefication analysis for control. The alpha rarefication analysis was used to determine if the richness of samples has been fully observed or sequenced. The alpha diver- sity rarefication plot was included (S2 File -> alpha_rarefaction -> rarefaction.qzv) to illus- trate different rarefication measurements such as Shannon and number of OTUs. It showed that richness of all treatments was achieved with sequence depth of about 10000 reads. All treatments received sequencing depth that was much more 70,000. This suggested that the S3 File. Analysis of OTU abundance of all identified OTUs in the single-mix treatment. S3 File. Analysis of OTU abundance of all identified OTUs in the single-mix treatment. ANOVA (S3 File -> anova_ocs.txt) ad Kruskal-Wallis (S3 File -> kw_ocs.txt) tests were per- formed for the single-mix treatment to find bacterial types that significantly differed in ASV counts across time points. The list of significantly changing bacterial types was included (S3 File -> singleMix_sig.csv) as well as summary statistics (average and standard deviation of abundance) of significantly changing bacteria categorized by their phyla classifications (S3 File -> singleMix_significant_taxa_stats.csv). (ZIP) S4 File. The alpha rarefication analysis for single-mix. The alpha rarefication analysis was used to determine if the richness of samples has been fully observed or sequenced. The alpha diversity rarefication plot was included (S4 File -> alpha_rarefaction -> rarefaction.qzv) to illustrate different rarefication measurements such as Shannon and number of OTUs. It showed that richness of all treatments was achieved with sequence depth of about 10000 reads. All treatments received sequencing depth that was much more 70,000. This suggested that the analysis discovered all microbes in these samples. Statistical tests of Shannon indices were included (S4 File -> Shannon_group_significance.qzv). PC analysis of reference sequence cat- egorized by samples was included in the output qzv file (S4 File -> beta_rarefaction -> weight- ed_unifrac.qzv) using weighted UniFrac distance. (ZIP) S5 File. Analysis of OTU abundance of all identified OTUs in the multi-mix treatment. ANOVA (S5 File -> anova_ocs.txt) ad Kruskal-Wallis (S5 File -> kw_ocs.txt) tests were per- formed for the multi-mix treatment to find bacterial types that significantly differed in ASV counts across time points. The list of significantly changing bacterial types was included (S5 File -> multiMix_sig.csv) as well as summary statistics (average and standard deviation of abundance) of significantly changing bacteria categorized by their phyla classifications (S5 File -> multiMix_significant_taxa_stats.csv). (ZIP) S6 File. The alpha rarefication analysis for multi-mix. The alpha rarefication analysis was used to determine if the richness of samples has been fully observed or sequenced. The alpha diversity rarefication plot was included (S6 File -> alpha_rarefaction -> rarefaction.qzv) to illustrate different rarefication measurements such as Shannon and number of OTUs. It showed that richness of all treatments was achieved with sequence depth of about 10000 reads. All treatments received sequencing depth that was much more 70,000. This suggested that the analysis discovered all microbes in these samples. Statistical tests of Shannon indices were included (S6 File -> Shannon_group_significance.qzv). S3 Table. Aitchison distance tables of normalized distances between different treatments within each individual time point. (TIF) S3 Table. Aitchison distance tables of normalized distances between different treatments within each individual time point. (TIF) S2 File. The alpha rarefication analysis for control. The alpha rarefication analysis was used to determine if the richness of samples has been fully observed or sequenced. The alpha diver- sity rarefication plot was included (S2 File -> alpha_rarefaction -> rarefaction.qzv) to illus- trate different rarefication measurements such as Shannon and number of OTUs. It showed that richness of all treatments was achieved with sequence depth of about 10000 reads. All treatments received sequencing depth that was much more 70,000. This suggested that the PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 18 / 22 PLOS ONE Cover cropping and soil microbiome analysis discovered all microbes in these samples. Statistical tests of Shannon indices were included (S2 File -> Shannon_group_significance.qzv). PC analysis of reference sequence cat- egorized by samples was included in the output qzv file (S2 File -> beta_rarefaction -> weight- ed_unifrac.qzv) using weighted UniFrac distance. (ZIP) S3 File. Analysis of OTU abundance of all identified OTUs in the single-mix treatment. PC analysis of reference sequence cat- egorized by samples was included in the output qzv file (S6 File -> beta_rarefaction -> weight- ed_unifrac.qzv) using weighted UniFrac distance. (ZIP) S7 File. List of unique bacterial types for each time point. For every time point, there was a full list of all the bacterial types that were unique to each treatment summarized in a csv file (ex. “S7 File -> tp1 -> C_unique.csv” was the list of unique bacterial types for the control PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 19 / 22 PLOS ONE Cover cropping and soil microbiome treatment at time point 1) including the identifier code, taxonomy code, ASV count for each bacterial type. Relatively unique populations were also listed for future references. (ZIP) treatment at time point 1) including the identifier code, taxonomy code, ASV count for each bacterial type. Relatively unique populations were also listed for future references. (ZIP) Acknowledgments We are grateful to Dr. Emily McLean for assistance with statistical analysis, and Dr. Latonia Taliaferro-Smith for her mentorship. Author Contributions Conceptualization: Linda Wu, Daniel Parson, Sarah C. Fankhauser. Conceptualization: Linda Wu, Daniel Parson, Sarah C. Fankhauser. Data curation: Zengyan Wang, Magdy S. Alabady, Sarah C. Fankhauser. Data curation: Zengyan Wang, Magdy S. Alabady, Sarah C. Fankhauser. Formal analysis: Charlotte H. Wang, Linda Wu, Zengyan Wang, Magdy S. Alabady, Sarah C. Fankhauser. Funding acquisition: Sarah C. Fankhauser. Funding acquisition: Sarah C. Fankhauser. Investigation: Charlotte H. Wang, Sarah C. Fankhauser. Investigation: Charlotte H. Wang, Sarah C. Fankhauser. Project administration: Sarah C. Fankhauser. Resources: Sarah C. Fankhauser. Resources: Sarah C. Fankhauser. Supervision: Sarah C. Fankhauser. Writing – original draft: Charlotte H. Wang, Linda Wu. Writing – original draft: Charlotte H. Wang, Linda Wu. Writing – review & editing: Charlotte H. Wang, Linda Wu, Zainab Molumo, Sarah C. Fankhauser. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 References bioinformatics.babraham.ac.uk/projects/trim_galore/ 15. Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. DADA2: High-resolution sample inference from Illumina amplicon data. Nature Methods. 2016 Jul; 13(7):581–3. https://doi.org/ 10.1038/nmeth.3869 PMID: 27214047 16. Desantis TZ, Hugenholtz P, Larsen N, Rojas M, Brodie EL, Keller K, et al. Greengenes, a Chimera- Checked 16S rRNA Gene Database and Workbench Compatible with ARB. Applied and Environmental Microbiology. 2006; 72(7):5069–72. https://doi.org/10.1128/AEM.03006-05 PMID: 16820507 17. Glo¨ckner FO, Yilmaz P, Quast C, Gerken J, Beccati A, Ciuprina A, et al. 25 years of serving the commu- nity with ribosomal RNA gene reference databases and tools. Journal of Biotechnology. 2017; 261:169–76. https://doi.org/10.1016/j.jbiotec.2017.06.1198 PMID: 28648396 18. Knight R, Vrbanac A, Taylor BC, Aksenov A, Callewaert C, Debelius J, et al. Best practices for analysing microbiomes. Nature Reviews Microbiology. 2018 Jul; 16(7):410–22. https://doi.org/10.1038/s41579- 018-0029-9 PMID: 29795328 19. Mert MC, Filzmoser P, Hron K. Error Propagation in Isometric Log-ratio Coordinates for Compositional Data: Theoretical and Practical Considerations. Math Geosci. 2016; 48(8):941–61. https://doi.org/10. 1007/s11004-016-9646-x PMID: 28316755 20. alpha_diversity.py–Calculate alpha diversity on each sample in an otu table, using a variety of alpha diversity metrics—Homepage [Internet]. [cited 2019 Dec 6]. Available from: http://qiime.org/scripts/ alpha_diversity.html 21. Shannon CE, Weaver W. The mathematical theory of communication. Urbana: University of Illinois Press; 1999. 22. Gaby JC, Buckley DH. A comprehensive aligned nifH gene database: a multipurpose tool for studies of nitrogen-fixing bacteria. Database. 2014; 2014. 23. Mazzoncini M, Tek BS, Paolo B, Antichi D, and Risaliti R. Long-Term Effect of Tillage, Nitrogen Fertiliza- tion and Cover Crops on Soil Organic Carbon and Total Nitrogen Content. Soil and Tillage Research. 2011; 114:165–74. 24. Schlesinger WH, and Andrews JA. Soil Respiration and the Global Carbon Cycle. Biogeochemistry. 2000; 48:7–20. 25. Sage RF, and Zhu XG. Exploiting the Engine of C(4) Photosynthesis. Journal of Experimental Botany. 2011; 62:2989–3000. https://doi.org/10.1093/jxb/err179 PMID: 21652533 26. Drinkwater LE, Wagoner P, and Sarrantonio M. Legume-Based Cropping Systems Have Reduced Car- bon and Nitrogen Losses. Nature. 1998; 396:262. 27. Allen DE, Singh BP, and Dalal RC. Soil Health Indicators Under Climate Change: A Review of Current Knowledge. In Soil Health and Climate Change. 2011:25–45 28. Drenovsky RE, Vo D, Graham KJ, and Scow KM. Soil Water Content and Organic Carbon Availability Are Major Determinants of Soil Microbial Community Composition. Microbial Ecology. 2004; 48:424– 30. https://doi.org/10.1007/s00248-003-1063-2 PMID: 15692862 29. Alde´n L, Demoling F, and Bååth E. Rapid Method of Determining Factors Limiting Bacterial Growth in Soil. References 1. Buckley DH, and Schmidt TM. The Structure of Microbial Communities in Soil and the Lasting Impact of Cultivation. Microbial Ecology. 2001; 42:11–21. https://doi.org/10.1007/s002480000108 PMID: 12035077 2. Tilak KVBR, Ranganayaki N, Pal KK, De R, Saxena AK, Nautiyal CS, et al. Diversity of plant growth and soil health supporting bacteria. Current science. 2005;136–150. 3. Nair A and Ngouajio M. Soil microbial biomass, functional microbial diversity, and nematode community structure as affected by cover crops and compost in an organic vegetable production system. Applied Soil Ecology. 2012; 58:45–55. 4. Haramoto ER., and Gallandt ER. Brassica Cover Cropping for Weed Management: A Review. Renew- able Agriculture and Food Systems. 2004; 19:187–98. 5. Locke MA, Zablotowicz RM, Steinriede RW, Testa S, Reddy KN. Conservation Management in Cotton Production: Long-Term Soil Biological, Chemical, and Physical Changes. Soil Science Society of Amer- ica Journal. 2013; 77:974. 6. Chu M, Jagadamma S, Walker FR, Eash NS, Buschermohle MJ, and Duncan LA. Effect of Multispecies Cover Crop Mixture on Soil Properties and Crop Yield. Agricultural and Environmental Letters. 2017;2. 7. Vukicevich E, Lowery T, Bowen P, U´ rbez-Torres JR, Hart M. Cover crops to increase soil microbial diversity and mitigate decline in perennial agriculture. A review. Agronomy for Sustainable Develop- ment. 2016; 36(3). 8. Creamer N, and Baldwin K. Summer Cover Crops \| NC State Extension Publications [Internet]. Summer Cover Crops \| NC State Extension Publications. [cited 2019May26]. Available from: https://content.ces. ncsu.edu/summer-cover-crops 9. Ruwanza S. The Edge Effect on Plant Diversity and Soil Properties in Abandoned Fields Targeted for Ecological Restoration. Sustainability. 2019; 11:140. 20 / 22 PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 PLOS ONE Cover cropping and soil microbiome 10. Shiri J, Keshavarzi A, Kisi O, Iturraran-Viveros U, Bagherzadeh A, Mousavi R, et al. Modeling soil cation exchange capacity using soil parameters: Assessing the heuristic models. Computers and Electronics in Agriculture. 2017; 135:242–51. 11. Islam K, Stine M, Gruver J, Samson-Liebig S, Weil R. Estimating active carbon for soil quality assess- ment: A simplified method for laboratory and field use. American Journal of Alternative Agriculture. 2003; 18(1):3–17. 12. Klindworth A, Pruesse E, Schweer T, Peplies J, Quast C, Horn M, et al. Evaluation of general 16S ribo- somal RNA gene PCR primers for classical and next-generation sequencing-based diversity studies. Nucleic Acids Research. 2012; 41(1). 13. Illumina. 16S Metagenomic Sequencing Library Preparation. 2013; 1–28. 14. Babraham Bioinformatics—Trim Galore! [Internet]. [cited 2019 Dec 6]. Available from: http://www. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 References Applied and Environmental Microbiology. 2001; 67:1830–38. https://doi.org/10.1128/AEM.67.4. 1830-1838.2001 PMID: 11282640 30. Kowalchuk GA, Buma DS, Boer WD, Klinkhamer PGL, and van Veen JA. Effects of Above-Ground Plant Species Composition and Diversity on the Diversity of Soil-Borne Microorganisms. Antonie van Leeuwenhoek, 2002; 81:509. https://doi.org/10.1023/a:1020565523615 PMID: 12448746 31. Zavalloni C, Alberti G, Biasiol S, Vedove GD, Fornasier F, Liu J, et al. Microbial Mineralization of Biochar and Wheat Straw Mixture in Soil: A Short-Term Study. Applied Soil Ecology. 2011; 50:45–51. 21 / 22 PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 PLOS ONE Cover cropping and soil microbiome 32. Chen C, Han YH, Chen XL, and Huang ZQ. Meta-Analysis Shows Positive Effects of Plant Diversity on Microbial Biomass and Respiration. Nature Communications. 2019; 10:1332. https://doi.org/10.1038/ s41467-019-09258-y PMID: 30902971 33. Bardgett RD and Shine A. Linkages between Plant Litter Diversity, Soil Microbial Biomass and Ecosys- tem Function in Temperate Grasslands. Soil Biology and Biochemistry. 1999; 31:317–21. 34. Li L, Xu M, Ali ME, Zhang W, Duan Y, and Li D. Factors Affecting Soil Microbial Biomass and Functional Diversity with the Application of Organic Amendments in Three Contrasting Cropland Soils during a Field Experiment. PLOS ONE. 2018; 13. 35. Xun W, Huang T, Zhao J, Ran W, Wang B, Shen Q, et al. Environmental Conditions Rather than Micro- bial Inoculum Composition Determine the Bacterial Composition, Microbial Biomass and Enzymatic Activity of Reconstructed Soil Microbial Communities. Soil Biology and Biochemistry. 2015; 90:10–18. 36. Robin T and Or D. Biophysical Processes Supporting the Diversity of Microbial Life in Soil. FEMS Micro- biology Reviews. 2017; 41:599–623. https://doi.org/10.1093/femsre/fux039 PMID: 28961933 37. Shin NR, Whon TW, & Bae JW. Proteobacteria: microbial signature of dysbiosis in gut microbiota. Trends in Biotechnology. 2015; 33:496–503. https://doi.org/10.1016/j.tibtech.2015.06.011 PMID: 26210164 38. Wolińska A, Kuźniar A, Zielenkiewicz U, Izak D, Szafranek-Nakonieczna A, Banach A, et al. Bacteroi- detes as a sensitive biological indicator of agricultural soil usage revealed by a culture-independent approach. Applied Soil Ecology. 2017; 119:128–137. 39. Fuerst JA. (2017). Chapter 1—Planctomycetes—New Models for Microbial Cells and Activities. In Kurt- bo¨ke I. (Ed.), Microbial Resources (pp. 1–27). 40. Kusserow K and Gulder TAM. Complete Genome Sequence of Actinomadura Parvosata Subsp. Kist- nae, A Rich Source of Novel Natural Product (Bio-)Chemistry. Journal of Genomics. 2017; 5:75–76. https://doi.org/10.7150/jgen.19673 PMID: 28698739 41. Tveit AT, Hestnes AG, Robinson SL, Schintlmeister A, Dedysh SN, Jehmlich N, et al. Widespread Soil Bacterium That Oxidizes Atmospheric Methane. PLOS ONE \| https://doi.org/10.1371/journal.pone.0232453 May 5, 2020 References Proceedings of the National Academy of Sciences of the United States of America. 2019; 116:8515–24. https://doi.org/10.1073/pnas.1817812116 PMID: 30962365 42. Shu C, Qi GF, Ma GQ, and Zhao XY. Biochar Amendment Controlled Bacterial Wilt through Changing Soil Chemical Properties and Microbial Community. Microbiological Research. 2020; 231:126373. https://doi.org/10.1016/j.micres.2019.126373 PMID: 31739260 43. Beulig F, Heuer VB, Akob DM, Viehweger B, Elvert M, Herrmann M, et al. Carbon Flow from Volcanic CO 2 into Soil Microbial Communities of a Wetland Mofette. The ISME Journal. 2015; 9:746–59. https:// doi.org/10.1038/ismej.2014.148 PMID: 25216086 44. Sheng Q, Xing K, Jiang JH, Xu LH, and Li WJ. Biodiversity, Bioactive Natural Products and Biotechno- logical Potential of Plant-Associated Endophytic Actinobacteria. Applied Microbiology and Biotechnol- ogy. 2011; 89:457–73. https://doi.org/10.1007/s00253-010-2923-6 PMID: 20941490 45. Santos MOD, Barreto-Garcia PAB, Gama-Rodrigues EF, MODS, Barreto-Garcia PAB, and Gama- Rodrigues EF. Soil Microbial Biomass as an Edge Effect Indicator in Semi-Deciduous Seasonal Forest Fragments. Floresta e Ambiente. 2019; 26. 22 / 22
https://openalex.org/W2782534363	https://europepmc.org/articles/pmc5896467?pdf=render	English	null	Impact of ambient gases on the mechanism of [Cs<sub>8</sub>Nb<sub>6</sub>O<sub>19</sub>]-promoted nerve-agent decomposition	Chemical science	2,018	cc-by	12,321	Received 21st November 2017 Accepted 5th January 2018 DOI: 10.1039/c7sc04997h rsc.li/chemical-science Chemical Science EDGE ARTICLE ases on the mechanism of ed nerve-agent oll, b Diego Troya,b Daniel L. Collins-Wildman,c and Djamaladdin G. Musaev ab O2, CO2 and SO2 on the structure, stability and decontamination studied computationally and experimentally. It was found that strongly than it adsorbs water and Sarin (GB) and that these nation. The impacts of diamagnetic CO2 and SO2 molecules damentally diﬀerent from that of NO2 radical. At ambient t NO2 radical to Cs8Nb6O19 conferred partial radical character ronger coordination of the second NO2 adsorbent to form pecies. Moreover, at low temperatures, NO2 radicals formed akly interacted with Cs8Nb6O19. It was found that both in the molecules, GB decontamination by the Cs8Nb6O19 species olving: (a) the adsorption of water and the nerve agent on of a water molecule on a basic oxygen atom of the of the nascent OH group to the phosphorus center of Sarin, ed pentacoordinated-phosphorus intermediate, followed by d formation of POM-bound isopropyl methyl phosphonic acid cid (MPFA), respectively. The presence of the ambient gas ermediate stationary points relative to the asymptote of the drolysis barrier. These changes closely correlate with the most energetically stable intermediates of the GB hydrolysis to be Cs8Nb6O19/X-MPFA-(i-POH) and Cs8Nb6O19/X-(i-MPA)- mbient gas molecules. The high stability of these intermediates ding between the adsorbates and the protonated [Cs8Nb6O19/ H) and regeneration of the catalyst required deprotonation of on of the phosphonic acids i-MPA and MPFA. This catalyst othermic process, which is the rate-limiting step of the GB oth in the absence and presence of ambient gas molecules. oduction View Article Online View Journal \| View Issue ases on the mechanism of ed nerve-agent oll, b Diego Troya,b Daniel L. Collins-Wildman,c and Djamaladdin G. Musaev ab O2, CO2 and SO2 on the structure, stability and decontamination studied computationally and experimentally. It was found that strongly than it adsorbs water and Sarin (GB) and that these nation. The impacts of diamagnetic CO2 and SO2 molecules damentally diﬀerent from that of NO2 radical. At ambient t NO2 radical to Cs8Nb6O19 conferred partial radical character ronger coordination of the second NO2 adsorbent to form pecies. Moreover, at low temperatures, NO2 radicals formed akly interacted with Cs8Nb6O19. It was found that both in the molecules, GB decontamination by the Cs8Nb6O19 species olving: (a) the adsorption of water and the nerve agent on of a water molecule on a basic oxygen atom of the of the nascent OH group to the phosphorus center of Sarin, ed pentacoordinated-phosphorus intermediate, followed by d formation of POM-bound isopropyl methyl phosphonic acid cid (MPFA), respectively. The presence of the ambient gas ermediate stationary points relative to the asymptote of the drolysis barrier. These changes closely correlate with the most energetically stable intermediates of the GB hydrolysis to be Cs8Nb6O19/X-MPFA-(i-POH) and Cs8Nb6O19/X-(i-MPA)- mbient gas molecules. The high stability of these intermediates ding between the adsorbates and the protonated [Cs8Nb6O19/ H) and regeneration of the catalyst required deprotonation of on of the phosphonic acids i-MPA and MPFA. This catalyst othermic process, which is the rate-limiting step of the GB oth in the absence and presence of ambient gas molecules. oduction View Article Online View Journal \| View Issue This journal is © The Royal Society of Chemistry 2018 Alexey L. Kaledin,a Darren M. Driscoll, b Diego Troya,b Daniel L. Collins-Wildman,c Craig L. Hill, c John R. Morrisb and Djamaladdin G. Musaev ab The impact of ambient gas molecules (X), NO2, CO2 and SO2 on the structure, stability and decontamination activity of Cs8Nb6O19 polyoxometalate was studied computationally and experimentally. It was found that Cs8Nb6O19 absorbs these molecules more strongly than it adsorbs water and Sarin (GB) and that these interactions hinder nerve agent decontamination. The impacts of diamagnetic CO2 and SO2 molecules on polyoxoniobate Cs8Nb6O19 were fundamentally diﬀerent from that of NO2 radical. At ambient temperatures, weak coordination of the ﬁrst NO2 radical to Cs8Nb6O19 conferred partial radical character on the polyoxoniobate and promoted stronger coordination of the second NO2 adsorbent to form a stable diamagnetic Cs8Nb6O19/(NO2)2 species. Moreover, at low temperatures, NO2 radicals formed stable dinitrogen tetraoxide (N2O4) that weakly interacted with Cs8Nb6O19. It was found that both in the absence and presence of ambient gas molecules, GB decontamination by the Cs8Nb6O19 species proceeds via general base hydrolysis involving: (a) the adsorption of water and the nerve agent on Cs8Nb6O19/(X), (b) concerted hydrolysis of a water molecule on a basic oxygen atom of the polyoxoniobate and nucleophilic addition of the nascent OH group to the phosphorus center of Sarin, and (c) rapid reorganization of the formed pentacoordinated-phosphorus intermediate, followed by dissociation of either HF or isopropanol and formation of POM-bound isopropyl methyl phosphonic acid (i-MPA) or methyl phosphonoﬂuoridic acid (MPFA), respectively. The presence of the ambient gas molecules increases the energy of the intermediate stationary points relative to the asymptote of the reactants and slightly increases the hydrolysis barrier. These changes closely correlate with the Cs8Nb6O19–X complexation energy. The most energetically stable intermediates of the GB hydrolysis and decontamination reaction were found to be Cs8Nb6O19/X-MPFA-(i-POH) and Cs8Nb6O19/X-(i-MPA)- HF both in the absence and presence of ambient gas molecules. The high stability of these intermediates is due to, in part, the strong hydrogen bonding between the adsorbates and the protonated [Cs8Nb6O19/ X/H]+-core. Desorption of HF or/and (i-POH) and regeneration of the catalyst required deprotonation of the [Cs8Nb6O19/X/H]+-core and protonation of the phosphonic acids i-MPA and MPFA. This catalyst regeneration is shown to be a highly endothermic process, which is the rate-limiting step of the GB hydrolysis and decontamination reaction both in the absence and presence of ambient gas molecules. Impact of ambient gases on the mechanism of [Cs8Nb6O19]-promoted nerve-agent decomposition†‡ Alexey L. Kaledin,a Darren M. Driscoll, b Diego Troya,b Daniel L. Collins-Wildman Craig L. Hill, c John R. Morrisb and Djamaladdin G. Musaev ab Cite this: Chem. Sci., 2018, 9, 2147 EDGE ARTICLE Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Impact of ambient gases on the mechanism of [Cs8Nb6O19]-promoted nerve-agent decomposition†‡ Alexey L. Kaledin,a Darren M. Driscoll, b Diego Troya,b Daniel L. Collins-Wildman,c Craig L. Hill, c John R. Morrisb and Djamaladdin G. Musaev *ab aC. L. Emerson Center for Scientic Computation and Department of Chemistry, Emory University, Atlanta, Georgia, 30322, USA. E-mail: dmusaev@emory.edu bDepartment of Chemistry, Virginia Tech, Blacksburg, Virginia, 24061, USA. E-mail: jmorris@vt.edu cDepartment of Chemistry, Emory University, Atlanta, Georgia, 30322, USA. E-mail: chill@emory.edu † Dedicated to the memory of Prof. Keiji Morokuma. ‡ Electronic supplementary information (ESI) available: (1) The calculated transition states, intermediates and products of the GB hydrolysis and their important geometry parameters (in ˚A) for X ¼ SO2, (2) the calculated adsorption energies (in kcal mol1) of NO2 radicals to Cs8Nb6O19, (3) Cartesian coordinates for all reported structures in xyz format. (structure.xyz). See DOI: 10.1039/c7sc04997h aC. L. Emerson Center for Scientic Computation and Department of Chemistry, Emory University, Atlanta, Georgia, 30322, USA. E-mail: dmusaev@emory.edu bDepartment of Chemistry, Virginia Tech, Blacksburg, Virginia, 24061, USA. E-mail: jmorris@vt.edu cDepartment of Chemistry Emory University Atlanta Georgia 30322 USA E mail aC. L. Emerson Center for Scientic Computation and Department of Chemistry, Emory University, Atlanta, Georgia, 30322, USA. E-mail: dmusaev@emory.edu bDepartment of Chemistry, Virginia Tech, Blacksburg, Virginia, 24061, USA. E-mail: jmorris@vt.edu cDepartment of Chemistry, Emory University, Atlanta, Georgia, 30322, USA. E-mail: chill@emory.edu † D di t d t th f P f K iji M k aC. L. Emerson Center for Scientic Computation and Department of Chemistry, Emory University, Atlanta, Georgia, 30322, USA. E-mail: dmusaev@emory.edu Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Recently, the use of POMs to catalyze nerve agent decon- tamination has attracted wide attention, in part because POMs are molecular representations of metal oxides and are thus far more amenable than the latter to extensive synthetic composi- tional alteration and characterization at the molecular level.20,21 Polyoxoniobates (PONbs), including [Nb6O19]8, are eﬀective OP nerve agent hydrolysis compounds because their high negative charge densities (negative charge per polyanion oxygen) render them highly basic and nucleophilic. Thus, it is not surprising that the synthesis and in-depth analysis of the structures and reactivities of various (alkali and organic) salts of PONbs continue to be the focus of extensive studies.22,23 These studies show that the structures and, consequently, the catalytic activ- ities of these materials for nerve agent decontamination depend on many factors, including (but not limited to) the nature of the counter-cation, the pH of the solution, the aggregate state (powder or solid-state material) of the catalyst, the real-time environmental conditions, and the nature and concentration of ambient gas molecules. Although PONb catalysts have been shown to react with CWAs, the chemistry has yet to be characterized in the presence of ambient gases (for example, NO2, CO2 and SO2), which may aﬀect the stabilities, structural motifs, and activities of the PONb catalysts. Because this issue is vital to the application of PONbs as decontamination catalysts in real conditions, this paper probes the impact of the common battleeld contami- nants NO2, CO2 and SO2 on the structure, stability and decon- tamination activity of the exemplary PONb species Cs8Nb6O19. This study addresses in depth the eﬀects of these ambient gases on the structures of the catalysts and the base hydrolysis mechanism for Sarin degradation using density functional theory (DFT) calculations and infrared (IR) spectroscopy. Earlier research on Lindqvist hexaniobate alkali salts (M8Nb6O19, M ¼ Li, K, Cs) reported rapid hydrolysis of the OP agent Sarin (GB, propan-2-yl methylphosphonouoridate, see Scheme 1) both in aqueous solution and at the gas–surface interface.11 Small-angle X-ray scattering (SAXS) measurements showed aggregation of the OP compounds on the poly- oxoniobate (PONb), which led to the suggestion that the reac- tion follows a general base hydrolysis mechanism.12 Our subsequent computational study on the mechanism of decomposition of GB by a Cs-salt of PONb, Cs8Nb6O19 (or CsPONb), conrmed the general base hydrolysis mechanism of this reaction at the gas–surface interface.24 Chemical Science Chemical Science the serine OH to the phosphorus atom of the nerve agent. Thus, an atomistic/molecular level understanding of the hydrolysis of OP compounds by nucleophilic addition and other processes may lead to the development of more eﬀective materials and catalysts for nerve agent decontamination. Ongoing research eﬀorts have identied several organic and inorganic materials, including metal–organic frameworks (MOFs, especially UiO-66, NU-1000 and MOF-808),5–10 polyoxometalates (POMs),11–14 MOF/ POM hybrid materials,15,16 zirconium hydroxide,17 zeolites,18 and organic polymers,19 as eﬀective OP hydrolysis materials. Briey, we have found that GB degradation by Cs8Nb6O19 includes the following elementary steps (see Scheme 1): (a) the adsorption of water and the nerve agent on the Cs8Nb6O19 species, (b) concerted dissociation of the adsorbed water molecule on a basic oxygen atom of the polyoxoniobate and nucleophilic addition of the nascent OH group to the phos- phorus center of the nerve agent, (c) rapid reorganization of the resulting pentacoordinated phosphorus intermediate by disso- ciation of either HF or isopropanol, and formation of POM- bound isopropyl methyl phosphonic acid (i-MPA) or methyl phosphonouoridic acid (MPFA), respectively. The calculations showed that the phosphonic acids i-MPA and MPFA are strongly bound to the protonated [Cs8Nb6O19H]+-core through hydrogen bonds and electrostatic interactions with the Cs counter-ions, suggesting that full catalyst regeneration may require addi- tional treatment and depends on the nature of the counter- cations as well as the real-time (ambient) experimental conditions. This journal is © The Royal Society of Chemistry 2018 Introduction The design of materials that can rapidly, fully, and catalytically decontaminate chemical warfare agents (CWAs) and other toxic compounds is an increasingly active area of research and one that presents some questions in fundamental chemistry.1–5 As suggested by enzymatic chemistry, some of the most eﬀective strategies for CWA destruction involve catalyzed hydrolysis reactions.4 Specically, it is well established that the P–X bonds (X ¼ F, CN, SR, etc.) of organophosphorus (OP) nerve agents rapidly inactivate acetylcholinesterase (a serine hydrolase), the enzyme that facilitates hydrolysis of the neurotransmitter acetylcholine in the nervous system. This inactivation occurs through rapid nucleophilic addition and irreversible binding of bDepartment of Chemistry, Virginia Tech, Blacksburg, Virginia, 24061, USA. E-mail: jmorris@vt.edu cDepartment of Chemistry, Emory University, Atlanta, Georgia, 30322, USA. E-mail: chill@emory.edu † Dedicated to the memory of Prof. Keiji Morokuma. ‡ Electronic supplementary information (ESI) available: (1) The calculated transition states, intermediates and products of the GB hydrolysis and their important geometry parameters (in ˚A) for X ¼ SO2, (2) the calculated adsorption energies (in kcal mol1) of NO2 radicals to Cs8Nb6O19, (3) Cartesian coordinates for all reported structures in xyz format. (structure.xyz). See DOI: 10.1039/c7sc04997h This journal is © The Royal Society of Chemistry 2018 Chem. Sci., 2018, 9, 2147–2158 \| 2147 Edge Article View Article Online Edge Article View Article Online A. Structure of Cs8Nb6O19 in the presence of gaseous CO2, NO2 and SO2 Here, we divide our discussion into two subsections. First, we discuss the interaction of diamagnetic CO2 and SO2 molecules with Cs8Nb6O19; then, we present our ndings on the interac- tions between NO2 radical and Cs8Nb6O19. A1. Structure of Cs8Nb6O19 in the presence of gaseous CO2 and SO2. Our calculations reveal that Cs8Nb6O19 very strongly binds the ambient gas molecules CO2 and SO2 at various (terminal Ot and/or bridging Om) sites to form Cs8Nb6O19/X adducts. As reported in Table 1, where we present the calculated adsorption energies, there is a clear preference of adsorption of both CO2 and SO2 at the terminal oxygen site rather than at the bridging oxygen site. Indeed, the calculated enthalpies and free energies of adsorption (presented as DH/DG) of Cs8Nb6O19 + X / Cs8Nb6O19/X upon coordination of X to the Ot and Om sites are 29.0/23.2 and 16.7/10.8 kcal mol1 for X ¼ CO2 and 47.6/40.3 and 34.1/26.6 kcal mol1 for X ¼ SO2, respectively. Scheme 1 Schematics of: (a) the mechanism of Sarin (GB) hydrolysis (reported in ref. 24) promoted by the Cs8Nb6O19 polyoxoniobate and (b) the Cs8Nb6O19 polyoxoniobate. The factors that impact the strength of the Cs8Nb6O19/X bonding and site-preference of absorption were analyzed by investigating the resulting geometries of these complexes. From Fig. 1, where we present the most important geometries of the Scheme 1 Schematics of: (a) the mechanism of Sarin (GB) hydrolysis (reported in ref. 24) promoted by the Cs8Nb6O19 polyoxoniobate and (b) the Cs8Nb6O19 polyoxoniobate. This journal is © The Royal Society of Chemistry 2018 2148 \| Chem. Sci., 2018, 9, 2147–2158 Chemical Science View Article Online Chemical Science View Article Online View Article Online This journal is © The Royal Society of Chemistry 2018 Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. ( ) \| \| The strong coordination of CO2 to Cs8Nb6O19 is also sup- ported by IR spectroscopic studies, where the adsorption of CO2 onto Cs8Nb6O19 was investigated by recording the infrared spectra before, during, and aer exposure of Cs8Nb6O19 to a constant ux of CO2. Upon adsorption of CO2 on Cs8Nb6O19, three prominent features appeared in the infrared spectrum (Fig. 2, i) at 1659 cm1, 1290 cm1 and 1229 cm1. Interestingly, we observed no feature in the infrared spectrum around 2300 cm1, which would be associated with a linear CO2 molecule bound to the POM surface. The 1229 cm1 and 1290 cm1 bands are consistent with the IR-inactive symmetric stretch from the gas-phase (n1(C–O)), which becomes IR active upon binding to the POM. The 1659 cm1 band is likely related to the antisymmetric n3(C–O) stretch of CO2/CO3 (occurs at 2349 cm1 in the gas-phase25 and signicantly redshis upon adsorption). The presence of both the n1 and n3 bands in the infrared spectrum indicates a bent structure of the adsorbate at the surface. Upon evacuation of CO2 from the chamber, the 1229 cm1 spectroscopic feature diminishes, indicating that this band corresponds to the vibrational motion of weakly bound CO2 species on the surface (Fig. 2, ii); however, the two other features (1659 and 1290 cm1) persist until annealing the Cs8Nb6O19 sample at 423 K (Fig. 2, iii). The elevated Cs8Nb6O19/X(Ot) and Cs8Nb6O19/X(Om) isomeric species for X ¼ CO2 and SO2 (for full geometries of these species, see the ESI‡), we make the following conclusions: (a) The CO2 molecule in Cs8Nb6O19/CO2(Ot) is bound to Cs8Nb6O19 via multiple interactions, including two (C]O)CO2/ Cs interactions and one C–Ot interaction. The long calculated bond distances for the Cs–O1, Cs–O2 and Cs–C interactions (3.23, 3.25 and 3.28 ˚A, respectively) indicate they are relatively weak. In contrast, the C–Ot interaction is stronger, with a bond distance of 1.39 ˚A. As a result of this strong interaction, the Nb– Ot distance is elongated from 1.80 ˚A to 2.02 ˚A (see Fig. 1). Thus, while the interactions with the Cs cations provide additional stability to the complex Cs8Nb6O19/CO2(Ot), the primary (a) The CO2 molecule in Cs8Nb6O19/CO2(Ot) is bound to Cs8Nb6O19 via multiple interactions, including two (C]O)CO2/ Cs interactions and one C–Ot interaction. The long calculated bond distances for the Cs–O1, Cs–O2 and Cs–C interactions (3.23, 3.25 and 3.28 ˚A, respectively) indicate they are relatively weak. Edge Article Edge Article Chemical Science interaction occurs between the C atom of CO2 and the terminal oxygen atom of Cs8Nb6O19. Table 1 Adsorption energies (total electronic E, enthalpy H and Gibbs free energy G in kcal mol1) deﬁned as energy diﬀerences of the complex Cs8Nb6O19/X and its two separated fragments, Cs8Nb6O19 + X, where X ¼ CO2, NO2 and SO2, as shown in the leftmost column. The superscripts “t” and “m” indicate the position where the molecule is adsorbed on Cs8Nb6O19; CsN indicates NO2 adsorption to two Cs counter-cations in a symmetric manner Table 1 Adsorption energies (total electronic E, enthalpy H and Gibbs free energy G in kcal mol1) deﬁned as energy diﬀerences of the complex Cs8Nb6O19/X and its two separated fragments, Cs8Nb6O19 + X, where X ¼ CO2, NO2 and SO2, as shown in the leftmost column. The superscripts “t” and “m” indicate the position where the molecule is adsorbed on Cs8Nb6O19; CsN indicates NO2 adsorption to two Cs counter-cations in a symmetric manner In the Cs8Nb6O19/CO2(Om) isomer, where CO2 interacts with a bridging oxygen, the C–Om bond distance is longer (calculated to be 1.44 ˚A) than the C–Ot bond distance in Cs8Nb6O19/CO2(Ot), while the Cs–O1 and Cs–O2 interactions are slightly stronger (based on the calculated Cs–O1 and Cs–O2 distances). The above presented geometry parameters of Cs8Nb6O19/CO2(Ot) and Cs8PONb/CO2(Om) not only explain the calculated energy diﬀerence between these species, but are also consistent with the amounts of charge transfer from Cs8Nb6O19 to CO2: in the Cs8Nb6O19/CO2(Ot) and Cs8Nb6O19/CO2(Om) complexes, almost 0.8 \|e\| and 0.4 \|e\| negative charge is transferred from Cs8Nb6O19 to CO2, respectively. Furthermore, the putative CO3-group in Cs8Nb6O19/CO2(Ot) has a total negative charge of 1.52 \|e\|. Cs8Nb6O19/X(Ot) and Cs8Nb6O19/X(Om) isomeric species for X ¼ CO2 and SO2 (for full geometries of these species, see the ESI‡), we make the following conclusions: (a) The CO2 molecule in Cs8Nb6O19/CO2(Ot) is bound to Cs8Nb6O19 via multiple interactions, including two (C]O)CO2/ Cs interactions and one C–Ot interaction. The long calculated bond distances for the Cs–O1, Cs–O2 and Cs–C interactions (3.23, 3.25 and 3.28 ˚A, respectively) indicate they are relatively weak. In contrast, the C–Ot interaction is stronger, with a bond distance of 1.39 ˚A. As a result of this strong interaction, the Nb– Ot distance is elongated from 1.80 ˚A to 2.02 ˚A (see Fig. 1). Edge Article Thus, while the interactions with the Cs cations provide additional stability to the complex Cs8Nb6O19/CO2(Ot), the primary DE DH DG Ot: Cs8Nb6O19 + CO2 29.1 29.0 23.2 Om: Cs8Nb6O19 + CO2 16.8 16.7 10.8 Ot: Cs8Nb6O19 + SO2 48.2 47.6 40.3 Om: Cs8Nb6O19 + SO2 34.7 34.1 26.6 CsN: Cs8Nb6O19 + NO2 21.7 22.2 22.1 Ot: Cs8Nb6O19 + NO2 18.8 18.7 13.9 CsN and Ot: Cs8Nb6O19 + 2NO2 77.4 74.6 58.0 Cs8Nb6O19/CO2(Ot) + H2O 24.2 22.4 11.9 Cs8Nb6O19/SO2(Ot) + H2O 22.7 21.0 9.7 Cs8Nb6O19 + H2O 24.6 23.6 17.3 Cs8Nb6O19/H2O + GB 18.8 17.4 4.2 Cs8Nb6O19/H2O/CO2(Ot) + GB 18.0 17.4 2.4 Cs8Nb6O19/H2O/SO2(Ot) + GB 17.4 16.4 3.7 Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. DE DH DG Ot: Cs8Nb6O19 + CO2 29.1 29.0 23.2 Om: Cs8Nb6O19 + CO2 16.8 16.7 10.8 Ot: Cs8Nb6O19 + SO2 48.2 47.6 40.3 Om: Cs8Nb6O19 + SO2 34.7 34.1 26.6 CsN: Cs8Nb6O19 + NO2 21.7 22.2 22.1 Ot: Cs8Nb6O19 + NO2 18.8 18.7 13.9 CsN and Ot: Cs8Nb6O19 + 2NO2 77.4 74.6 58.0 Cs8Nb6O19/CO2(Ot) + H2O 24.2 22.4 11.9 Cs8Nb6O19/SO2(Ot) + H2O 22.7 21.0 9.7 Cs8Nb6O19 + H2O 24.6 23.6 17.3 Cs8Nb6O19/H2O + GB 18.8 17.4 4.2 Cs8Nb6O19/H2O/CO2(Ot) + GB 18.0 17.4 2.4 Cs8Nb6O19/H2O/SO2(Ot) + GB 17.4 16.4 3.7 . Downloaded on 12/04/2018 14:46:10. e Commons Attribution 3.0 Unported Licence. Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. In contrast, the C–Ot interaction is stronger, with a bond distance of 1.39 ˚A. As a result of this strong interaction, the Nb– Ot distance is elongated from 1.80 ˚A to 2.02 ˚A (see Fig. 1). Thus, while the interactions with the Cs cations provide additional stability to the complex Cs8Nb6O19/CO2(Ot), the primary Fig. 1 Adsorption of the ambient gas molecules CO2 and SO2 on Cs8Nb6O19 at the Ot sites (left) and bridging sites (right) with their important geometry parameters (in ˚A). Fig. 2 Infrared spectra recorded during and after the adsorption of CO2 onto Cs8Nb6O19 at 300 K. (i) Adsorption of 100 mTorr of CO2. (ii) After CO2 evacuation. (iii) After CO2 evacuation and thermal treatment at 423 K. Fig. 2 Infrared spectra recorded during and after the adsorption of CO2 onto Cs8Nb6O19 at 300 K. (i) Adsorption of 100 mTorr of CO2. (ii) After CO2 evacuation. (iii) After CO2 evacuation and thermal treatment at 423 K. Fig. 1 Adsorption of the ambient gas molecules CO2 and SO2 on Cs8Nb6O19 at the Ot sites (left) and bridging sites (right) with their important geometry parameters (in ˚A). This journal is © The Royal Society of Chemistry 2018 Chem. Sci., 2018, 9, 2147–2158 \| 2149 Edge Article View Article Online View Article Online Chemical Science temperature required to fully remove the CO2 suggests that the molecules responsible for these infrared bands are strongly bound to Cs8Nb6O19. The experimentally observed IR features are in full agreement with the DFT calculations (harmonic, un- scaled). Indeed, the two prominent IR active features at 1290 and 1659 cm1, which are ascribed to the bending and asym- metric CO stretch motions of a bent CO2 molecule, are calcu- lated to be 1316 and 1317 cm1 (for the bend) and 1702 and 1719 cm1 (for the asymmetric stretch) in Cs8Nb6O19/CO2(Ot) and Cs8Nb6O19/CO2(Om), respectively. Thus, the above pre- sented experimental and computational analysis shows that a large fraction of the adsorbed CO2 strongly binds to Cs8Nb6O19. Once at the surface, the molecule adopts a bent geometry, which activates the n1 vibrational motion toward absorption of infrared radiation. Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. gas-phase and at low temperatures, NO2 radicals are in equi- librium with dinitrogen tetraoxide, N2O4, while higher temperatures shithe equilibrium towards nitrogen dioxide.26 gas phase and at low temperatures, NO2 radicals are in equi librium with dinitrogen tetraoxide, N2O4, while higher temperatures shithe equilibrium towards nitrogen dioxide.26 The calculations show that N2O4 is planar, with an N–N bond distance of 1.85 ˚A, which is signicantly longer than the average N–N single bond length of 1.45 ˚A; this species has a dimeriza- tion energy of DE/DH/DG ¼ 19.8/17.3/5.6 kcal mol1. Unlike NO2, N2O4 is diamagnetic and coordinates to the Cs8Nb6O19 catalyst (see Fig. 3) with a Cs8Nb6O19–N2O4 interaction energy of DE/DH/DG ¼ 25.3/25.3/18.5 kcal mol1. Secondly, as shown in Table 1 and Fig. 3, the “free” NO2 radical (at ambient temperature) can interact with polyoxoniobate to form the [Cs8Nb6O19]/NO2 adduct. This adduct exists in two energetically stable isomeric forms, Cs8Nb6O19/NO2(Ot) and Cs8Nb6O19/ NO2(CsN); the most favorable form is Cs8Nb6O19/NO2(CsN), where the N atom of NO2 interacts with two Cs centers (with Cs1–N and Cs2–N distances of 3.42 and 3.44 ˚A). To our surprise, these Cs–N interactions lead to electron transfer from Cs8Nb6O19 to NO2 in Cs8Nb6O19/NO2(CsN), as evidenced by population analysis: the calculated Mulliken spin/charge is NO2(CsN) ¼ 0.34/0.65 \|e\|. Thus, as a result of these interac- tions, almost 0.65 \|e\| charge is transferred from Cs8Nb6O19 to NO2; and, consequently, the Cs8Nb6O19 unit develops partial radical character. Further analysis shows that most of the unpaired spin of Cs8Nb6O19 is located on the internal OIN-center (0.24 \|e\|) and the bridging Om (0.14 \|e\|) located close to the Cs atoms coordinated to NO2, while the remaining spin is delocalized on all other atoms of the polyoxoniobate. Similarly, but to a lesser extent, NO2(Ot) has a 0.67/0.41 \|e\| spin/charge distribution in the Cs8Nb6O19/NO2(Ot) isomer. The calculated Cs8Nb6O19–NO2 binding energies are DH/DG ¼ 22.2/22.1 and 18.7/13.9 kcal mol1 in Cs8Nb6O19/NO2(CsN) and Cs8Nb6O19/NO2(Ot), respectively. This journal is © The Royal Society of Chemistry 2018 Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. (b) The geometric features of Cs8Nb6O19/SO2(Ot) and Cs8Nb6O19/SO2(Om) adducts are similar to those of their CO2 analogs: among the interactions, the strongest are the S–Ot and S–Om interactions, with 1.70 and 1.79 ˚A bond distances, respectively. These geometry parameters are consistent with the greater stability of the Cs8Nb6O19/SO2(Ot) isomer. Furthermore, comparison of the calculated Cs8Nb6O19–X binding energies shows that SO2 interacts with Cs8Nb6O19 much more strongly, by nearly a factor of two, than CO2. However, both molecules clearly persist on the POM at ambient and well above ambient temperatures. Further calculations showed that single Cs8Nb6O19 species can bind several CO2 and SO2 molecules. As seen in Table 2, where we have summarized the adsorption energies as a func- tion of the number of molecules adsorbed at the six Ot sites, there is a pronounced monotonic convergence of the electronic and enthalpy binding energies up to n ¼ 6. The free energies, on the other hand, reveal thermodynamic instability for CO2 adsorption at larger values of n, suggesting that only the Cs8Nb6O19/(CO2)n species with n # 4 are viable. The Cs8Nb6O19/ (SO2)n species may still be stable for n > 6. Fig. 3 The calculated most energetically stable structures of NO2 and N2O4 adsorbed on Cs8Nb6O19, with their important geometry parameters (in ˚A). A2. Structure of Cs8Nb6O19 in the presence of NO2 radicals. As one might expect, the interaction of Cs8Nb6O19 with NO2 radicals is conceptually diﬀerent than those discussed above for the diamagnetic CO2 and SO2 molecules. Indeed, at rst, in the Table 2 Adsorption of n(X) (X ¼ CO2 and SO2 and n ¼ 1 to 6) on Cs8Nb6O19 at its Ot-sites, reported as incremental DZn,n1(Cs8Nb6O19/ X) ¼ Zn(Cs8Nb6O19/X) Zn1(Cs8Nb6O19/X) Z(X) energies (where Z ¼ E, H or G) (in kcal mol1) n X ¼ CO2 X ¼ SO2 Z ¼ E Z ¼ H Z ¼ G Z ¼ E Z ¼ H Z ¼ G 1 29.1 29.0 23.2 48.2 47.6 40.3 2 22.9 21.8 11.2 43.3 42.4 27.6 3 19.6 18.9 9.6 38.8 37.1 27.3 4 13.3 12.5 0.7 34.2 32.9 19.8 5 7.3 6.6 4.4 26.9 25.9 14.9 6 2.2 2.0 6.9 29.0 27.9 16.9 Fig. 3 The calculated most energetically stable structures of NO2 and N2O4 adsorbed on Cs8Nb6O19, with their important geometry parameters (in ˚A). Fig. Edge Article Consequently, the acquired partial radical character of Cs8Nb6O19 in the Cs8Nb6O19/NO2(CsN) complex signicantly increases its NO2-aﬃnity, enabling strong coordination of another (second) NO2 radical to the Ot-site of polyoxoniobate and formation of the most thermodynamically favorable Cs8Nb6O19/ [NO2(CsN)NO2(Ot)] singlet species. The calculated energy of the reaction Cs8Nb6O19/NO2(CsN) + NO2 / Cs8Nb6O19/[NO2(CsN) NO2(Ot)] is DE/DH/DG ¼ 55.8/52.4/35.9 kcal mol1, while the energy of the Cs8Nb6O19/NO2(Ot) + NO2 / Cs8Nb6O19/ [NO2(CsN)NO2(Ot)] reaction is DE/DH/DG ¼ 63.8/61.5/ 52.8 kcal mol1 (see ESI‡ for more details). Thus, removal of NO2(Ot) and/or NO2(CsN) from Cs8Nb6O19/[NO2(CsN)NO2(Ot)] with two NO2-adsorbates requires greater energy than removing them from Cs8Nb6O19/NO2(Ot) and/or Cs8Nb6O19/NO2(CsN). The overall energy of the reaction Cs8Nb6O19 + 2NO2 / Cs8Nb6O19/ [NO2(CsN)NO2(Ot)] is DE/DH/DG ¼ 77.4/74.6/58.0 kcal mol1 (see Table 1). To summarize, a relatively weak coordination of the rst NO2 radical to Cs8Nb6O19 at the Cs sites promotes stronger coordination of the second NO2 molecule at the Ot site of Cs8Nb6O19. As a result, the coordination of the NO2 radicals to Cs8Nb6O19 is substantially stronger than that of CO2, yet slightly weaker than that of one SO2 molecule. Cs8Nb6O19/[NO2(CsN)NO2(Ot)], the coordination of the NO2 radical positioned at Ot to Cs8Nb6O19 is substantially stronger than that of CO2, yet is slightly weaker than that of one SO2 molecule. These conclusions from our computations are fully sup- ported by experiments. As with CO2, the adsorption of NO2 onto Cs8Nb6O19 was probed with the use of infrared spectroscopy. Fig. 4 shows spectra for the interaction between NO2 and Cs8Nb6O19. The adsorption of NO2 on the surface resulted in two infrared vibrational features – one at 1668 cm1 and another at 1240 cm1 (Fig. 4, i). These features are attributed to adsorbate N–O stretches as opposed to POM motions, which appear below 1000 cm1. Analysis of the calculated [Cs8Nb6O19]–NO2 complexes suggests that the 1668 cm1 feature is likely due to an asymmetric NO stretch originating from the Cs8Nb6O19/N2O4 complex, which the calculations (unscaled) show to be at 1813 cm1. For the other NO2 complexes, the highest frequency NO stretch is found below 1600 cm1, i.e. at a lower frequency relative to the experimental peak. Taking into account the usual anharmonic correction, these modes will be found to be even further redshied to lower frequencies. Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. 3 The calculated most energetically stable structures of NO2 and N2O4 adsorbed on Cs8Nb6O19, with their important geometry parameters (in ˚A). This journal is © The Royal Society of Chemistry 2018 2150 \| Chem. Sci., 2018, 9, 2147–2158 Chemical Science View Article Online Chemical Science View Article Online This journal is © The Royal Society of Chemistry 2018 Edge Article However, if we consider that the calculated asym- metric (strongly IR active) NO stretch of a free NO2 radical is 1752 cm1, and the known experimental value25 is 1618 cm1, the resulting frequency scale factor of 0.92 brings the asym- metric NO stretch of the Cs8Nb6O19/N2O4 complex to exactly 1668 cm1. Moreover, if the measured 1668 and 1240 cm1 peaks originate from the same thermal mixture of diamagnetic complexes, we expect that the 1240 cm1 peak, which persists at higher temperatures (Fig. 4, iii), is due to the Cs8Nb6O19/ [NO2(CsN)NO2(Ot)] complex, which is much more stable to thermal perturbation than Cs8Nb6O19/N2O4. The latter imme- diately transforms to Cs8Nb6O19/[NO2(CsN)NO2(Ot)] upon N–N activation by heating. Indeed, there is a group of IR-active NO stretching motions in the Cs8Nb6O19/[NO2(CsN)NO2(Ot)] complex in the 1346 to 1361 cm1 (1238 to 1252 cm1 scaled) region, which captures the measured 1240 cm1 peak. B. Hydrolysis of Sarin by Cs8Nb6O19/X species (where X ¼ CO2 and SO2) As we have shown previously24 and have briey discussed above (see Scheme 1), Sarin hydrolysis by Cs8Nb6O19 is a multistep process, with coordination of a water molecule to the catalyst as the rst step. The calculations (see Table 1) show that coordination of H2O to Cs8Nb6O19 to form Cs8Nb6O19/H2O is exothermic/exergonic by (presented as DH/DG) 23.6/ 17.3 kcal mol1. In Cs8Nb6O19/H2O, the two hydrogens of the water molecule interact with one bridging (Om) and one terminal (Ot) oxygen atom of the polyoxometalate. Thus, the data presented above show that in the presence of ambient gas molecules of CO2, NO2 and SO2, Cs8Nb6O19 will absorb these molecules more strongly than the water and GB molecules required for hydrolysis of Sarin (see Table 1). This is expected to impact the hydrolysis of Sarin by Cs8Nb6O19 in the following two ways: rst, because the ambient gas molecules coordinate to catalytically active Ot-centers, they block these catalytically active centers, hindering water and Sarin coordi- nation, and may alter the previously reported mechanism of Sarin hydrolysis by Cs8Nb6O19. Second, the interaction of an ambient gas molecule with Cs8Nb6O19 may change the elec- tronic properties of the polyoxoniobate: this is expected to only impact the calculated energetics of the Sarin hydrolysis and to not signicantly change the nature of the previously reported intermediates and transition state structures. Water molecule coordination to the adduct formed when CO2 (or SO2) binds to the Ot-center of Cs8Nb6O19 is a few kcal mol1 less than that for “free” Cs8Nb6O19; the energies are 22.4/11.9 and 21.0/9.7 kcal mol1 for X ¼ CO2 and SO2, respectively. This eﬀect is more pronounced for the Cs8Nb6O19/SO2 adduct than for Cs8Nb6O19/CO2. Furthermore, as seen in Fig. 6, the water coordination motif in Cs8Nb6O19/CO2/H2O is diﬀerent from that in Cs8Nb6O19/H2O: because CO2 occupies the Ot-position in Cs8Nb6O19/CO2, the H2O molecule is H-bonded to one bridging (Om) and one CO (O1) oxygen atom (instead of Ot). The calculated Om–H1 and O1–H2 bond distances are 1.76 and 1.85 ˚A, respectively. Following the formation of the Cs8Nb6O19/X/H2O complex, the addition of Sarin to this intermediate occurs. Previously, we examined several approaches of Sarin to Cs8Nb6O19 and found that nerve agent decomposition occurs when Sarin approaches the hydrated Cs8Nb6O19 segment with its O(sp2) and O(sp3) atoms (labeled in Fig. Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Furthermore, experiments clearly show that NO2 is strongly bound to the POM, i.e. the 1240 cm1 peak is present even aer gas phase evacuation (Fig. 4, ii) and upon heating the POM to 423 K (Fig. 4, iii). In fact, thermal treatment up to 600 K was required to fully desorb the species, i.e. to remove NO2 radicals. This suggests that NO2 binds more strongly to the POM than CO2, which is consistent with the computational data for the Cs8Nb6O19/[NO2(CsN)NO2(Ot)] complex presented above (see Table 1 and the ESI‡). Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Sci., 2018, 9, 2147–2158 \| 2151 This journal is © The Royal Society of Chemistry 2018 Edge Article View Article Online Edge Article View Article Online View Article Online Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. The reported highly stable Cs8Nb6O19/[NO2(CsN)NO2(Ot)] complex with two NO2 fragments can also be formed via N–N bond activation of the coordinated N2O4 molecule by the poly- oxometalate catalyst. The former pathway (i.e. stepwise addition of two NO2 radicals to polyoxoniobate) may be valid at ambient temperature, while the latter process may occur at low temperature. In this paper, we did not study the N–N activation barrier; however, we found that the complex Cs8Nb6O19/ [NO2(CsN)NO2(Ot)] (see Fig. 3) lies signicantly lower in energy than the Cs8Nb6O19 + N2O4 dissociation limit, by DE/DH/DG ¼ 57.6/57.3/52.4 kcal mol1, and DE/DH/DG ¼ 37.8/40.0/ 46.8 kcal mol1 lower than the Cs8Nb6O19/N2O4 intermediate. In order to better understand the factors impacting the strength of the Cs8Nb6O19/NO2 interaction, we also analyzed the geometry of the complex Cs8Nb6O19/[NO2(CsN)NO2(Ot)]. As seen in Fig. 3, where we present the most important geometries of the Cs8Nb6O19/NO2(Ot), Cs8Nb6O19/NO2(CsN) and Cs8Nb6O19/ [NO2(CsN)NO2(Ot)] species (for full geometries of these species, see the ESI‡), in Cs8Nb6O19/[NO2(CsN)NO2(Ot)], the Nb–Ot bond is elongated to 2.47 ˚A; concurrently, the Nb–OIN bonds (which are 2.34, 2.44 and 2.02 ˚A in these complexes, respectively) and N–Ot bonds (which are 1.28 and 2.25 ˚A in Cs8Nb6O19/[NO2(CsN) NO2(Ot)] and Cs8Nb6O19/NO2(Ot), respectively) are formed. As a result, the complex Cs8Nb6O19/[NO2(CsN)NO2(Ot)] has one NO3 d fragment (d ¼ 0.4) and one NO2 d fragment (d ¼ 0.94) coordinated to the Cs-cations. Fig. 4 Infrared spectra recorded during and after the adsorption of NO2 onto Cs8Nb6O19 at 300 K. (i) Adsorption of 35 mTorr of NO2. (ii) After NO2 evacuation. (iii) After NO2 evacuation and thermal treatment at 423 K. Above, we have shown that: (1) at ambient temperatures, a relatively weak coordination of the rst NO2 radical to Cs8Nb6O19 confers partial radical character on the poly- oxoniobate and promotes a stronger coordination of the second NO2 radical to form a stable diamagnetic Cs8Nb6O19/[NO2(CsN) NO2(Ot)] complex, and (2) at low temperatures, coordination of a weakly stable N2O4 molecule to Cs8Nb6O19 followed by facile N–N bond activation leads to the same Cs8Nb6O19/[NO2(CsN) NO2(Ot)] complex. Regardless of the formation mechanisms, in Fig. 4 Infrared spectra recorded during and after the adsorption of NO2 onto Cs8Nb6O19 at 300 K. (i) Adsorption of 35 mTorr of NO2. (ii) After NO2 evacuation. (iii) After NO2 evacuation and thermal treatment at 423 K. Chem. This journal is © The Royal Society of Chemistry 2018 Chemical Science Below, we test the rst hypothesis by (a) studying the full potential energy surfaces of GB hydrolysis by Cs8Nb6O19/X, where X ¼ CO2 and SO2, and (b) comparing these new ndings with our previous results on the same reaction in the absence of ambient gas molecules. For the sake of simplicity, we discuss in detail only the reaction mechanism (as well as the structures of the pre-reaction complexes, intermediates, transition states and products) for X ¼ CO2 and compare these ndings with those (previously reported) in the absence of ambient gas molecules. In addition, we briey discuss, where appropriate, our ndings for X ¼ SO2 (full potential energy surfaces are available in the ESI‡). The reactivities of the NO2 and other radical species coordinated to Cs8Nb6O19 will be reported elsewhere. Thus, we present two spectra (shown in Fig. 5); one corre- sponds to low temperature, Fig. 5(A), which is the sum of Cs8Nb6O19/N2O4 and Cs8Nb6O19/[NO2(CsN)NO2(Ot)], and one corresponds to high temperature, Fig. 5(B), which is pure Cs8Nb6O19/[NO2(CsN)NO2(Ot)]. The frequency axis is scaled by the same factor of 0.92. The calculation is consistent with experiments with regard to the disappearance of the 1668 cm1 peak and the persistence of the 1240 cm1 peak. However, the peak at 1470 cm1 (aer scaling) in the calculated spectra, which is attributed to a local NO stretch of the NO3 unit in Cs8Nb6O19/[NO2(CsN)NO2(Ot)], is absent in the room- temperature experiments. This suggests that the formation of NO3 requires a signicant amount of thermal energy. B. Hydrolysis of Sarin by Cs8Nb6O19/X species (where X ¼ CO2 and SO2) Similar structures for X ¼ SO2 are presented in the ESI.‡ the coordination of GB to Cs8Nb6O19/H2O, in Cs8Nb6O19/CO2/ H2O/GB, the nerve agent is coordinated to one of the Cs centers of the Cs8Nb6O19-core with its P]O4 double bond. Several H- bonds also exist between the GB ligand and the [Cs8Nb6O19/ CO2/H2O]-fragment. In this complex, the calculated Cs3–O4 bond distance is 3.13 ˚A and the nascent P–O3(OH2) bond distance is 3.12 ˚A. the coordination of GB to Cs8Nb6O19/H2O, in Cs8Nb6O19/CO2/ H2O/GB, the nerve agent is coordinated to one of the Cs centers of the Cs8Nb6O19-core with its P]O4 double bond. Several H- bonds also exist between the GB ligand and the [Cs8Nb6O19/ CO2/H2O]-fragment. In this complex, the calculated Cs3–O4 bond distance is 3.13 ˚A and the nascent P–O3(OH2) bond distance is 3.12 ˚A. R-F_X are 7.8/7.5 and 8.4/8.8 kcal mol1 for X ¼ CO2 and SO2, respectively. These values are slightly larger than the values of 6.8/6.1 kcal mol1 calculated for the reaction in the absence of these ambient gas molecules; this suggests that common battleeld contaminants may impair the hydrolytic decompo- sition of nerve agents under operational conditions. The hydrolysis product is a pentacoordinated-phosphorus complex P5-F_X with a trigonal bipyramidal structure around the central phosphorus atom. In our previous paper, we showed that this intermediate exhibits multiple isomeric forms.24 Here, we discuss only the energetically most favorable form, which is directly connected to the transition state TS-F_X. For example, In the next step, hydrolysis of the coordinated water mole- cule occurs between the coordinated gas molecule X, the bridging oxygen (Om) of the Cs8Nb6O19-core and the phosphorus center of Sarin. The transition state associated with this process, TS-F_CO2, is shown in Fig. 6 (for TS-F_SO2, see the ESI‡). As seen in this gure, at TS-F_CO2, the breaking O3–H1 bond of the water molecule is elongated to 1.16 ˚A, and the forming Om–H1 bond distance becomes 1.25 ˚A. In addition, Nb1–Om and Nb2–Om bonds are slightly elongated and the Cs3– O4 bond is slightly shortened. Importantly, the Cs3-center of the Cs8Nb6O19 core also interacts with the oxygen (O3) of water and provides additional support for hydrolysis. Here, the other coordinates of interest are the P–O3(H2O) bond and the P–F bond. Fig. 7 Potential energy proﬁle of the hydrolysis of Sarin (GB) by Cs8Nb6O19 and Cs8Nb6O19/X(Ot), where X ¼ CO2 or SO2. This journal is © The Royal Society of Chemistry 2018 B. Hydrolysis of Sarin by Cs8Nb6O19/X species (where X ¼ CO2 and SO2) 6 and below as O5 and O4, respectively).24 The additional stabilization of the complex arises from the long- range O4/Cs3 ionic interaction. Thus, for the purposes of modeling the decomposition of Sarin in the presence of carbon and sulfur dioxide, it is suﬃcient to examine the most ener- getically favorable pathway, similar to that previously reported for the case with no ambient gas molecules. In keeping with the previously established shorthand notation, the pre-reaction complex of this reaction pathway is labeled as R–F_X. Fig. 5 The calculated harmonic IR spectra (all in the NO stretch region), scaled by 0.92, of (A) the sum of the Cs8Nb6O19/N2O4 and Cs8Nb6O19/[NO2(CsN)NO2(Ot)] species; (B) the pure Cs8Nb6O19/ [NO2(CsN)NO2(Ot)] species. The present calculations show that the coordination of Sarin (GB) to Cs8Nb6O19/X/H2O is exothermic by 18.0/2.4 and 17.4/3.8 kcal mol1 for X ¼ CO2 and SO2, respectively. Inspection of the structure of the Cs8Nb6O19/X/H2O/GB, R–F_X, intermediate reveals a non-covalently bonded [Cs8Nb6O19/CO2/ H2O]–GB complex. For example, as seen in Fig. 6, in contrast to Fig. 5 The calculated harmonic IR spectra (all in the NO stretch region), scaled by 0.92, of (A) the sum of the Cs8Nb6O19/N2O4 and Cs8Nb6O19/[NO2(CsN)NO2(Ot)] species; (B) the pure Cs8Nb6O19/ [NO2(CsN)NO2(Ot)] species. This journal is © The Royal Society of Chemistry 2018 2152 \| Chem. Sci., 2018, 9, 2147–2158 Chemical Science View Article Online Edge Article Fig. 6 Calculated pre-reaction complexes, transition states and products of GB hydrolysis by Cs8Nb6O19/CO2 (i.e. the reaction Cs8Nb6O19/CO2 + H2O + GB / R-F_CO2 / TS-F_CO2 / P5-F_CO2) and their important geometry parameters (in ˚A). Similar structures for X ¼ SO2 are presented in the ESI.‡ Edge Article Chemical Science ed on 08 January 2018. Downloaded on 12/04/2018 14:46:10. censed under a Creative Commons Attribution 3.0 Unported Licence. Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Fig. 6 Calculated pre-reaction complexes, transition states and products of GB hydrolysis by Cs8Nb6O19/CO2 (i.e. the reaction Cs8Nb6O19/CO2 + H2O + GB / R-F_CO2 / TS-F_CO2 / P5-F_CO2) and their important geometry parameters (in ˚A). Similar structures for X ¼ SO2 are presented in the ESI ‡ Fig. 6 Calculated pre-reaction complexes, transition states and products of GB hydrolysis by Cs8Nb6O19/CO2 (i.e. the reaction Cs8Nb6O19/CO2 + H2O + GB / R-F_CO2 / TS-F_CO2 / P5-F_CO2) and their important geometry parameters (in ˚A). B. Hydrolysis of Sarin by Cs8Nb6O19/X species (where X ¼ CO2 and SO2) DE and DG are the changes in the electronic and Gibbs free energies and are calculated relative to the reactants Cs8Nb6O19/X(Ot) + H2O + GB. As seen in Fig. 6, P–O3(H2O) undergoes a major reduction from 3.12 ˚A in R-F_CO2 to 1.96 ˚A in TS-F_CO2. Its P-F counter- part, located trans to the water-activated molecule, extends from the typical single bond in R-F_CO2, with an increase from 1.62 ˚A to 1.71 ˚A in TS-F_CO2. Similar geometry changes at the hydro- lysis transition state were observed for X ¼ SO2 (see the ESI‡). As seen in Fig. 7, the calculated DE/DG barrier heights relative to Fig. 7 Potential energy proﬁle of the hydrolysis of Sarin (GB) by Cs8Nb6O19 and Cs8Nb6O19/X(Ot), where X ¼ CO2 or SO2. DE and DG are the changes in the electronic and Gibbs free energies and are calculated relative to the reactants Cs8Nb6O19/X(Ot) + H2O + GB. This journal is © The Royal Society of Chemistry 2018 Chem. Sci., 2018, 9, 2147–2158 \| 2153 View Article Online Chemical Science as seen in Fig. 6, the formed P–O3 bond in P5-F_CO2 contracts to 1.79 ˚A, while its P-F counterpart, located trans to the activated water molecule, extends to 1.76 ˚A. The broken O3–H1 bond is elongated to 1.64 ˚A and the Nb1–Om and Nb2–Om bonds are elongated to 2.20 and 2.13 ˚A, respectively. Concurrently, the Cs3–O3 bond of 3.11 ˚A is formed to provide additional stabili- zation to the pentacoordinated-phosphorus complex, similar to that previously reported in P5-F.24 Based on the calculated Mulliken charge distribution, the resulting P5-F_X complexes can be labeled as a [(GBOH)–(Cs8Nb6O19H/X)+] ion-pair system. a H-bonding network. This process depends on multiple factors (including, but not limited to, the concentration of water in the system and the reaction temperature) and was not studied in this paper. Another possible mechanism of HF and/or isopropanol formation is direct removal of the proton from the OmH1-group of the catalyst by the uoride and/or isopropoxide ligands, respectively. As shown previously for the Cs8Nb6O19 catalyst (i.e., in the absence of ambient gas molecules),24 these processes occur with very small energy barriers which have no contribu- tion to the overall outcome of the decontamination reaction but lead to the most energetically stable intermediates, Cs8Nb6O19- (i-MPA)-HF and Cs8Nb6O19-(MPFA)-(i-POH), respectively (see Fig. 8). Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. As shown in Fig. 7, the hydrolysis of Sarin, i.e. the reaction Cs8Nb6O19/X + H2O + GB / R-F_X / TS-F_X / P5-F_X for X ¼ none, CO2 or SO2, is exergonic by 50.1/27.9, 38.8/11.1 and 40.5/14.7 kcal mol1 (presented as DH/DG) and proceeds over energy barriers of 6.8/6.1, 7.5/7.5 and 8.4/8.8 kcal mol1 (calculated relative to the pre-reaction intermediate R-F_X), respectively. The reaction R-F_X / TS-F_X / P5-F_X is exothermic for X ¼ none and SO2 by 6.8/6.4 and 0.4/ 1.2 kcal mol1, respectively, but is endothermic by 3.4/ 3.2 kcal mol1 for X ¼ CO2. These energy values allow us to conclude that the presence of ambient gas molecules increases the energies of the stationary points relative to the asymptote of the reactants. This is likely the result of a diﬀerent charge distribution on Cs8Nb6O19/X relative to Cs8Nb6O19, where X acquires a net negative charge of 0.8 to 0.7 \|e\|, as we showed above. Furthermore, X coordinates to the Ot reactive center and disables its hydrolytic activity. The presence of ambient gases increases the hydrolysis barrier by 1.0 to 2.5 kcal mol1, and this change in the energy barrier is closely associated with the Cs8Nb6O19–X complexation energy: the stronger the Cs8Nb6O19– X bond, the higher the barrier for Sarin hydrolysis. As seen in Fig. 8, at the transition state TS2-F_(i-POH)_CO2 associated with the formation of i-POH and MPFA, the activated P–O5(Osp2) bond extends to 2.19 ˚A from 1.68 ˚A in P5-F_CO2, with simultaneous formation of a double H-bond network (O5–H1 ¼ 1.63 ˚A and O1–H2 ¼ 1.57 ˚A) as a precursor to the 2154 \| Chem. Sci., 2018, 9, 2147–2158 B. Hydrolysis of Sarin by Cs8Nb6O19/X species (where X ¼ CO2 and SO2) Here, we performed an extensive search to locate the transition states TS2-F_HF_X and TS2-F_(i-POH)_X that lead to either HF and i-MPA or isopropanol (i-POH) and MPFA from the most stable pentacoordinated intermediate P5-F_X (where X ¼ CO2 or SO2). Ultimately, we were able to locate only the TS2-F_(i- POH)_X transition state (see Fig. 8). The search for the HF formation transition state TS2-F_HF_X was unsuccessful and always led to either the Cs8Nb6O19/X-(i-MPA)-HF intermediate, its derivative F/H+/Om intermediate, or the pre-reaction complex P5-F_X. For example, in Fig. 8, we present the inter- mediates, transition states and products involved in penta- coordinated P5-F_CO2 intermediate dissociation alone, with their important geometry parameters (for those of X ¼ SO2, see the ESI‡). The relative energies of these species, calculated from the P5-F_X pre-reaction complex, are given in Fig. 9. This journal is © The Royal Society of Chemistry 2018 Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. As seen in Fig. 9, the formed Cs8Nb6O19/X-MPFA-(i-POH) is the energetically lowest structure on the potential energy surface of the entire GB hydrolysis and decontamination reac- tion by Cs8Nb6O19 both in the absence and presence of ambient gas molecules. The intermediate Cs8Nb6O19/X-(i-MPA)-HF is found to be only slightly higher in energy. As we mentioned previously,24 the high stability of these intermediates is due in part to the strong hydrogen bonds between the adsorbates and the Cs8Nb6O19/X-core; however, it is also due to the additional stabilizing interactions between the Cs counter-ions and the electronegative atoms of the nerve-agent fragments. Fig. 9 Potential energy proﬁles of the decomposition reaction of pentacoordinated-phosphorus intermediate P5-F_X, where X ¼ CO2 and SO2 (i.e. reactions: P5-F_X / Cs8Nb6O19/X-(MPFA)-(i-POH) / Cs8Nb6O19/X + MPFAH + (i-POH) and P5-F_X / Cs8Nb6O19/X- (MPA)-HF / Cs8Nb6O19/X + MPAH + HF). DE and DG are the changes in the electronic and Gibbs free energies and are calculated relative to the reactants Cs8Nb6O19/X(Ot) + H2O + GB. charge-preserving double proton transfer. In other words, as the proton of the OmH1-group moves to Sarin, the proton on the PO3H1 group of Sarin migrates back to the Cs8Nb6O19/X-core to form [HXOt]q, where X ¼ CO2. As a result, the formed P–O3(H2O) bond is shortened from 1.79 ˚A in P5-F_CO2 to 1.61 ˚A in the transition state. Desorption of HF and isopropanol from Cs8Nb6O19/X-(i-MPA)- HF and Cs8Nb6O19/X-MPFA-(i-POH) requires 15.2/12.7 and 20.0/ 5.3 kcal mol1 energy, respectively, for X ¼ CO2. Dissociation of HF and isopropanol only slightly modies the geometries of the products Cs8Nb6O19/X-(i-MPA) and Cs8Nb6O19/X-MPFA frag- ments (for example, see Fig. 8 for X ¼ CO2) compared with the Cs8Nb6O19/X-(i-MPA)-HF and Cs8Nb6O19/X-MPFA-(i-POH) adducts; therefore, this will not be discussed in detail. As seen in Fig. 9, the calculated energy barriers from P5-F_X are 7.0/7.6, 5.4/5.4 and 3.6/6.5 kcal mol1 for X ¼ none, CO2 and SO2. Thus, the presence of these gas molecules in the reaction mixture slightly reduces the pentacoordinated P5-F_X inter- mediate dissociation barrier (by 2 to 4 kcal mol1). Further- more, the changes in the i-POH and MPFA formation energy barriers correlate with the Cs8Nb6O19–X complexation energy: the stronger the Cs8Nb6O19–X bond, the smaller the i-POH and MPFA formation energy barriers. However, both i-MPA in Cs8Nb6O19/X-(i-MPA) and MPFA in Cs8Nb6O19/X-MPFA are strongly bound to the protonated [Cs8Nb6O19/X]H+-core. C. Pentacoordinated P5-F_X intermediate dissociation Once the pentacoordinated P5-F_X species is formed, the reaction can proceed along several paths, as discussed in our previous paper.24 In order to evaluate the role of the catalyst in the course of the reaction, as above, here we discuss in detail only those pathways that directly involve the catalyst. These are the HF and isopropanol elimination and, ultimately, desorption pathways. The accompanying products of these paths are iso- propyl methyl phosphonic acid (i-MPA) and methyl phospho- nouoridic acid (MPFA), respectively. It is evident that in order to form HF or isopropanol, protonation of the uoride or oxygen centers of the isopropoxy ligand is required. Further- more, in order to facilitate regeneration of the catalyst, the ultimate proton source should be the OmH1-group of the [Cs8Nb6O19H/X]+ cation. However, these processes are expected to be very complex and may proceed via multiple mechanisms. One of these processes could involve any surrounding water molecules, which are expected to be present in real experi- mental conditions. In this mechanism, a water molecule located close to the uoride or oxygen atoms of the iopropoxy ligand is expected to donate its proton to these groups (to form HF and/or isopropanol, respectively) and compensate by removing the proton from the OmH1-group of the catalyst via Fig. 8 The calculated transition states, intermediates and products of the dissociation of pentacoordinated P5-F_CO2 (i.e. reactions: P5- F_CO2 / Cs8Nb6O19/CO2-(MPFA)-(i-POH) / Cs8Nb6O19/CO2 + MPFAH + (i-POH) and P5-F_X / Cs8Nb6O19/CO2-(MPA)-HF / Cs8Nb6O19/CO2 + MPAH + HF) with their important geometry parameters (in ˚A). Similar structures for X ¼ SO2 are presented in the ESI.‡ Fig. 8 The calculated transition states, intermediates and products of the dissociation of pentacoordinated P5-F_CO2 (i.e. reactions: P5- F_CO2 / Cs8Nb6O19/CO2-(MPFA)-(i-POH) / Cs8Nb6O19/CO2 + MPFAH + (i-POH) and P5-F_X / Cs8Nb6O19/CO2-(MPA)-HF / Cs8Nb6O19/CO2 + MPAH + HF) with their important geometry parameters (in ˚A). Similar structures for X ¼ SO2 are presented in the ESI.‡ This journal is © The Royal Society of Chemistry 2018 2154 \| Chem. Sci., 2018, 9, 2147–2158 Chemical Science View Article Online View Article Online Edge Article Chemical Science Fig. 9 Potential energy proﬁles of the decomposition reaction of pentacoordinated-phosphorus intermediate P5-F_X, where X ¼ CO2 and SO2 (i.e. reactions: P5-F_X / Cs8Nb6O19/X-(MPFA)-(i-POH) / Cs8Nb6O19/X + MPFAH + (i-POH) and P5-F_X / Cs8Nb6O19/X- (MPA)-HF / Cs8Nb6O19/X + MPAH + HF). Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Thus, regeneration of the catalyst requires deprotonation of the [Cs8Nb6O19/X]H+-core and protonation of the phosphonic acids i-MPA and MPFA. This step of the reaction, which forms a nal decontaminated form of the GB and re-generated catalyst, is found to be highly endothermic/endergonic, i.e. 6.4/6.5 and 2.8/1.9 kcal mol1 for i-MPAH and MPFAH formation, respectively. Overall, the last steps of the reaction, i.e. the reactions Cs8Nb6O19/X-(i-MPA)-HF / Cs8Nb6O19/X + HF + (i-MPAH) and Cs8Nb6O19/X-MPFA-(i- POH) / Cs8Nb6O19/X + (i-POH) + MPFAH, are highly prohibi- tive and require energies of 80.3/60.2 (X ¼ none), 64.4/36.3 (X ¼ CO2), and 66.0/39.4 (X ¼ SO2) kcal mol1 and 81.0/60.4 (X ¼ none), 71.2/43.6 (X ¼ CO2), and 63.3/36.4 (X ¼ SO2) kcal mol1, respectively. Furthermore, deprotonation of the [Cs8Nb6O19/X] H+-core and protonation of the phosphonic acids i-MPA and MPFA is expected to be very complex and may proceed via several pathways depending on the reaction conditions. One of these may involve any surrounding water molecules, in real experimental conditions, via a concerted protonation- deprotonation mechanism involving the hydrogen-bonded water-based network. However, this process depends on multiple factors (including, but not limited to, the concentra- tion of water in the system and the reaction temperature) and was not studied in this paper. We also failed to locate a transi- tion state associated with the direct deprotonation of Comparison of the calculated energetics for the dissociation of the pentacoordinated intermediate with those for hydrolysis (i.e. formation of the pentacoordinated intermediate) show that, in general, the hydrolysis step is a rate-determining step for all reported species, and the presence of ambient gas molecules increases this energy barrier only slightly, by 2 to 4 kcal mol1. Furthermore, the height of this rate-determining energy barrier correlates with the Cs8Nb6O19–X complexation energy: the stronger the Cs8Nb6O19–X bond, the more diﬃcult the hydrolysis of GB by the Cs8Nb6O19 catalyst. This journal is © The Royal Society of Chemistry 2018 C. Pentacoordinated P5-F_X intermediate dissociation DE and DG are the changes in the electronic and Gibbs free energies and are calculated relative to the reactants Cs8Nb6O19/X(Ot) + H2O + GB. prefers to form an [F/H+] ion-pair prior to dissociation. Similarly, in the Cs8Nb6O19/CO2-(MPFA)-(i-POH) complex, the (i- POH) molecule and MPFA-fragment are hydrogen bonded to another Ot atom and the OH-unit of the HOCOOt fragment, respectively: the calculated H1–O6 and O3–H2 bond distances are 1.73 and 1.67 ˚A, respectively. In addition, there are elec- trostatic interactions between the O center of the isopropanol molecule and two Cs cations of the Cs8Nb6O19/CO2 core, with Cs1–O5 and Cs3–O5 bond distances of 3.19 ˚A and 3.31 ˚A, respectively. The formed P–O3 fragment also interacts with the Cs1-center, with a Cs1–O3 distance of 3.16 ˚A. Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. A. Computational and experimental procedures A1. Computational methodology. A major computational challenge in the present work is to properly describe the non- covalent interactions involving the various ions in the systems studied: the Cs+ counter-cations, [Nb6O19]8 with its large negative charge, and the H+ and OH ions resulting from heterolytic water dissociation. These interactions are expected to be well described by the M06-L density functional,27 a pure density functional designed for transition metal bonding and non-covalent interactions. Therefore, all presented calculations have been carried out with the M06-L density functional, as implemented in the Gaussian09 code.28 In these calculations, we used the 6-31++G(d,p) basis set for the elements S, P, F, O, C, H and the Lanl2dz basis set with corresponding Hay–Wadt eﬀective core potentials for Nb and Cs, as implemented in Gaussian09. The sets of diﬀuse functions (++) were added specically to obtain proper descriptions of the diﬀuse charge densities and long-range interactions. All reported stationary points were conrmed to have either all real frequencies (minima) or one imaginary frequency (transition states). The latter were further veried to connect the corresponding minima by IRC calculations. All reported enthalpy and Gibbs free energies were computed at a temperature of 298.15 K and 1 atm pressure. (1) Cs8Nb6O19 absorbs ambient gas molecules of X ¼ CO2, NO2 and SO2 more strongly than it absorbs water or Sarin (GB) molecules. The calculated Cs8Nb6O19–X binding energy follows the trend for DG (X ¼ CO2) < DG (NO2) < DG (SO2). (2) The impacts of the diamagnetic CO2 and SO2 molecules on polyoxoniobate Cs8Nb6O19 are fundamentally diﬀerent than that of the NO2 radical. At ambient temperatures, weak coor- dination of the rst NO2 radical to Cs8Nb6O19 confers partial radical character on the polyoxoniobate and promotes a stronger coordination of the second NO2 radical to form a stable diamagnetic Cs8Nb6O19/(NO2)2 species; meanwhile, at low temperatures, NO2 radicals form weakly stable dinitrogen tetraoxide (N2O4), which interacts weakly with Cs8Nb6O19. Chemical Science Chemical Science [Cs8Nb6O19/X]H+ and protonation of the phosphonic acids because of the high stability of the corresponding pre-reaction complexes Cs8Nb6O19/X-(i-MPA) and Cs8Nb6O19/X-MPFA. The solution to this issue requires special comprehensive experi- mental and computational studies, which are in progress. (6) Desorption of HF or/and (i-POH) and regeneration of the catalyst requires deprotonation of the [Cs8Nb6O19/X/H]+-core with protonation of the phosphonic acids i-MPA and MPFA. Regeneration of the catalyst is a highly endergonic process and is the rate-limiting step for GB hydrolytic decontamination, both in the absence and presence of ambient gas molecules. Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. This paper, for the rst time, addresses the impact of environmentally-signicant ambient gas molecules, NO2, CO2 and SO2, on the structure, stability and decontamination activity of a basic polyoxometalate species. Specically, Cs8Nb6O19 in the presence of these gases has been studied in depth by comple- mentary computational and experimental approaches. It was found that: D. Catalyst regeneration Thus, the facile dissociation of the P5_F_X species, as reported previously for P5_F, yields HF and i-MPA and/or i-POH and MPFA products. All these species are initially bound to Cs8Nb6O19, as shown in Fig. 8 for X ¼ CO2 and the ESI‡ for X ¼ SO2. For example, in the Cs8Nb6O19/CO2-(i-MPA)-HF complex, the (i-MPA)-fragment forms a hydrogen bond with the Om and OH-unit of the HOCOOt fragment, respectively. The HF mole- cule, on the other hand, in the presence of an X molecule, This journal is © The Royal Society of Chemistry 2018 Chem. Sci., 2018, 9, 2147–2158 \| 2155 Edge Article View Article Online Edge Article View Article Online 2156 \| Chem. Sci., 2018, 9, 2147–2158 A. Computational and experimental procedures (3) Similar to the case without ambient gas molecules, re- ported previously,24 in the presence of X, GB hydrolysis by Cs8Nb6O19/X proceeds via general base hydrolysis involving: (a) adsorption of water and the nerve agent on the Cs8Nb6O19/X catalyst, (b) concerted hydrolysis of the adsorbed water mole- cule on a basic oxygen atom of the polyoxoniobate and nucle- ophilic addition of the nascent OH group to the phosphorus center of the nerve agent, (c) rapid reorganization of the resulting pentacoordinated-phosphorus intermediate followed by dissociation of either HF or isopropanol with formation of POM-bound isopropyl methyl phosphonic acid (i-MPA) or methyl phosphonouoridic acid (MPFA), respectively. A2. Experimental methodology. The synthesis of Cs8Nb6- O19$14H2O followed a known literature procedure. A solution of cesium hydroxide (14.6 g, 50% by weight) was heated to 90 C in an Erlenmeyer ask. 5 g of hydrous, amorphous niobium oxide was added in small portions, with full dissolution of each portion before addition of the subsequent portion. Evaporative crystallization yielded giant hexagonal crystals. Infrared spectroscopic experiments were performed in a stainless-steel high-vacuum chamber with a base pressure of 1 108 Torr. The Cs8Nb6O19 sample was pressed, as a 7 mm diameter disk, into a tungsten grid, which was then clamped onto a sample mount coupled to a precision manipulator. An empty region of the grid was used to monitor the gas phase species in the chamber and was also employed as a background for surface adsorption and desorption studies. The grid was resistively heated, and the temperature was monitored via a K-type thermocouple spot-welded adja- cent to the sample. A PID controller mediated the sample temperature to within 1 K. Details of the vacuum chamber and sample mount can be found in a previous publication.29 An FTIR spectrometer (Thermo, Nicolet, Nexus 470 FTIR) with an external liquid-N2-cooled MCT-A detector and a spectral resolution of 2 cm1 was used for collection of the infrared spectra. (4) Cs8Nb6O19 adsorbs ambient gas molecules X at its basic Ot (or Om) reactive centers, which shields them from involve- ment in the base hydrolysis. As a result, one of the O centers of the coordinated ambient gas molecule becomes an active hydrolysis center. This increases the energies of the stationary points relative to the asymptote of the reactants and increases the hydrolysis barrier. This journal is © The Royal Society of Chemistry 2018 Notes and references G. Izzet and A. Proust, Chem. Sci., 2013, 4, 1737; (g) A. E. Kuznetsov, Y. V. Geletii, C. L. Hill, K. Morokuma and 1 (a) F. M. Raushel, Nature, 2011, 469, 310; (b) M. Enserink, Science, 2013, 341, 1050; (c) M. B. D. Nikitin, P. K. Kerr and A. Feickert, Syria's Chemical Weapons: Issues for Congress, 2013, report # R42848. 15 15 J. Song, Z. Luo, D. K. Britt, H. Furukawa, O. M. Yaghi, K. I. Hardcastle and C. L. Hill, J. Am. Chem. Soc., 2011, 133, 16839. 16 F. J. Ma, S. X. Liu, C. Y. Sun, D. D. Liang, G. J. Ren, F. Wei, Y. G. Chen and Z. M. Su, J. Am. Chem. Soc., 2011, 133, 4178. 2 (a) N. Sharma and R. Kakkar, Adv. Mater. Lett., 2013, 4, 508; (b) N. Munro, Environ. Health Perspect., 1994, 102, 18; (c) W. W. George, in Nanoscale Materials in Chemistry: Environmental Applications, American Chemical Society, 2010, vol. 1045, pp. 125–136. 17 T. J. Bandosz, M. Laskoski, J. Mahle, G. Mogilevsky, G. W. Peterson, J. A. Rossin and G. W. Wagner, J. Phys. Chem. C, 2012, 116, 11606. 18 S. W. Yang, D. C. Doetschman, J. T. Schulte, J. B. Sarnbur, C. W. Kanyi and J. D. Fox, Microporous Mesoporous Mater., 2006, 92, 56. 3 R. H. Pastel and E. C. Ritchie, in Psychological Responses to the New Terrorism: A NATO-Russia Dialogue, IOS Press, Amsterdam, 2005, pp. 9–24. 4 (a) Y. J. Yang, K. Kim, O. G. Tsay, D. A. Atwood and D. G. Churchill, Chem. Rev., 2015, 115, PR1; (b) Y. C. Yang, 19 L. Bromberg, H. Schreuder-Gibson, W. R. Creasy, D. J. McGarvey, R. A. Fry and T. A. Hatton, Ind. Eng. Chem. Res., 2009, 48, 1650. J. A. Baker and J. R. Ward, Chem. Rev., 1992, 92, 1729. 20 (a) M. K. Kinnan, W. R. Creasy, L. B. Fullmer, H. L. Schreuder-Gibson and M. Nyman, Eur. J. Inorg. Chem., 2014, 2014, 2361; (b) F.-J. Ma, S.-X. Liu, C.-Y. Sun, D.-D. Liang, G.-J. Ren, F. Wei, Y.-G. Chen and Z.-M. Su, J. Am. Chem. Soc., 2011, 133, 4178; (c) D. M. Mizrahi, S. Saphier and I. Columbus, J. Hazard. Mater., 2010, 179, 495–499; (d) N. M. Okun, J. C. Tarr, D. A. Hilleshiem, L. Zhang, K. I. Hardcastle and C. L. Hill, J. Mol. Catal. A: Chem., 2006, 246, 11; (e) N. M. Okun, T. M. Notes and references Anderson and C. L. Hill, J. Mol. Catal. A: Chem., 2003, 197, 283; (f) F. Carniato, C. Bisio, R. Psaro, L. Marchese and M. Guidotti, Angew. Chem., Int. Ed., 2014, 53, 10095. 5 J. B. DeCoste and G. W. Peterson, Chem. Rev., 2014, 114, 5695. 6 S. Y. Moon, G. W. Wagner, J. E. Mondloch, G. W. Peterson, J. B. DeCoste, J. T. Hupp and O. K. Farha, Inorg. Chem., 2015, 54, 10829. 7 Y. Liu, S. Y. Moon, J. T. Hupp and O. K. Farha, ACS Nano, 2015, 9, 12358. 8 J. E. Mondloch, M. J. Katz, W. C. Isley, P. Ghosh, P. L. Liao, W. Bury, G. W. Wagner, M. G. Hall, J. B. DeCoste, G. W. Peterson, R. Q. Snurr, C. J. Cramer, J. T. Hupp and O. K. Farha, Nat. Mater., 2015, 14, 512. 9 M. J. Katz, J. E. Mondloch, R. K. Totten, J. K. Park, S. T. Nguyen, O. K. Farha and J. T. Hupp, Angew. Chem., Int. Ed., 2014, 53, 497. 21 21 (a) T. Yamase and M. T. Pope, in Nanostructure Science and Technology, ed. D. J. Lockwood, Kluwer Academic/Plenum Publishers, New York, vol. 2, 2002; (b) M. T. Pope, in Comprehensive Coordination Chemistry II: From Biology to Nanotechnology, ed. A. G. Wedd, Elsevier Ltd., Oxford, UK, 2004, vol. 4, pp. 635–678; (c) H. N. Miras, J. Yan, D.-L. Long and L. Cronin, Chem. Soc. Rev., 2012, 41, 7403; (d) H. Lv, Y. V. Geletii, C. Zhao, J. W. Vickers, G. Zhu, Z. Luo, J. Song, T. Lian, D. G. Musaev and C. L. Hill, Chem. Soc. Rev., 2012, 41, 7572. 10 R. Gaillac, P. Pullumbi, K. A. Beyer, K. W. Chapman, D. A. Keen, T. D. Bennett and F.-X. Coudert, Nat. Mater., 2017, 16, 1149–1154. 11 M. K. Kinnan, W. R. Creasy, L. B. Fullmer, H. L. Schreuder- Gibson and M. Nyman, Eur. J. Inorg. Chem., 2014, 2014, 2361. 12 W. Guo, H. Lv, K. P. Sullivan, W. O. Gordon, A. Balboa, G. W. Wagner, D. G. Musaev, J. Bacsa and C. L. Hill, Angew. Chem., Int. Ed., 2016, 55, 7403. 22 13 (a) D. M. Mizrahi, S. Saphier and I. Columbus, J. Hazard. Mater., 2010, 179, 495; (b) N. M. Okun, J. C. Tarr, D. A. Hilleshiem, L. Zhang, K. I. Hardcastle and C. L. Hill, J. Mol. Catal. A: Chem., 2006, 246, 11; (c) N. M. Open Access Article. Published on 08 January 2018. Downloaded on 12/04/2018 14:46:10. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. C. Bo and J. M. Poblet, Chem. Soc. Rev., 2012, 41, 7537; (f) B. Matt, X. Xiang, A. L. Kaledin, N. N. Han, J. Moussa, H. Amouri, S. Alves, C. L. Hill, T. Q. Lian, D. G. Musaev, A. Computational and experimental procedures These changes are closely correlated with the Cs8Nb6O19–X complexation energy; the stronger the Cs8Nb6O19–X bond, the higher the barrier for Sarin hydrolysis. (4) Cs8Nb6O19 adsorbs ambient gas molecules X at its basic Ot (or Om) reactive centers, which shields them from involve- ment in the base hydrolysis. As a result, one of the O centers of the coordinated ambient gas molecule becomes an active hydrolysis center. This increases the energies of the stationary points relative to the asymptote of the reactants and increases the hydrolysis barrier. These changes are closely correlated with the Cs8Nb6O19–X complexation energy; the stronger the Cs8Nb6O19–X bond, the higher the barrier for Sarin hydrolysis. (5) The most energetically stable products of the GB hydro- lysis and decontamination reaction are Cs8Nb6O19/X-MPFA-(i- POH) and Cs8Nb6O19/X-(i-MPA)-HF both in the absence and presence of ambient gas molecules. The high stability of these intermediates is due in part to the strong hydrogen bonds between the adsorbates and the protonated [Cs8Nb6O19/X/H]+- core and to interactions between the Cs counterions and the electronegative atoms of the adsorbates. This journal is © The Royal Society of Chemistry 2018 2156 \| Chem. Sci., 2018, 9, 2147–2158 Chemical Science View Article Online Chemical Science View Article Online View Article Online Edge Article Chemical Science Chemical Science Conﬂicts of interest T. M. Anderson and C. L. Hill, J. Mol. Catal. A: Chem., 2003, 197, 283; (d) F. Carniato, C. Bisio, R. Psaro, L. Marchese and M. Guidotti, Angew. Chem., Int. Ed., 2014, 53, 10095. There are no conicts to declare. Acknowledgements 14 14 As an example for computational studies of the POMs see: (a) J. M. Poblet, X. Lopez and C. Bo, Chem. Soc. Rev., 2003, 32, 297; (b) P. Miro, J. M. Poblet, J. B. Avalos and C. Bo, Can. J. Biochem., 2009, 87, 1296; (c) C. Bo and J. M. Poblet, Isr. J. Chem., 2011, 51, 228; (d) X. Lopez, P. Miro, J. J. Carbo, A. Rodriguez-Fortea, C. Bo and J. M. Poblet, Theor. Chem. Acc., 2011, 128, 393; (e) X. Lopez, J. J. Carbo, C. Bo and J. M. Poblet, Chem. Soc. Rev., 2012, 41, 7537; (f) B. Matt, X. Xiang, A. L. Kaledin, N. N. Han, J. Moussa, H. Amouri, S. Alves, C. L. Hill, T. Q. Lian, D. G. Musaev, G. Izzet and A. Proust, Chem. Sci., 2013, 4, 1737; (g) A. E. Kuznetsov, Y. V. Geletii, C. L. Hill, K. Morokuma and D. C. Musaev, J. Am. Chem. Soc., 2009, 131, 6844. This material is based upon work supported by the U. S. Army Research Laboratory and the U. S. Army Research Oﬃce under grant number W911NF-15-2-0107. The authors are grateful for the support of the Defence Threat Reduction Agency. The authors gratefully acknowledge NSF MRI-R2 grant (CHE- 0958205 for D. G. M.) and the use of the resources of the Cherry Emerson Center for Scientic Computation. Notes and references Okun, 22 (a) M. R. Antonio, M. Nyman and T. M. Anderson, Angew. Chem., Int. Ed., 2009, 121, 6252; (b) M. Nyman, T. M. Alam, F. Bonhomme, M. A. Rodriguez, C. S. Frazer and M. E. Welk, J. Cluster Sci., 2006, 17, 197; (c) M. Nyman, This journal is © The Royal Society of Chemistry 2018 Chem. Sci., 2018, 9, 2147–2158 \| 2157 Edge Article View Article Online View Article Online Edge Article View Article Online View Article Online Chemical Science Dalton Trans., 2011, 40, 8049; (d) E. M. Villa, C. A. Ohlin, E. Balogh, T. M. Anderson, M. D. Nyman and W. H. Casey, Angew. Chem., Int. Ed., 2008, 47, 4844; (e) E. Balogh, T. M. Anderson, J. R. Rustad, M. Nyman and W. H. Casey, Inorg. Chem., 2007, 46, 7032. B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery Jr, J. E. Peralta, F. ß. Ogliaro, M. J. Bearpark, J. Heyd, E. N. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari, A. P. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi, N. Rega, N. J. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, A. Dapprich, A. D. Daniels, ˜n. n. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski and D. J. Fox, Gaussian, Inc.: Wallingford, CT, USA, 2009. T. M. Anderson, J. R. Rustad, M. Nyman and W. H. Casey, Inorg. Chem., 2007, 46, 7032. 23 M. Nyman, F. Bonhomme, T. M. Alam, M. A. Rodriguez, 23 M. Nyman, F. Bonhomme, T. M. Alam, M. A. Rodriguez, B. R. Cherry, J. L. Krumhansl, T. M. Nenoﬀ and A. M. Sattler, Science, 2002, 297, 996. y g B. R. Cherry, J. L. Krumhansl, T. M. Nenoﬀ and A. M. Sattler, Science, 2002, 297, 996. 24 R. C. Chapleski Jr, D. G. Musaev, C. L. Hill and D. Troya, J. Phys. Chem. C, 2016, 120, 16822. 25 T. 2158 \| Chem. Sci., 2018, 9, 2147–2158 This journal is © The Royal Society of Chemistry 2018 Notes and references Shimanouchi, Tables of molecular vibrational frequencies, National Bureau of Standards, U.S. Govt. Print. Oﬀ., Washington, 1972. 26 (a) H. A. Bent, Inorg. Chem., 1963, 2, 747; (b) E. Wiberg, N. Wiberg and A. F. Holleman, Inorganic Chemistry, Academic Press, San Diego, 2001, ISBN 0-12-352651-5. 27 Y. Zhao and D. G. Truhlar, J. Chem. Phys., 2006, 125, 194101. 29 P. Basu, T. H. Ballinger and J. T. Yates, Rev. Sci. Instrum., 1988, 59, 1321. 28 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, 28 M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, 2158 \| Chem. Sci., 2018, 9, 2147–2158 This journal is © The Royal Society of Chemistry 2018
https://openalex.org/W2052313033	https://thescipub.com/pdf/ajmsp.2014.11.17.pdf	English	null	HAEMATOLOGICAL PROFILE OF PREGNANT WOMEN INFECTED WITH MALARIA PARASITES AT FEDERAL TEACHING HOSPITAL ABAKALIKI, EBONYI STATE	American journal of microbiology	2,014	cc-by	4,165	American Journal of Microbiology 5 (1): 11-17, 2014 ISSN: 1948-982x © 2014 A.M. Nnnaemeka et al., This open access article is distributed under a Creative Commons Attribution (CC-BY) 3.0 license doi:10.3844/ajmsp.2014.11.17 Published Online 5 (1) 2014 (http://www.thescipub.com/ajm.toc) American Journal of Microbiology 5 (1): 11-17, 2014 ISSN: 1948-982x © 2014 A.M. Nnnaemeka et al., This open access article is distributed under a Creative Commons Attribution (CC-BY) 3.0 license doi:10.3844/ajmsp.2014.11.17 Published Online 5 (1) 2014 (http://www.thescipub.com/ajm.toc) ABSTRACT Malaria in pregnancy is a major public health problem in endemic areas of sub-Saharan Africa and has important consequences on birth outcome. There are subtle and substantial changes in hematological parameters of malaria in pregnancy. This work is designed to ascertain the impact of malaria in pregnant women visiting Federal Teaching Hospital Abakaliki II. Out of the 100 pregnant women screened for malaria, 44 (44.0%) were positive for malaria parasite. Of the 44.0% pregnant women positive for malaria parasite, the age range of 26-30 years (31.8%) were more infected with malaria parasite, followed by 21-25 years (22.7), while 41-45 (4.5%) years were the least infected. Pregnant women with no formal education were most infected (36.4%), followed by primary education (27.3%), while secondary education showed the lowest rate of malaria in pregnancy (13.6%). Housewives reported the highest cases of malaria in pregnancy, while student reported the lowest (9.2%). Married women showed the highest cases of malaria in pregnancy (79.5%), while widow reported the lowest (2.3). Christian reported in the highest cases of pregnancy in malaria (46.0%) and Muslim the least (40.0%). The mean values of the haematological parameters of pregnant women with respect to parasitaemia were Heamoglobin (9.78±37.45 g/dL), Packed cell volume (31.56±2721.14%), White blood count (8.58±50.06×103/mm2), Neutrophils (57.96±1004.97%), Lymphocyte (28.24±1392.97%), Mesophils (7.28±110.49%), Eosinophils (3.62±1156.91%) and Platelets (141.88±133873.07×109/l). This study have shown that the adverse consequences of malaria in pregnancy has great impact on heamatological parameters which may affect not only the neonate and infant but also increase the risk of non communicable diseases when the child grows into an adult and the risk of low birth weight in the next generation. Keywords: Malaria, Haematological Profile, Pregnant Women, Abakaliki Science Publications HAEMATOLOGICAL PROFILE OF PREGNANT WOMEN INFECTED WITH MALARIA PARASITES AT FEDERAL TEACHING HOSPITAL ABAKALIKI, EBONYI STATE 1Alo Moses Nnnaemeka, 2Okonkwo Eucharia Chinyere, 3Anyim Chukwudi and 4I. Ugah Uchenna P. malariae cause a generally milder form of malaria that is rarely fatal (Beare et al., 2006). 1Alo Moses Nnnaemeka, 2Okonkwo Eucharia Chinyere, 3Anyim Chukwudi and 4I. Ugah Uchenna 1Department of Microbiology, Federal University, Ndufu-Alike Ikwo, Ebonyi State, Nigeria 2Department of Applied Microbiology, Ebonyi State University, Abakaliki, Ebonyi State, Nigeria 3Department of Applied Microbiology and Brewing, Nnamdi Azikiwe University, Awka, Nigeria 4Department of Medical Biochemistry, Federal University, Ndufu-Alike Ikwo, Ebonyi State, Nigeria Received 2014-05-13; Revised 2014-06-06; Accepted 2014-07-16 2.4. Blood Collection, Staining and Microscopy Two milliliter of blood was collected intravenously under a sterile condition. The blood samples were put in EDTA bottles, labeled and sent to Department of Medical Laboratory Science Laboratory, Ebonyi State University, Abakaliki, where thick was prepared and stained with 10% Giemsa as described by Ochei and Kolhatkar (2007). The blood films were examined microscopically using 40× and 100× objectives (with oil immersion) and 7× eye piece. Despite numerous studies conducted over the last decades, Malaria In Pregnancy (MIP) remains an important public health problem that has proved difficult to tackle (DeBeaudrap et al., 2013). The relationship between malaria in pregnancy and its outcome on birth in endemic areas such as Nigeria 1. INTRODUCTION Corresponding Author: Alo Moses Nnnaemeka, Department of Microbiology, Federal University, Ndufu-Alike Ikwo, Ebonyi State, Nigeria 1. INTRODUCTION Malaria is widespread in tropical and subtropical regions in a broad band around the equator, including much of Sub-Saharan Africa, Asia and the Americas. Five species of Plasmodium can infect and be transmitted by humans. The vast majority of deaths are caused by P. falciparum while P. vivax, P. ovale and g y is rarely fatal (Beare et al., 2006). It has been recognized for nearly a century that pregnant women are especially prone to severe malaria (Nayak et al., 2009). Approximately 50 million pregnant women are exposed to malaria each year (Gamble et al., 2009). The burden of Malaria In Pregnancy (MIP) remains high in endemic areas, where despite Science Publications AJM 11 Alo Moses Nnnaemeka et al. / American Journal of Microbiology 5 (1): 11-17, 2014 considerable immunity, pregnant women continue to have symptomatic and asymptomatic parasitaemia resulting in adverse pregnancy outcomes (Olukemi et al., 2011). Malaria and pregnancy usually affect the course of each other adversely. The physiological changes of pregnancy and pathological changes due to malaria have a deleterious effect on the course of each other. In endemic areas, clinical episodes of malaria are more frequent and more severe during pregnancy and mortality rate is higher among them as compared to non- pregnant (Ramsay, 2003). continues to be a subject of research. Hence this present work is carried out to ascertain the haematological profile of pregnant women infected with malaria parasites at Federal Teaching Hospital Abakaliki, Ebonyi State. 2.3. Ethical Consideration Consent was also obliged and obtained from the husbands of the pregnant women. Informed consent was also obtained from all study subjects. Ethical clearance was obtained from Federal Teaching Hospital Abakaliki II (FETHA II), Ebonyi State. Semi-structured questionnaires were administered to obtain vital information such as age, level of education, marital status and religion. 2.1. Study Area The study was carried out at Federal Teaching Hospital Abakaliki II (FETHA II), Ebonyi State. The study area is located between Latitude 06° 4΄N and longitude 08° 5΄E and rainfall pattern is bimodal (April- July), September-November with a short spell sometimes in August. The annual rainfall is between 1000-1500 mm. The vegetation of the area is predominantly derived Savannah. The mean annual temperature is about 24°C and the relative humidity is between 60-80%. Many studies from areas with different malaria transmission patterns have investigated the consequences of malaria in pregnancy on both maternal health and birth outcomes (DeBeaudrap et al., 2013). Malaria in pregnancy has been associated with significant degree of intrauterine growth restriction, 36% of preterm deliveries, 30% of preventable low birth weight deliveries, 14% of low birth weight deliveries and 15% of maternal anaemia (Steketee et al., 2001). While the consequences of MIP on maternal health are dominated by anaemia, data on malaria-related maternal mortality are sparse (Desai et al., 2007). 2.2. Study Population A total of 100 pregnant women were randomly selected from FETHA II between December, 2013 and March, 2014. The sample population was selected irrespective of age, level of education, marital status, occupation and religion. The World Health Organization recommends the use of Intermittent Presumptive Treatment with sulphadoxine pyrimethamine (IPTsp), household use of Insecticide Treated Nets (ITNs) and effective and prompt case management as malaria control strategies in pregnancy (WHO, 2012). In areas of stable malaria transmission in sub-Saharan Africa, ITNs are highly effective in reducing childhood mortality and morbidity frommalaria (Lengeler, 2004). Although ITNs are being promoted as a major tool in the fight against malaria in pregnancy, the available evidence about their effect in pregnancy appears inconsistent (Gamble et al., 2009). Other malaria control measures recommended include personal protection measures against vectors such as use of residual sprays, window screening and mosquito repellent creams. In Nigeria, traditional remedies against malaria have always been employed, though with unproven efficacy, while chemoprophylaxis with weekly pyrimethamine and chloroquine which were widely utilized in several African countries are no longer efficacious because of emergence of resistance (WHO, 2013; Nahlen et al., 2012; Sirima et al., 2003). Science Publications 3. RESULTS Out of the 100 pregnant women screened for malaria parasite, 44 (44.0%) pregnant women were positive for malaria parasite as shown in Table 1. Out of the 44 pregnant women positive for malaria parasite (as shown in Table 1), the age range of 26-30 years were more infected with malaria parasite 14 (31.8%), followed by 21-25 years age range 10 (22.7%), while pregnant women within the age range of 41-45 years were the least infected with malaria parasite 2 (4.5%). The level of education of the pregnant women with respect to infection of malaria parasite showed that pregnant women with no formal education were most infected with malaria parasite 16 (36.4%), followed by primary education 12 (27.3%), while secondary education showed the lowest rate of malaria in pregnancy 6 (13.6%) as shown in Table 2. A total of 100 women were enrolled for this study, of which 44.0% were positive for malaria infections detected by thick film as shown in Table 1. The result obtained in this work is higher than the prevalence rate of 28.0% observed in Mbarara District, Southwestern Uganda by DeBeaudrap et al. (2013). Most of the pregnant women within the age range of 26-30 years were more infected with malaria parasite (31.8%), followed by 21-25 years age range (22.7%), while pregnant women within the age range of 41-45 years were the least infected with malaria parasite (4.5%) as shown in Table 1. This is in line with the work of Matthew et al. (2013) who reported that MiP mostly occurred between the age range of 20-29 years (48.0%). With respect to occupation, housewives reported the highest cases of malaria in pregnancy 13 (29.5%), followed by petty traders 11 (25.0%), civil servants 10 (22.7%), casual labourers 6 (13.6%), students reported the lowest 4 (9.2%) as shown in Table 3. In terms of marital status, married women reported the highest cases of malaria in pregnancy 35 (79.5%), followed by unmarried women 5 (11.4%), divorced/separated 3 (6.8%) and widow reported the lowest 1 (2.3%) as shown in Table 4. With respect to religion, Christians reported the highest case of malaria in pregnancy 33 (75.00%) and Muslim the least 11 (25.0%). 2.5. Complete Blood Count Complete Blood Count (CBC) absolute count of white blood cell (total and differential), heamoglobin estimation, AJM 12 Alo Moses Nnnaemeka et al. / American Journal of Microbiology 5 (1): 11-17, 2014 platelets and lymphocyte where determined using an automated full counter (Abacus Junior Analyzer). pregnant counterparts (Boel et al., 2012). Maternal, placental or foetal malaria infection during pregnancy adversely affects development and survival of foetus through low birth weight, maternal anemia and possibly abortion and stillbirth. These malaria induced medical problems constitute major clinical, public health and research challenges (Murphy and Breman, 2001). This infection can aggravate other infections; dual infection has additional detrimental effects on maternal and infant survival (Ticconi et al., 2003). Malaria also remains the preventable cause of low birth weight deliveries worldwide (Saba et al., 2008). Despite numerous studies conducted over the last decades, Malaria in Pregnancy (MiP) remains an important public health problem that has proved difficult to tackle (DeBeaudrap et al., 2013). 3. RESULTS The rate of Malaria in Pregnancy (MiP) in association with occupation, it is notice that most of the study population with MiP were mostly the housewives (29.5%), followed by petty traders (25.0%), civil servants (22.7%), casual labourers (13.6%), while students (9.2%) the least (Table 2). This work is similar to that reported by Matthew et al. (2013) where occupations of pregnant women are mostly as full-time housewives (54.0%). The haematological parameters (heamoglobin, packed cell volume, white blood count, neutrophils, lymphocytes, mesophils, eosinophils, basophils and platelets) of the pregnant women attending FETHA II were also ascertained as shown in Table 4. Table 1. Prevalence of malaria parasite in pregnancy with respect to demographic data (age) Age range Number Number (yeasrs) examined positive (%) <21 10 4 (9.2) 21-25 21 10 (22.7) 26-30 30 14 (31.8) 31-35 16 8 (18.2) 36-40 13 6 (13.6) 41-45 10 2 (4.5) Total 100 44 (44.0) Table 1. Prevalence of malaria parasite in pregnancy with respect to demographic data (age) 2.6. Statistical Analysis Percentage, mean and Standard Deviation (SD) were used to analyze data obtained in this study. Science Publications Alo Moses Nnnaemeka et al. / American Journal of Microbiology 5 (1): 11-17, 2014 Alo Moses Nnnaemeka et al. / American Journal of Microbiology 5 (1): 11-17, 2014 With respect to religion of MiP, Christians (75.0%) were the highest of MiP and Muslim the least (25.0%). This is result is associated with religion (Christianity) of the most of the people in Abakaliki Metropolis (Table 5). A well-known risk factor for Malaria in Pregnancy (MiP) is level of education (Steketee et al., 2001; Desai et al., 2007). Malaria in Pregnancy (MiP) was recorded highest in pregnant women with no formal education (36.4%), followed by those who attended primary schools only (27.3%), higher institution (22.7%), this was followed by prevalence of malaria among pregnant women that had secondary education were recorded the lowest (13.6%) (Table 3). Hence low education levels were dependently associated with malaria during pregnancy in FETHA II. These findings further support the notion that it is essential to scale up malaria prevention efforts in more isolated and deprived communities as recently highlighted in a meta-analysis of datasets from 25 African countries (Eisele et al., 2012). Similarly, reported that most infected with malaria infection were the pregnant women that had secondary education (38.0%), followed by those with adult literacy (30.0%); which is likely to be those no formal education in this study). Pregnancy causes significant changes in metabolism, fluid balance, organ function and blood circulation which are driven by estrogen and the presence of the feto- placental unit. These dramatic changes influence a wide variety of hematological parameters. Acknowledge of these changes is essential when interpreting the result of hematological investigation to diagnose or monitor illness pregnant woman (Elgari, 2013). With respect to marital status of MiP, married pregnant women were the highest of MiP (79.5%), followed by unmarried (11.4%) while widow recorded the least (2.3%) as shown in Table 4. Table 5. Prevalence of malaria parasite in pregnancy with respect to demographic data (religion) Table 5. Prevalence of malaria parasite in pregnancy with respect to demographic data (religion) Religion Number examined Number positive (%) Christianity 72 33 (46.0) Muslim 28 11 (40.0) Total 100 44 (44.0) Table 2. Prevalence of malaria parasite in pregnancy with respect to demographic data (level of education) Level of Number Number education examined positive (%) No formal education 38 16 (36.4) Primary 25 12 (27.3) Secondary 16 6 (13.6) Higher Institution 21 10 (22.7) Total 100 44 (44.0) Table 3. 4. DISCUSSION Malaria remains a major public health problem in sub Saharan Africa and the extent of utilisation of malaria preventive measures may impact on the burden of malaria in pregnancy (Tongo et al., 2011). Pregnant women are more susceptible to malaria than their non- AJM 13 13 Alo Moses Nnnaemeka et al. / American Journal of Microbiology 5 (1): 11-17, 2014 Science Publications 6. ACKNOWLEDGEMENT We wish to aknoledge the staff of Microbiology Unit, Federal Teaching Hospital Abakaliki II for their assistance during the period of this work. The decrease in PCV ascertained in this work, may be due to increase in plasma volume during pregnancy. Hence there is need for adequate management of their blood profiles with dietary supplementation. A study showed marked decrease in PCV in the third trimester of pregnancy might be attributed to maternal diabetes (Pilsczek et al., 2008). 5. CONCLUSION The World Health Organization has suggested that anemia is present in pregnancies when Hb concentration is less than 11 g/dL (Milman et al., 2007). The study revealed significant decreases in haemoglobin (9.78±37.45 g/dL), packed cell volume (31.56±2721.14%) and platelets (141.88±133873.07 ×109/l) (Table 6) respectively of pregnant women compared to the control (Table 7). However, mean numeric values for most of the hematological profiles were below the normal range values for pregnant women reported (Abbassi-Ghanavati and Greer, 2010). Malaria in Pregnancy (MiP) adversely affects the pregnancy outcome. It is likely to increases the risk of spontaneous abortion, stillbirths, premature delivery and low birth weight. Anaemia (low haemoglobin, <11g/dL) in pregnancy is associated with adverse consequences both for the mother and the foetus. This study have shown that the adverse consequences of MiP has great impact on heamatological parameters which may affect not only the neonate and infant but also increase the risk of non communicable diseases when the child grows into an adult and the risk of low birth weight in the next generation. The findings of this work are consistent with previous study which reported that the decreases in hemoglobin and packed cell volume concentration are common findings during pregnancy and results from increased plasma volume combined poor iron intake (Bashiri et al., 2003; Ruchi et al. 2013; Elgari, 2013). Alo Moses Nnnaemeka et al. / American Journal of Microbiology 5 (1): 11-17, 2014 Prevalence of malaria parasite in pregnancy with respect to demographic data (occupation) Number Number Occupation examined positive (%) Housewife 25 13 (29.5) Civil servant 23 10 (22.7) Petty trader 21 11 (25.0) Student 14 4 (9.2) Casual labourer 17 6 (13.6) Total 100 44 (44.0) Table 4. Prevalence of malaria parasite in pregnancy with respect to demographic data (marital status) Number Number Marital status examined positive (%) Married 66 35 (79.5) Unmarried 10 5 (11.4) Divorced/separated 15 3 (6.8) Widow 9 1 (2.3) Total 100 44 (44.0) Table 2. Prevalence of malaria parasite in pregnancy with respect to demographic data (level of education) Table 6. Haematological parameters of pregnant women with respect to parasitaemia Table 6. Haematological parameters of pregnant women with respect to parasitaemia Parameters Mean values Heamoglobin (g/dL) 9.78±37.45 Packed cell volume (%) 31.56±2721.14 White blood count (×103/mm2) 8.58±50.06 Neutrophils (%) 57.96±1004.97 Lymphocyte (%) 28.24±1392.97 Mesophils (%) 7.28±110.49 Eosinophils (%) 3.62±1156.91 Basophils (%) 0 Platelets (×109/l) 141.88±133873.07 Values were mean ± Standard Deviation (SD) Table 3. Prevalence of malaria parasite in pregnancy with respect to demographic data (occupation) Table 7. Mean haematological parameters Parameters Mean values Heamoglobin (g/dL) ≥11 Packed cell volume (%) 38.75±3.70 White blood count (×103/mm2) 4.93±0.90 Neutrophils (%) 44.63±13.4 Mesophils (%) 44.86±12.50 Eosinophils (%) 6.32±3.40 Basophils (%) 1.30±0.52 Platelets (×109/l) 260.0±66.0 Values were mean ± Standard Deviation (SD) Table 7. Mean haematological parameters AJM 14 Alo Moses Nnnaemeka et al. / American Journal of Microbiology 5 (1): 11-17, 2014 Alo Moses Nnnaemeka et al. / American Journal of Microbiology 5 (1): 11-17, 2014 Alo Moses Nnnaemeka et al. / American Journal of Microbiology 5 (1): 11-17, 2014 Science Publications 7. REFERENCES Dis., 7: 93-104. PMID: 17251080 Nayak, K.C., M.P. Khatri, B.K. Gupta, P. Sirohi and V. Choudhary et al., 2009. Spectrum of vivax malaria in pregnancy and its outcome: A hospital based study. Vector Borne Dis., 46: 299-302. PMID: 19959857 Eisele, T.P., D.A. Larsen, P.A. Anglewicz, J. Keating and J. Yukich et al., 2012. Malaria prevention in pregnancy, birth weight and neonatal mortality: A meta-analysis of 32 national cross-sectional datasets in Africa. Lancet Infect. Dis., 12: 942- 949. PMID: 22995852 Ochei, J.O. and A.A. Kolhatkar, 2007. Medical Laboratory Science: Theory and Practice. 1st Edn., Tata McGraw-Hill Publishing Company Limited, New Delhi, New York, USA., pp: 963. Olukemi, O.T., E.O. Adebola and O.A. Olusegun, 2011. Utilization of malaria preventive measures during pregnancy and birth outcomes in Ibadan, Nigeria. BMC Pregnancy Childbirth. 11: 60-60. DOI: 10.1186/1471-2393-11-60 Elgari, M.M., 2013. Evaluation of hematological parameters of sudanese pregnant women attending at Omdurman Al Saudi Maternity Hospital. Egypt. Acad. J. Biol. Sci., 5: 37-42. Gamble, C.L., J.P. Ekwaru and F.O. Kuile, 2009. Insecticide-Treated Nets for Preventing Malaria in Pregnancy (Review). 1st Edn., John Wiley and Sons, Ltd Publication, pp: 38. Pilsczek, F.H., W. Renn, H. Hardin and R.M. Schmülling, 2008. Clinical laboratory values during diabetic pregnancies. J. Ayuba Med. College Abbottabad, 20: 3-6. PMID: 19024174 Lengeler, C., 2004. Insecticide-treated bed nets and curtains for preventing malaria. Cochrane Database Systematic Rev., 24: 2-2. DOI: 10.1002/14651858 Pitkin, R.M. and D.L. Witte, 1979. Platelet and leucocyte count in pregnancy. J. Am. Med. Assoc.. 242: 2696- 2698. DOI: 10.1001/jama.242.24.2696 Ramsay, S., 2003. Preventing malaria in pregnancy. Lancet, 3: 4-4. PMID: 12505015 Matthews, J.H., S. Benjamin, D.S. Gill and N.A. Smith, 1990. Pregnancy associated thrombocytopenia: Definition, incidence and natural history. Acta Haematol., 84: 24-24. DOI: 10.1159/000205022 Rouse, D.J., J. Owen and R.L. Goldenberg, 1998. Routine maternal platelet count: An assessment of a echnologically driven screening practice. Am. J. Obstetric Gynecol., 179: 573-573. DOI: 10.1016/S0002-9378(03)00912-8 Matthew, A.K., O.O.A. James, D. Bolanle and O.O. Stephen, 2013. Hamatological parameters and malaria parasite infection among pregnant women in Northwest Nigeria. Asian Pacific J. Tropical Dis., 3: 47-50. DOI: 10.1016/S2222- 1808(13)60010-9 Roy, L., P.E. Duffy, C. Antony and D.W. Taylor, 2007. Malaria in pregnancy: Pathogenesis and immunity. Lancet Infect. Dis., 7: 105-117. DOI: 10.1016/S1473-3099(07)70022-1 Ruchi, K., B. Pradeep and K. Varun, 2013. A study of erythron status in pregnant and non-pregnant age matched females of Jodhpur. IOSR J. Pharmacy, 3: 35-39. Milman, N., T. Bergholt, K. 7. REFERENCES Abbassi-Ghanavati, M. and L.G. Greer, 2010. Reference Table of Normal Laboratory Values in Uncomplicated Pregnancies. In: Williams Obstetrics, Cunningham, F.G., K.J. Leveno, S. Bloom, J.C. Hauth and D.J. Rouse et al. (Eds.), McGraw Hill Professional, New York, ISBN-10: 0071702857, pp: 1404-1404. Significant decreases in platelet count of out pregnant women obtained in this work compared to control in agreement with study reported that: Although platelet counts remain in the normal pregnant range in most women during uncomplicated pregnancies (Matthews et al., 1990). Mean platelet counts of pregnant women may be slightly lower than in healthy non pregnant women (Verdy et al., 1997). Bashiri, A.B., E.E. Sheiner and M. Mazor, 2003. Anemia during pregnancy and treatment with intravenous iron: Review of literature. Eur. J. Obstetric Gynecol. Reproductive Biol., 110: 2-7. PMID: 12932861 White blood cells are responsible for body defense. During pregnancy, WBC is reported to be elevated (Pitkin and Witte, 1979). We found significant increased in white blood count (8.58±50.06% ×103/mm2), neutrophils (57.96±1004.97%), lymphocytes (28.24±1392.97%), mesophils (7.28±110.49%) and eosinophils (3.62±1156.91%) significant higher compared to that of the controls. The finding in agreement with previous study reported (Rouse et al., 1998). Increase in these haematological parameters may be as a result of the body building the immunity of the fetus and it is achieved by a state of selective immune tolerance, in the presence of a strong antimicrobial immunity (Elgari, 2013). The result of this work agrees with previous work by Roy et al. (2007) that reported a total leukocyte count rising in early pregnancy which remained elevated through pregnancy. Beare, N.A., T.E. Taylor, S.P. Harding, S. Lewallen and M.E. Molyneux, 2006. Malarial retinopathy: A newly established diagnostic sign in severe malaria. Am. J. Tropical Med. Hygiene, 75: 790-797. PMID: 17123967 Boel, M.E., M.J. Rijken, B.J. Brabin, F. Nosten and R. McGready, 2012. The epidemiology of postpartum malaria: Asystematic review. Malaria J., 11: 1-7. DOI: 10.1186/1475-2875-11-114 DeBeaudrap, P., E. Turyakira, L.J. White, C. Nabasumba and B. Tumwebaze et al., 2013. Impact of malaria during pregnancy on pregnancy outcomes in a Ugandan prospective cohort with intensive malaria screening and prompt treatment. Malaria J., 12: 139- 139. DOI: 10.1186/1475-2875-12-139 Science Publications AJM 15 Alo Moses Nnnaemeka et al. / American Journal of Microbiology 5 (1): 11-17, 2014 Desai, M., F.O. Ter Kuile, F. Nosten, R. McGready and K. Asamoa et al., 2007. Epidemiology and burden of malaria in pregnancy. Lancet Infect. Science Publications Science Publications 7. REFERENCES Byg, L. Eriksen and A.M. Hvas, 2007. Reference intervals for haematological variables during normal pregnancy and postpartum in 434 healthy Danish women. Eur. J. Haematol., 79, 39-46. PMID: 17598837 Saba, N., A. Sultana and I. Mahsud, 2008. Outcome and complications of malaria in pregnancy. Gomal J. Med. Sci., 6: 98-101. DOI: 10.5001/omj.2012.52 Murphy, S.C. and J.G. Breman, 2001. Gaps in the childhood malaria burden in Africa: Cerebral malaria, neurological sequelae, anemia, respiratory distress, hypoglycemia and complications of pregnancy. Am. J. Tropical Med. Hygiene, 64: 57-67. PMID: 11425178 Sirima, S.B., R. Sawadogo, A.C. Moran, A. Konate and A. Diarra et al., 2003. Failure of a chloroquine chemoprophylaxis program to adequately prevent malaria during pregnancy in Koupela District, Burkina Faso. Clin. Infect. Dis., 36: 1374-1382. DOI: 10.1086/375077 Nahlen, B.L., T. Alakija, O. Ogunbode, O. Adetoro and A. Akintunde et al., 2012. Lack of pyrimethamine prophylaxis in pregnant Nigerian women. Lancet. 8667: 830-834. DOI: 10.1016/S0140-6736(89)92998-X Steketee, R.W., B.L. Nahlen, M.E. Parise and C. Menendez, 2001. The burden of malaria in pregnancy in malaria-endemic areas. Am. J. Tropical Med. Hygiene, 64: 28-35. DOI: org/10.4236/ojpm.2013.32035 Science Publications AJM 16 Alo Moses Nnnaemeka et al. / American Journal of Microbiology 5 (1): 11-17, 2014 Ticconi, D., M. Mapfumo, M. Dorrucci, N. Naha and E. Tarira et al., 2003. Effect of maternal HIV and malaria infection on pregnancy and perinatal outcome in Zimbabwe. J. Acquire Immune Deficiency Syndromes, 34: 289-94. DOI: 10.1097/00126334-200311010-00005 Verdy, E., V. Bessous and M. Dreyfus, 1997. Longitudinal analysis of platelet count and volume in normal pregnancy. Thromb Haemost. 77: 806. DOI: 10.1016/j.amepre.2009.12.018 WHO, 2012. A strategic framework for malaria prevention and control during pregnancy in the African region. World Health Organization. Tongo, O.O., A.E. Orimadegun and O.O. Akinyinka, 2011. Utilisation of malaria preventive measures during pregnancy and birth outcomes in Ibadan, Nigeria. BMC Pregnancy Childbirth, 11: 1-7. DOI: 10.1186/1471-2393-11-60 WHO, 2013. Expert committee on malaria 18th report. World Health Organization. AJM Science Publications 17 Science Publications Science Publications
https://openalex.org/W4385573494	https://aclanthology.org/2022.nlp4pi-1.14.pdf	English	null	“Am I Answering My Job Interview Questions Right?”: A NLP Approach to Predict Degree of Explanation in Job Interview Responses	null	2,022	cc-by	5,120	“Am I Answering My Job Interview Questions Right?”: A NLP Approach to Detecting the Degree of Explanation in Job Interview Responses Raghu D. Verrap Texas A&M University raghudv@tamu.edu Ehsanul Haque Nirjhar Texas A&M University nijrhar71@tamu.edu Theodora Chaspari Texas A&M University chaspari@tamu.edu Ani Nenkova Adobe Research nenkova@adobe.com that can facilitate training in a safe environment on specific verbal and nonverbal behaviors and can help individuals effectively adapt to cognitively demanding and socially challenging interview situ- ations (Hemamou et al., 2019b). This technology can further contribute to an inclusive workforce. Since the employment interview comprises the first step of the job hiring process, intelligent interview training augmented with NLP can detect linguistic and semantic communicative behaviors that might jeopardize candidates’ performance in the inter- view, suggest the exact modifications needed to effectively communicate their skills, and facilitate access to training material and information in a personalized manner (Marienko et al., 2020). Abstract Providing the right amount of explanation in an employment interview can help the interviewee effectively communicate their skills and experi- ence to the interviewer and convince that she/he is the right candidate for the job. This paper examines natural language processing (NLP) approaches, including word-based tokenization, lexicon-based representations, and pre-trained embeddings with deep learning models, for de- tecting the degree of explanation in a job inter- view response. These are exemplified in a study of 24 military veterans who are the focal group of this study, since they can experience unique challenges in job interviews due to the unique verbal communication style that is prevalent in the military. Military veterans participated in mock interviews with industry recruiters and data from these interviews were transcribed and analyzed. Results indicate that the feasi- bility of automated NLP methods for detect- ing the degree of explanation in an interview response. Features based on tokenizer analy- sis are the most effective in detecting under- explained responses (i.e., 0.29 F1-score), while lexicon-based methods depict the higher per- formance in detecting over-explanation (i.e., 0.51 F1-score). Findings from this work lay the foundation for the design of intelligent as- sistive technologies that can provide personal- ized learning pathways to job candidates, es- pecially those belonging to sensitive or under- represented populations, and helping them suc- ceed in employment job interviews, ultimately contributing to an inclusive workforce. Military veterans is a group that can particularly benefit from assistive interview training technolo- gies. “Am I Answering My Job Interview Questions Right?”: A NLP Approach to Detecting the Degree of Explanation in Job Interview Responses In many countries around the world, mili- tary veterans face major barriers to participating in the civilian workforce after separation from ac- tive duty (McAllister et al., 2015; Ahern et al., 2015). The military background and training of most veterans is significantly different compared to the general job candidate population, who usually comprise of relatively younger fresh college gradu- ates. Military veterans often find it challenging to clearly articulate their strengths and “brag" about their achievements in the civilian employment in- terview setting. Particularly, they can experience unique verbal communication gaps, such as ineffec- tive translation of relevant military experience and technical skills, over-explaining their responses, and excessive use of military jargon, that hamper them from successfully obtaining a job in the civil- ian workforce (Roy et al., 2020). Intelligent job in- terview training systems can potentially track these linguistic behaviors of interest and provide military veterans the right feedback at the right time. Proceedings of the Second Workshop on NLP for Positive Impact (NLP4PI), pages 122 - 129 December 7, 2022 ©2022 Association for Computational Linguistics 1 Introductioni Artificial intelligence (AI) can empower a plethora of assistive tools for enhancing one’s visual, hearing, communication, cognitive, and motor skills (Zdravkova, 2022). By automating natural language processing (NLP) and understanding, AI technologies can enable individuals who belong to sensitive populations, to better express themselves or better understand the world around them. In- telligent interview training is one such technology We conduct a linguistic analysis of veterans’ re- sponses in civilian interview settings. We focus on the degree of explanation in the response, since 122 this construct is particularly relevant to the inter- view success and unexplored by previous work, and particularly we investigate a range of NLP sys- tems to detect over/under-explained, succinct, and comprehensive responses (Hagen et al., 2022). To accomplish this task, we examine NLP systems that rely on text tokenization, lexicon-based analysis, and deep learning methods. These are evaluated on transcripts from mock interviews between 24 military veterans and 5 industry recruiters. A total of 163 responses provided during the interviews were coded by third-party annotators with respect to the degree of explanation. Results indicate the feasibility of automated NLP analysis for detect- ing the outcome of interest. Particularly, features based on tokenizer analysis are the most effective in detecting under-explained responses (i.e., 0.29 F1-score), while lexicon-based methods depict the higher performance in detecting over-explanation (i.e., 0.51 F1-score). Challenges that were met dur- ing data analysis, namely, the small data sample, subjectivity in coding, and uneven class distribu- tions, are described. Discussion of these results further provides ways in which the proposed NLP analysis can contribute to the design of assistive technologies for interview training. virtual reality system that simulated various inter- view settings, including the interviewer’s propen- sity toward the interviewee (i.e., friendly, neutral, unfriendly) and the physical space of the interview (e.g., break room, office) (Hartholt et al., 2019). A user would interact with the training system by starting from easy to more challenging scenarios. No additional feedback was provided to the user. Another line of work has evaluated interviewees based on multimodal data that were mostly col- lected in an asynchronous manner. Chen et al. estimated applicants’ personality traits based on the audiovisual analysis of monologue job inter- views (Chen et al., 2017). Linguistic analysis was conducted with a Bag-Of-Words text representa- tion. Hemamou et al. 1 Introductioni designed a hierarchical at- tention model, called “HireNet" that predicted the hirability of an interviewee based on asynchronous video interviewing. HireNet relied on multimodal information from text, audio, and video (Hemamou et al., 2019a,b). Similarly, Ngugen & Gatricia- Perez and Muralidhar et al. analyzed acoustic and visual cues of video resumes and examined their effectiveness in estimating the candidate’s hireabil- ity and social and communication skills (Nguyen and Gatica-Perez, 2016; Muralidhar et al., 2016). Finally, Naim et al. analyzed interviewees’ per- formance in mock job interviews using their fa- cial expressions (e.g., smiles, head gestures, facial tracking points), language (e.g., word counts, topic modeling), and prosodic information (e.g., pitch, in- tonation, and pauses). Results presented in the MIT Interview Dataset suggest that the use of unique words and personal pronouns, and the degree of speech fluency significantly affect one’s interview performance (Naim et al., 2016). 2 Related Work Prior work in assistive technologies for interview training has focused on helping users demonstrate effective social skills and positive personality cues. The TARDIS project, for example, designed a game simulation platform through which interviewees in- teracted with a virtual agent in an effort to improve social cues and affective expressions during the interview (Anderson et al., 2013; Gebhard et al., 2018). The system automatically detected and an- alyzed smiles, head nods, and body movements, which were used by a machine learning algorithm to classify the mental state (e.g., stressed, bored, hesitant) and affective state (e.g., positive/negative mood) of the user. During the virtual interview, the user received credits in the game when depicting behaviors that were deemed as effective for the in- terview. At the end, users received a series of statis- tics for each of the focal behaviors, which were also visualized over time. MACH—My Automated Conversation coacH is another automated interview training system that provided feedback to the user regarding their performance based on the analysis of facial expressions, speech, and prosody (Hoque et al., 2013). Similarly, Hartholt et al. designed a The contributions of this paper in comparison to prior work are: (1) While previous work focuses on global characteristics of the interviewee (e.g., per- sonality, social/communication skills) and overall descriptors of the interview outcome (e.g., hire- ability, performance), this paper provides a closer study to turn-level behaviors that can affect the job interview outcome, thus laying out the foundation toward intelligent assistive technologies that can analyze micro-level data and provide users with detailed feedback at the turn-level; (2) In contrast to the majority of prior work, this paper analyzes data from synchronous interactions between an in- terviewer and an interviewee, which are more dy- namic and diverse; and (3) Prior work has mostly 123 Degree of Explanation No. of Samples Under-explained 16 Succinct 67 Comprehensive 58 Over-explained 17 Total Samples 158 Table 1: Distribution of classes characterizing the degree of explanation to an interview question. Degree of Explanation No. of Samples focused on college students or fresh college gradu- ates, while this research investigates a unique pop- ulation that comprises of military veterans facing unique challenges when preparing for a job inter- view, thus outlining unique design characteristics when it comes to creating assistive technologies for this population. Table 1: Distribution of classes characterizing the degree of explanation to an interview question. authors’ university. We use data from an ongoing research study with U.S. military veterans who participated in a mock job interview conducted by experienced interview- ers from the industry. Currently, 24 participants completed the study. Data from one participant is excluded from this paper due to technical issues in pre-processing. The average age of participants was 36.4 years (stand. dev. = 10.6 years), and two out of the 24 participants were female. The study was conducted in a hybrid format, where the in- terviewees (i.e., military veterans) were present in the lab, and the interviewers (i.e., industry experts) were connected via Zoom video conferencing. In order to obtain naturalistic conversational data in the mock job interview, we created customized job postings tailored to each participant’s résumé, which were shared with both the interviewees and the interviewers. Interviewees were instructed to think that they applied for the aforementioned job and they were participating in the corresponding job interview. The interviewers were instructed to conduct the interview based on the job posting, and ask questions in a similar fashion as they would normally do as part of their job role. The aver- age length of the interviews was about 18 minutes (stand. dev. = 6.4 minutes). Audio and video of the interviews were recorded, while the tran- scripts of the interviews were obtained by the au- tomatic speech recognition functionality provided via Zoom. Transcripts were manually checked for errors, such as spelling mistakes, incomprehensible words, disfluencies, and non-verbal vocalizations. Next, interviews were checked manually to mark the start and end timestamps of each question and their corresponding responses. If the interviewer provided any prompts or asked for additional in- formation after a response, these turns were con- sidered as a part of the response to the original question. In total, 163 responses to the interview questions from the participants were recorded and were used for further analysis. This study has been approved by the institutional review board of the 3.1 Data Collection authors’ university. 3.2 Behavioral Annotation In order to label the degree of explanation in the responses to the interview questions, behavioral annotation was performed by three third-party an- notators, who were undergraduate students in psy- chology and had previous experience in behavioral coding and annotation tasks. Consistently with previous work (Busso et al., 2016; Lefter et al., 2014), annotators were asked to watch the individ- ual questions and the corresponding responses from the interview and rate the degree of explanation in each response into the following four possible cate- gories. Under-explained (Class 0): Short response that does not fully answer the interviewer’s ques- tion. Such responses might end abruptly; Succinct (Class 1): Concise and to-the-point responses that answer the interviewer’s question fully and briefly; Comprehensive (Class 2): Detailed response that answers the fully answers the question; and Over- explained (Class 3): Very long response to the question with excess verbiage and too much detail that potentially affects the coherence of the answer. The numerical labels are assigned based on the expected increasing order in response length for each of these categories (i.e., succinct responses are expected to be shorter compared to compre- hensive ones). The annotation process resulted in a moderate annotator agreement of Fleiss’ κ = 0.437 (Fleiss, 1971; Hallgren, 2012). After the annotation, five responses yielded labels with com- plete disagreement. These were excluded from the rest of the analysis, which renders the sample size, N = 158. The final labels were obtained by aggregating annotations through majority voting. Table 1 shows the distribution of labels obtained from this aggregation. It is to be noted that both “Under-explained” and “Over-explained” classes are minority classes, although they are the classes of interest, since these types of responses tend to contribute most to perceived hireability and job interview performance. 124 4 Methods binary classification tasks are unbalanced, the F1- score is used as evaluation metric for the following systems. F1-score is reported for each class using a leave-one-subject-out cross-validation. Accord- ing to this, the responses from one interviewee are included in the test set and the responses from the remaining interviewees are included in the train set, with this procedure repeating until all interviewees are part of the test set. Since the numbers of samples belonging to the classes of interest (i.e., “Under-explained”, “Over- explained”) is much lower compared to the major- ity classes, it would be counter-productive to for- mulate the target problem as a 4-way classification task. To resolve this issue, we examine the associa- tion between the response length and the explana- tion labels. Intuitively, we anticipate that responses belonging to the “Under-Explained" and “Suc- cinct" classes will have significantly shorter length compared to the ones belonging to the “Compre- hensive" and “Over-Explained" classes. Response length is measured in terms of word count (i.e., the number of words in the response) and response du- ration (i.e., the duration of the response in seconds). Both these measures exhibit significantly high Pear- son’s correlation coefficients with the explanation labels (i.e., r = 0.68, p < 0.01 for word count, r = 0.66, p < 0.01 for response duration). This suggests that the shorter responses tend to fall into “Under-explained” and “Succinct” categories, while the longer responses belong to the “Comprehen- sive” and “Over-explained” classes. To further con- firm this, a binary classification task is conducted to identify whether a response falls into the short (i.e., “Under-explained”, “Succinct”) or long (i.e., “Comprehensive”, “Over-explained”) category. For this purpose, a logistic regression model with re- sponse length as feature and with leave-one-subject- out cross-validation is used, which resulted in an macro-average F1-score of 0.87. This suggests that we can simply classify the responses into the short (i.e., “Under-explained”, “Succinct”) or long (i.e., “Comprehensive”, “Over-explained”) category be- fore estimating the original classes. Therefore, to estimate the degree of explanation, in the follow- ing analysis, we formulate two binary classification problems (i.e., “Under-explained” vs. “Succinct”, “Comprehensive” vs. “Over-explained”) instead of a 4-class problem. 4.1 Tokenizer We extract the linguistic information from the par- ticipants’ responses to the interview questions us- ing NLTK tokenizer (Bird et al., 2009). The NLTK tokenizer breaks each response into chunks at the word-level that can be considered as discrete el- ements. Tokens are generated from the response text without any truncation and padding. A total of 510 tokens with frequency more than three are selected as features for conventional machine learn- ing models. The frequency of the corresponding tokens serves as the feature vector of length 510 to a decision tree model that conducts the binary classification tasks. 4.2 Lexicon-based method In order to identify the psycholinguistic content of the participants’ responses to the interview ques- tions, we employ the Linguistic Inquiry and Word Count (LIWC) toolbox (Pennebaker et al., 2015). This tool measures the count (or percentage) of words from several constructs, known as LIWC categories. The LIWC categories include gen- eral descriptors (e.g., word count, words per sen- tence), summary variables (e.g., analytical think- ing, clout), standard linguistic dimensions (e.g., pronouns, verbs), psychological constructs (e.g., af- fect, cognition), personal concern constructs (e.g., work, leisure), informal language marker (e.g., filler words, assents), and punctuation (e.g., pe- riods, commas). Overall, we obtain 93 LIWC fea- tures from each sample, that comprise the input features of a binary decision tree. We pursue three different approaches for these binary classification tasks. The first approach em- ploys a tokenizer that breaks text into word tokens, followed by a decision tree that conducts the binary classification task. The second approach utilizes a lexicon-based model of psycholinguistic speech attributes, followed by a decision tree. The third approach leverages a transformer-based model pre- trained on a large corpus of English text in self- supervised manner. Since the classes of each of the 4.3 Deep learning method 4.3 Deep learning method We further explore the use of deep learning mod- els for the considered binary classification tasks. We use the RoBERTa-base (Liu et al., 2019) as the backbone network, a popular transformer-based model (Vaswani et al., 2017) pre-trained on a large corpus of English text in self-supervised manner. The input of this model comprises of the segments resulting from the Tokenizer (Section 4.1), namely, the first 510 tokens. The input is connected to two 125 fully connected layers with 768 nodes each, ReLU activation, and dropout, following by the final out- put layer. As the dataset is highly unbalanced, we perform undersampling on the majority class and oversampling on the minority class. In addition, we freeze the initial 75% layers of the RoBERTa base pre-trained model. The model is trained for 20 epochs with a learning rate of 10−5. ences which are typically associated with male ref- erences. Finally, the Tokenizer method achieves the highest F1-score for the “Under-explained” class, potentially because these types of responses depict distinctive patterns with respect to the frequency of tokens compared to the “Succinct" class. 6 Discussion The increasingly complex and demanding employ- ment market and future workforce requires ma- ture handling of content and emotions by the job candidates, therefore failing to explain one’s skills or over-sharing information can be detrimental to succeeding in the employment interview (Cismas, 2021). Results from this study indicate that various types of NLP techniques can be effective in auto- matically identifying the degree of explanation in job interview responses, which can be particularly valuable when designing training technologies to prepare candidates for future employment. While previous work has focused on behavioral impres- sions that can affect the overall outcome of the interview (Anderson et al., 2013; Gebhard et al., 2018; Hoque et al., 2013; Hartholt et al., 2019), this paper focuses on linguistic behaviors at the turn- level, which can serve as the foundation for provid- ing tangible low-level feedback to the interviewee. Training technologies that rely on automated NLP systems, such as the ones examined in this paper, can help pinpoint exact turns in the dialog that ef- fectively serve the job interview outcome (i.e., suc- cinct, comprehensive responses), as well as turns that might hurt the interview outcome (i.e., under- explaining, over-explaining). Intelligent cognitive enhancement technologies can potentially assist job candidates in helping them effectively communi- cate their skills to the interviewers. Such technolo- gies need to rely on robust NLP approaches, that are adequately generalizable to unseen users and new contexts and depict reliable performance, espe- cially for the detection of classes of interest, such as the under-explaining and over-explaining classes in our case. In addition, NLP technologies need to be effectively meshed with human-computer interac- tion (HCI) interfaces, in order to provide feedback in the right form (e.g., visual, tactile) and the right time (e.g., during practice, post-practice). In addi- tion to detecting points of improvement, explaining their role in interview performance and suggest- ing appropriate changes to those responses would pave the way for personalized learning pathways. It is also essential to consider the degree of expla- 5 Experiments Results obtained by the different NLP systems are summarized in Table 2. The F1-score for the “Suc- cinct” and “Comprehensive” classes is significantly higher than the other two, since these are the major- ity classes. The deep learning method that relies on the RoBERTa model further achieves higher score than the Tokenizer and Lexicon-based methods for the “Succinct” and “Comprehensive” classes. This is anticipated as these two classes have a relatively high number of samples, thus the deep learning model can effectively learn their linguistic repre- sentation. Meanwhile, the lexicon-based features achieve the highest performance for the “Over- explained" class, which might be due to the fact that these two types of responses can be effectively dif- ferentiated via psycholinguistic dimensions. Statis- tical analysis via t-tests between the two classes of interest indicates that comprehensive responses de- pict significantly more positive emotional tone com- pared to over-explained responses (µ3 = 56.83%, µ4 = 44.47%, p < 0.05), where µ3 and µ4 are the mean values of the comprehensive and over- explained responses, respectively. This might be attributed to the fact that over-explained responses merely report content without depicting one’s af- fective view. Comprehensive responses also in- clude a significantly larger percentage of long words (i.e., words greater than six letters) com- pared to over-explained responses (µ3 = 17.14%, µ4 = 13.67%, p < 0.01) and significantly more work-relevant words (µ3 = 4.88%, µ4 = 3.39%, p < 0.05). This indicates that comprehensive re- sponses are characterized by more complex expres- sion (Smith-Keiling and Hyun, 2019) and commu- nicate one’s work-related experiences. On the con- trary, over-explained responses have a significantly larger number of male references compared to com- prehensive ones (µ3 = 27.24%, µ4 = 67.55%, p < 0.05) and include more past tense verbs (µ3 = 3.87%, µ4 = 5.39%, p < 0.05), poten- tially because over-explained responses are overly focused on one’s immersion to past military experi- 126 Methods F1-score Under-explained Succinct Comprehensive Over-explained Tokenizer 0.29 0.81 0.74 0.26 Lexicon-based method 0.22 0.78 0.83 0.51 Deep learning method 0.27 0.89 0.84 0.39 Table 2: F1-score for each class of interest obtained by the considered methods. Table 2: F1-score for each class of interest obtained by the considered methods. adequately generalizing to other individuals and populations. Acknowledgements The authors would like to acknowledge the Na- tional Science Foundation (NSF) for funding this work (NSF #1956021, NSF #1955721, the Don & Ellie Knauss Veteran Resource and Support Center (VRSC) at Texas A&M University for their partner- ship in this research, and undergraduate students, Luis Garcia, Li Wen Jan, Albin Kyle Myscich, and Felicity Woodson for their contribution in data tran- scription, annotation, and pre-processing. Ethics Statement The authors of this paper strove to maintain high- est standards of professional conduct and ethical practice when conducting this work via respect- ing and maintaining the privacy of the participants of this study and security of the data and disclos- ing all pertinent system capabilities and limitations. This work is guided by the values of equality, in- clusiveness, and respect for others, since it aims to render assistive interview technologies accessi- ble to populations such as military veterans who have traditionally faced challenges in entering the workforce and have not actively been the focus of prior studies in computing that have examined the automated processing of interview data. 5 Experiments In addition, due to the demographics of the region from which the data was sampled, the current dataset is highly skewed toward White male participants. As part of our future work, we will be verifying those findings with additional data that will include more diverse participants, which will allow us to make these technologies truly inclusive to all people. Second, the moderate agreement level (i.e., κ = 0.437) will be addressed via adjudica- tion meetings. Third, this work takes into account the interviewee’s response in isolation without con- sidering the content of the question. Future work will incorporate the interview context, turn-taking between interviewer and interviewee, and acous- tic information from speech, which is expected to yield improved performance. nation in the context of other linguistic behaviors (e.g., excessive use of military jargon, ineffective translation of military experience to the civilian job context), gestures (e.g., rigidity in posture), and vocal expressions (e.g., voice loudness), which will allow us to design technologies that can assist vet- eran interviewees in a holistic manner. User studies are needed to be conducted so that we can better un- derstand the effectiveness of these technologies in the overarching goal of assisting military veterans to succeed in civilian job interviews. 7 Conclusion We examined linguistic behaviors of military vet- erans that are indicative of the degree of explana- tion in job interview responses. We investigated different types of linguistic descriptors, ranging from word-based tokenization and lexicon-based representations, to pre-trained embeddings with deep learning models. Our results indicate that pre- trained embeddings are effective in detecting suc- cinct and comprehensive responses, which contain the majority of samples. Lexicon-based features can reliably detect over-explained responses, po- tentially because of their unique psycholinguistic characteristics related to affect, work experience, and complex expression. Finally, under-explained answers are best recognized via the token-based ap- proach, which might be due to the fact that these are characterized by significantly different frequency of tokens compared to the succinct responses. Re- sults from this study lay the foundation toward intelligent interview training technologies that pro- vide personalized learning by detecting verbal be- haviors important for the job interview, explaining their role to the user, and suggesting appropriate changes that can effectively help users secure their desired job. References Jennifer Ahern, Miranda Worthen, Jackson Masters, Sheri A Lippman, Emily J Ozer, and Rudolf Moos. 2015. The challenges of afghanistan and iraq veter- ans’ transition from military to civilian life and ap- proaches to reconnection. PloS one, 10(7):e0128599. Léo Hemamou, Ghazi Felhi, Jean-Claude Martin, and Chloé Clavel. 2019a. Slices of attention in asyn- chronous video job interviews. In 2019 8th Interna- tional Conference on Affective Computing and Intel- ligent Interaction (ACII), pages 1–7. IEEE. Keith Anderson, Elisabeth André, Tobias Baur, Sara Bernardini, Mathieu Chollet, Evi Chryssafidou, Ionut Damian, Cathy Ennis, Arjan Egges, Patrick Gebhard, et al. 2013. The tardis framework: intelligent vir- tual agents for social coaching in job interviews. In International Conference on Advances in Computer Entertainment Technology, pages 476–491. Springer. Léo Hemamou, Ghazi Felhi, Vincent Vandenbuss- che, Jean-Claude Martin, and Chloé Clavel. 2019b. Hirenet: A hierarchical attention model for the auto- matic analysis of asynchronous video job interviews. In Proceedings of the AAAI conference on artificial intelligence, volume 33, pages 573–581. Steven Bird, Ewan Klein, and Edward Loper. 2009. Nat- ural language processing with Python: analyzing text with the natural language toolkit. " O’Reilly Media, Inc.". Mohammed Hoque, Matthieu Courgeon, Jean-Claude Martin, Bilge Mutlu, and Rosalind W Picard. 2013. Mach: My automated conversation coach. In Pro- ceedings of the 2013 ACM international joint confer- ence on Pervasive and ubiquitous computing, pages 697–706. Carlos Busso, Srinivas Parthasarathy, Alec Burmania, Mohammed AbdelWahab, Najmeh Sadoughi, and Emily Mower Provost. 2016. Msp-improv: An acted corpus of dyadic interactions to study emotion per- ception. IEEE Transactions on Affective Computing, 8(1):67–80. Iulia Lefter, Gertjan J Burghouts, and Leon JM Rothkrantz. 2014. An audio-visual dataset of human– human interactions in stressful situations. Journal on Multimodal User Interfaces, 8(1):29–41. Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Man- dar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019. Roberta: A robustly optimized bert pretraining ap- proach. arXiv preprint arXiv:1907.11692. Lei Chen, Ru Zhao, Chee Wee Leong, Blair Lehman, Gary Feng, and Mohammed Ehsan Hoque. 2017. Au- tomated video interview judgment on a large-sized corpus collected online. In 2017 Seventh Interna- tional Conference on Affective Computing and Intel- ligent Interaction (ACII), pages 504–509. IEEE. Maiia Marienko, Yulia Nosenko, and Mariya Shyshkina. 2020. Personalization of learning using adaptive technologies and augmented reality. arXiv preprint arXiv:2011.05802. Suzana Carmen Cismas. 2021. Strategies to enhance students’ employability and job interview abilities by didactic role-plays. Reading Multiculturalism. Limitations The results of this work should be considered in the light of the following limitations. First, while it is difficult to obtain large-scale corpora from real-life interpersonal interactions, the relatively small size of the dataset prevents results of this study from 127 ACM International Conference on Intelligent Virtual Agents, pages 238–240. References Human and Social Perspectives, page 23. Charn P McAllister, Jeremy D Mackey, Kaylee J Hack- ney, and Pamela L Perrewé. 2015. From combat to khakis: An exploratory examination of job stress with veterans. Military Psychology, 27(2):93–107. Joseph L Fleiss. 1971. Measuring nominal scale agree- ment among many raters. Psychological bulletin, 76(5):378. Skanda Muralidhar, Laurent Son Nguyen, Denise Frauendorfer, Jean-Marc Odobez, Marianne Schmid Mast, and Daniel Gatica-Perez. 2016. Training on the job: Behavioral analysis of job interviews in hospitality. In Proceedings of the 18th acm international conference on multimodal interaction, pages 84–91. Patrick Gebhard, Tanja Schneeberger, Elisabeth An- dré, Tobias Baur, Ionut Damian, Gregor Mehlmann, Cornelius König, and Markus Langer. 2018. Seri- ous games for training social skills in job interviews. IEEE Transactions on Games, 11(4):340–351. Ellen Hagen, Md Nazmus Sakib, Neha Rani, Ehsanul Haque Nirjhar, Ani Nenkova, Theodora Chaspari, Sharon Lynn Chu, Amir Behzadan, and Winfred Arthur, Jr. 2022. Interviewer perceptions of veterans in civilian employment interviews and sug- gested interventions. International Military Testing Association. Iftekhar Naim, Md Iftekhar Tanveer, Daniel Gildea, and Mohammed Ehsan Hoque. 2016. Automated analysis and prediction of job interview performance. IEEE Transactions on Affective Computing, 9(2):191– 204. Kevin A Hallgren. 2012. Computing inter-rater relia- bility for observational data: an overview and tuto- rial. Tutorials in quantitative methods for psychology, 8(1):23. Laurent Son Nguyen and Daniel Gatica-Perez. 2016. Hirability in the wild: Analysis of online conversa- tional video resumes. IEEE Transactions on Multi- media, 18(7):1422–1437. James W Pennebaker, Ryan L Boyd, Kayla Jordan, and Kate Blackburn. 2015. The development and psycho- metric properties of liwc2015. Technical report. Arno Hartholt, Sharon Mozgai, and Albert" Skip" Rizzo. 2019. Virtual job interviewing practice for high- anxiety populations. In Proceedings of the 19th 128 Deborah Roy, Jana Ross, and Cherie Armour. 2020. Making the transition: How finding a good job is a risky business for military veterans in northern ire- land. Military Psychology, 32(5):428–441. Beverly L Smith-Keiling and Hye In F Hyun. 2019. Applying a computer-assisted tool for semantic anal- ysis of writing: Uses for stem and ell. Journal of microbiology & biology education, 20(1):70. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. Advances in neural information processing systems, 30. Katerina Zdravkova. 2022. The potential of artificial intelligence for assistive technology in education. References In Handbook on Intelligent Techniques in the Educa- tional Process, pages 61–85. Springer. 129
https://openalex.org/W3203870982	https://www.frontiersin.org/articles/10.3389/fphar.2021.773228/pdf	English	null	Editorial: Inflammation and Fibrosis in the Gastrointestinal Tract and Liver: Mechanisms and Targets	Frontiers in pharmacology	2,021	cc-by	3,398	INTRODUCTION Inﬂammation of the gastrointestinal tract and liver can be caused by a variety of factors including genetic diseases, immune mediated conditions, environmentally induced reactions, toxic exposure, tissue injury, abnormalities in the gut-liver axis, and general infectious diseases. The consequence of inﬂammation is wound healing following ﬁbrous tissue repair and regeneration. Although it occurs in all organs, each organ shows a speciﬁc pattern of reactivity. †Deceased Specialty section: This article was submitted to Gastrointestinal and Hepatic Pharmacology, a section of the journal Frontiers in Pharmacology Received: 09 September 2021 Accepted: 13 September 2021 Published: 24 September 2021 Inﬂammation and Fibrosis in the Gastrointestinal Tract and Liver: Mechanisms and Targets Inﬂammation and Fibrosis in the Gastrointestinal Tract and Liver: Mechanisms and Targets Editorial: Inﬂammation and Fibrosis in the Gastrointestinal Tract and Liver: Mechanisms and Targets Ralf Weiskirchen 1, Dario Sorrentino 2† and Wolfgang R. Stremmel 3 1Institute for Molecular Pathobiochemistry, Experimental Gene Therapy and Clinical Chemistry (IFMPEGKC), RWTH University Hospital Aachen, Aachen, Germany, 2IBD Center, Division of Gastroenterology, Virginia Tech Carilion School of Medicine, Roanoke, VA, United States, 3Practice for Internal Medicine, Baden-Baden, Germany 1Institute for Molecular Pathobiochemistry, Experimental Gene Therapy and Clinical Chemistry (IFMPEGKC), RWTH University Hospital Aachen, Aachen, Germany, 2IBD Center, Division of Gastroenterology, Virginia Tech Carilion School of Medicine, Roanoke, VA, United States, 3Practice for Internal Medicine, Baden-Baden, Germany Keywords: liver, gastrointestinal tract, IBD, inﬂammation, ﬁbrosis, therapy LIVER INFLAMMATION AND FIBROSIS An et al. discuss the main roles of interleukins (ILs) in the context of hepatic inﬂammation and how this knowledge can be used for the development of therapeutic drugs. In particular, they highlight the dynamics by which interleukins mediate the cross-talk between hepatic cells during liver inﬂammation and injury. It becomes clear that abrogating IL signaling might prevent the progression of hepatic ﬁbrogenesis in the early stages of liver injury. However, as the authors discuss there are some practical problems that need to be addressed to move IL-targeted therapies from the experimental stage to the clinical setting. Edited by and reviewed by: Angelo A. Izzo, University of Naples Federico II, Italy Edited by and reviewed by: Angelo A. Izzo, University of Naples Federico II, Italy In this Research Topic, leading experts from the ﬁeld of Gastroenterology and Hepatology share new ﬁndings and current concepts in the pathogenesis and management of inﬂammation and ﬁbrosis in the gastrointestinal tract and liver. In total about 200 basic scientists and clinicians from eight countries (China, Egypt, Germany, India, Japan, Spain, Italy, and USA) contributed 28 originals or review articles about current perspectives and ﬁndings in corresponding diseases. These articles cover a wide range of aspects in the pathogenesis of liver inﬂammation and ﬁbrosis, gastrointestinal disease, and inﬂammatory bowel disease (IBD). The contributions show that these fascinating ﬁelds have made conceptual advances that in the near future will hopefully lead to novel treatment options. Correspondence: Ralf Weiskirchen rweiskirchen@ukaachen.de *Correspondence: Ralf Weiskirchen rweiskirchen@ukaachen.de EDITORIAL published: 24 September 2021 doi: 10.3389/fphar.2021.773228 Citation: Weiskirchen R, Sorrentino D and Stremmel WR (2021) Editorial: Inﬂammation and Fibrosis in the Gastrointestinal Tract and Liver: Mechanisms and Targets. Front. Pharmacol. 12:773228. doi: 10.3389/fphar.2021.773228 In an experimental animal study, Han et al. investigated the impact of a traditional medicinal herbal extract derived from Chicory (Cichorium pumilum Jacq) in a rat model of liver ﬁbrosis induced by chronic colitis. The authors demonstrate that this herb can promote probiotic growth and prevent liver ﬁbrosis. In addition, the authors demonstrate that Lactucin that is one of the bitter September 2021 \| Volume 12 \| Article 773228 1 Frontiers in Pharmacology \| www.frontiersin.org Weiskirchen et al. Editorial: Gastrointestinal Tract and Liver tasting sesquiterpene lactone of Chicory is able to inhibit lipopolysaccharide-induced inﬂammatory responses in vitro and activation of the Mitogen-activated protein kinase and Akt signaling. Therefore, the authors proposed Lactucin as one of the most signiﬁcant anti-inﬂammatory compounds in Chicory extracts. Xiang et al. analyzed the effects of kaempferol on the liver X receptor α (LXRα)/lysophosphatidylcholine acyltransferase 3 (LPCAT3) pathway in livers of mice fed a high fat diet inducing NASH. They could show that this natural ﬂavonol reduces endoplasmic reticulum stress and inﬂammation as assessed in vitro and in vivo by reduced expression of LXRα, LPCAT3 and genes involved in the initiation or execution of endoplasmic reticulum stress. Based on their results the authors conclude that the LXRα/LPCAT network offers new potential targets for NASH treatment. Another compound boosting endogenous antioxidant mechanisms in the context of acetaminophen (APAP)-induced liver damage was identiﬁed by Rahman et al. In their study, the authors provide evidence that carveol improves liver detriments and liver metabolic deﬁcits in mice subjected to APAP, while the nuclear factor erythroid 2-related factor 2 (Nrf2) inhibitor all- trans retinoic acid exaggerated APAP toxicity. Mechanistically, the authors found that carveol treatment increases Nrf2 expression that combats various reactive oxygen species and other stress kinases. Additional, comprehensible in-silico docking studies showed that carveol might have an afﬁnity to the active catalytic pockets of COX-2, HO-1, IL-1, NF-κB, iNOS, Nrf2 and TNF-α that all impact inﬂammation. Acharya et al. summarized recent aspects in the understanding of hepatic ﬁbrosis including relevant disease- associated pathways, cellular and molecular drivers of ﬁbrogenesis, and how this knowledge can be translated into therapy of respective disease. Citation: The review shows that hepatic ﬁbrogenesis is a highly complex and orchestrated process that offers many therapeutic opportunities for impeding or reversing this disease process. The Nrf2 pathway was also investigated in an experiment study by Zhao et al., who analyzed the effect of bicyclol in cholestatic mice caused by bile duct ligation. Interestingly, it was found that bicyclol acts as a signiﬁcant hepatoprotective agent that attenuates liver damage by increasing the levels of hydropholic bile acids such as α-muricholic acid (MCA) and β-MCA. Moreover, bicyclol promoted autophagy and activated Nrf2 expression and antioxidant downstream genes. The sodium-glucose cotransporter two inhibitor dapagliﬂoxin was shown by Li et al. ameliorates hepatic steatosis by decreasing the de novo lipogenesis enzyme acetyl-CoA carboxylase 1 (ACC1), increasing the fatty oxidation enzyme acyl-CoA oxidase 1 (ACOX1), and inducing autophagy via the AMPK- mTOR pathway. This may be in future used as therapeutic approach to treat NAFLD by a functional mechanism. The group of Ye et al. showed that Salidroside inhibits carbon tetrachloride-induced liver ﬁbrosis in mice by suppression of JNK activation and modulating the sphingosine kinase 1/sphingosine- 1-phosphate signaling pathway for inhibition of Akt. This reduces activation and migration of hepatic stellate cells (HSC) as the executive players in ﬁbrosis. Salidroside alleviates by this mechanism liver injury, hepatocyte apoptosis and consequently liver ﬁbrosis. Pu et al. provide evidence that the 5-lipoxygenase (5-LO) that catalyzes the oxidation of essential fatty acid substrates into inﬂammatory leukotrienes is a key enzyme involved in mediating hepatic ﬁbrosis. The authors could show that genetic ablation or pharmacological inhibition of 5-LO is therapeutically beneﬁcial to block hepatic ﬁbrosis in vitro and in vivo. In line, patients suffering from non-alcoholic steatohepatitis and liver ﬁbrosis showed elevated levels of 5- LO suggesting that strategies to target 5-LO may offer new therapeutic avenues. A review article by Drescher et al. discusses developmental differences between individual immune cells and their role in health and disease, with a focus on liver diseases. They summarized ﬁndings showing there exists a fragile balance between T helper 17 (TH17) and regulatory T cells (TReg) that is critical in maintaining immune homeostasis during acute and chronic inﬂammation. A new network in the pathogenesis of hepatic ﬁbrosis was introduced by Shouman et al. Frontiers in Pharmacology \| www.frontiersin.org GASTROINTESTINAL TRACT These promising results should now be conﬁrmed in randomized trials with larger numbers of patients. A thematically related study by Ren et al. demonstrated that the COX-2 inhibitor rutaecaprine has gastroprotective effects on ethanol-induced acute gastric mucosal injury in mice. In the respective model, rutaecaprine augmented cellular antioxidant capacity, most likely by triggering antioxidant and anti-apoptosis defense capacities that results in inhibition of the inﬂammatory acting NF-κB, pathway. Chen et al. performed an ultra-high performance liquid chromatography-quadrupole time-of-ﬂight mass spectrometry (UPLC-Q-TOF/MS)-based study to investigate the ameliorative effects of palmatine on Heliobacter pylori-induced chronic atrophic gastritis in rats using serum and urine metabolomics. Interestingly, this alkaloid attenuated pathological damage and inﬂammation and improved integrity of the gastric mucosal epithelial barrier. Mechanistically, the authors suggested that the therapeutic effects of palmatine are majorly attributable to its capacity to impact the metabolism of taurine, subtaurine, glycerol phospholipid and the mutual transformation of pentose and glucoronide. Al-Bawardy et al. highlighted the relevance of continued development of new therapeutic strategies and modiﬁcations of existing therapies to improve the outcome of IBD. In particular, they suggested small molecule inhibitors targeting the Janus kinase or the sphingosine-1-phosphate receptor that both critically contribute to the pathogenesis of IBD as promising therapeutic targets. Most important is that most of these agents can be administered orally and have a favorable safety proﬁle compared to established anti-TNF agents. Yang et al. found that Gremlin 1 (GREM1) is signiﬁcantly increased in mouse ﬁbrotic colon. In their original study, the authors demonstrated that GREM1 promoted the proliferations and activation of intestinal ﬁbroblasts by activating the VEGFR2 receptor and triggering fatty acid oxidation. In line with these ﬁndings, the authors could block intestinal ﬁbrosis progression in vivo by blocking the GREM1-VEGFR2 axis. Ayyar and Moss summarized the actual knowledge on extracellular vesicles including exosomes in intestinal inﬂammation. The compiled information illustrates that exosomes in particular and extracellular vesicles in general are capable of modulating gene expression and cellular functions, thereby critically impacting inﬂammation, immune responses, and composition of gut microbiota. As such these vesicles might be suitable biomarkers of IBD and molecular structures for therapeutic targeting speciﬁc cargos to the inﬂamed gastrointestinal tract. Citation: This overview of new therapeutic approaches with novel biologicals is helpful for clinicians treating IBD. performed and intensive bioinformatics analysis and provide evidence that IL-6 is a good predictive factor for risk assessment. of AICAR on alleviation of liver injury was signiﬁcantly weakened suggesting that AICAR protects against PALI by interfering with Nrf2-triggered processes. Schmidt et al. evaluated the clinical efﬁcacy of tumor necrosis factor antibodies on chronic IBD outcome. In total one third of the patients do not respond and in many cases drug efﬁcacy is lost over time. Contraindications, adverse events and intolerance have to be considered when these biologicals are administered. This overview of new therapeutic approaches with novel biologicals is helpful for clinicians treating IBD. GASTROINTESTINAL TRACT Lamas-Paz et al. identiﬁed a novel mechanism that triggers acute alcoholic hepatic injury in the context of the gut-liver axis. The authors could demonstrate mice subjected to acute alcohol injury develop signiﬁcant alterations in the gut microbiota and in intestinal epithelial barrier function. Interestingly, these alterations are triggered by yet unknown mediators released from extracellular vesicles produced by intestinal epithelial cells provoking deleterious effects on hepatocyte viability and steatosis. helpful for clinicians treating IBD. The group of Zhong et al. showed that the serine-threonine kinase inhibitor rapamycin is effective in the treatment of upper gastrointestinal Crohn’s disease-related stricture formation, but has no effect in lower gastrointestinal Crohn’s disease. The drug was recently shown to reduce intestinal ﬁbrosis by inhibiting CX3Cr1-mTOR-induced autophagy in mononuclear phagocytes and upregulating the IL-23/IL-22 axis. This was now evaluated by a Gastrointestinal obstruction symptoms score and diet score, which both showed improvement in upper gastrointestinal Crohn’s disease, but not in lower gut stricture disease. Adverse events were recorded in 40% of the patients, mostly mouth ulcers. However, no death or serious opportunistic infections were observed. This therapy may avoid surgical or endoscopic interventions. These promising results should now be conﬁrmed in randomized trials with larger numbers of patients. Al-Bawardy et al. highlighted the relevance of continued development of new therapeutic strategies and modiﬁcations of existing therapies to improve the outcome of IBD. In particular, they suggested small molecule inhibitors targeting the Janus kinase or the sphingosine-1-phosphate receptor that both critically contribute to the pathogenesis of IBD as promising therapeutic targets. Most important is that most of these agents can be administered orally and have a favorable safety proﬁle compared to established anti-TNF agents. The group of Zhong et al. showed that the serine-threonine kinase inhibitor rapamycin is effective in the treatment of upper gastrointestinal Crohn’s disease-related stricture formation, but has no effect in lower gastrointestinal Crohn’s disease. The drug was recently shown to reduce intestinal ﬁbrosis by inhibiting CX3Cr1-mTOR-induced autophagy in mononuclear phagocytes and upregulating the IL-23/IL-22 axis. This was now evaluated by a Gastrointestinal obstruction symptoms score and diet score, which both showed improvement in upper gastrointestinal Crohn’s disease, but not in lower gut stricture disease. Adverse events were recorded in 40% of the patients, mostly mouth ulcers. However, no death or serious opportunistic infections were observed. This therapy may avoid surgical or endoscopic interventions. Frontiers in Pharmacology \| www.frontiersin.org Citation: They evaluated antisense oligonucleotides directed against the tissue factor in rats in which liver ﬁbrosis were induced by a single administration of Diethylnitrosamine followed by repeated doses of carbon tetrachloride once weekly for 6 weeks. The authors could demonstrate that the blockage of tissue factor expression resulted in a signiﬁcant downregulation of the protease- activated receptor 1 (PAR1) and toll-like receptor 4 (TLR4) that are both critical drivers in hepatic inﬂammation and ﬁbrosis. Zhu et al. focused on Hengshun aromatic vinegar as a food additive to improve NAFLD. It improved cell viability and attenuated cell damage. Serum levels of triglycerides, transaminases (ALT, AST) and malondialdehyde (MDA) dropped. Inﬂammation and lipogenesis were reduced by interference with metabolic key regulators, i.e. by enhancing silent information regulator of transcription 1 (Sirt1). Thus, it represents a new assistant strategy to ﬁght NAFLD. Roeb and Weiskirchen summarized ﬁndings from a selective literature search on topics related to non-alcoholic fatty liver disease (NAFLD), fructose and ﬁbrosis. The study showed that the rate of overweight and obesity is signiﬁcantly higher in adult and pediatric patients suffering from non-alcoholic steatohepatitis (NASH). Although it is presently not known whether this is due to an excess of energy or the particular metabolism of fructose, the authors suggest that reduction in sugar consumption in conjunction with avoidance of foods enriched in saturated fats and weight loss is recommended for NAFLD and NASH patients. Kong et al. investigated the role of the AMP analog 5- aminoimidazole-4-carboxamide (AICAR) which acts as an activator of the AMP-kinase (AMPK) on the pathogenesis of acute pancreatitis-associated liver injury (PALI). In a rat model of sodium taurocholate-induced acute pancreatitis, they found that AICAR attenuated PALI and restored liver function. Moreover, the compound activated the AMPK/ Nrf2 signaling pathway, while the AMPK inhibitor Compound C aggravated PALI. In Nrf2 null mice, the effect September 2021 \| Volume 12 \| Article 773228 Frontiers in Pharmacology \| www.frontiersin.org 2 Editorial: Gastrointestinal Tract and Liver Weiskirchen et al. performed and intensive bioinformatics analysis and provide evidence that IL-6 is a good predictive factor for risk assessment. Schmidt et al. evaluated the clinical efﬁcacy of tumor necrosis factor antibodies on chronic IBD outcome. In total one third of the patients do not respond and in many cases drug efﬁcacy is lost over time. Contraindications, adverse events and intolerance have to be considered when these biologicals are administered. ACKNOWLEDGMENTS The authors are grateful to all authors that have provided excellent contributions to this Research Topic and the kind support of the Editorial ofﬁce team of Frontiers that helped to arrange a strict review process and assisted to enable reliable correspondence with the authors. Moreover, we acknowledge the excellent and efﬁcient work of the expert reviewers who reviewed submissions in a timely, fair, and constructive manner. In addition, we like to thank Drs. Leo A. van Grunsven (Vrije University Brussel, Belgium), Stefano Fiorucci (University of Perugia, Italy), Luca Antonioli (University of Pisa, Italy), and Barbara Romano (University of Naples Federico II, Italy), who edited contributions in which we had conﬂicts of interest. The study by Nguyen et al. demonstrates the importance of the noncanonical NF-κB signaling pathway in IBD patients. In their study, the authors analyzed the expression of 88 target genes known to be associated with noncanonical NF-κB signaling in biopsy specimens that were collected during colonoscopy. Importantly, the expression of a number of genes was associated with increased gastrointestinal inﬂammation and in particular in patients unresponsive to anti-TNF agents. Although the study is based on a relatively small sample size, the study suggests that the noncanonical NF-κB signaling is an understudied pathway that critically impacts the pathogenesis of IBD. Conﬂict of Interest: The authors declare that the research was conducted in the absence of any commercial or ﬁnancial relationships that could be construed as a potential conﬂict of interest. Publisher’s Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their afﬁliated organizations, or those of the publisher, the editors, and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. All these 28 articles show that inﬂammation and ﬁbrosis of the gastrointestinal tract and liver are complex problems. Currently, there are a number of real breakthroughs in this research area, but of course scientists and clinicians are still at the beginning in the identiﬁcation of suitable therapeutic targets. Since the frequency of these malignancies and their complications currently increase dramatically, new strategies anti-inﬂammatory and anti-ﬁbrotic therapies are urgently needed. Copyright © 2021 Weiskirchen, Sorrentino and Stremmel. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). INFLAMMATORY BOWEL DISEASE Ferretti et al. provide an update on current sonographic methods to discriminate inﬂammation and ﬁbrosis in Crohn’s disease, mainly focusing on studies investigating pathological features or the response to treatment as direct and indirect reference parameters. In their contribution, the authors highlight strengths and weaknesses of individual ultrasound methods and conclude that despite the promising results obtained with novel sonographic techniques such as contrast- enhanced ultrasound (CEUS) and sonoelastrography, several limitations must be addressed before using these techniques in routine clinical practice and trials. Enck and Klosterhalfen addressed the interesting question why in trials of irritable bowel syndrome (IBS) and IBD the placebo rates are remarkably high. Based on a meta-analysis, they report that the spontaneous improvement accounts for 50% of the placebo effect. Moreover, nocebo effects are easily detectable in randomized control trials (RCTs), whereas adverse events are difﬁcult to see in respective reports. Chen Y et al. developed a novel prediction model useful to estimate the risk of primary non-response rate to inﬂiximab therapy and select the optimal treatment for individual patients suffering from Crohn’s disease. In their retrospective study including 322 Crohn’s disease patients and data from the Gene Expression Omnibus (GEO) database, the authors Matsumoto et al. investigated potential therapeutic effects of the RXR agonist Net-3IB that was previously developed in their team in an experimental model of colitis induced through the adoptive transfer of CD45RBhighCD4+ cells. The authors showed September 2021 \| Volume 12 \| Article 773228 Frontiers in Pharmacology \| www.frontiersin.org 3 Weiskirchen et al. Editorial: Gastrointestinal Tract and Liver AUTHOR CONTRIBUTIONS that Net-3IB ameliorates colitis by inhibiting both the expansion of Th1 cells and the activation of inﬂammatory macrophages locally in the colon. Thus, the small molecule inhibitor appears to be a promising candidate for the treatment of IBD. All authors contributed equally to the review and editorial process for this article Research Topic. Ke and colleagues addressed the question whether the anti- inﬂammatory effect of metformin in the intestine is due to a change of the microbiota. They examined this in the experimental colitis model of dextran sulfate sodium (DSS)-induced ulcerative colitis. Indeed the inﬂammation was attenuated by metformin. An increased expression of mucin 2 was noted. The gut microbiota changed under metformin by increasing protective Lactobacillus and Akkermansia species. With antibiotic exposure the anti-inﬂammatory and mucus-protecting effect of metformin was abolished. It is concluded that metformin-induced changes of the microbiota are therapeutically effective in ulcerative colitis. Frontiers in Pharmacology \| www.frontiersin.org September 2021 \| Volume 12 \| Article 773228 ACKNOWLEDGMENTS The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. September 2021 \| Volume 12 \| Article 773228 Frontiers in Pharmacology \| www.frontiersin.org
https://openalex.org/W4246145856	https://www.sciendo.com/pdf/10.2478/bsrj-2020-0011	English	null	Editorial for the special issue: “Novel Solutions and Novel Approaches in Operational Research”	Business systems research	2,020	cc-by	3,512	Abstract This special issue of Business Systems Research (SI of the BSR) is co-published by the Slovenian Society INFORMATIKA – Section for Operational Research (SSI -SOR) and highlights recent advances in Operations Research and Management Science (OR /MS), with a focus on linking OR /MS with other areas of quantitative and qualitative methods in a multidisciplinary framework. Eleven papers selected for this SI of the BSR present improvements and new techniques (methodology) in Operations Research (OR) and their application in various fields of economics, business, spatial science, smart mobility, higher education, human resources, environment, agriculture and social networks. Keywords: interdisciplinary research, operations research, risk and uncertainty, statistical analysis, machine learning, multi-criteria decision making, big data, location-allocation, fuzzy logic, graph theory, project management, system dynamics, simulation methods. Citation: Drobne, S., Dumičić, K., Zadnik Stirn, L. (2020). “Editorial for the special issue: “Novel solutions and novel approaches in Operational Research”, Vol.11, No. 2, 1-6. DOI: 10.2478/bsrj-2020-0011 Samo Drobne University of Ljubljana, Faculty of Civic and Geodetic Engineering, Ljubljana, Slovenia Ksenija Dumičić University of Zagreb, Faculty of Economics and Business, Zagreb, Croatia Lidija Zadnik Stirn University of Ljubljana, Biotechnical Faculty, Ljubljana, Slovenia Samo Drobne University of Ljubljana, Faculty of Civic and Geodetic Engineering, Ljubljana, Slovenia Ksenija Dumičić University of Zagreb, Faculty of Economics and Business, Zagreb, Croatia Lidija Zadnik Stirn University of Ljubljana, Biotechnical Faculty, Ljubljana, Slovenia Samo Drobne University of Ljubljana, Faculty of Civic and Geodetic Engineering, Ljubljana, Slovenia Ksenija Dumičić University of Zagreb, Faculty of Economics and Business, Zagreb, Croatia Lidija Zadnik Stirn University of Ljubljana, Biotechnical Faculty, Ljubljana, Slovenia University of Ljubljana, Biotechnical Faculty, Ljubljana, Slovenia Business Systems Research \| Vol. 11 No. 2 \|2020 Business Systems Research \| Vol. 11 No. 2 \|2020 Editorial process The term operations research (OR), or often management science (MS), refers to a scientific approach to decision making that attempts to determine how best to design or operate a system, usually under conditions that require the allocation of resources in business language (Winston, 2003). OR is a decision support discipline and as such is concerned with the development of systems to help decision makers to solve problems and make decisions. Within decision support topics, OR offers data analysis, simulation, modelling techniques, and software tools (Mladenić et al., 2003; Rubio et al., 2014). The applications of OR in real word problems are very numerous and in very different fields such as industrial engineering, management, economics, production, government, health care, transport, geographic information systems, scheduling, marketing, inventory, environment and others (Cochran et al., 2011). The applications of OR allow complex 1 Business Systems Research \| Vol. 11 No. 2 \|2020 Business Systems Research \| Vol. 11 No. 2 \|2020 problems to be structured transparently and flexibly in a realistic context, introducing both quantitative (e.g. financial ratios) and quantitative criteria into the evaluation process (Figueira et al., 2005). OR has had an important impact on improving the efficiency of organizations and has contributed to increasing productivity and social welfare. The International Federation of Operational Research Societies (IFORS) and the Association of European Operational Research Societies (EURO) are umbrella organizations for OR societies worldwide, representing more than 50 national societies, including Slovenian Society INFORMATIKA - Section for Operational Research (SSI-SOR). The main events organized by SSI-SOR are the international symposia. The 15th International Symposium on Operations Research, called SOR'19, took place in Bled, Slovenia, from 25 to 27 September 2019. SOR'19 was the scientific event in the field of Operations Research, another in the traditional series of biennial international OR conferences, organized in Slovenia by SSI -SOR. The main objective of SOR'19 was to promote the knowledge, interest and education of OR in Slovenia, in Europe and worldwide, in order to build the intellectual and social capital that is essential for maintaining the identity of OR, especially in a time when interdisciplinary cooperation is proclaimed to be significantly important for solving problems in the current challenging times. Furthermore, the SSI -SOR agreed to cooperate with different disciplines, i.e. Contributions In accordance with the goals and editorial policy of BSR, the papers published in BSR are intended to present original theoretical and empirical advances in business and economic systems using a wide range of methodological approaches, primarily from the fields of operations research/analytics, management science and statistics. The eleven papers accepted by BSR for this SI fulfil these objectives. In the first paper, entitled “Green Practices as a Path towards the Sustainability: Evidence from Portuguese Companies”, Alves, Silva and Rodrigues consider the problem of the impact the companies cause on the environment and the society. They evaluate the level of environmental practices in micro, small, medium and large companies in the northern region of Portugal. The results reveal that the environmental issues are not yet properly addressed by treated companies; especially small companies face several berries to implement green actions, mainly those related to certification. The value of the paper is its contribution to new insights on how the handled companies have been implementing sustainable practices, as well which practices they still need to develop to reach higher level of strategic, including financial, policies, and green and sustainable practices. In the second paper, entitled "Comparison of Two Network-Theory-Based Methods for detecting Functional Regions", authors Drobne, Garre, Hontoria and Konjar analyse two methods for modelling functional regions based on graph theory. In the case study of Slovenia, functional regions are calculated using the Walktrap algorithm and a proprietary, so-called, chain approach. They analyse the quality of the two regionalisation methods using the fuzzy set theory with its revised approach. After the results of the case study of Slovenia, the authors concluded that the Walktrap algorithm functionally calculates more closed regions (more workers find work in the home region) than their chain method. In the third paper, entitled “Deep Learning Predictive Models for Terminal Call Rate Prediction during the Warranty Period”, Ferencek, Kofjač, Škraba, Sašek and Kljajić Borštnar present the problem of production companies to optimize their costs by minimizing the amount of funds to be reserved for product repairs during the warranty period. The research extends previous research by providing new insight into machine learning models, and offers additional understanding and validation on how data quality can affect those models. A few new predictive models based on different neural network architectures were developed. They were implemented on a case of a company in the field of home appliances. Business Systems Research \| Vol. 11 No. 2 \|2020 The success of the SI of the BSR should be seen as a result of the joint efforts. The guest editors would like to thank the authors for their well-written contributions and the reviewers for their careful evaluation of the submissions and their thoughtful and constructive comments. Last but not least, the guest editors express their deep appreciation and gratitude to the editor-in-chief, Professor Mirjana Pejić Bach, PhD for her generosity, service and commitment in inviting us as guest editors of SI of the BSR. Editorial process to strike a balance between the depth of theoretical knowledge in OR and the understanding of theory, methods and problems in other areas inside and outside OR. About 115 participants from research institutes, universities, governmental institutions, private and public companies from 16 countries around the world took part in SOR'19. 106 papers were presented, written by 203 authors and co-authors. The papers were accepted after a blind peer review process by two independent reviewers selected from SOR'19 Program Committee and by reviewers appointed by SSI -SOR. At SOR'19 it was agreed that the special issue (SI) of the BSR would be published, so the Call for Papers for this SI was already published during this symposium in Bled in September 2019. The call was addressed to the participants of SOR'19 as well as to other researchers from the area of OR. The submitted papers should present developments and new techniques in OR methods/models and their practical applications in the fields of economics, business, finance, organization, management, social sciences, environment, transport and other fields. p Several contributions have been received. Some of them are extended journal versions of short SOR'19 papers from proceedings (Zadnik Stirn et al., 2019). Each submission to the SI of the BSR was first blind reviewed by the guest editors and then by two independent experts. Eleven contributions were selected for this special issue of the BSR. They owe their practical orientation and the consistent emphasis on model formulation and modelling. In addition, they go beyond a mere presentation of algorithms and reinforce the features and coverage of the latest developments in optimization, simulation and decision analysis. The selected contributions deal with developments and techniques in OR and their practical application in the fields of business, economics, spatial science and location, environment and social sciences. The topics covered in the selected contributions represent interdisciplinary research and include, from a methodological point of view, multi-criteria decision making, fuzzy logic, neural networks, machine learning, predictive models, risk and uncertainty, and big data, while from an application perspective, they include business process modelling, organisational performance, strategic planning, financial applications, farm tourism, project management, pension expectations, mental wellbeing of employees, smart mobility, and higher education. The case studies origin from five countries: Portugal, Croatia, Slovenia, Norway and Hungary. 2 Business Systems Research \| Vol. 11 No. 2 \|2020 that process mining coupled with simulation models offers a suitable innovation environment. Practical implications in the smart parking case refer to better use of (public) resources, and point to the constraints in terms of the non-existence of specific event data, while In the case of higher education the results indicate a better prediction of student behaviour. In the fifth paper, entitled “Portfolio Optimization Efficiency Test Considering Data Snooping Bias”, Kresta and Wang are interested in the evaluation of strategy portfolio performance. The general approach for such evaluation is testing whether the strategy outperforms the benchmark. Such approach does not answer whether the overperformance is high enough to be considered as significant or whether it is just due to the randomness in data. Consequently, the authors propose an alternative approach based on the statistical test in order to evaluate the efficiency of the portfolio optimization strategies in view of the risk. The proposed approach is demonstrated on the Markowitz minimum variance model and the fuzzy probabilities minimum variance model. The results confirm that minimizing the variance of portfolio return in-sample also lowers the out-of-sample risk measures, and that the analysed strategies lower the risk of the portfolio during the market’s decline in 94% of the time in the 2009- 2019 period. p g p Examining the mental wellbeing of employees, which is crucial for the long-term success of an organization, the authors Lipovac, Hajdu, Wie and Nyrud offer to the readers the sixth paper, entitled “Improving Mental Wellbeing in Organizations with Targeted Psychosocial Interventions. The presented research is looked at within individual-level interventions approach. The authors created a social network based on the data collected on 414 employees from 14 nursing homes in Norway where the nodes represent employees and edges describe the connections between the nodes, while the edge probability represents the connection strength between two nodes. The optimization environment computes the possible intervention scenarios and maximizes the overall wellbeing by minimizing the scores of the nodes (i.e., reversed wellbeing score) with the set of employees receiving the intervention. Interventions were provided to either contagious or randomly selected individuals. The results show that selectively targeting highly contagious individuals could be an efficient approach to improving wellbeing in organizations. g g In the seventh paper, entitled “Selection of Project Managers: An Overview”, the author Šiško Kuliš presents the problem of selecting a project manager. Contributions The results show that the best two models, deep neural network with 6 layers and a convolutional neural network differed in 1% when predicting products’ terminal call rate at 12 months. In the next paper, entitled “Process Mining Contributions to Discrete-event Simulation Modelling”, Jadrić, Ninčević Pašalić and Ćukušić state that the technological advance supported by information systems generates event logs which contain important information about the performance of a business process. Event logs are analysed using process mining techniques. The aim of the paper is to demonstrate and assess the potential of using process mining results as an input for discrete-event simulation modelling. The process mining procedure is employed on two datasets. The results show 3 Business Systems Research \| Vol. 11 No. 2 \|2020 Business Systems Research \| Vol. 11 No. 2 \|2020 The paper is seen as an overview paper focusing on the existing studies/methodology. Project manager selection is concentrated on two approaches, a traditional one and on a modern based on multi-criteria decision making. The traditional approach is supported by structured interviews with the goal to project the candidate's behaviour in new and unknown circumstances. A new, modern, approach is based on psychometric testing and multi- criteria methods, among which AHP has an important place. Current characteristics of the process of selecting project manager in Croatia are investigated. A special attention is devoted to the certification of managers as certificate is the basic criteria during project manager selection process. Important for the future research in the field is also the enormous list of up-to-date references. Then follows the paper entitled “Pension Pessimism in the Young Generation: Basics or Instincts to Blame?”, written by Kovács and Vaskövi, which is dealing with pension expectations that are so important for national economies, decision makers and individuals. The paper focuses Hungarian young generation’s reasons regarding pessimistic attitude towards state pension expectations. A non-representative sample research is applied, resulting with 250 filled questionnaires. The surveys data were analysed using multidimensional statistical method, above all factor analysis, to test different hypotheses connected to financial literacy and gender differences in the 4 Business Systems Research \| Vol. 11 No. 2 \|2020 Business Systems Research \| Vol. 11 No. 2 \|2020 pension scheme. The survey results were used also to make a comparative analysis with the ten biases, called instincts by Rosling (2018), in order to find the behavioural aspects lying behind the pessimistic attitude of most respondents. The overall results reveal general pessimism among Hungarian university students towards the social security benefits, due to the general pessimism in the ‘overdramatic worldview’. Further, the paper, entitled “A System Dynamics Approach to Decision- making Tools in Farm Tourism Development”, Žibert, Rozman, Škraba and Prevolšek, establish that nowadays more and more agricultural holdings decide on developing market-oriented multi-function farming. The focus of their research is the development of rural tourism as one of socio-economic activities. The authors generate a qualitative causal loop model and a system dynamics model for the simulation of transition of farming establishments into tourist farms for the purpose of increasing income through the diversification. Business Systems Research \| Vol. 11 No. 2 \|2020 The validation of the model was upgraded with the Mean Squared Error auxiliary variable and Cumulative Mean Squared Error level element, using Powersim Solver with Genetic Algorithms. Different parameter values were used in eight simulation scenarios. The case study of Slovenia has been used in order to explore the scenarios for farm tourism development. It was discovered that transition to diverse farms relies on subsidies that are the main driving force for the transition. In the tenth paper, entitled “Standard Project Risk Analysis Approach”, Žužek, Rihar, Berlec and Kušar, state that the companies, if they want to stay effective, have to be able to adapt to the competitive environment and to essentially manage the risk. The authors consider different risk analysis methods/tools: qualitative, semi-quantitative, quantitative and a risk matrix which is in the quantitative case extended into a continuous graph, called a risk map. First, the major risk factors were identified and assigned to individual activities. The risk events were linked to their impacts, and the risk event probabilities, based on the experience with similar projects, were assessed. Finally, a risk map was generated. The results show that separate treatment of the risk event and the impact advantageous clarifies the cause and the effect, and thus allows for a separate planning of preventive and corrective measures. p g p In the last paper of this SI, entitled “An Investigation of Business Process Maturity: Report on Croatian Companies”, Milanović Glavan, considers business process maturity which is an extensive version of business process orientation (BPO). Reaching higher stages of maturity means higher levels of process skills for the company while companies are viewed as a mixture of unified business processes. In every maturity level, it is of crucial importance to recognize and improve key turning points, i.e., maturity components that lead companies to the next level. The aim of the research was to provide a report on BPO maturity of Croatian companies and to stress the importance of key turning points. The level of maturity was investigated, the turning points were addressed by using cluster analysis method, and finally the most critical maturity components for each maturity level were determined. The results of the cluster analysis show that companies in Croatia have to improve all key turning points, with a special emphasis on the strategic view. Business Systems Research \| Vol. 11 No. 2 \|2020 p y g p p p g It might be concluded that the high quality and up-to-date challenging topics of the SI of BSR papers would be interesting to both, the scientific and the professional audience, since possible influence on theory and applications are visible. Ljubljana, Zagreb, October 2020 Ljubljana, Zagreb, October 2020 Ljubljana, Zagreb, October 2020 Samo Drobne Ksenija Dumičić Lidija Zadnik Stirn 5 5 Business Systems Research \| Vol. 11 No. 2 \|2020 Business Systems Research \| Vol. 11 No. 2 \|2020 References 1. Cochran, J.J., Cox, L.A., Keskinocak, P., Kharoufeh, J.P., Smith, C. (2011), “Wiley Encyclopedia of Operations Research and Management Science”, Wiley, New Jersey. 1. Cochran, J.J., Cox, L.A., Keskinocak, P., Kharoufeh, J.P., Smith, C. (2011), “Wiley Encyclopedia of Operations Research and Management Science”, Wiley, New Jersey. 2. Figueira, J., Greco, S., Ehrgott, M. (2005),“Multicriteria Decision Analysis”, Springer, New York. 3. Mladenić, D., Lavrač, N., Bohanec. M., Moyle, S. (2003), “Data Mining and Decision Support”, Kluwer, Boston. 4. Rubio, S., & Jiménez-Parra, B. (2014), “Reverse logistics: Overview and challenges for supply chain management”, International Journal of Engineering Business Management, 6, 12. 5 Winston W L (2003) “Operations Research: Applications and Algorithms” Duxbury 4. Rubio, S., & Jiménez-Parra, B. (2014), “Reverse logistics: Overview and challenges for supply chain management”, International Journal of Engineering Business Management, 6, 12. 4. Rubio, S., & Jiménez-Parra, B. (2014), “Reverse logistics: Overview and challenges for supply chain management”, International Journal of Engineering Business Management, 6, 12. 5. Winston, W. L. (2003), “Operations Research: Applications and Algorithms”, Duxbury chain management”, International Journal of Engineering Business Management, 6, 12. 5. Winston, W. L. (2003), “Operations Research: Applications and Algorithms”, Duxbury g g g g 5. Winston, W. L. (2003), “Operations Research: Applications and Algorithms”, Duxbury 6. Zadnik Stirn L., J. Žerovnik, M. Kljajić Borštnar, S. Drobne (2019): “The 14th International Symposium on Operational Research SOR'17, Proceedings”, Bled, Slovenia, September 25-27, 2020. Ljubljana: Slovenian Society Informatika (SSI), Section for Operational Research (SOR). 6. Zadnik Stirn L., J. Žerovnik, M. Kljajić Borštnar, S. Drobne (2019): “The 14th International Symposium on Operational Research SOR'17, Proceedings”, Bled, Slovenia, September 25-27, 2020. Ljubljana: Slovenian Society Informatika (SSI), Section for Operational Research (SOR). Samo Drobne Samo Drobne Samo Drobne Member of Management Board of Slovenian Society INFORMATIKA – Section for Operational Research (SSI-SOR); Secretary of Slovenian Society INFORMATIKA – Section for Operational Research (SSI-SOR); Co-editor of several proceedings of the international symposia on operations research in Slovenia (Proceedings of SOR); Member of Editorial board of Geodetski vestnik – an open access journal of the Association of Surveyors of Slovenia; Member of Editorial board of Journal Communications - Scientific Letters of the University of Žilina. Editor can be contacted at samo.drobne@fgg.uni-lj.si Ksenija Dumičić Ksenija Dumičić Elected Member of International Statistical Institute (ISI) and Member of ISI Sections: International Association of Survey Statisticians (IASS) and International Association of Statistical Education (IASE); Member of: Royal Statistical Society (RSS), American Statistical Association (ASA), American Society for Quality (ASQ) and ASQ Statistics Division, Croatian Operational Research Society (CRORS) and Croatian Biometric Society (HBMD); Chair of Women in Statistics Section at Croatian Statistical Association (CSA); Management Committee Member of ISI Committee on Women in Statistics (CW-ISI); Editorial Board Member of: Croatian Operational Research Review (CRORR), Croatian Review of Economic, Business and Social Statistics (CREBSS) and Proceedings of the Faculty of Economics and Business in Zagreb; Program Committee Member of International Symposium on Operations Research in Slovenia - SOR, Slovenia. Editor can be contacted at kdumicic@net.efzg.hr Lidija Zadnik Stirn President of Slovenian Society INFORMATIKA – Section for Operational Research (SSI- SOR); Vice-president of Slovenian Society INFORMATIKA (SSI); Representative of SSI-SOR in International Federation of Operational Research Societies (IFORS); Representative of SSI-SOR in Association of European Operational Research Societies (EURO); Co-editor of Central European Journal of Operations Research (CEJOR); Co-editor of several proceedings of the international symposia on operations research in Slovenia (Proceedings of SOR); Member of Editorial board of Croatian Operational Research Review (CrORR). Editor can be contacted at lidija.zadnik@bf.uni-lj.si 6
https://openalex.org/W4380681234	https://riis.essnortecvp.pt/index.php/RIIS/article/download/227/192	Portuguese	null	Atitudes dos Estudantes de Enfermagem Perante a Pessoa com Doença Mental	Revista de investigação & inovação em saúde	2,023	cc-by	9,391	Como referenciar: Ponte, D., Tavares, C., Alves, P., Quesado, P.A., Ferreira, A., & Quesado, A. (2023). Atitudes dos estudantes de enfermagem perante a pessoa com doença mental. Revista de Investigação & Inovação em Saúde, 5(1), 35-48. https://doi:10.37914/riis.v6i1.227 ABSTRACT k Keywords: attitude; social stigma; nursing students; mental disorders g g Author contribution: Study conception and design, Data collection, Data analysis and interpretation, Drafting of the article, Critical revision of the article; RN, Enfermeira Especialista em Enfermagem de Saúde Mental e Psiquiatria no Hospital do Divino Espírito Santo de Ponta Delgada, EPER - https://orcid.org/0000-0001-5797-5472 -- Author contribution: Study conception and design, Data collection, Data analysis and interpretation, Drafting of the article, Critical revision of the article; * RN, Enfermeiro Especialista em Enfermagem de Saúde Mental e Psiquiatria no Instituto São João de Deus - Casa de Saúde de São Miguel - https://orcid.org/0000-0002-4713-1409 - Author contribution: Study conception and design, Data collection, Data analysis and interpretation, Drafting of the article, Critical revision of the article; **RN, Pós-graduada em Terapia Assistida por Animais - Agrupamento de Centros de Saúde (ACES) de Entre Douro e Vouga II – Aveiro Norte - https://orcid.org/0000-0002-7890- 251X - Author contribution: Data analysis and interpretation, Drafting of the article, Critical revision of the article; RN, Pós-graduada em Terapia Assistida por Animais - Agrupamento de Centros de Saúde (ACES) de Entre Douro e Vouga II – Aveiro Norte - https://orcid.org/0000-0002-7890- 251X - Author contribution: Data analysis and interpretation, Drafting of the article, Critical revision of the article; *MSc, PhD student em Enfermagem no ICS/UCP-Porto - Professor Adjunto Enfermagem na Escola Superior de Saúde Norte da Cruz Vermelha Portuguesa - https://orcid.org/0000- 0001-5008-3746 - Author contribution: Study conception and design, Data collection, Data analysis and interpretation, Drafting of the article, Critical revision of the article; **** PhD, na Escola Superior de Saúde da Universidade de Aveiro https://orcid.org/0000-0003-2234-4720 - Author contribution: Study conception and design, Data collection, Data analysis and interpretation, Drafting of the article, Critical revision of the article; RESUMO Enquadramento: as atitudes dos estudantes de enfermagem perante a pessoa com doença mental podem influenciar as suas aprendizagens e o desenvolvimento de competências, comprometendo a prestação de cuidados como futuros enfermeiros. Objetivo: conhecer as atitudes dos estudantes de enfermagem perante a pessoa com doença mental e verificar a correlação existente entre a frequência das unidades curriculares de Saúde Mental e Psiquiatria e as suas atitudes perante a pessoa com doença mental. Metodologia: estudo quantitativo, descritivo-correlacional. Amostra de conveniência com 47 estudantes do curso de licenciatura em enfermagem de uma Escola Superior de Saúde da região norte de Portugal. Dados colhidos através de questionário on-line, constituído pelo Attribution Questionaire (AQ-27) (Sousa et al., 2008). Recorreu-se à análise estatística descritiva e inferencial, através do Statistical Package for the Social Sciences (versão 26). Resultados: verificou-se a presença de estigma moderado nas atitudes dos estudantes de enfermagem. Observaram-se diferenças estatisticamente significativas entre o semestre do curso sobre as categorias Irritação [X2 (3) =14,416; P=0,002], Perigosidade [X2 (3) =11,650; P=0,009] e Medo [X2 (3) =12,523; P=0,006] e Pena [F (3,43) = 5,471; P=0,003]. Conclusão: o ensino teórico e prático revelou diminuição de atitudes estigmatizantes, bem como a experiência prévia em Saúde Mental. estigmatizantes, bem como a experiência prévia em Saúde Mental. Palavras-chave: atitude; estigma social; estudantes de enfermagem; doença mental g , p p Palavras-chave: atitude; estigma social; estudantes de enfermagem; doença mental Autor de correspondência: Ana Quesado E-mail: ana.quesado@ua.pt Las actitudes de los estudiantes de enfermería hacia la persona con enfermedad mental Diana Ponte, Cláudia Tavares, Pedro Alves, Paula Alexandra Quesado, António Ferreira*, Ana Quesado**** https://orcid.org/0000-0003-2234-4720 - Author contribution: Study conception and design, Data collection, Data analysis and interpretation, Drafting of the article, Critical revision of the article; ATITUDES DOS ESTUDANTES DE ENFERMAGEM PERANTE A PESSOA COM DOENÇA MENTA Nursing students’ attitudes towards people with mental illness ARTIGO DE INVESTIGAÇÃO RIIS \| vol.6(1), 35-48 ATITUDES DOS ESTUDANTES DE ENFERMAGEM PERANTE A PESSOA COM DOENÇA MENTAL Nursing students’ attitudes towards people with mental illness Las actitudes de los estudiantes de enfermería hacia la persona con enfermedad mental RIIS \| vol.6(1), 35-48 ABSTRACT k Background: the attitudes of nursing students towards people with mental illness can influence their learning and the development of skills, compromising the provision of care as future nurses. Objectives: to know the attitudes of nursing students towards the person with mental illness and to verify the existing correlation between the attendance to the curricular units of Mental Health and Psychiatry and their attitudes towards the person with mental illness. Methodology: quantitative, descriptive-correlational study. Convenience sample with 47 undergraduate nursing students, from a Higher Education Institution in the northern region of Portugal. Data collected through an online questionnaire, consisting of the Attribution Questionaire (AQ-27) (Sousa et al., 2008). Descriptive and inferential statistical analysis was used, using the Statistical Package for the Social Sciences (version 26). Results: the presence of moderate stigma was verified in the attitudes of nursing students. There were statistically significant differences between the semester of the course on the categories Irritation [X2 (3) =14,416; P=0,002], Hazard [X2 (3) =11,650; P=0,009] and Fear [X2 (3) =12,523; P=0,006] and Pena [F (3,43) = 5,471; P=0,003]. Conclusion: theoretical and practical teaching revealed a decrease in stigmatizing attitudes, as well as previous experience in Mental Health. RN, Enfermeira Especialista em Enfermagem de Saúde Mental e Psiquiatria no Hospital do Divino Espírito Santo de Ponta Delgada, EPER -https://orcid.org/0000-0003-3283-7107 - Author contribution: Study conception and design, Data collection, Data analysis and interpretation, Drafting of the article, Critical revision of the article; * RN, Enfermeira Especialista em Enfermagem de Saúde Mental e Psiquiatria no Hospital do Divino Espírito Santo de Ponta Delgada, EPER - https://orcid.org/0000-0001-5797-5472 -- Author contribution: Study conception and design, Data collection, Data analysis and interpretation, Drafting of the article, Critical revision of the article; * RN, Enfermeiro Especialista em Enfermagem de Saúde Mental e Psiquiatria no Instituto São João de Deus - Casa de Saúde de São Miguel - https://orcid.org/0000-0002-4713-1409 - Author contribution: Study conception and design, Data collection, Data analysis and interpretation, Drafting of the article, Critical revision of the article; RN, Pós-graduada em Terapia Assistida por Animais - Agrupamento de Centros de Saúde (ACES) de Entre Douro e Vouga II – Aveiro Norte - https://orcid.org/0000-0002-7890- 251X - Author contribution: Data analysis and interpretation, Drafting of the article, Critical revision of the article; *MSc, PhD student em Enfermagem no ICS/UCP-Porto - Professor Adjunto Enfermagem na Escola Superior de Saúde Norte da Cruz Vermelha Portuguesa - https://orcid.org/0000- 0001-5008-3746 - Author contribution: Study conception and design, Data collection, Data analysis and interpretation, Drafting of the article, Critical revision of the article; **** PhD, na Escola Superior de Saúde da Universidade de Aveiro Recebido para publicação: 25/03/2022 Aceite para publicação: 02/02/2023 INTRODUÇÃO Uma boa saúde mental é reconhecida como um ativo importante e um recurso fundamental para o bem-estar da população e para o desenvolvimento social e económico (Conselho Nacional de Saúde, 2019). Apesar disso, a pessoa com doença mental é vítima de estigma e discriminação. As atitudes negativas que a população em geral apresenta são partilhadas por profissionais de saúde e estudantes do ensino superior, em particular os que frequentam a área da saúde (Corrigan et al., 2011; Happell et al., 2019). Nos profissionais de saúde estas atitudes têm um impacto superior, pois representam uma barreira no acesso aos cuidados de saúde e à perceção e aceitação da doença por parte das pessoas com doença mental (Querido et al., 2020). Embora em escassa quantidade, na literatura verifica-se que as atitudes perante a pessoa com doença mental têm sido alvo de estudo nos estudantes de enfermagem. O impacto das atitudes, a importância de aumentar a literacia em saúde mental e o combate ao estigma são temáticas pertinentes e prioritárias. Neste sentido, desenvolveu-se este estudo de investigação cujos objetivos foram conhecer as atitudes dos estudantes de enfermagem perante a pessoa com doença mental e verificar a correlação existente entre a frequência das unidades curriculares de Saúde Mental e Psiquiatria e as atitudes dos estudantes perante a pessoa com doença mental. Os estudantes de enfermagem, antes de iniciarem o seu percurso académico, trazem consigo, na maioria das vezes, atitudes negativas e carência de conhecimento acerca da doença mental, que poderão influenciar o desenvolvimento das suas competências, com impacto na sua futura prática profissional (Martinho et al., 2014; Querido et al., 2020). Durante o curso os estudantes de enfermagem têm contacto e prestam cuidados a pessoas com doença mental através da realização do Ensino Clínico. Ramos (2021) acrescenta ao referido anteriormente, que cuidar de pessoas com perturbações na área da saúde mental acarreta exigências psicológicas que podem provocar desgaste emocional de quem cuida. Neste contexto complexo, os estudantes durante o ensino clínico nesta área, necessitam de uma supervisão que forneça instrumentos que lhe RESUMEN Tendo o docente/supervisor um papel fundamental na adequação das estratégias de ensino/aprendizagem aos conteúdos e oportunidades das experiências clínicas, e sejam promotoras de momentos de reflexão capazes de desmistificar ideias pré-concebidas neste âmbito (Ramos, 2021). RESUMEN Marco contextual: las actitudes de los estudiantes de enfermería hacia las personas con enfermedad mental pueden influir en su aprendizaje y desarrollo de habilidades, comprometiendo la prestación de cuidados como futuros enfermeros. Objetivos: conocer las actitudes de los estudiantes de enfermería hacia la persona con enfermedad mental y verificar la correlación existente entre la asistencia a las unidades curriculares de Salud Mental y Psiquiatría y sus actitudes hacia la persona con enfermedad mental. Metodología: estudio cuantitativo, descriptivo-correlacional. Muestra de conveniencia con 47 estudiantes de pregrado en enfermería de una Institución de Educación Superior en la región norte de Portugal. Datos recolectados a través de un cuestionario en línea, consistente en el Cuestionario de Atribución (AQ-27) (Sousa et al., 2008). Se utilizó análisis estadístico descriptivo e inferencial, utilizando el Paquete Estadístico para las Ciencias Sociales (versión 26). Resultados: se verificó la presencia de estigma moderado en las actitudes de los estudiantes de enfermería. Se verificó la presencia de estigma moderado. Hubo diferencias estadísticamente significativas entre el semestre de la carrera en las categorías Irritación [X2 (3) =14,416; P=0,002], Riesgo [X2 (3) =11,650; P=0,009] y Miedo [X2(3)=12,523; P=0,006] y Peña [F (3,43) = 5,471; p=0,003]. Conclusión: la enseñanza teórica y práctica reveló una disminución de las actitudes estigmatizantes, así como la experiencia previa en Salud Mental. Autor de correspondência: Ana Quesado E-mail: ana.quesado@ua.pt Palabras Clave: actitudes; estigma social; estudiantes de enfermería; trastornos mentales 35 RIIS Revista de Investigação & Inovação em Saúde https://doi:10.37914/riis.v6i1.227 https://doi:10.37914/riis.v6i1.227 Atitudes dos estudantes de enfermagem perante a pessoa com doença mental l é reconhecida como um ecurso fundamental para o o e para o desenvolvimento nselho Nacional de Saúde, pessoa com doença mental discriminação. As atitudes ção em geral apresenta são nais de saúde e estudantes permitam lidar de forma apropriada com os sentimentos e emoções, como refletir sobre as suas atitudes perante a pessoa com doença mental. Tendo o docente/supervisor um papel fundamental na adequação das estratégias de ensino/aprendizagem aos conteúdos e oportunidades das experiências clínicas, e sejam promotoras de momentos de reflexão capazes de desmistificar ideias pré-concebidas neste âmbito (Ramos, 2021). e enfermagem perante a pessoa com doença mental m o o , l s o s permitam lidar de forma apropriada com os sentimentos e emoções, como refletir sobre as suas atitudes perante a pessoa com doença mental. ENQUADRAMENTO/ FUNDAMENTAÇÃO TEÓRICA Atualmente a sociedade aparenta estar consciente sobre o que deve ser feito, nas dimensões médica, psicológica e social, pelas pessoas com doença mental. Não obstante, permanecem ainda alguns preconceitos, estigmas e atitudes negativas contra estas pessoas (Ramos, 2021). A perspetiva da população em geral aponta para a pessoa portadora de doença mental como sendo perigosa, preguiçosa e imprevisível (Xavier et al., 2013 as cit. 36 RIIS Revista de Investigação & Inovação em Saúde RIIS RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde nfermagem perante a pessoa com doença mental assume-se a necessidade do combate ao estigma como uma preocupação significativa em saúde pública. As atitudes de atitudes negativas perante a pessoa com doença mental são consideradas fatores de previsão de comportamentos discriminatórios relacionados ao estigma (Ferreira, 2018). A escassez ou distorção da informação contribui para a manutenção destas atitudes e crenças irrealistas associadas à doença mental. O aumento de conhecimento acerca da doença mental, para além de contribuir para a redução do estigma, aumenta a literacia. O conceito de literacia em saúde mental, na atualidade, refere-se a: entender como obter e manter a saúde mental positiva; compreender transtornos mentais e os seus tratamentos; diminuir o estigma relacionado a transtornos mentais; melhorar a eficácia da procura de ajuda (saber quando e onde procurar ajuda e desenvolver competências destinadas a melhorar os cuidados de saúde mental e as capacidades de autogestão) (Kutcher et al., 2015 as cit in Loureiro & Freitas, 2020). Atitudes dos estudantes de enfermagem perante a pessoa com doença mental in Querido et al., 2016). Estas atitudes negativas estendem-se a profissionais de saúde, incluindo enfermeiros (Happell et al., 2019). A formação inadequada e a falta de preparação para trabalhar com esta população pode ser considerada uma das causas para as atitudes negativas dos profissionais de saúde (Ferreira, 2018). assume-se a necessidade do combate ao estigma como uma preocupação significativa em saúde pública. As atitudes de atitudes negativas perante a pessoa com doença mental são consideradas fatores de previsão de comportamentos discriminatórios relacionados ao estigma (Ferreira, 2018). A escassez ou distorção da informação Em Portugal, as perturbações psiquiátricas têm uma prevalência de 22,9%, colocando o país num preocupante segundo lugar entre os países europeus (Conselho Nacional de Saúde, 2019). ENQUADRAMENTO/ FUNDAMENTAÇÃO TEÓRICA Embora alguns estudos enfatizam a importância da componente teórica na melhoria das atitudes perante as pessoas com doença mental (Happell et al., 2019), outros demonstraram que as atitudes com base na experiência direta são os determinantes mais importantes (Happell et al., 2019; Jingjing et al., 2013, as cit in Granados- Gámez et al., 2017). Markström et al. (2009) corroboram esta opinião ao concluir que os estudantes revelaram menores níveis de estigmatização após o ensino clínico em saúde mental; vão mais além, ao concluir que a componente prática, pode ter, até certo ponto, um efeito desestigmatizante nas atitudes. A investigação de Querido et al. (2016) revela uma correlação negativa significativa entre os anos de frequência do curso de licenciatura em enfermagem e o estigma. Com recurso à escala AQ- 27, conclui que os estudantes finalistas apresentaram menos atitudes estigmatizantes comparativamente aos iniciados, relacionada, provavelmente, com a educação na Escola de Saúde e com o contacto com as pessoas com doença mental ao longo dos anos. Querido et al. (2020) li t i tâ i d i t õ al., 2010; Querido et al., 2020; São João et al., 2017). al., 2010; Querido et al., 2020; São João et al., 2017). O estudo de Marques et al. (2010) comparou as atitudes face à doença mental dos diferentes estudantes de cursos de saúde, com recurso à escala Attribution Questionnaire (AQ-27) de Corrigan et al. (2003), revelando a presença de estigma acentuado, com predominância de estereótipos de pena, coação, perigosidade, evitamento, medo, segregação e ajuda. As atitudes estigmatizantes diminuíram com o decorrer do curso, aparentemente devido ao contato com pessoas com doença mental. A problemática em questão também tem sido estudada nos estudantes de enfermagem (Querido et al., 2020; Querido et al., 2016). Os estudantes de enfermagem trazem uma série de atitudes pré- concebidas em relação à doença mental, antes de iniciar o seu percurso académico. Numa fase mais inicial partilham de conceções estigmatizantes que poderão influenciar as suas aprendizagens, o desenvolvimento de competências e, futuramente, a prestação de cuidados enquanto enfermeiros (Martinho et al., 2014). O estudo de Wedgeworth et al. (2019) confirma as perceções negativas dos estudantes de enfermagem em relação a pessoas com doença mental, o medo de interação com estes e a preocupação em aprender como comunicar. ENQUADRAMENTO/ FUNDAMENTAÇÃO TEÓRICA A nível nacional, verifica-se a presença de forte estigma e discriminação associado à doença mental, pelo que urge a importância do desenvolvimento de políticas de combate ao estigma e promoção da saúde mental. O estigma em relação a pessoas com problemas de saúde mental consiste em atitudes de desaprovação social com base em certos aspetos pessoais, características, crenças ou comportamentos que estão em conflito com a norma social e cultural (Observatório de Estigma na Doença Mental, 2021; Programa Nacional de Saúde Mental, 2017). A aprendizagem em saúde mental visa não só a aquisição do conhecimento para a atuação nesse campo, mas, de igual modo, a desconstrução de ideias pré-concebidas e a criação de condições para que novos conceitos, conhecimentos e atitudes possam se desenvolver (Mendes, et al., 2018; Thongpriwan et al, 2015). Em Portugal, o estudo das atitudes perante a doença mental também tem sido alvo de investigação no público em geral, nas famílias das pessoas com doença mental, nos estudantes, particularmente em estudantes de enfermagem, bem como nos profissionais de saúde (Ramos, 2021; Marques et Autores como Gronholm et al., (2017) descrevem o estigma como um fenómeno multifacetado, com grave impacto na discriminação experimentada e antecipada, em combinação, e com múltiplas consequências: acesso precário a cuidados de saúde mental e físicos, reduzida expetativa de vida, exclusão do ensino superior e de acesso a emprego, maior risco de contato com o sistema de justiça criminal, vitimização, pobreza e dificuldade no acesso a habitação. Tais consequências foram descritas, por algumas pessoas, como piores que a experiência da doença mental em si. Assim, 37 RIIS Revista de Investigação & Inovação em Saúde RIIS RIIS RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde nfermagem perante a pessoa com doença mental antes de iniciarem o ensino clínico em saúde mental, descrevem uma série de emoções tais como ansiedade, medo, inquietação e rejeição. Estas encontram-se relacionadas com mitos, estereótipos e atitudes negativas e ignorância ou ideias pré-concebidas sobre possíveis episódios de agressividade com que se podem vir a deparar (Estevez et al., 2017, as cit in Querido et al., 2020). ENQUADRAMENTO/ FUNDAMENTAÇÃO TEÓRICA De acordo com os mesmos autores é fundamental lidar com atitudes negativas, sentimentos desagradáveis e ansiedade dos estudantes de enfermagem, relativos à pessoa com doença mental, antes do seu primeiro contacto com estas em contexto clínico, promovendo experiências educacionais que desenvolvam interesse e competência na área da prestação de cuidados em saúde mental. semelhantes à do presente estudo, quer na su versão original, quer na versão portuguesa (te apresentado boas propriedades psicométricas Assim, a colheita de dados foi realizada através d um questionário de autopreenchiment disponibilizado através da plataforma on-lin Microsof Forms®, aplicado durante os meses d maio e julho de 2021. O referido questionário e constituído por duas partes: a primeira co õ i d áfi d ENQUADRAMENTO/ FUNDAMENTAÇÃO TEÓRICA Contudo, estudos apontam que os estudantes de enfermagem que têm uma experiência pessoal de doença mental através de família ou amigos ostentam perspetivas e opiniões de maior aceitação (Granados-Gámez et al., 2017). Frequentemente, os estudantes de enfermagem, 38 RIIS Revista de Investigação & Inovação em Saúde RIIS RIIS RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde nfermagem perante a pessoa com doença mental semelhantes à do presente estudo, quer na sua versão original, quer na versão portuguesa (tem apresentado boas propriedades psicométricas). Assim, a colheita de dados foi realizada através de um questionário de autopreenchimento, disponibilizado através da plataforma on-line Microsof Forms®, aplicado durante os meses de maio e julho de 2021. O referido questionário era constituído por duas partes: a primeira com questões sociodemográficas e a segunda com a versão Portuguesa do Attribution Questionnaire (AQ-27) (Corrigan et al., 2003), traduzido e validado por Sousa et al. (2008). A parte sociodemográfica recolheu dados relativos a: género, idade, estado civil, relação com a Escola Superior de Saúde, ano e semestre do curso, estatuto trabalhador estudante, experiência profissional na área da saúde mental e psiquiatria, se tem familiar ou não familiar com doença mental, relação com a pessoa e regularidade de contacto. O AQ-27 permite a avaliação das atitudes face à pessoa com doença mental, baseando-se em nove estereótipos relacionados com as pessoas com doença mental: Responsabilidade (pessoas com doença mental podem controlar os seus sintomas e são responsáveis por ter a doença), Pena (pessoas com doença mental são dominadas pelo seu próprio transtorno e, portanto, merecem preocupação e pena), Irritação (pessoas com doença mental são culpadas por terem a doença e provocam ira e irritação) Perigosidade (pessoas Atitudes dos estudantes de enfe combate ao estigma. De acordo com os mesmos autores é fundamental lidar com atitudes negativas, sentimentos desagradáveis e ansiedade dos estudantes de enfermagem, relativos à pessoa com doença mental, antes do seu primeiro contacto com estas em contexto clínico, promovendo experiências educacionais que desenvolvam interesse e competência na área da prestação de cuidados em saúde mental. s v a A u d M m c Atitudes dos estudantes de enfermagem perante a pessoa com doença mental combate ao estigma. METODOLOGIA Foi realizado um estudo quantitativo, descritivo, correlacional, transversal. A população selecionada englobou os estudantes de enfermagem de uma Escola Superior de Saúde da região norte e a amostra foi selecionada por um método de amostragem não probabilística de conveniência. Foram colocadas as seguintes hipóteses: H1: Existe correlação entre a frequência das unidades curriculares de Saúde Mental e Psiquiatria e as atitudes dos estudantes de enfermagem perante a pessoa com doença mental; O AQ-27 permite a avaliação das atitudes face à pessoa com doença mental, baseando-se em nove estereótipos relacionados com as pessoas com doença mental: Responsabilidade (pessoas com doença mental podem controlar os seus sintomas e são responsáveis por ter a doença), Pena (pessoas com doença mental são dominadas pelo seu próprio transtorno e, portanto, merecem preocupação e pena), Irritação (pessoas com doença mental são culpadas por terem a doença e provocam ira e irritação), Perigosidade (pessoas com doença mental não estão seguras), Medo (pessoas com doença mental são perigosas), Ajuda (pessoas com doença mental precisam de assistência), Coação (pessoas com doença mental H2: Existe correlação entre a familiaridade e contato próximo com pessoas com doença mental e as atitudes dos estudantes de enfermagem perante a pessoa com doença mental. Considerando a finalidade deste estudo, o Attribition Questionnaire (AQ-27) criado por Corrigan et al. (2003) foi considerado o instrumento mais adequado, visto estar projetado para medir o estigma público em relação à doença mental, já estar traduzido e validado para Portugal, assim como, já ter sido utilizado em populações 39 RIIS RIIS Revista de Investigação & Inovação em Saúde RIIS RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde Atitudes dos estudantes de enfermagem perante a pessoa com doença mental participar na gestão do tratamento), ão (pessoas com doença mental são para instituições localizadas longe da ade) e Evitamento (pessoas com doença ão vivem em sociedade) (Corrigan et al, AQ-27 é composto por uma vinheta sobre ente com esquizofrenia, seguida de 27 es que serão pontuadas numa escala tipo 9 pontos, em que 1 corresponde a ao objetivo deste estudo selecionou-se a vinheta mais neutra e cuja finalidade se direciona apenas à avaliação do estigma, assim a utilizada foi a seguinte: "O José é um homem com 30 anos de idade, solteiro e com esquizofrenia. Às vezes ouve vozes e fica perturbado. METODOLOGIA O José vive sozinho num apartamento e trabalha como estafeta num grande escritório de advogados. Já foi internado seis vezes devido à sua doença". s de enfermagem perante a pessoa com doença mental to), são da nça t al, bre 27 tipo e a ao objetivo deste estudo selecionou-se a vinheta mais neutra e cuja finalidade se direciona apenas à avaliação do estigma, assim a utilizada foi a seguinte: "O José é um homem com 30 anos de idade, solteiro e com esquizofrenia. Às vezes ouve vozes e fica perturbado. O José vive sozinho num apartamento e trabalha como estafeta num grande escritório de advogados. Já foi internado seis vezes devido à sua doença". Atitudes dos estudantes de devem participar na gestão do tratamento), Segregação (pessoas com doença mental são enviadas para instituições localizadas longe da comunidade) e Evitamento (pessoas com doença mental não vivem em sociedade) (Corrigan et al, 2003). O AQ-27 é composto por uma vinheta sobre um paciente com esquizofrenia, seguida de 27 afirmações que serão pontuadas numa escala tipo Likert de 9 pontos, em que 1 corresponde a “nenhum ou nada” e o valor 9 a “muito ou completamente”. Os autores associaram alguns destes construtos a atitudes discriminatórias (Responsabilidade, Perigosidade, Medo, Irritação, Coação, Segregação e Evitamento) e os outros a atitudes de proximidade e assistência (Ajuda e Pena). Cada um destes fatores tem uma pontuação possível de 3 a 27. Os itens nos fatores Ajuda e Evitamento recebem pontuação inversa. Os resultados são calculados considerando os valores médios (não a soma deles) obtidos para os itens que compõem cada um dos fatores (Responsabilidade, Perigosidade, Medo, Irritação, Coação, Segregação, Evitamento, Ajuda e Pena). Quanto maior a pontuação do fator maior a sua contribuição para o estigma. Considera-se um nível de estigma baixo se a pontuação obtida for inferior a 11, moderado em pontuações entre 12 a 19 e estigma elevado se os valores obtidos forem superiores a 20 (Corrigan et al., 2003). Este questionário contém várias vinhetas alternativas, correspondendo a variações nas características da doença mental avaliada, principalmente quanto à id d E i h t lt ti it Após a obtenção da autorização do Conselho de Direção da Escola Superior de Saúde onde foi realizado o estudo, procedeu-se à recolha de dados através do envio de email convite para participação no estudo e com a hiperligação para o questionário online em formato Microsoft Forms®. METODOLOGIA Para as correlações foi usado o teste paramétrico ANOVA, quando existe normalidade na distribuição da variável; quando não estavam cumpridos os pressupostos dos testes paramétricos METODOLOGIA No sentido de garantir a proteção dos dados pessoais foi solicitada colaboração ao Conselho de Direção da Escola Superior de Saúde para a disseminação desse e-mail convite pelos estudantes. A introdução ao questionário integrou o consentimento informado, garantindo o tratamento ético dos participantes, a confidencialidade dos dados pessoais e respetivo anonimato, respeitando a Declaração de Helsínquia. A garantia da participação livre e voluntária foi obtida através da resposta a uma questão de carácter obrigatório, em que só acediam aos questionários os estudantes que respondessem afirmativamente considerando estar informados e aceitar participar no estudo. O estudo teve parecer favorável da Comissão de Ética de uma Escola Superior de Saúde (Parecer nº 026/2020) e encontra-se inscrito na Unidade de Investigação de uma Escola Superior de Saúde da região norte do país. A utilização do AQ-27 Após a obtenção da autorização do Conselho de Direção da Escola Superior de Saúde onde foi realizado o estudo, procedeu-se à recolha de dados através do envio de email convite para participação no estudo e com a hiperligação para o questionário online em formato Microsoft Forms®. Após a obtenção da autorização do Conselho de Direção da Escola Superior de Saúde onde foi realizado o estudo, procedeu-se à recolha de dados através do envio de email convite para participação no estudo e com a hiperligação para o questionário online em formato Microsoft Forms®. No sentido de garantir a proteção dos dados pessoais foi solicitada colaboração ao Conselho de Direção da Escola Superior de Saúde para a disseminação desse e-mail convite pelos estudantes. A introdução ao questionário integrou o consentimento informado, garantindo o tratamento ético dos participantes, a confidencialidade dos dados pessoais e respetivo anonimato, respeitando a Declaração de Helsínquia. A garantia da participação livre e voluntária foi obtida através da resposta a uma questão de carácter obrigatório, em que só acediam aos questionários os estudantes que respondessem afirmativamente considerando estar informados e aceitar participar no estudo. O estudo teve parecer favorável da Comissão de Ética de uma Escola Superior de Saúde (Parecer nº 026/2020) e encontra-se inscrito na Unidade de Investigação de uma Escola Superior de Saúde da região norte do país. METODOLOGIA A utilização do AQ-27 40 RIIS RIIS Revista de Investigação & Inovação em Saúde RIIS RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde Atitudes dos estudantes de enfermagem perante a pessoa com doença mental encontrava-se previamente autorizada pelo autor da tradução e validação para a população portuguesa. Para a análise estatística dos dados utilizou-se o programa estatístico Statistical Package for Social Sciences (SPSS) versão 26, recorreu-se à estatística descritiva e inferencial. A normalidade dos dados foi avaliada pelo teste de Shapiro-Wilk. Para as correlações foi usado o teste paramétrico ANOVA, quando existe normalidade na distribuição da variável; quando não estavam cumpridos os pressupostos dos testes paramétricos (normalidade e homogeneidade) foi realizado o teste não paramétrico Kruskal-Wallis. O nível de significância foi estabelecido em p <0,05. RESULTADOS Obtiveram-se 47 respostas aos questionários, o que correspondeu a uma taxa de resposta de 25,1%. Os dados de caraterização sociodemográfica dos participantes apresentam-se na tabela 1. Tabela 1 Caracterização Sociodemográfica da Amostra Variável Operacionalização % n Género Masculino Feminino 2,1 97,9 1 46 Estado civil Casado Solteiro União de facto 4,3 91,5 4,3 2 43 2 Idade 18 - 22 anos 23 - 27 anos 28 - 32 anos 33 - 37 anos 38 - 42 anos 43 - 47 anos 78,7 8,5 2,1 4,3 4,3 2,1 37 4 1 2 2 1 Semestre letivo 2.º Semestre 4.º Semestre 6.º Semestre 8.º Semestre 38,3 34,0 17,0 10,6 18 16 8 5 Estatuto trabalhador-estudante Não Sim 74,5 21,3 35 10 Em síntese, verificou-se que a maioria dos participantes eram do género feminino (97,9%, amostra verificou-se que 21,3% (n=10) eram trabalhadores-estudantes. Verificou-se que cerca Atitudes dos estudantes de enfermagem perante a pessoa com doença mental mente autorizada pelo autor idação para a população (normalidade e homogeneidade) foi realizado o teste não paramétrico Kruskal-Wallis. O nível de significância foi estabelecido em p <0,05. nfermagem perante a pessoa com doença mental (normalidade e homogeneidade) foi realizado o teste não paramétrico Kruskal-Wallis. O nível de significância foi estabelecido em p <0,05. encontrava-se previamente autorizada pelo autor da tradução e validação para a população portuguesa. Para a análise estatística dos dados utilizou-se o programa estatístico Statistical Package for Social Sciences (SPSS) versão 26, recorreu-se à estatística descritiva e inferencial. A normalidade dos dados foi avaliada pelo teste de Shapiro-Wilk. RESULTADOS Obtiveram-se 47 respostas aos questionários, o que correspondeu a uma taxa de resposta de 25,1%. Os dados de caraterização sociodemográfica dos participantes apresentam-se na tabela 1. Caracterização Sociodemográfica da Amostra Variável Operacionalização % n Género Masculino Feminino 2,1 97,9 1 46 Estado civil Casado Solteiro União de facto 4,3 91,5 4,3 2 43 2 Idade 18 - 22 anos 23 - 27 anos 28 - 32 anos 33 - 37 anos 38 - 42 anos 43 - 47 anos 78,7 8,5 2,1 4,3 4,3 2,1 37 4 1 2 2 1 Semestre letivo 2.º Semestre 4.º Semestre 6.º Semestre 8.º Semestre 38,3 34,0 17,0 10,6 18 16 8 5 Estatuto trabalhador-estudante Não Sim 74,5 21,3 35 10 amostra verificou-se que 21,3% (n=10) eram trabalhadores-estudantes. Verificou-se que cerca de 12,8% (n=6) dos participantes já tinham exercido funções na Área da Saúde Mental e Psiquiatria. Relativamente à questão se tem algum familiar com doença mental, 42,6% (n=20) respondeu afirmativamente, sendo o grau de parentesco com mais relevância o pai ou mãe Em síntese, verificou-se que a maioria dos participantes eram do género feminino (97,9%, n=46) e 78,7% (n=37) eram solteiros. Em relação à idade, 72,4% (n=34) dos inquiridos apresentaram idade entre os 18 e 22 anos, sendo a idade média de 22 anos (DP=6,56). A maior parte dos participantes encontrava-se a frequentar o segundo semestre (38,3%, n=14). Da totalidade da 41 RIIS Revista de Investigação & Inovação em Saúde RIIS RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde Atitudes dos estudantes de enfermagem perante a pessoa com doença mental (10,6%, n=5). No que diz respeito a “outra” relação familiar destaca-se o parentesco com a avó (4,3%, n=2) e a tia (6,4%, n=3). Dos que responderam ter algum familiar com doença mental, a maior percentagem (17,0%, n=8) referiu contactar que a relação com essa pessoa não familiar era de vizinho, seguida de amigo (17,0%, n=8), em que a maior percentagem (31,9%, n=15) referiu contactar raramente com essa pessoa e 10,6% (n=5) referiu semanalmente. O AQ-27 avalia 9 dimensões das atitudes estigmatizantes, cujos valores globais revelaram a presença de um estigma moderado, apresentados na tabela 2. RESULTADOS O teste de comparações post- hoc de Teste de Tukey mostrou que a dimensão Pena dos estudantes do 2.º semestre era diferente do 8.º semestre, assim como, entre os estudantes Relacionando os níveis de estigma com o semestre de curso verificou-se que os estudantes numa fase 42 RIIS Revista de Investigação & Inovação em Saúde RIIS RIIS RIIS Revista de Investigação & Inovação em Saúde Atitudes dos estudantes de enfermagem perante a pessoa com doença mental do 4.º semestre e do 8.º semestre. desenvolvimento das suas competências específicas para a prestação de cuidados de Enfermagem de Saúde Mental e Psiquiatria. Para além disso, é uma das áreas menos preferidas por esses estudantes como possibilidade de carreira profissional (Mendes et al., 2018). do 4.º semestre e do 8.º semestre. desenvolvimento das suas competências específicas para a prestação de cuidados de Enfermagem de Saúde Mental e Psiquiatria. Para além disso, é uma das áreas menos preferidas por esses estudantes como possibilidade de carreira profissional (Mendes et al., 2018). Relativamente aos estudantes com experiencia prévia na área da saúde mental e psiquiatria, o teste de Kruskal-Wallis mostrou que há efeito do grupo sobre as dimensões: Irritação [X2 (1)=1,974; p=0,160], Perigosidade [X2 (1)=1,370; p=0,242], Medo [X2 (1)=1,967; p=0,161], Ajuda [X2 (1)=0,189; p=0,664], Coação [X2 (1)=0,189; p=0,664], Segregação [X2 (1)=0,597; p=0,440], Evitamento [X2 (1)=0,388; p=0,533]. Verificando- se uma redução significativa do estigma nestas dimensões, em relação aos que nunca trabalharam na área. A ANOVA mostrou que as dimensões Responsabilidade e Pena não sofreram efeito neste grupo. Na avaliação inicial verificamos a presença de estigma moderado por parte dos estudantes de enfermagem, em que a Ajuda e a Coação foram as atitudes negativas com maior relevância. Também a Pena e o Evitamento foram reveladores de estigma, o que é corroborado pelo estudo de Granados-Gamez et al. (2017) que utilizaram a mesma escala e obtiveram resultados idênticos. A dimensão Pena, influenciada e relacionada com o cariz protetor, despoleta nos estudantes de enfermagem atitudes de simpatia e paternalismo, apoiada na ideia de que pessoas com doença mental são dominadas pela sua patologia, pelo que carecem de preocupação e pena (Querido et al., 2016). Este fato pode espelhar a baixa literacia em saúde mental da amostra, uma vez que esta atitude negativa é sustentada na literatura pela falta de informação. RESULTADOS O teste de Kruskal-Wallis mostrou que há efeito do grupo que tem familiar com doença mental na dimensão Irritação [X2 (1)=4,526; p=0,033] e que não existe efeito da regularidade de contacto em nenhuma dimensão. O mesmo teste mostrou não existir efeito de grupo que conhece alguém, não familiar, com doença mental em qualquer das dimensões. Todavia, na análise dos dados obtidos, observa-se diferenças entre os níveis de estigma dos estudantes do 2.º ao 8.º semestre, o que poderá atestar a diminuição das atitudes estigmatizantes ao longo do curso. Verifica-se também uma redução na Pena entre o 2.º semestre e o 8.º semestre, bem como entre o 4.º e o 8.º semestre, o que pode ser explicado quer pela passagem pela teoria quer pelo ensino clínico na área da saúde mental, uma vez que, neste estabelecimento de ensino, a unidade curricular de saúde mental e RESULTADOS Por fim, à questão se conhece alguém, não familiar, com doença mental, 68,1% (n=32) respondeu afirmativamente, a maioria (31,9%, n=15) referiu Média e Desvio Padrão dos Estudantes de Enfermagem no AQ-27 Média Desvio Padrão Mínimo Máximo Responsabilidade 7,1277 0,40535 3 13 Pena 17,4681 0,74616 5 27 Irritação 6,8936 0,55522 3 18 Perigosidade 8,1489 0,68415 3 23 Medo 7,9149 0,70732 3 23 Ajuda 23,3191 0,45835 16 27 Coação 23,3191 0,45835 16 27 Segregação 7,7447 0,68853 3 25 Evitamento 18,8511 0,83015 8 27 TOTAL 13,4208 0,614842 Média e Desvio Padrão dos Estudantes de Enfermagem no AQ-27 inicial apresentavam valores superiores de estigma em todas as categorias. O teste de Kruskal-Wallis mostrou que há efeito do semestre do curso sobre as categorias Irritação [X2 (3)=14,416; P=0,002], Perigosidade [X2 (3)=11,650; P=0,009] e Medo [X2 (3)=12,523; P=0,006]. A comparações post hoc mostraram diferenças significativas nas categorias referidas entre o 4.º e o 6.º semestre. Na análise ao conjunto dos estudantes, ilustrada na Na análise ao conjunto dos estudantes, ilustrada na tabela 2, verificou-se um menor estigma na categoria Irritação e maior estigma relativamente à Ajuda e Coação. Foram também observados níveis elevados de estigma na Pena e Evitamento, com valores mínimos de 5 e 8, respetivamente e máximos de 27. Na análise ao conjunto dos estudantes, ilustrada na tabela 2, verificou-se um menor estigma na categoria Irritação e maior estigma relativamente à Ajuda e Coação. Foram também observados níveis elevados de estigma na Pena e Evitamento, com valores mínimos de 5 e 8, respetivamente e máximos de 27. De acordo com os valores do teste de normalidade de Shapiro-Wilk (p<0,05) verificou-se que não existia normalidade na distribuição das categorias Irritação, Perigosidade, Medo, Ajuda, Coação, Segregação e Evitamento, pelo que se procedeu à realização do teste de correlação não paramétrico Kruskal-Wallis. Para as dimensões Pena e Responsabilidade, pela aplicação do teste de correlação paramétrico ANOVA, verificou-se existir efeito de semestre na Pena [F (3,43)=5,471; P=0,003] mas não na Responsabilidade. Atitudes dos estudantes de enfermagem perante a pessoa com doença mental doença mental, mas, por outro, a permanência de estereótipos negativos, como o contágio, a imprevisibilidade do comportamento, a incompetência e a infantilidade, conduzindo ao distanciamento social destes indivíduos. Querido et al. (2020), que no seu estudo verificaram um aumento de estereótipo relativos à Ajuda e Coação, vão mais além e referem que tal pode indicar um sentimento de pena e de controlabilidade, influenciada pelos contextos clínicos, ainda estruturados nos modelos mais tradicionais e o desempenho do papel de supervisores de estudantes. A Coação foi outra dimensão que nos parece pesar nas atitudes negativas dos estudantes e que não sofreu influência dos semestres do curso. Acreditamos que tal facto pode dever-se à pressão dos estudantes de enfermagem sobre as pessoas com doença mental, no incentivo constante e permanente à adesão ao tratamento, pois, tal como referem Querido et al. (2016), é dado ênfase à importância da adesão ao tratamento ao longo do curso de enfermagem, como medida de obtenção de ganhos em saúde. Outro fator que poderá estar a influenciar os resultados nos domínios da Ajuda, Coação, Segregação e Evitamento seria a experiência que os estudantes tiveram nos contextos clínicos, as atitudes que observaram nos enfermeiros perante a pessoa com doença mental. Pode ainda ser reflexo do modelo hospitalocêntrico do cuidado à pessoa com doença mental e do próprio ensino de enfermagem. Poderá ser outro fator influenciador, o papel que os supervisores clínicos desempenharam no desenvolvimento do pensamento crítico-reflexivo psiquiatria é ministrada no 5.º semestre e o ensino clínico no 6.º semestre. Ao analisarmos a influência do ensino clínico nas atitudes dos estudantes de enfermagem, verifica- se um decréscimo com predominância nos domínios Pena, Irritação, Perigosidade e Medo, resultados semelhantes aos encontrados nos estudos de Bingham & O’Brien (2017) e de Martinez-Martinez et al. (2019), com a aplicação da AQ-27 antes e após uma experiência educativa e clínica. De facto, segundo Querido et al. (2016), os estudantes finalistas expressam menos Pena, Perigosidade, Medo e Evitamento, aparentemente devido ao contato com pessoas com doença mental. Estes resultados parecem-nos indicar que o ensino clínico contribui para uma perspetiva mais positiva, o que é corroborado por vários estudos, na medida em que o ambiente prático e a formação promovem uma mudança nas atitudes dos estudantes, afetando positivamente e de forma significativa as suas perceções e atitudes face à pessoa com doença mental (Querido et al., 2020; Querido et al., 2016). DISCUSSÃO A amostra recolhida é limitada no que se refere à representatividade da população de estudantes de enfermagem, contudo as características sociodemográficas, à semelhança de outros estudos, revelaram um predomínio de indivíduos do sexo feminino, jovens adultos e solteiros. Os próprios estudantes de enfermagem compartilham estereótipos e conceções estigmatizantes sobre a doença mental que poderão influenciar as suas aprendizagens e o 43 RIIS RIIS RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde Atitudes dos estudantes de enfermagem perante a pessoa com doença mental Atitudes dos estudantes de enfermagem perante a pessoa com doença mental Atitudes dos estudantes de enfermagem perante a pessoa com doença mental e na desconstrução das ideias pré-concebidas dos estudantes no que se refere à assistência à pessoa com doença mental, a sua participação na gestão do tratamento, a necessidade de institucionalização destas pessoas e a sua integração na sociedade (Ramos, 2021) . e na desconstrução das ideias pré-concebidas dos estudantes no que se refere à assistência à pessoa com doença mental, a sua participação na gestão do tratamento, a necessidade de institucionalização destas pessoas e a sua integração na sociedade (Ramos, 2021) . experienciado na prática com pessoas com doença mental (Bingham & O'Brien, 2017). Granados-Gamez et al. (2017) desenvolveram um estudo semelhante, também com recurso ao AQ- 27, que contrapõem as hipóteses por nós colocadas, uma vez que revelou não existir redução de estigma em nenhum fator após o contato quer com a teoria quer com a prática clínica. integração na sociedade (Ramos, 2021) . À semelhança do que acontece na prática, o ensino teórico na unidade curricular de Enfermagem de Saúde Mental e Psiquiátrica também parece influenciar as atitudes negativas dos estudantes. Se analisarmos o programa do plano de estudos praticado na escola onde decorreu o presente estudo, deparamo-nos com conteúdos programáticos específicos relativos à doença mental, como a exposição dos atuais problemas e necessidades em saúde mental, facilitando a redução do estigma, o que poderá ter contribuído para uma redução das atitudes de irritação, perigosidade, medo e pena. Para além disso, a alternância entre períodos de ensino na escola com aulas teóricas, teórico-práticas e práticas laboratoriais e períodos de ensino clínico constituem momentos privilegiados para o desenvolvimento de aprendizagens, consolidação de conhecimentos e reflexão sobre as práticas. Em concordância com a bibliografia, estudos demonstraram que os alunos tendem a ter atitudes mais favoráveis em relação à pessoa com doença mental quando têm mais horas de preparação teórica nessa área científica (Mendes et al., 2018). Contudo, a educação formal por si só pode não ser suficiente para conduzir uma mudança nas atitudes, pelo que a associação com a prática clínica potencia o desenvolvimento de À semelhança do que acontece na prática, o ensino teórico na unidade curricular de Enfermagem de Saúde Mental e Psiquiátrica também parece influenciar as atitudes negativas dos estudantes. Relativamente aos resultados das atitudes estigmatizantes em função do nível de contato com familiar com doença mental, verifica-se uma redução apenas no fator Irritação. Atitudes dos estudantes de enfermagem perante a pessoa com doença mental Particularmente, autores como Markström et al. (2009) concluem que, após a realização da prática clínica, os estudantes consideram as pessoas com doença mental menos perigosas do que julgavam ser. Contudo, atitudes como a Ajuda, Coação, Segregação e Evitamento não sofreram alterações ao longo dos semestres, o que nos permite afirmar que o ensino teórico e prático não teve qualquer influência nestas atitudes negativas. Este fato poderá indicar, por um lado, atitudes de proximidade e assistência para com a pessoa com 44 RIIS RIIS RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde Atitudes dos estudantes de enfermagem perante a pessoa com doença mental Atitudes dos estudantes de enfermagem perante a pessoa com doença mental estigma na maioria das dimensões, com exceção da Responsabilidade e Pena. Tal pode ser explicado pelo fato do contato em termos práticos surtir um efeito passível de mudar crenças e atitudes perante a pessoa com doença mental, conduzindo a uma melhor disposição para cuidar dessa população (Granados-Gamez et al., 2017). O estudo de Thongpriwan et al. (2015) aborda a questão da experiência prévia em saúde mental e revela que os estudantes que têm esta experiência apresentam melhor preparação e menor ansiedade em relação à doença mental, mas contrapõem os nossos resultados ao afirmar que não existem diferenças significativas no que respeita a estereótipos negativos em relação aos que não tiveram experiência prévia. O estudo levado a cabo por Martinez-Martinez et al. (2019) também contrapõe os nossos resultados, afirmando não existir influencia desta variável na redução do estigma. O desenvolvimento de estratégias de redução de estigma numa fase precoce do curso de enfermagem, poderia impulsionar o autoconhecimento e a autoavaliação das crenças e atitudes dos estudantes de enfermagem, incentivando e proporcionando espaços de reflexão, com partilha de sentimentos e ideias. A integração de estudantes em programas de luta contra o estigma e envolvimento ativo em processos de mudança de atitudes, recentemente estimulados com a criação do Observatório de Estigma na Doença Mental, permitiria obter precocemente conhecimentos sobre a temática, bem como a proximidade com pessoas e grupos com doença mental em espaços abertos. Os resultados obtidos podem também refletir a necessidade de mudança do paradigma conceptual das unidades curriculares de Saúde Mental contextualizado com uma prática de luta contra o estigma e discriminação, assim como de mudança de representações e atitudes face à doença mental, que espelhe a centralidade na pessoa com doença mental e promova a redução de atitudes negativas, como a ajuda, coação, segregação e evitamento, como observado no nosso estudo. Para além disso, a influência que o enfermeiro supervisor exerce sobre os estudantes é fundamental na modificação de atitudes discriminatórias e estereótipos que os estudantes possam apresentar, uma vez que estes internalizam os valores e comportamentos a que estão expostos. Assim, acreditamos que a reflexão individual de cada enfermeiro possa influenciar os estudantes em contexto de ensino clínico. Ao longo do estudo verificou-se que o número Atitudes dos estudantes de enfermagem perante a pessoa com doença mental A literatura aponta que a familiaridade contribui, na generalidade, para atitudes mais positivas, em particular no que respeita à Pena, encontrando-se negativamente relacionada com sentimentos de raiva e medo (Corrigan et al., 2003). Os mesmos autores defendem que a familiaridade predispõe ainda a atitudes de Ajuda e Evitamento para com a pessoa com doença mental, diminuindo o nível de estigma quando a familiaridade aumenta. Inversamente ao verificado na familiaridade, o contato com pessoas não familiares com doença mental não produz efeito no grupo em nenhuma dimensão, ao contrário do apontado em vários estudos (Granados-Gamez et al., 2017) e colocado por nós como hipótese. Resultados semelhantes foram obtidos no estudo de Ferreira (2018), ao apontar que a convivência com pessoas com doença mental por si só não parece constituir condição suficiente para produzir alterações nas atitudes. Os estudantes com experiência profissional prévia em saúde mental apresentaram redução do 45 RIIS RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde Atitudes dos estudantes de enfermagem perante a pessoa com doença mental reduzido da amostra se revelou uma limitação, na medida em que poderá ter influenciado os resultados obtidos, admitindo também a possibilidade de erro estatístico. Para além disso, a distribuição não uniforme da amostra pelos vários semestres poderá ter condicionado os resultados. Licenciatura em Enfermagem, com a respetiva avaliação de impacte. Licenciatura em Enfermagem, com a respetiva avaliação de impacte. reduzido da amostra se revelou uma limitação, na medida em que poderá ter influenciado os resultados obtidos, admitindo também a possibilidade de erro estatístico. Para além disso, a distribuição não uniforme da amostra pelos vários semestres poderá ter condicionado os resultados. q p resultados obtidos, admitindo também a possibilidade de erro estatístico. Para além disso, a distribuição não uniforme da amostra pelos vários semestres poderá ter condicionado os resultados. Acreditamos também que o preenchimento dos questionários on-line pode ter constituído uma limitação, na medida em que pode ter condicionado a própria motivação dos estudantes no seu preenchimento e, assim, limitado o tamanho da amostra. Os resultados desta investigação demonstram que apesar dos estudantes de enfermagem terem contacto com a área da Saúde Mental e proximidade com pessoas com doença mental em Ensino Clínico, só por si não é condição suficiente para a redução de estigma, podendo este influenciar o seu desempenho na prestação de cuidados. Considerando-se importante a implementação de estratégias de luta contra o estigma desde o início do curso de Enfermagem, no sentido de promover o desenvolvimento pessoal dos estudantes, reduzindo os preconceitos e receios sobre estas pessoas, aumentando a sua segurança e motivação na prestação de cuidados. Desta forma, sugere-se a continuidade de estudos sobre esta temática com inclusão de amostras mais alargadas e diversificadas, abranjam outros estabelecimentos de ensino, bem como outros cursos de saúde, a fim de se efetuar a comparação entre os estudantes das diferentes áreas da saúde, assim como, o desenvolvimento de projetos de luta contra o estigma e descriminação com integração curricular nos cursos da área da saúde, principalmente na https://doi.org/10.2307/1519806 Corrigan, P., Roe, D. & Tsnag, W. (2011). Challenging the Stigma of Mental Illness: Lessons or Therapists and Advocates. Oxford, England: Wiley-Blackwell. Ferreira, M. (2018). Influência do ensino de saúde mental na modificação de atitudes estigmatizantes de alunos técnicos de enfermagem. [Masters dissertation, Escola Politécnica de Saúde Joaquim Venâncio]. Repositório Institucional da Escola Politécnica de Saúde Joaquim Venâncio, Fundação Oswaldo Cruz. Ferreira, M. (2018). Influência do ensino de saúde mental na modificação de atitudes estigmatizantes de alunos técnicos de enfermagem. [Masters dissertation, Escola Politécnica de Saúde Joaquim Venâncio]. Repositório Institucional da Escola Politécnica de Saúde Joaquim Venâncio, Fundação Oswaldo Cruz. REFERÊNCIAS BIBLIOGRÁFICAS Bingham, H. & O’Brien, A. (2017). Educational intervention to decrease stigmatizing attitudes of undergraduate nurses towards people with mental illness. International Journal of Mental Health Nursing. 27(1), 311-319. https://doi.org/10.1111/inm.12322 https://doi.org/10.1111/inm.12322 Conselho Nacional de Saúde (2019). Sem mais tempo a perder – Saúde mental em Portugal: um desafio para a próxima década. Lisboa: CNS. https://www.cns.min-saude.pt/wp- content/uploads/2019/12/SEM-MAIS-TEMPO-A- PERDER.pdf Corrigan, P., Markowitz, F. E., Watson, A., Rowan, D., & Kubiak, M. A. (2003). An Attribution Model of Public Discrimination Towards Persons with Mental Illness. Journal of Health and Social Behavior, 44(2), 162–179. CONCLUSÃO Apesar das limitações do estudo foi possível conhecer as atitudes dos estudantes de enfermagem perante a pessoa com doença mental, assim como, a correlação existente entre a frequência das unidades curriculares de Saúde Mental e Psiquiatria e as suas atitudes perante a pessoa com doença mental. Assim, da análise dos resultados conclui-se que a influência de conhecimentos teóricos e práticos na área da saúde mental e psiquiatria foi a que se revelou mais significativa na redução das atitudes estigmatizantes dos estudantes. Ao longo do estudo verificou-se que o número Ao longo do estudo verificou-se que o número 46 RIIS Revista de Investigação & Inovação em Saúde RIIS RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde Atitudes dos estudantes de enfermagem perante a pessoa com doença mental Programa Nacional de Saúde Mental (2017) Lisboa: Direção-Geral da Saúde. https://nocs.pt/wp- content/uploads/2017/11/DGS_PNSM_2017.10.0 9_v2.pdf Loureiro, L., & Freitas, P. (2020). Literacia em saúde mental dos jovens estudantes de enfermagem na integração ao ensino superior. Revista Portuguesa de Enfermagem de Saúde Mental (24), 34-42. https://doi.org/10.19131/rpesm.0279 Loureiro, L., & Freitas, P. (2020). Literacia em saúde mental dos jovens estudantes de enfermagem na integração ao ensino superior. Revista Portuguesa de Enfermagem de Saúde Mental (24), 34-42. https://doi.org/10.19131/rpesm.0279 Querido, A., Tomás, C, Carvalho, D., Gomes, J. & Cordeiro, M. (2020). Impacto de uma intervenção no estigma em saúde mental e ansiedade intergrupal. Acta Paulista de Enfermagem. 33, 1-9. http://dx.doi.org/10.37689/actaape/2020AO0226 Markström, U., Gyllensten, A. L., Bejerholm, U., Björkman, T., Brunt, D., Hansson, L., Leufstadius, C., Sandlund, M., Svensson, B., Ostman, M., & Eklund, M. (2009). Attitudes towards mental illness among health care students at Swedish universities--a follow-up study after completed clinical placement. Nurse Education Today, 29(6), 660– 665. https://doi.org/10.1016/j.nedt.2009.02.006 Querido, A., Tomás, C. & Carvalho, D. (2016). O Estigma face à doença mental nos estudantes de saúde. Revista Portuguesa de Enfermagem de Saúde Mental, Especial 3, 67-72. http://dx.doi.org/10.19131/rpesm.0120 Ramos, A. A. M. L. (2021). O ensino clínico de enfermagem de saúde mental e Psiquiátrica: proposta de uma matriz para o acompanhamento dos estudantes do curso de Licenciatura. [Doctoral dissertation, Universidade Católica Portuguesa]. Repositório Institucional da Universidade Católica Portuguesa. Marques, A., Barbosa, T., & Queirós, C. (2010, June 17-19). O Estigma na Doença Mental Perspectivado por Futuros Profissionais de Saúde Mental. [Poster presentation]. III Congreso de la Federación Española de Asociaciones de Rehabilitación Psicosocial, Valladolid. https://repositorio- aberto.up.pt/handle/10216/44483 Marques, A., Barbosa, T., & Queirós, C. (2010, June 17-19). O Estigma na Doença Mental Perspectivado por Futuros Profissionais de Saúde Mental. [Poster presentation]. III Congreso de la Federación Española de Asociaciones de Rehabilitación Psicosocial, Valladolid. https://repositorio- aberto.up.pt/handle/10216/44483 https://www.arca.fiocruz.br/bitstream/handle/ici ct/26956/Marcela_Ferreira_EPSJV_Mestrado_201 8.pdf?sequence=2&isAllowed=y Granados-Gámez, G., Rodríguez, M., Granados, A. & Márquez-Hernández, V. (2017). Attitudes and Beliefs of Nursing Students Toward Mental Disorder: The Significance of Direct Experience With Patients. Perspect Psychiatr Care, 53(2),135- 143. https://doi.org/10.1111/ppc.12147 Gronholm, P., Henderson, C., Deb, T. & Thornicroft, G. (2017). Interventions to reduce discrimination and stigma: The state of the art. Social Psychiatry and Psychiatric Epidemiology, 52, 249– 258. https://doi.org/10.1007/s00127-017-1341-9 47 RIIS RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde 82. http://pepsic.bvsalud.org/pdf/smad/v14n2/en_03 .pdf Happell, B., Platania-Phung, C., Scholz, B., Bocking, J., Horgan, A., Manning, F., Doody, R., Hals, E. Granerud, A., Lahti, M., Pullo, J., Vatula, A., Ellilä, H., Vaart, K., Allon, J., Griffin, M., Russell, S., MacGabhann, L., Bjornsson, E. & Biering, P. (2019) Nursing student attitudes to people labelled with ‘mental illness’ and consumer participation: A survey-based analysis of findings and psychometric properties. Nurse Education Today, 76, 89–95. https://doi.org/10.1016/j.nedt.2019.02.003 Happell, B., Platania-Phung, C., Scholz, B., Bocking, J., Horgan, A., Manning, F., Doody, R., Hals, E. Granerud, A., Lahti, M., Pullo, J., Vatula, A., Ellilä, H., Vaart, K., Allon, J., Griffin, M., Russell, S., MacGabhann, L., Bjornsson, E. & Biering, P. (2019) Nursing student attitudes to people labelled with ‘mental illness’ and consumer participation: A survey-based analysis of findings and psychometric properties. Nurse Education Today, 76, 89–95. https://doi.org/10.1016/j.nedt.2019.02.003 Observatório de Estigma na Doença Mental (2021). O Observatório. https://labrp.pt/observatorio/index.php Programa Nacional de Saúde Mental (2017) Lisboa: Direção-Geral da Saúde. https://nocs.pt/wp- content/uploads/2017/11/DGS_PNSM_2017.10.0 9_v2.pdf aberto.up.pt/handle/10216/44483 São João, R., Coelho, T., Ferreira, C., Castelo, A., & Massano, M. (2017). Estigma na Doença Mental: Estudo Observacional e Piloto em Portugal. Revista da UIIPS, 5 (2), 171-185. Martinez-Martinez, C., Sanchez-Martinez, V., Sales-Orts, R., Dinca, A., Richart-Martinez, M. & Ramos-Pichardo, J. (2019). Effectiveness of direct contact intervention with people with mental illness to reduce stigma in nursing students. International Journal of Mental Health Nursing 28 (3), 735-743. https://doi.org/10.1111/inm.12578 Sousa S., Queirós C., Marques A., Rocha N., & Fernandes A. (2008). Versão preliminar portuguesa do Attribution Questionnaire (AQ-27). [adaptada com autorização de P. Corrigan]. Porto: Faculdade de Psicologia e Ciências de Educação da Universidade do Porto / ESTSPIPP Martinho, J., Pires, R., Carvalho, J., & Pimenta, G. (2014). Formação e Desenvolvimento de Competências de Estudantes de Enfermagem em contexto de ensino clínico em saúde mental e psiquiatria. Revista Portuguesa de Enfermagem de Saúde Mental, (Ed. Esp. 1) 97-102. https://comum.rcaap.pt/bitstream/10400.26/363 52/1/Júlia%20M.-09.pdf Martinho, J., Pires, R., Carvalho, J., & Pimenta, G. (2014). Formação e Desenvolvimento de Competências de Estudantes de Enfermagem em contexto de ensino clínico em saúde mental e psiquiatria. Revista Portuguesa de Enfermagem de Saúde Mental, (Ed. Esp. 1) 97-102. https://comum.rcaap.pt/bitstream/10400.26/363 52/1/Júlia%20M.-09.pdf Thongpriwan, V., Leuck, S., Powell, R., Young, S., Schuler, S. & Hughes, R. (2015) Undergraduate Nursing Students: Attitudes Toward Mental Health Nursing. Nurse Education Today, 35 (8), 948-953. Thongpriwan, V., Leuck, S., Powell, R., Young, S., Schuler, S. & Hughes, R. (2015) Undergraduate Nursing Students: Attitudes Toward Mental Health Nursing. Nurse Education Today, 35 (8), 948-953. http://dx.doi.org/10.1016/j.nedt.2015.03.011 Wedgeworth, M., Ford, C. & Tice, J. (2019). “I’m scared”: Journaling Uncovers Student Perceptions Prior to a Psychiatric Clinical Rotation. Journal of the American Psychiatric Nurses Association. 26(5). 1-7. https://doi.org/10.1177/1078390319844002 Mendes, A., Marques, M., Monteiro, A., Barroso, T. & Quaresma, M. (2018). Educação em enfermagem de saúde mental e psiquiatria no curso de licenciatura em enfermagem. SMAD. Revista eletrónica saúde mental álcool e drogas, 14 (2), 73- 48 RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde RIIS Revista de Investigação & Inovação em Saúde
https://openalex.org/W2139126860	https://aacr.figshare.com/articles/journal_contribution/Supplementary_Figure_Legends_1-6_from_i_De_novo_i_Lipogenesis_Protects_Cancer_Cells_from_Free_Radicals_and_Chemotherapeutics_by_Promoting_Membrane_Lipid_Saturation/22385486/1/files/39830930.pdf	English	null	<i>De novo</i> Lipogenesis Protects Cancer Cells from Free Radicals and Chemotherapeutics by Promoting Membrane Lipid Saturation	Cancer research	2,010	cc-by	797	Supplementary Figure Legends for Supplementary Figure Legends for De novo lipogenesis protects cancer cells from free radicals and chemotherapeutics by promoting membrane lipid saturation Evelien Rysman, Koen Brusselmans, Katryn Scheys, Leen Timmermans, Rita Derua, Sebastian Munck, Paul P. Van Veldhoven, David Waltregny, Veerle W. Daniels, Jelle Machiels, Frank Vanderhoydonc, Karine Smans, Etienne Waelkens, Guido Verhoeven, and Johannes V. Swinnen§ These authors contributed equally to the work. § To whom correspondence should be addressed. E-mail: J h S i @ d k l b * These authors contributed equally to the work. § To whom correspondence should be addressed. E-mail: § To whom correspondence should be addressed. E-mail: Johan.Swinnen@med.kuleuven.be phosphatidylethanolamine (PE) and phosphatidylserine (PS) species LNCaP cells were treated with soraphen (100 nM) or vehicle (control) for 72 hr. Lipid extracts were prepared and subjected to ESI-MS/MS. Phosphatidylcholine (PC), phosphatidylethanolamine (PE) and phosphatidylserine (PS) were detected by scanning for phosphocholine (m/z 184) in the positive ion mode, the neutral loss of phosphoethanolamine (m/z 141) in the positive ion mode and the neutral loss of serine (m/z 87) in the negative ion mode, respectively. The m/z of major species and their species assignment are indicated. Std refers to the lipid standards. The relative changes in the cellular content of major phospholipid species in LNCaP cells in response to soraphen treatment were recorded in the MRM mode. Lipid profiling was performed in three pairs of samples (1-3). Changes are expressed relative to the control and are color coded as indicated by the scale bar. The heat maps show the fold change of the soraphen/control ratio (expressed as log2) for different PC, PE or PS species. Figure S1. Effect of soraphen on intact phosphatidylcholine (PC), phosphatidylethanolamine (PE) and phosphatidylserine (PS) species 1 1 Figure S3. Effect of lipogenesis on the cell’s sensitivity to oxidative stress- induced cell death Figure S3. Effect of lipogenesis on the cell’s sensitivity to oxidative stress- induced cell death PC-3, BT474 and HCT116 cells were treated with soraphen (100 nM) or vehicle (control) for 72 hr. During the last 24 hr, cells were exposed to 300 µM (PC-3), 200 µM (BT474) or 400 µM (HCT116) H2O2. Cells were collected and stained with trypan blue to assess the cell viability. Data represent means ± SE (n=3). Significantly different (p<0.05) from control without H2O2 exposure. #Significantly different (p<0.05) from both soraphen and H2O2 alone. Figure S2. RNA interference of FASN in LNCaP cells and of ACC in HCT116 cells LNCaP and HCT116 cells were transfected with 50 nM siRNA targeting fatty acid synthase (siFASN) or acetyl-coA carboxylase-α (siACC) or with a control siRNA (siCtrl). Expression of mRNA was analyzed by quantitative RT-PCR and protein expression was measured by western blotting analysis 72 to 96 hours after transfection. To analyze lipogenic activity, 2-[14C]-acetate was added during the last 4 hours to measure [14C]- incorporation in total lipids. Data represent means ± SE. Significantly different (p<0.05) from control. 2 2 farnesylated GFP BT474 cells were transfected with a farnesylated GFP construct and treated with soraphen (100 nM) or vehicle (control). After 72 hr, the fluorescence was bleached in a specific region. The fluorescence recovery was analyzed at the indicated time points. Data represent means ± SE (n=17-19). Values from soraphen-treated cells were significantly different from control (p<0.05) starting from 2 s after bleaching. 3 Figure S5. Inhibition of fatty acid synthesis in 22Rv1 and PC-3 cells affects the accumulation of doxorubicin Figure S5. Inhibition of fatty acid synthesis in 22Rv1 and PC-3 cells affects the accumulation of doxorubicin (A) Effect of soraphen on doxorubicin accumulation in 22Rv1 cells. 22Rv1 cells were cultured with or without soraphen (100 nM) for 72 hr. Doxorubicin (20 µM) was added, and 2 hr later, the accumulation of doxorubicin was visualized by fluorescence microscopy. (B) Quantification of the effect of soraphen on doxorubicin accumulation in 22Rv1 and PC-3 cells. Cells were treated with or without soraphen (100 nM) for 72 hr. Two hr prior to analysis, 20 µM of doxorubicin was added to the cells. Cellular extracts were prepared and doxorubicin content was fluorimetrically determined. Data represent means ± SE (n=4). *Significantly different (p<0.05) from control. Figure S6. CI analysis of soraphen- and doxorubicin-mediated cytotoxicity in LNCaP cells LNCaP cells were treated with increasing concentrations of soraphen (for 72 hr) or doxorubicin (for 24 hr) alone, or simultaneously with a fixed ratio of both compounds (doxorubicin was added during the last 24 hr). The cells were collected and stained with trypan blue. The combined effect at four different ratios (80:1; 40:1; 20:1 and 10:1) was assessed using CI analysis. Experimental CIs were plotted against the fractional inhibition (fraction affected; Fa), as measured by the % of trypan blue-positive cells. The experimental CIs correspond to soraphen and doxorubicin values, as indicated in the table. 4 4
https://openalex.org/W4283124066	https://discovery.ucl.ac.uk/id/eprint/10150955/1/Dubis_The%20Natural%20History%20of%20CNGB1-Related%20Retinopathy_VoR.pdf	English	null	The Natural History of CNGB1-Related Retinopathy: A Longitudinal Phenotypic Analysis	International journal of molecular sciences	2,022	cc-by	8,844	Article Daniel J. Jackson 1, Adam M. Dubis 2 and Mariya Moosajee 1,3,4,5,* 1 Institute of Ophthalmology, UCL, London EC1V 9EL, UK; daniel.jackson.21@ucl.ac.uk 2 Global Business School for Health, UCL, London WC1E 6BT, UK; a.dubis@ucl.ac.uk 3 Moorﬁelds Eye Hospital NHS Trust, London EC1V 2PD, UK 4 Great Ormond Street Hospital for Children NHS Trust, London WC1N 3JH, UK 5 The Francis Crick Institute, Imperial College London, London NW1 1AT, UK * Correspondence: m.moosajee@ucl.ac.uk 1 Institute of Ophthalmology, UCL, London EC1V 9EL, UK; daniel.jackson.21@ucl.ac.uk 2 Global Business School for Health, UCL, London WC1E 6BT, UK; a.dubis@ucl.ac.uk 3 Moorﬁelds Eye Hospital NHS Trust, London EC1V 2PD, UK 4 Great Ormond Street Hospital for Children NHS Trust, London WC1N 3JH, UK 5 The Francis Crick Institute, Imperial College London, London NW1 1AT, UK * Correspondence: m.moosajee@ucl.ac.uk 1 Institute of Ophthalmology, UCL, London EC1V 9EL, UK; daniel.jackson.21@ucl.ac.uk 2 Global Business School for Health, UCL, London WC1E 6BT, UK; a.dubis@ucl.ac.uk 3 Moorﬁelds Eye Hospital NHS Trust, London EC1V 2PD, UK 4 Great Ormond Street Hospital for Children NHS Trust, London WC1N 3JH, UK 5 The Francis Crick Institute, Imperial College London, London NW1 1AT, UK * Correspondence: m.moosajee@ucl.ac.uk Abstract: Cyclic nucleotide-gated channel β 1 (CNGB1) encodes a subunit of the rod cyclic nucleotide- gated channel. Pathogenic variants in CNGB1 are responsible for 4% of autosomal recessive retinitis pigmentosa (RP). Several treatment strategies show promise for treating inherited retinal degener- ations, however relevant metrics of progression and sensitive clinical trial endpoints are needed to assess therapeutic efﬁcacy. This study reports the natural history of CNGB1-related RP with a longitudinal phenotypic analysis of 33 molecularly-conﬁrmed patients with a mean follow-up period of 4.5 ± 3.9 years (range 0–17). The mean best corrected visual acuity (BCVA) of the right eye was 0.31 ± 0.43 logMAR at baseline and 0.47 ± 0.63 logMAR at the ﬁnal visit over the study period. The ellipsoid zone (EZ) length was measurable in at least one eye of 23 patients and had a mean rate of constriction of 178 ± 161 µm per year (range 1.0–661 µm), with 57% of patients having a decrease in EZ length of greater than 250 µm in a simulated two-year trial period. Hyperautoﬂuorescent outer ring (hyperAF) area was measurable in 17 patients, with 10 patients not displaying a ring phenotype. The results support previous ﬁndings of CNGB1-related RP being a slowly progressive disease with patients maintaining visual acuity. International Journal of Molecular Sciences International Journal of Molecular Sciences International Journal of Molecular Sciences International Journal of Molecular Sciences Citation: Jackson, D.J.; Dubis, A.M.; Moosajee, M. The Natural History of CNGB1-Related Retinopathy: A Longitudinal Phenotypic Analysis. Int. J. Mol. Sci. 2022, 23, 6785. https://doi.org/10.3390/ ijms23126785 Citation: Jackson, D.J.; Dubis, A.M.; Moosajee, M. The Natural History of CNGB1-Related Retinopathy: A Longitudinal Phenotypic Analysis. Int. J. Mol. Sci. 2022, 23, 6785. https://doi.org/10.3390/ ijms23126785 Keywords: CNGB1; retinitis pigmentosa; natural history Academic Editors: Tamar Ben-Yosef and Susanne Roosing Article Prospective deep phenotyping studies assessing multimodal retinal imaging and functional measures are now required to determine clinical endpoints to be used in a trial. 1. Introduction CNGB1 (MIM #600724) is located on chromosome 16, with 5657 bp (cDNA) and is formed of 33 exons. It encodes the 240-kDa β-subunit of the cyclic nucleotide-gated ion channel located in rod photoreceptor plasma membranes, consisting of a glutamic acid rich protein (GARP) and channel domain [1]. The CNG channels in rods form heterote- tramers consist of 3 α-subunits (CNGA1) and 1 β-subunit (CNGB1), whereas the cone channel is formed by 2 α-subunits (CNGA3) and 2 β-subunits (CNGB3), and both translate light-mediated changes of second-messenger cyclic guanosine monophosphate into voltage signals [2–4]. Pathogenic variants in CNGB1 account for up to 4% of autosomal recessive retinitis pigmentosa (RP) cases (RP45 #613767) [5–9]. Patients present with nyctalopia in childhood due to the initial loss of rod photoreceptors, followed by progressive con- striction of the visual ﬁeld. Fundus appearance is typical of RP bone spicule pigmentary deposits and retinal vessel attenuation. Spectral-domain optical coherence tomography (SD-OCT) shows progressive loss of the ellipsoid zone (EZ) and fundus autoﬂuorescence (FAF) demonstrates a hyperautoﬂuorescent (hyperAF) ring around the fovea and areas of hypoautoﬂuorescence corresponding to the atrophic areas and pigment deposition. Despite symptoms developing early in life, visual acuity is usually well-preserved into adulthood, Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- iations. Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional afﬁl- iations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). https://www.mdpi.com/journal/ijms Int. J. Mol. Sci. 2022, 23, 6785. https://doi.org/10.3390/ijms23126785 Int. J. Mol. Sci. 2022, 23, 6785 2 of 9 hence patients potentially have a longer therapeutic window for intervention following disease onset compared to other forms of RP [10]. p [ ] There are 84 known disease-causing CNGB1 variants including 24 missense vari- ants, 21 nonsense, 19 splicing defects, 10 small deletions, 1 small insertion, 1 small insertion–deletion, 7 small duplications, and 1 gross deletion, and this is comprehen- sively discussed in a recent review [11]. The known variants span the entire gene with no known cluster regions. No genotype-phenotype correlations have been identified to date, however there are only limited reports describing CNGB1-RP phenotypes [5,10–18]. Animal modelling of CNGB1-related RP has successfully recapitulated the human dis- ease. 1. Introduction In mice, Cngb1-X26 model, generated by excising exon 26, signiﬁcantly impaired rod function was detected at age 2–3 weeks, followed later by cone dysfunction from six months [19]. Similarly, a canine model has been identiﬁed with a spontaneous CNGB1 mu- tation, c.2387delA;2389_2390insAGCTAC in exon 26, leading to a frameshift and premature stop codon, which closely resembles the Cngb1-X26 mouse and human RP phenotype [20]. Gene augmentation therapy using adeno-associated viral (AAV) vectors has successfully rescued the phenotypes resulting in restoration of structural retinal integrity [12,21]. In the mouse model, a 221 bp long SV40 polyA sequence with a 471 bp mouse rhodopsin promoter was packaged into an AAV8 capsid and injected into the subretinal space of two week old mice [21]. Restoration of rod-driven light responses was observed in the treated mice as well as superior performance in vision-guided behavioural tests. In the canine model, an AAV5 vector was used to deliver canine CNGB1 under control of a human GRK1 promoter. CNGB1 expression was detected at three months and sustained for 23 months following subretinal injection and a sustained improvement in rod-mediated electroretinogram (ERG) was observed [12]. The identiﬁcation of clinical trial endpoints can be a challenge in subsets of RP with slow disease progression. Visual acuity is commonly not a suitable metric to assess thera- peutic efﬁcacy, as has been found with other related studies assessing rod-predominant degeneration [22,23]. Multimodal imaging has been central to assess disease status, with SD-OCT being one of the most valuable methods for quantifying progressive changes involving the EZ [24–27]. Constriction of hyperAF rings have correlated with disease pro- gression in other RP subtypes [22,28], and have shown to have detectable changes in area over a 2–5 year period in a small CNGB1-related RP cohort of three patients [12]. Central retinal thickness is not a suitable metric due to confounding factors including epiretinal membrane (ERM) and cystoid macular oedema (CMO), prevalent in these patients [11]. This is in line with a previous FDA assessment of age-related macular degeneration (AMD), whereby considerable variability has been previously shown in metrics such as visual acuity, retinal thickness, and visual ﬁelds, whereas EZ length and hyperAF ring area may be more useful in measuring photoreceptor degeneration [29]. Longitudinal natural history studies combining structural and functional outcome metrics are required to deﬁne prognosis. CNGB1-related RP is a clinical trial priority given the preclinical success of AAV gene replacement. 1. Introduction Here we report the natural history of the largest patient cohort of CNGB1-related RP over a 17 year follow-up period. 2.1. Demographics and Visual Acuity The protein consists of a glutamic-acid rich protein domain (GARP), a calmodulin-binding domain (CaM), 6 transmembrane (TM) domains, and a cyclic nucle- otide binding domain (CNDB). Mean best corrected visual acuity (BCVA) analysis from the right eye of the 33 pa- tients at baseline was 0.29 ± 0.41 logarithm of the minimum angle of resolution (logMAR) Figure 1. CNGB1 gene and protein schematic with mutations, corresponding to canonical transcrip ENST00000251102, NM_001297.5. The protein consists of a glutamic-acid rich protein domain (GARP), a calmodulin-binding domain (CaM), 6 transmembrane (TM) domains, and a cyclic nucle otide binding domain (CNDB). Figure 1. CNGB1 gene and protein schematic with mutations, corresponding to canonical transcript ENST00000251102, NM_001297.5. The protein consists of a glutamic-acid rich protein domain (GARP), a calmodulin-binding domain (CaM), 6 transmembrane (TM) domains, and a cyclic nucleotide binding domain (CNDB). (GARP), a calmodulin-binding domain (CaM), 6 transmembrane (TM) domains, and a cyclic nucle- otide binding domain (CNDB). Mean best corrected visual acuity (BCVA) analysis from the right eye of the 33 pa- tients at baseline was 0.29 ± 0.41 logarithm of the minimum angle of resolution (logMAR) Mean best corrected visual acuity (BCVA) analysis from the right eye of the 33 pa tients at baseline was 0.29 ± 0.41 logarithm of the minimum angle of resolution (logMAR (range 0.0–2.2). Twenty-nine patients had at least one follow-up visit and had a mean BCVA of 0.31 ± 0.43 logMAR in their right eye at baseline and 0.47 ± 0.63 logMAR at the most recent (final) visit over a duration of 4.5 ± 3.9 years. Correlation between the change in BCVA at baseline and the most recent follow-up visit demonstrated a trend of worsen ing BCVA over time (Figure 2A). The change in BCVA of each patient at baseline and mos recent visit as a function of age is illustrated in Figure 2B. Mean best corrected visual acuity (BCVA) analysis from the right eye of the 33 patients at baseline was 0.29 ± 0.41 logarithm of the minimum angle of resolution (logMAR) (range 0.0–2.2). Twenty-nine patients had at least one follow-up visit and had a mean BCVA of 0.31 ± 0.43 logMAR in their right eye at baseline and 0.47 ± 0.63 logMAR at the most recent (ﬁnal) visit over a duration of 4.5 ± 3.9 years. 2.1. Demographics and Visual Acuity This study identiﬁed 33 patients from 32 families, of which 61% (20/33) were female. The mean age at diagnosis was 40.8 ± 17.1 years (range 10–81 years). The mean follow-up period was 4.5 ± 3.9 years (range 0–17) with a mean number of visits per patient of 5 ± 3.7 (range 1–15 visits). Of these, 21% (7/33) had a family history of RP and 6% (2/33) were from consanguineous families. The cohort consisted of families with diverse ethnic origins: 17 (51.2%) were White British, 10 (30.3%) were Middle Eastern, 2 (6.1%) South Asian, 2 (6.1%) White Other, 1 (3.0%) North African, 1 (3.0%) Black African. The presenting complaint was nyctalopia in 88% (21/24) with symptom onset in childhood for the majority. Of the 33 patients, 18 (54.5%) had homozygous mutations (9 missense, 4 splice site, 3 nonsense, Int. J. Mol. Sci. 2022, 23, 6785 3 of 9 etero- 1 deletion, 1 duplication) with the remaining 15 (45.4%) having compound heterozygous mutations (Supplementary Table S1). There were 2 novel variants (c.1936C>T, c.290G>C). Figure 1 details the locations of the mutations, which are distributed across the GARP and channel domains. sense, 1 deletion, 1 duplication) with the remaining 15 (45.4%) having compound hetero zygous mutations (Supplementary Table S1). There were 2 novel variants (c.1936C>T c.290G>C). Figure 1 details the locations of the mutations, which are distributed across the GARP and channel domains. GARP and channel domains. mutations (Supplementary Table S1). There were 2 novel variants (c.1936C>T, c.290G>C). Figure 1 details the locations of the mutations, which are distributed across the GARP and channel domains. c.290G>C). Figure 1 details the locations of the mutations, which are distributed across the GARP and channel domains. Figure 1. CNGB1 gene and protein schematic with mutations, corresponding to canonical transcript ENST00000251102, NM_001297.5. The protein consists of a glutamic-acid rich protein domain (GARP), a calmodulin-binding domain (CaM), 6 transmembrane (TM) domains, and a cyclic nucle- otide binding domain (CNDB). Figure 1. CNGB1 gene and protein schematic with mutations, corresponding to canonical transcript ENST00000251102, NM_001297.5. The protein consists of a glutamic-acid rich protein domain (GARP), a calmodulin-binding domain (CaM), 6 transmembrane (TM) domains, and a cyclic nucleotide binding domain (CNDB). Figure 1. CNGB1 gene and protein schematic with mutations, corresponding to canonical transcript ENST00000251102, NM_001297.5. 2.1. Demographics and Visual Acuity Correlation between the change in BCVA at baseline and the most recent follow-up visit demonstrated a trend of worsening BCVA over time (Figure 2A). The change in BCVA of each patient at baseline and most recent visit as a function of age is illustrated in Figure 2B. (range 0.0–2.2). Twenty-nine patients had at least one follow-up visit and had a mean BCVA of 0.31 ± 0.43 logMAR in their right eye at baseline and 0.47 ± 0.63 logMAR at the most recent (final) visit over a duration of 4.5 ± 3.9 years. Correlation between the change in BCVA at baseline and the most recent follow-up visit demonstrated a trend of worsen- ing BCVA over time (Figure 2A). The change in BCVA of each patient at baseline and most recent visit as a function of age is illustrated in Figure 2B. Figure 2. Best corrected visual acuity (BCVA) changes in patients with CNGB1-related RP. (A) Pear- son correlation between the changes in BCVA from the right eye at baseline to the most recent fol- low-up. The dashed line represents no change, while the solid correlation line illustrates the change seen in patients. (B) demonstrates the change in BCVA of each patient at baseline and most recent Figure 2. Best corrected visual acuity (BCVA) changes in patients with CNGB1-related RP. (A) Pear- son correlation between the changes in BCVA from the right eye at baseline to the most recent fol- low-up. The dashed line represents no change, while the solid correlation line illustrates the change seen in patients. (B) demonstrates the change in BCVA of each patient at baseline and most recent Figure 2. Best corrected visual acuity (BCVA) changes in patients with CNGB1-related RP. (A) Pearson correlation between the changes in BCVA from the right eye at baseline to the most recent follow-up. The dashed line represents no change, while the solid correlation line illustrates the change seen in patients. (B) demonstrates the change in BCVA of each patient at baseline and most recent visual acuity as a function of age. The majority of patients maintain visual acuity of greater than 0.5 logMAR in at the right eye over the study period. 2.2. SD-OCT Ellipsoid Zone and Associated Features SD-OCT images were analysed from 65 eyes of 33 patients. 2.1. Demographics and Visual Acuity Best corrected visual acuity (BCVA) changes in patients with CNGB1-related RP. (A) Pearson correlation between the changes in BCVA from the right eye at baseline to the most recent follow-up. The dashed line represents no change, while the solid correlation line illustrates the change seen in patients. (B) demonstrates the change in BCVA of each patient at baseline and most recent visual acuity as a function of age. The majority of patients maintain visual acuity of greater than 0.5 logMAR in at the right eye over the study period. Figure 2. Best corrected visual acuity (BCVA) changes in patients with CNGB1-related RP. (A) Pear- son correlation between the changes in BCVA from the right eye at baseline to the most recent fol- low-up. The dashed line represents no change, while the solid correlation line illustrates the change Figure 2. Best corrected visual acuity (BCVA) changes in patients with CNGB1-related RP. (A) Pearson correlation between the changes in BCVA from the right eye at baseline to the most recent follow-up. The dashed line represents no change, while the solid correlation line illustrates the change seen in patients. (B) demonstrates the change in BCVA of each patient at baseline and most recent visual acuity as a function of age. The majority of patients maintain visual acuity of greater than 0.5 logMAR in at the right eye over the study period. 2.1. Demographics and Visual Acuity Epiretinal membrane was present in 47.0% (31/65) of the eyes in 17 patients (51.5%) in at least one scan over the mean 4.5 year follow-up period. Cystoid macular oedema was present in 19.7% (13/65) of eyes in Figure 2 Best corrected visual acuity (BCVA) changes in patients with CNGB1 related RP (A) Pear Figure 2. Best corrected visual acuity (BCVA) changes in patients with CNGB1-related RP. (A) Pear- son correlation between the changes in BCVA from the right eye at baseline to the most recent fol- low-up. The dashed line represents no change, while the solid correlation line illustrates the change Figure 2. Best corrected visual acuity (BCVA) changes in patients with CNGB1-related RP. (A) Pearson correlation between the changes in BCVA from the right eye at baseline to the most recent follow-up. The dashed line represents no change, while the solid correlation line illustrates the change seen in patients. (B) demonstrates the change in BCVA of each patient at baseline and most recent visual acuity as a function of age. The majority of patients maintain visual acuity of greater than 0.5 logMAR in at the right eye over the study period. Fi 2 B d i l i (BCVA) h i i i h CNGB1 l d RP (A) P Figure 2. Best corrected visual acuity (BCVA) changes in patients with CNGB1-related RP. (A) Pear- son correlation between the changes in BCVA from the right eye at baseline to the most recent fol- low-up. The dashed line represents no change, while the solid correlation line illustrates the change Figure 2. Best corrected visual acuity (BCVA) changes in patients with CNGB1-related RP. (A) Pearson correlation between the changes in BCVA from the right eye at baseline to the most recent follow-up. The dashed line represents no change, while the solid correlation line illustrates the change seen in patients. (B) demonstrates the change in BCVA of each patient at baseline and most recent visual acuity as a function of age. The majority of patients maintain visual acuity of greater than 0.5 logMAR in at the right eye over the study period. Figure 2. Best corrected visual acuity (BCVA) changes in patients with CNGB1-related RP. (A) Pear- son correlation between the changes in BCVA from the right eye at baseline to the most recent fol- low-up. The dashed line represents no change, while the solid correlation line illustrates the change Figure 2. Figure 2. Best corrected visual acuity (BCVA) changes son correlation between the changes in BCVA from th seen in patients. (B) demonstrates the change in BCVA 2.2. SD-OCT Ellipsoid Zone and Associated Features g g y low-up. The dashed line represents no change, while the solid correlation line illustrates the change seen in patients. (B) demonstrates the change in BCVA of each patient at baseline and most recen SD-OCT images were analysed from 65 eyes of 33 patients. Epiretinal membrane was present in 47.0% (31/65) of the eyes in 17 patients (51.5%) in at least one scan over the mean 4.5 year follow-up period. Cystoid macular oedema was present in 19.7% (13/65) of eyes in Int. J. Mol. Sci. 2022, 23, 6785 4 of 9 was 7 patients (21.2%). A lamellar hole was present in one eye, so this eye was excluded from further analysis. in 7 patients (21.2%). A lamellar hole was present in one eye, so this eye was excluded from further analysis. Th EZ bl i 23 ti t f th i ht EZ l th d t h y The EZ was measurable in 23 patients from the right eye. EZ length measured at each visit is shown in Figure 3A. Nine were excluded from the SD-OCT analysis as the EZ extended beyond the boundaries of the image; these nine patients had a mean age of 52.5 ± 13.5 years. The mean rate of constriction was 178 µm ± 161 µm per year (range 1.0–661 µm per year). There was no correlation between rate of EZ length constriction with age (R2 = 0.009 p = 0.48) (Figure 3B). Figure 3C represents the correlation of EZ length against age for each patient who had three visits, each with an interval of 10–14 months. This was best described by a linear or exponential decay, but not logarithmic trend (Supplementary Figure S1). A ﬁtted linear trend shows a detectable decrease in EZ length evident within a clinical trial setting for all the patients represented (R2 = 0.194, p = 0.004). Figure 3D demonstrates EZ length correlation with age for the subtypes of patients with homozygous CNGB1 mutations. From these data, no speciﬁc genotype/phenotype correlations can be inferred, possibly due to limited patient numbers and follow-ups coupled with a diverse range of CNGB1 variants. The EZ was measurable in 23 patients from the right eye. EZ length measured at each visit is shown in Figure 3A. Figure 2. Best corrected visual acuity (BCVA) changes son correlation between the changes in BCVA from th seen in patients. (B) demonstrates the change in BCVA 2.2. SD-OCT Ellipsoid Zone and Associated Features Nine were excluded from the SD-OCT analysis as the EZ ex- tended beyond the boundaries of the image; these nine patients had a mean age of 52.5 ± 13.5 years. The mean rate of constriction was 178 µm ± 161 µm per year (range 1.0–661 µm per year). There was no correlation between rate of EZ length constriction with age (R2 = 0.009 p = 0.48) (Figure 3B). Figure 3C represents the correlation of EZ length against age for each patient who had three visits, each with an interval of 10–14 months. This was best described by a linear or exponential decay, but not logarithmic trend (Supplementary Fig- ure S1). A fitted linear trend shows a detectable decrease in EZ length evident within a clinical trial setting for all the patients represented (R2 = 0.194, p = 0.004). Figure 3D demon- strates EZ length correlation with age for the subtypes of patients with homozygous CNGB1 mutations. From these data, no specific genotype/phenotype correlations can be inferred, possibly due to limited patient numbers and follow-ups coupled with a diverse range of CNGB1 variants. Figure 3. Ellipsoid zone (EZ) length analysis in patients with CNGB1-related retinitis pigmentosa. (A) Correlation between EZ length and age for each patient at each visit showing a reduction in EZ length. (B) EZ constriction rate (change in size divided by time between visits), which appears to be constant with age. (C) Correlation between EZ length and age for patients who had three visits, each visit between 10–14 months apart (n = 14) with a fitted linear trend that was statistically significant using Pearson correlation (R2 = 0.194, p = 0.004). Fifty seven percent (8/14) had a reduction of EZ length of 250 µm or more during the two year simulated trial period. (D) Correlation between EZ Figure 3. Ellipsoid zone (EZ) length analysis in patients with CNGB1-related retinitis pigmentosa. (A) Correlation between EZ length and age for each patient at each visit showing a reduction in EZ length. (B) EZ constriction rate (change in size divided by time between visits), which appears to be constant with age. (C) Correlation between EZ length and age for patients who had three visits, each visit between 10–14 months apart (n = 14) with a ﬁtted linear trend that was statistically signiﬁcant using Pearson correlation (R2 = 0.194, p = 0.004). Figure 2. Best corrected visual acuity (BCVA) changes son correlation between the changes in BCVA from th seen in patients. (B) demonstrates the change in BCVA 2.2. SD-OCT Ellipsoid Zone and Associated Features (C) Correlation between EZ length and age for patients who had three visits, each visit between 10–14 months apart (n = 14) with a fitted linear trend that was statistically significant using Pearson correlation (R2 = 0.194, p = 0.004). Fifty seven percent (8/14) had a reduction of EZ length of 250 µm or more during the two year simulated trial period. (D) Correlation between EZ Figure 3. Ellipsoid zone (EZ) length analysis in patients with CNGB1-related retinitis pigmentosa. (A) Correlation between EZ length and age for each patient at each visit showing a reduction in EZ length. (B) EZ constriction rate (change in size divided by time between visits), which appears to be constant with age. (C) Correlation between EZ length and age for patients who had three visits, each visit between 10–14 months apart (n = 14) with a ﬁtted linear trend that was statistically signiﬁcant using Pearson correlation (R2 = 0.194, p = 0.004). Fifty seven percent (8/14) had a reduction of EZ length of 250 µm or more during the two year simulated trial period. (D) Correlation between EZ length and age for each patient per visit for each homozygous mutation subtype (insertion/deletion ‘Indel’, splice site, missense, and nonsense). Figure 2. Best corrected visual acuity (BCVA) changes son correlation between the changes in BCVA from th seen in patients. (B) demonstrates the change in BCVA 2.2. SD-OCT Ellipsoid Zone and Associated Features Fifty seven percent (8/14) had a reduction of EZ length of 250 µm or more during the two year simulated trial period. (D) Correlation between EZ length and age for each patient per visit for each homozygous mutation subtype (insertion/deletion ‘Indel’, splice site, missense, and nonsense). Figure 3. Ellipsoid zone (EZ) length analysis in patients with CNGB1-related retinitis pigmentosa. (A) Correlation between EZ length and age for each patient at each isit showing a reduction in EZ Figure 3. Ellipsoid zone (EZ) length analysis in patients with CNGB1-related retinitis pigmentosa Figure 3. Ellipsoid zone (EZ) length analysis in patients with CNGB1-related retinitis pigmentosa. (A) Correlation between EZ length and age for each patient at each visit showing a reduction in EZ length. (B) EZ constriction rate (change in size divided by time between visits), which appears to be constant with age. (C) Correlation between EZ length and age for patients who had three visits, each visit between 10–14 months apart (n = 14) with a fitted linear trend that was statistically significant using Pearson correlation (R2 = 0.194, p = 0.004). Fifty seven percent (8/14) had a reduction of EZ length of 250 µm or more during the two year simulated trial period. (D) Correlation between EZ Figure 3. Ellipsoid zone (EZ) length analysis in patients with CNGB1-related retinitis pigmentosa. (A) Correlation between EZ length and age for each patient at each visit showing a reduction in EZ length. (B) EZ constriction rate (change in size divided by time between visits), which appears to be constant with age. (C) Correlation between EZ length and age for patients who had three visits, each visit between 10–14 months apart (n = 14) with a ﬁtted linear trend that was statistically signiﬁcant using Pearson correlation (R2 = 0.194, p = 0.004). Fifty seven percent (8/14) had a reduction of EZ length of 250 µm or more during the two year simulated trial period. (D) Correlation between EZ length and age for each patient per visit for each homozygous mutation subtype (insertion/deletion ‘Indel’, splice site, missense, and nonsense). Figure 3. Ellipsoid zone (EZ) length analysis in patients with CNGB1-related retinitis pigmentosa. (A) Correlation between EZ length and age for each patient at each visit showing a reduction in EZ length. (B) EZ constriction rate (change in size divided by time between visits), which appears to be constant with age. 2.3. HyperAF Area The genotypes between the two groups were also spread across the GARP and channel domain of the CNGB1 gene, with no genotype-phenotype correlations identified. The hyperAF ring area showed a decline with age (Figure 4A) and hyperAF ring constriction rate was not correlated with age (r2 = 0.012 p = 0.667) with no clear inflec- tion point (Figure 4B). Figure 4. Fundus Autofluorescence (FAF) examples from patients with CNGB1-related retinitis pig- mentosa. (A–D) show a hyperautofluorescent (hyperAF) outer ring phenotype, while (E–H) do not exhibit a measurable ring pattern. (I,J) show the quantification of the hyperAF outer ring area on FAF. (I) Correlation of hyperAF ring area as a function of age. (J) Correlation of rate of constriction of hyperAF area as a function of age, which was not statistically significant (r2 = 0.012 p = 0.667) using Pearson correlation Figure 4. Fundus Autoﬂuorescence (FAF) examples from patients with CNGB1-related retinitis pigmentosa. (A–D) show a hyperautoﬂuorescent (hyperAF) outer ring phenotype, while (E–H) do not exhibit a measurable ring pattern. (I,J) show the quantiﬁcation of the hyperAF outer ring area on FAF. (I) Correlation of hyperAF ring area as a function of age. (J) Correlation of rate of constriction of hyperAF area as a function of age, which was not statistically signiﬁcant (r2 = 0.012 p = 0.667) using Pearson correlation. Figure 4. Fundus Autofluorescence (FAF) examples from patients with CNGB1-related retinitis pig- mentosa. (A–D) show a hyperautofluorescent (hyperAF) outer ring phenotype, while (E–H) do not exhibit a measurable ring pattern. (I,J) show the quantification of the hyperAF outer ring area on FAF. (I) Correlation of hyperAF ring area as a function of age. (J) Correlation of rate of constriction of hyperAF area as a function of age, which was not statistically significant (r2 = 0.012 p = 0.667) using P l ti Figure 4. Fundus Autoﬂuorescence (FAF) examples from patients with CNGB1-related retinitis pigmentosa. (A–D) show a hyperautoﬂuorescent (hyperAF) outer ring phenotype, while (E–H) do not exhibit a measurable ring pattern. (I,J) show the quantiﬁcation of the hyperAF outer ring area on FAF. (I) Correlation of hyperAF ring area as a function of age. (J) Correlation of rate of constriction of hyperAF area as a function of age, which was not statistically signiﬁcant (r2 = 0.012 p = 0.667) using Pearson correlation. 2.3. HyperAF Area Seventeen patients had a hyperAF outer ring phenotype that was measurable. There were a diverse range of FAF phenotypes that did not display a characteristic ring pattern in 10 patients (Figure 4). The difference between the FAF phenotypes was not due to Int. J. Mol. Sci. 2022, 23, 6785 5 of 9 There more advanced disease with older age, with the ‘ring’ phenotypes having a mean ± SD age of 45.9 ± 14.3 years (range 17–72) and the other phenotypes having a mean age of 54.5 ± 14.7 years (range 28–81). The genotypes between the two groups were also spread across the GARP and channel domain of the CNGB1 gene, with no genotype-phenotype correlations identiﬁed. The hyperAF ring area showed a decline with age (Figure 4A) and hyperAF ring constriction rate was not correlated with age (r2 = 0.012 p = 0.667) with no clear inﬂection point (Figure 4B). p ( g ) p yp advanced disease with older age, with the ‘ring’ phenotypes having a mean ± SD age of 45.9 ± 14.3 years (range 17–72) and the other phenotypes having a mean age of 54.5 ± 14.7 years (range 28–81). The genotypes between the two groups were also spread across the GARP and channel domain of the CNGB1 gene, with no genotype-phenotype correlations identified. The hyperAF ring area showed a decline with age (Figure 4A) and hyperAF ring constriction rate was not correlated with age (r2 = 0.012 p = 0.667) with no clear inflec- tion point (Figure 4B). more advanced disease with older age, with the ‘ring’ phenotypes having a mean ± SD age of 45.9 ± 14.3 years (range 17–72) and the other phenotypes having a mean age of 54.5 ± 14.7 years (range 28–81). The genotypes between the two groups were also spread across the GARP and channel domain of the CNGB1 gene, with no genotype-phenotype correlations identiﬁed. The hyperAF ring area showed a decline with age (Figure 4A) and hyperAF ring constriction rate was not correlated with age (r2 = 0.012 p = 0.667) with no clear inﬂection point (Figure 4B). p ( g ) p yp advanced disease with older age, with the ‘ring’ phenotypes having a mean ± SD age of 45.9 ± 14.3 years (range 17–72) and the other phenotypes having a mean age of 54.5 ± 14.7 years (range 28–81). 3. Discussion We report the ﬁndings of the largest retrospective natural history study involving 33 molecularly conﬁrmed patients with CNGB1-related RP. The mean age at diagnosis in our cohort was 40.8 years, yet the majority of patients reported nyctalopia from childhood. This suggests patients may not manifest further debilitating visual symptoms resulting in engagement with ophthalmic services until several decades later, supporting a lengthy window of therapeutic opportunity. Patient records were only pooled from a single tertiary centre and patients may have been assessed at other facilities prior to referral. The patients within our cohort maintained a BCVA of 0.3 logMAR or better after a mean 4.5 ± 3.9 years Int. J. Mol. Sci. 2022, 23, 6785 6 of 9 (range 0–17) follow-up. Thirteen patients had undergone cataract surgery in at least one eye, with a mean age of ﬁrst eye cataract surgery of 52.7 years (range 37–69) (n = 6), however, the age of surgery was unknown in seven patients. The slowly progressive nature and confounding variables, such as cataract surgery, exclude visual acuity from being a reliable clinical endpoint within a trial setting. Multimodal retinal imaging using SD-OCT EZ length and FAF hyperAF ring measure- ments showed most promise as clinical trial endpoints from this study. EZ length exhibited a detectable reduction within a two-year simulated trial period (Figure 2C). There was no clear inﬂection point to determine if the rate changed with age. ERM prevalence in our study is higher than in previously reported CNGB1 cohorts (34.8% in Nassisi et al. and 20% in Hull et al.), but has been reported in up to 64% from other RP cohorts, precluding the use of central retinal thickness as a reliable outcome measure [10,11,30]. Abnormal hyperAF ring phenotypes on FAF have been reported in 59–68% of RP patients, depending on genetic cause [31,32]. A recent review identiﬁed ring phenotypes on quantitative FAF in 79.3% of RP patients with CNGB1 mutations, similar to our cohort of 69.7% [11]. These differing FAF phenotypes did not appear correlated with age or genotype in this study. The use of wider ﬁeld imaging modalities may increase the number of patients with measurable FAF ring phenotypes. g p yp Using FAF and SD-OCT as a rapid, non-invasive and easily gradable method to determine clinical endpoints would be highly advantageous. 3. Discussion However, the majority of patients reported in this study were over the age of 40 years, therefore, extrapolation of ﬁndings to younger cohorts may be limited and reﬂects the slow progression of disease. Of 33 patients with imaging, nine were excluded from the SD-OCT analysis as their EZ extended beyond the boundaries of the image. Similarly for FAF, six patients were excluded as their hyperAF ring extended beyond the boundary and could not be measured. This will have important implications in identifying outcome metrics, as these patients with milder diseases will likely signiﬁcantly beneﬁt from therapy. A limitation of this study is the use of measurements from a single grader, and previous repeatability assessments have demonstrated an inter-observer variability in EZ length recordings of within approximately 250 µm [22]. Fifty-seven percent (8/14) of patients had a change in EZ length of greater than 250 µm during a two year trial simulation (Figure 3C), suggesting the change in this metric over this timeframe is greater than inter-observer differences in grading. Deep learning methods to accurately segment retinal layers have shown promise in other cases of RP and could be expanded to CNGB1-retinopathy, potentially overcoming the shortcomings of manual measurements, although these have only been in small datasets to date [33]. Despite collecting longitudinal follow-up of visual outcomes and multimodal reti- nal imaging, retrospective data from routine clinic visits has limitations in accurately determining important clinical details, such as age of symptom onset, as well as varying imaging protocols and the use of other functional modalities that can vary from visit to visit. Only 13 patients had visual ﬁelds measured with a variety of protocols, and nine patients had once only electrodiagnostic testing as part of their clinical care. Full-ﬁeld electroretinograms (ff-ERG) have shown demonstratable rod dysfunction in all patients, with scotopic responses attenuated in younger and extinguished in older patients from other CNGB1-RP cohorts [11,12]. Abnormal photopic responses have also been reported across different age groups [12]. Symptomatic visual ﬁeld loss occurs later in life, with confrontational visual ﬁeld testing showing variable peripheral loss in RP patients with CNGB1 variants [11]. Interestingly, formal kinetic perimetry has detected visual ﬁeld loss in a 12 year old patient [10]. Retinal sensitivity measures in the form of full ﬁeld dynamic and static perimetry or microperimetry can complement multimodal retinal imaging in tracking disease progression. 4.1. Subjects Potential subjects were identiﬁed from the prospectively consented Moorﬁelds Eye Hospital Inherited Eye Disease Database for structure/function of genetic diseases (Re- search Ethics Number: 12/LO/0141). Data for these studies are collected as part of stan- dard of care and retrospectively analysed. Patients with molecular genetic conﬁrmation of disease-causing variants in CNGB1 were identiﬁed. For all subjects, a full ophthalmic assessment was conducted at each visit as part of their clinical care, including best-corrected visual acuity and retinal imaging. Data was extracted from the electronic medical records of each patient at Moorﬁelds Eye Hospital and supplemented with written records, where appropriate. 4.2. Retinal Imaging Spectral domain optical coherence tomography (SD-OCT) and fundus autoﬂuores- cence (FAF) imaging was performed on all patients using the Heidelberg Spectralis (Hei- delberg Engineering, Heidelberg, Germany) with Automated Retinal Tracking. The central SD-OCT B scan was identiﬁed by a trained observer as having the least residual inner retinal tissue and thickest outer nuclear layer (ONL) presence. FAF imaging was per- formed using high power blue light autoﬂuorescence at 30 or 55◦depending on which best visualised the residual FAF area. EZ length and hyperAF area was measured using the integrated micrometre calliper tool with the inbuilt Heidelberg Eye Explorer software (Heyex, Heidelberg Engineering, Heidelberg, Germany) by a senior trained grader [38]. For analysis, scans whereby the EZ band or hyperAF ring extended beyond the boundaries of the image were excluded. 4.3. Statistics Given the high degree of human interocular symmetry, the measurements from the right eye were used for analyses. Most measurements occurred at yearly increments; for those that were not, decline was assumed to be linear over the period of time between observations and, therefore, the change was divided by time and thus the rate of change per year was calculated. Pearson correlations were used to assess the relatedness between progression rates and parameters. All statistical analyses were completed using GraphPad Prism (version 8.0.0 GraphPad Software, San Diego, CA, USA). Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/ijms23126785/s1. 3. Discussion Microperimetry has high reproducibility, as well as good interocular correlation, and can accurately detect disease progression in other forms of RP [34,35]. Full- ﬁeld stimulus testing may feasibly be used to detect measurable endpoints in patients with very impaired visual acuity and visual ﬁeld, although there are limitations in accessibility and careful patient instruction and examiner training are required to reduce confounding Int. J. Mol. Sci. 2022, 23, 6785 7 of 9 variables [36,37]. No genotype-phenotype correlations for CNGB1-retinopathy have been previously described and this was conﬁrmed within this study. However, recruiting more patients and undertaking a prospective deep phenotyping study may help reveal potential relationships. References 1. Zheng, J.; Trudeau, M.C.; Zagotta, W.N. Rod cyclic nucleotide-gated channels have a stoichiometry of three CNGA1 subunits and one CNGB1 subunit. Neuron 2002, 36, 891–896. [CrossRef] Rich, E.D.; Varnum, M.D. Subunit conﬁguration of heteromeric cone cyclic nucleotide-gated channels. Neur [CrossRef] 2. Peng, C.; Rich, E.D.; Varnum, M.D. Subunit conﬁguration of heteromeric cone cyclic nucleotide-gat 401–410. [CrossRef] 3. Shuart, N.G.; Haitin, Y.; Camp, S.S.; Black, K.D.; Zagotta, W.N. Molecular mechanism for 3:1 subunit stoichiometry of rod cyclic nucleotide-gated ion channels. Nat. Commun. 2011, 2, 457. [CrossRef] g U.B.; Seifert, R. Cyclic nucleotide-gated ion channels. Physiol. Rev. 2002, 82, 769–824. [CrossRef] [PubMed] g 4. Kaupp, U.B.; Seifert, R. Cyclic nucleotide-gated ion channels. Physiol. Rev. 2002, 82, 769–824. [Cros 4. Kaupp, U.B.; Seifert, R. Cyclic nucleotide-gated ion channels. Physiol. Rev. 2002, 82, 769–824. [CrossRef] [PubMed] 5 B il C H l C P D l V A d B D ill J Cl t M S ti f t ti i CNGB1 di 4. Kaupp, U.B.; Seifert, R. Cyclic nucleotide gated ion channels. Physiol. Rev. 2002, 82, 769 824. [CrossRef] [PubMed] 5. Bareil, C.; Hamel, C.P.; Delague, V.; Arnaud, B.; Demaille, J.; Claustres, M. Segregation of a mutation in CNGB1 encoding the beta-subunit of the rod cGMP-gated channel in a family with autosomal recessive retinitis pigmentosa. Hum. Genet. 2001, 108, 328–334. [CrossRef] 6. Ge, Z.; Bowles, K.; Goetz, K.; Scholl, H.P.; Wang, F.; Wang, X.; Xu, S.; Wang, K.; Wang, H.; Chen, R. NGS-based Molecular diagnosis of 105 eyeGENE(®) probands with Retinitis Pigmentosa. Sci. Rep. 2015, 5, 18287. [CrossRef] [PubMed] p g p T.; Golczak, M.; Palczewski, K. Retinopathy in mice induced by disrupted all-trans-retinal clearance. J. Biol 84–26693. [CrossRef] 7. Maeda, A.; Maeda, T.; Golczak, M.; Palczewski, K. Retinopathy in mice induced by disrupted all-t Chem. 2008, 283, 26684–26693. [CrossRef] 8. Xu, Y.; Guan, L.; Shen, T.; Zhang, J.; Xiao, X.; Jiang, H.; Li, S.; Yang, J.; Jia, X.; Yin, Y.; et al. Mutations of 60 known causative genes in 157 families with retinitis pigmentosa based on exome sequencing. Hum. Genet. 2014, 133, 1255–1271. [CrossRef] p g q g 9. Dryja, T.P.; Finn, J.T.; Peng, Y.W.; McGee, T.L.; Berson, E.L.; Yau, K.W. Mutations in the gene encoding the alpha subunit of the rod cGMP-gated channel in autosomal recessive retinitis pigmentosa. Proc. Natl. Acad Sci. USA 1995, 92, 10177–10181. [CrossRef] 10. 5. Conclusions We have demonstrated detectable changes on SD-OCT and FAF imaging in patients with CNGB1-related RP. The natural history conﬁrmed a slowly progressive disease whereby central visual acuity is maintained for a lengthy window of time compared with other RP subtypes. EZ length and hyperAF area may be the most sensitive measures of change and have potential as outcome metrics in future trials for subgroups of CNGB1- related RP patients who display the phenotype within measurable parameters. The ability of microperimetry, dynamic perimetry, full-threshold stimulus testing and electrodiagnostic testing to determine disease progression needs to be examined within a prospective deep phenotyping study to provide greater insights into the key functional clinical endpoints that will offer accurate efﬁcacy metrics in assessing future therapeutics. Int. J. Mol. Sci. 2022, 23, 6785 8 of 9 Author Contributions: Conceptualization, M.M., data collection D.J.J., data analysis D.J.J., A.M.D. and M.M., drafting of manuscript D.J.J., A.M.D. and M.M. All authors have read and agreed to the published version of the manuscript. Author Contributions: Conceptualization, M.M., data collection D.J.J., data analysis D.J.J., A.M.D. and M.M., drafting of manuscript D.J.J., A.M.D. and M.M. All authors have read and agreed to the published version of the manuscript. Funding: We are grateful to PTC Therapeutics Inc and the Wellcome Trust (205174/Z/16/Z) for funding this study. NIHR Biomedical Research Centre at Moorﬁelds Eye Hospital NHS Trust and UCL Institute of Ophthalmology (IS-BRC_1215-20002). Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Moorﬁelds Eye Hospital (Research Ethics Number: 12/LO/0141). Informed Consent Statement: Potential subjects were identiﬁed from the prospectively consented Moorﬁelds Eye Hospital Inherited Eye Disease Database for structure/function of genetic diseases (Research Ethics Number: 12/LO/0141). Data Availability Statement: The summarised data presented in this study are provided in Supple- mentary Figures S1 and S2. Full datasets are available on request from the corresponding author. Conﬂicts of Interest: The authors have no conﬂict of interest to declare. Conﬂicts of Interest: The authors have no conﬂict of interest to declare. References Hull, S.; Attanasio, M.; Arno, G.; Carss, K.; Robson, A.G.; Thompson, D.A.; Plagnol, V.; Michaelides, M.; Holder, G.E.; Henderson, R.H.; et al. Clinical Characterization of CNGB1-Related Autosomal Recessive Retinitis Pigmentosa. JAMA Ophthalmol. 2017, 135, 137–144. [CrossRef] 11. Nassisi, M.; Smirnov, V.M.; Solis Hernandez, C.; Mohand-Saïd, S.; Condroyer, C.; Antonio, A.; Kühlewein, L.; Kempf, M.; Kohl, S.; Wissinger, B.; et al. CNGB1-related rod-cone dystrophy: A mutation review and update. Hum. Mutat. 2021, 42, 641–666. [CrossRef] 12. Petersen-Jones, S.M.; Occelli, L.M.; Winkler, P.A.; Lee, W.; Sparrow, J.R.; Tsukikawa, M.; Boye, S.L.; Chiodo, V.; Capasso, J.E.; Becirovic, E.; et al. Patients and animal models of CNGβ1-deﬁcient retinitis pigmentosa support gene augmentation approach. J. Clin. Investig. 2018, 128, 190–206. [CrossRef] g [ ] 13. Charbel Issa, P.; Reuter, P.; Kühlewein, L.; Birtel, J.; Gliem, M.; Tropitzsch, A.; Whitcroft, K.L.; Bolz, H.J.; Ishihara, K.; MacLaren, R.E.; et al. Olfactory Dysfunction in Patients with CNGB1-Associated Retinitis Pigmentosa. JAMA Ophthalmol. 2018, 136, 761–769. [CrossRef] [ ] 14. Ba-Abbad, R.; Holder, G.E.; Robson, A.G.; Neveu, M.M.; Waseem, N.; Arno, G.; Webster, A.R. Isolated rod dysfunction associated with a novel genotype of CNGB1. Am. J. Ophthalmol. Case Rep. 2019, 14, 83–86. [CrossRef] 15. Alshamrani, A.A.; Raddadi, O.; Schatz, P.; Lenzner, S.; Neuhaus, C.; Azzam, E.; Abdelkader, E. Severe retinitis pigmentosa phenotype associated with novel CNGB1 variants. Am. J. Ophthalmol. Case Rep. 2020, 19, 100780. [CrossRef] 16. Kondo, H.; Qin, M.; Mizota, A.; Kondo, M.; Hayashi, H.; Hayashi, K.; Oshima, K.; Tahira, T.; Hayashi, K. A homozygosity-based search for mutations in patients with autosomal recessive retinitis pigmentosa, using microsatellite markers. Investig. Ophthalmol. Vis. Sci. 2004, 45, 4433–4439. [CrossRef] , , [ ] 17. Bocquet, B.; Marzouka, N.A.; Hebrard, M.; Manes, G.; Sénéchal, A.; Meunier, I.; Hamel, C.P. Homozygosity mapping in autosomal recessive retinitis pigmentosa families detects novel mutations. Mol. Vis. 2013, 19, 2487–2500. 9 of 9 Int. J. Mol. Sci. 2022, 23, 6785 18. Maranhao, B.; Biswas, P.; Gottsch, A.D.; Navani, M.; Naeem, M.A.; Suk, J.; Chu, J.; Khan, S.N.; Poleman, R.; Akram, J.; et al. Investigating the Molecular Basis of Retinal Degeneration in a Familial Cohort of Pakistani Decent by Exome Sequencing. PLoS ONE 2015, 10, e0136561. [CrossRef] [ ] 19. Hüttl, S.; Michalakis, S.; Seeliger, M.; Luo, D.G.; Acar, N.; Geiger, H.; Hudl, K.; Mader, R.; Haverkamp, S.; Moser, M.; et al. Impaired channel targeting and retinal degeneration in mice lacking the cyclic nucleotide-gated channel subunit CNGB1. J. References Neurosci. 2005, 25, 130–138. [CrossRef] 20. Winkler, P.A.; Ekenstedt, K.J.; Occelli, L.M.; Frattaroli, A.V.; Bartoe, J.T.; Venta, P.J.; Petersen-Jones, S.M. A large animal model for CNGB1 autosomal recessive retinitis pigmentosa. PLoS ONE 2013, 8, e72229. [CrossRef] 21. Koch, S.; Sothilingam, V.; Garcia Garrido, M.; Tanimoto, N.; Becirovic, E.; Koch, F.; Seide, C.; Beck, S.C.; Seeliger, M.W.; Biel, M.; et al. Gene therapy restores vision and delays degeneration in the CNGB1(−/−) mouse model of retinitis pigmentosa. Hum. Mol. Genet. 2012, 21, 4486–4496. [CrossRef] 22. Toms, M.; Dubis, A.M.; de Vrieze, E.; Tracey-White, D.; Mitsios, A.; Hayes, M.; Broekman, S.; Baxendale, S.; Utoomprurkporn, N.; Bamiou, D.; et al. Clinical and preclinical therapeutic outcome metrics for USH2A-related disease. Hum. Mol. Genet. 2020, 29, 1882–1899. [CrossRef] 23. Hagag, A.M.; Mitsios, A.; Narayan, A.; Abbouda, A.; Webster, A.R.; Dubis, A.M.; Moosajee, M. Prospective deep phenotyping of choroideremia patients using multimodal structure-function approaches. Eye 2021, 35, 838–852. [CrossRef] 24. Spaide, R.F.; Curcio, C.A. Anatomical correlates to the bands seen in the outer retina by optical coherence tomography: Literature review and model. Retina 2011, 31, 1609–1619. [CrossRef] 25. Tao, L.W.; Wu, Z.; Guymer, R.H.; Luu, C.D. Ellipsoid zone on optical coherence tomography: A review. Clin. Exp. Ophthalmol. 2016, 44, 422–430. [CrossRef] 26. Birch, D.G.; Locke, K.G.; Wen, Y.; Locke, K.I.; Hoffman, D.R.; Hood, D.C. Spectral-domain optical coherence tomography measures of outer segment layer progression in patients with X-linked retinitis pigmentosa. JAMA Ophthalmol. 2013, 131, 1143–1150. [CrossRef] [ ] 27. Strampe, M.R.; Huckenpahler, A.L.; Higgins, B.P.; Tarima, S.; Visotcky, A.; Stepien, K.E.; Kay, C.N.; Carroll, J. Intraobserver Repeatability and Interobserver Reproducibility of Ellipsoid Zone Measurements in Retinitis Pigmentosa. Transl. Vis. Sci. Technol. 2018, 7, 13. [CrossRef] , , [ ] 28. Lima, L.H.; Burke, T.; Greenstein, V.C.; Chou, C.L.; Cella, W.; Yannuzzi, L.A.; Tsang, S.H. Progressive constriction of the hyperautoﬂuorescent ring in retinitis pigmentosa. Am. J. Ophthalmol. 2012, 153, 718–727. [CrossRef] 29. Csaky, K.; Ferris, F., 3rd; Chew, E.Y.; Nair, P.; Cheetham, J.K.; Duncan, J.L. Report From the NEI/FDA Endpoints Workshop on Age-Related Macular Degeneration and Inherited Retinal Diseases. Investig. Ophthalmol. Vis. Sci. 2017, 58, 3456–3463. [CrossRef] 30. Liew, G.; Strong, S.; Bradley, P.; Severn, P.; Moore, A.T.; Webster, A.R.; Mitchell, P.; Kiﬂey, A.; Michaelides, M. Prevalence of cystoid 29. Csaky, K.; Ferris, F., 3rd; Chew, E.Y.; Nair, P.; Cheetham, J.K.; Duncan, J.L. Report From the NEI/FDA Endpoints Workshop on Age-Related Macular Degeneration and Inherited Retinal Diseases. Investig. References Ophthalmol. Vis. Sci. 2017, 58, 3456–3463. [CrossRef] macular oedema, epiretinal membrane and cataract in retinitis pigmentosa. Br. J. Ophthalmol. 2019, 103, 1163–1166. [CrossRef] 31. Murakami, T.; Akimoto, M.; Ooto, S.; Suzuki, T.; Ikeda, H.; Kawagoe, N.; Takahashi, M.; Yoshimura, N. Association between abnormal autoﬂuorescence and photoreceptor disorganization in retinitis pigmentosa. Am. J. Ophthalmol. 2008, 145, 687–694. [CrossRef] [PubMed] 32. Khateb, S.; Nassisi, M.; Bujakowska, K.M.; Méjécase, C.; Condroyer, C.; Antonio, A.; Foussard, M.; Démontant, V.; Mohand-Saïd, S.; Sahel, J.A.; et al. Longitudinal Clinical Follow-up and Genetic Spectrum of Patients with Rod-Cone Dystrophy Associated with Mutations in PDE6A and PDE6B. JAMA Ophthalmol. 2019, 137, 669–679. [CrossRef] [PubMed] 33. Camino, A.; Wang, Z.; Wang, J.; Pennesi, M.E.; Yang, P.; Huang, D.; Li, D.; Jia, Y. Deep learning for the segmentation of preserved photoreceptors on en face optical coherence tomography in two inherited retinal diseases. Biomed. Opt. Express 2018, 9, 3092–3105. [CrossRef] [PubMed] 34. Anikina, E.; Georgiou, M.; Tee, J.; Webster, A.R.; Weleber, R.G.; Michaelides, M. Characterization of Retinal Function Using Microperimetry-Derived Metrics in Both Adults and Children with RPGR-Associated Retinopathy. Am. J. Ophthalmol. 2021, 234, 81–90. [CrossRef] 35. Buckley, T.M.W.; Jolly, J.K.; Menghini, M.; Wood, L.J.; Nanda, A.; MacLaren, R.E. Test-retest repeatability of microperimetry in patients with retinitis pigmentosa caused by mutations in RPGR. Clin. Exp. Ophthalmol. 2020, 48, 714–715. [CrossRef] 35. Buckley, T.M.W.; Jolly, J.K.; Menghini, M.; Wood, L.J.; Nanda, A.; MacLaren, R.E. Test-retest repeatability of microperimetry in patients with retinitis pigmentosa caused by mutations in RPGR. Clin. Exp. Ophthalmol. 2020, 48, 714–715. [CrossRef] 36. Roman, A.J.; Cideciyan, A.V.; Wu, V.; Garafalo, A.V.; Jacobson, S.G. Full-ﬁeld stimulus testing: Role in the clinic and as an outcome measure in clinical trials of severe childhood retinal disease. Prog. Retin. Eye Res. 2022, 87, 101000. [CrossRef] 37. Roman, A.J.; Cideciyan, A.V.; Aleman, T.S.; Jacobson, S.G. Full-ﬁeld stimulus testing (FST) to quantify visual perception in l bli d did t f t t t t i l Physiol Meas 2007 28 N51 N56 [C R f] 35. Buckley, T.M.W.; Jolly, J.K.; Menghini, M.; Wood, L.J.; Nanda, A.; MacLaren, R.E. Test retest repeatability of microperimetry in patients with retinitis pigmentosa caused by mutations in RPGR. Clin. Exp. Ophthalmol. 2020, 48, 714–715. [CrossRef] 36. Roman, A.J.; Cideciyan, A.V.; Wu, V.; Garafalo, A.V.; Jacobson, S.G. Full-ﬁeld stimulus testing: Role in the clinic and as an outcome measure in clinical trials of severe childhood retinal disease. Prog. Retin. References Eye Res. 2022, 87, 101000. [CrossRef] patients with retinitis pigmentosa caused by mutations in RPGR. Clin. Exp. Ophthalmol. 2020, 48, 714–715. [CrossRef] 36. Roman, A.J.; Cideciyan, A.V.; Wu, V.; Garafalo, A.V.; Jacobson, S.G. Full-ﬁeld stimulus testing: Role in the clinic and as an outcome measure in clinical trials of severe childhood retinal disease. Prog. Retin. Eye Res. 2022, 87, 101000. [CrossRef] 37. Roman, A.J.; Cideciyan, A.V.; Aleman, T.S.; Jacobson, S.G. Full-ﬁeld stimulus testing (FST) to quantify visual perception in patients with retinitis pigmentosa caused by mutations in RPGR. Clin. Exp. Ophthalmol. 2020, 48, 714 715. [CrossRef] 36. Roman, A.J.; Cideciyan, A.V.; Wu, V.; Garafalo, A.V.; Jacobson, S.G. Full-ﬁeld stimulus testing: Role in the clinic and as an outcome measure in clinical trials of severe childhood retinal disease. Prog. Retin. Eye Res. 2022, 87, 101000. [CrossRef] 37. Roman, A.J.; Cideciyan, A.V.; Aleman, T.S.; Jacobson, S.G. Full-ﬁeld stimulus testing (FST) to quantify visual perception in severely blind candidates for treatment trials. Physiol. Meas. 2007, 28, N51–N56. [CrossRef] 36. Roman, A.J.; Cideciyan, A.V.; Wu, V.; Garafalo, A.V.; Jacobson, S.G. Full-ﬁeld stimulus testing: Role in the clinic and as an outcome measure in clinical trials of severe childhood retinal disease. Prog. Retin. Eye Res. 2022, 87, 101000. [CrossRef] g y 37. Roman, A.J.; Cideciyan, A.V.; Aleman, T.S.; Jacobson, S.G. Full-ﬁeld stimulus testing (FST) to quantify visual perception in severely blind candidates for treatment trials. Physiol. Meas. 2007, 28, N51–N56. [CrossRef] 37. Roman, A.J.; Cideciyan, A.V.; Aleman, T.S.; Jacobson, S.G. Full ﬁeld stimulus testing (FST) to q severely blind candidates for treatment trials. Physiol. Meas. 2007, 28, N51–N56. [CrossRef] 38. Gill, J.S.; Theofylaktopoulos, V.; Mitsios, A.; Houston, S.; Hagag, A.M.; Dubis, A.M.; Moosajee, M. Investigating Biomarkers for USH2A Retinopathy Using Multimodal Retinal Imaging. Int. J. Mol. Sci. 2022, 23, 4198. [CrossRef]
https://openalex.org/W4220670423	https://www.nature.com/articles/s41598-022-09131-x.pdf	English	null	Long-term nutritional trends in the Finnish population estimated from a large laboratory database from 1987 to 2020	Scientific reports	2,022	cc-by	7,325	www.nature.com/scientificreports www.nature.com/scientificreports Long‑term nutritional trends in the Finnish population estimated from a large laboratory database from 1987 to 2020 Tamara Tuuminen1,2, Mikko Sorsa2,3, Martin Tornudd2, Pertti Lauri Lähteenmäki1, Tuija Poussa4, Pyry Suonsivu1, Eeva Marja Pitkänen2, Erkki Antila1 & Kaarlo Jaakkola2 Tamara Tuuminen1,2, Mikko Sorsa2,3, Martin Tornudd2, Pertti Lauri Lähteenmäki1, Tuija Poussa4, Pyry Suonsivu1, Eeva Marja Pitkänen2, Erkki Antila1 & Kaarlo Jaakkola2 The assessments of malnutrition in adults with MUST or NRS-2002 criteria do not give a detailed insight into the sufficiency of micronutrients. Sufficiency assessment of essential micronutrients on the individual level can be achieved only with laboratory measurements. The aim of this study was to estimate long-term trends in micronutrient sufficiency in the Finnish population with regards to gender and sex covariates. We retrieved from the clinical laboratory database (n = 67,236) all results on whole blood Magnesium, (B-Mg), Manganese (B-Mn), Zinc (B-Zn), Selenium (B-Se) and Copper from erythrocytes (E-Cu) and fasting serum β-carotenes (fS-BKarot), vitamin A (fS-A-vit), coenzyme Q10 (Ubiquinone, fS-Q10) and serum vitamin D (S-D-25) from the database of clinical laboratory Mineraalilaboratorio Mila Oy from the years 1987–2020. A weak positive linear trend is seen for B-Mg, B-Zn and ln(fS-Q10) both for children and adults, but a moderate linear positive trend was observed for ln(S-D-25) based on correlation between calendar year and ln(S-D-25), R = 0.44 and 0.41, p < 0.001 for adults and children, respectively. Laboratory database is helpful to monitor the nutritional public policy to prevent hidden malnutrition in the society. 1 fic Reports \| (2022) 12:5008 \| https://doi.org/10.1038/s41598-022-09131-x Abbreviations AAS Atomic absorption spectroscopy A-vit Vitamin A B Blood BKarot β-Carotenes CLIA Chemiluminescence immunoassay Cu Copper DAD Diode array detection D-25 Vitamin D E Erythrocytes ECLIA Electro-chemiluminescence immunoassay fS Fasting serum HPLC High pressure liquid chromatography ICP-OES Inductively coupled plasma atomic emission spectroscopy LC–MS/MS Liquid chromatography-tandem mass spectrometry Mg Magnesium Mn Manganese Se Selenium Zn Zinc Q10 Coenzyme Q10, Ubiquinone UPLC Ultra performance liquid chromatography Malnutrition is a deficiency of not only macronutrients, such as proteins, fat, and carbohydrates but also insuf- ficiency of micronutrients such as vitamins, microelements, co-factor molecules, amino acids, and balanced fatty acids. Deficiency of micronutrients may impair inter alia immune and endocrine systems because many 1Medical Center Kruunuhaka Oy, Kaisaniemenkatu 1Ba, Helsinki, Finland. 2Mineraalilaboratorio Mila Oy, Helsinki, Finland. 3Solu Digital Oy, Helsinki, Finland. 4STAT-Consulting, Nokia, Finland. Scientific Reports \| (2022) 12:5008 Long‑term nutritional trends in the Finnish population estimated from a large laboratory database from 1987 to 2020 email: Tuuminen@gmail.com Abbreviations AAS Atomic absorption spectroscopy A-vit Vitamin A B Blood BKarot β-Carotenes CLIA Chemiluminescence immunoassay Cu Copper DAD Diode array detection D-25 Vitamin D E Erythrocytes ECLIA Electro-chemiluminescence immunoassay fS Fasting serum HPLC High pressure liquid chromatography ICP-OES Inductively coupled plasma atomic emission spectroscopy LC–MS/MS Liquid chromatography-tandem mass spectrometry Mg Magnesium Mn Manganese Se Selenium Zn Zinc Q10 Coenzyme Q10, Ubiquinone UPLC Ultra performance liquid chromatography Malnutrition is a deficiency of not only macronutrients, such as proteins, fat, and carbohydrates but also insuf- ficiency of micronutrients such as vitamins, microelements, co-factor molecules, amino acids, and balanced fatty acids. Deficiency of micronutrients may impair inter alia immune and endocrine systems because many 1Medical Center Kruunuhaka Oy, Kaisaniemenkatu 1Ba, Helsinki, Finland. 2Mineraalilaboratorio Mila Oy, Helsinki, Finland. 3Solu Digital Oy, Helsinki, Finland. 4STAT-Consulting, Nokia, Finland. email: Tuuminen@gmail.com Scientific Reports \| (2022) 12:5008 \| https://doi.org/10.1038/s41598-022-09131-x www.nature.com/scientificreports/ Table 1. Pearson’s correlation coefficients (R) between the calendar year and micronutrients in adults and children. Year vs. micronutrient Adults Children R P value N R P value N B-Mg 0.24 < 0.001 48,372 0.16 < 0.001 2,731 B-Mn − 0.03 < 0.001 39,433 − 0.02 0.34 2,145 B-Zn 0.19 < 0.001 51,396 0.17 < 0.001 3,452 E-Cu 0.02 < 0.001 42,741 − 0.03 0.10 2,357 fS-A-Vit 0.06 < 0.001 34,802 0.02 0.20 2,616 ln-(B-Se) 0.11 < 0.001 53,464 0.05 0.01 3,396 ln-(fS-Bkarot) 0.01 0.04 32,950 − 0.06 0.005 2,191 ln (fS-Q10) 0.19 < 0.001 39,693 0.14 < 0.001 2,425 ln (S-D-25) 0.44 < 0.001 19,693 0.41 < 0.001 1,586 Year vs. micronutrient Adults Children R P value N R P value N B-Mg 0.24 < 0.001 48,372 0.16 < 0.001 2,731 B-Mn − 0.03 < 0.001 39,433 − 0.02 0.34 2,145 B-Zn 0.19 < 0.001 51,396 0.17 < 0.001 3,452 E-Cu 0.02 < 0.001 42,741 − 0.03 0.10 2,357 fS-A-Vit 0.06 < 0.001 34,802 0.02 0.20 2,616 ln-(B-Se) 0.11 < 0.001 53,464 0.05 0.01 3,396 ln-(fS-Bkarot) 0.01 0.04 32,950 − 0.06 0.005 2,191 ln (fS-Q10) 0.19 < 0.001 39,693 0.14 < 0.001 2,425 ln (S-D-25) 0.44 < 0.001 19,693 0.41 < 0.001 1,586 Table 1. Pearson’s correlation coefficients (R) between the calendar year and micronutrients in adults and children. micronutrients have bioactive and immunomodulating properties. Macronutrients are the natural carriers of micronutrients. Deficiency of macronutrients may lead to low body mass, loss of muscle mass and disturbed energy supply1–3. Long‑term nutritional trends in the Finnish population estimated from a large laboratory database from 1987 to 2020 Persons with obesity are also categorised as malnourished.h micronutrients have bioactive and immunomodulating properties. Macronutrients are the natural carriers of micronutrients. Deficiency of macronutrients may lead to low body mass, loss of muscle mass and disturbed energy supply1–3. Persons with obesity are also categorised as malnourished.h There are several ways to study the nutritional status of a person and assess overt or hidden malnutrition. Detection of malnutrition is very important in highly vulnerable patient groups, patients with polymorbidities and chronic diseases such as e.g., cancers1,3. Nutritional assessment has become increasingly important in times of CoVID-19 pandemic to predict and prevent poor outcomes and reduce mortality in vulnerable populations2. As has been recently outlined, the screening and the assessment of malnutrition in out-patient adults should be initially assessed with the MUST criteria (MUST criteria: see https://www.bapen.org.uk/screening-and-must/ must-calculator2 or in hospitalized and geriatric patients and especially in polymorbid patients with the NRS- 2002 criteria (NRS-2002 criteria: https://www.mdcalc.com/nutrition-risk-screening-2002-nrs-20022. However, these methods are only descriptive screening methods and do not provide a detailed insight into the sufficiency of micronutrients. The only way to estimate the sufficiency of all essential micronutrients is to measure them in the laboratory from patients’ samples.The measurement of micronutrients on the individual level allows performing personalized medicine, to predict and prevent deficiencies and insufficiencies and thus improve the health of the population.High variability of the measurement techniques, reference values, or inconsistency of selection of bio- logical sample materials such as plasma(sera) or whole blood hampers however between-laboratory comparison.h g p p ( ) p y p The current literature reviews micronutrient monitoring in several vulnerable population groups, such as e.g. pregnant women4 and frailty5; in patients with chronic conditions, such as chronic liver diseases6, in metabolic syndrome including central obesity, insulin resistance, hypertension, glucose intolerance, and dyslipidemia7, the status after bypass bariatric surgery8, in conditions related to the bone metabolism9, etc. Excessive literature exists to confirm a positive correlation between the micronutrient sufficiency and the resilience to infections10,11, the somewhat neglected postulate that needs to be emphasized during the current CoVID-19 pandemics. Long‑term nutritional trends in the Finnish population estimated from a large laboratory database from 1987 to 2020 However, to the best of our knowledge, a retrospective review of the laboratory database to assess nutritional trends in the population from the public health perspective has not yet been published, at least the data covering three decades.l y g We postulated that laboratory data accumulated during 1987–2020 from our out-patients may reflect the trends in the general Finnish population. The primary aim of this study was to estimate long-term trends in of the levels of micronutrients through the laboratory data acquisition. The secondary aim was to estimate whether age or gender is associated with the level of the micronutrient variables. Resultsh This database includes a total of 67,236 samples from 1987 to 2020. Most samples were from females 45,696 (68%) and 21,540 (32%) samples from males. Median age of male and female subjects was 51 years (range 1–96 years). From adults (age ≥ 18 years) there were 63,113 (94%) samples and from children (age < 18 years) there were 4123 (6%) samples. The majority (59%) of samples was from the age-group 40–69 years. Linear trends in micronutrients. Table1 presents the results of the Pearson s correlation coefficients (R) that assesses a linear trend in the levels of micronutrient during 1987–2020 in adults and in children. Figure 1 illustrates the yearly levels of selected micronutrients.i y y As seen from the Table 1, all correlations were significant due to the large sample size. A weak positive linear trend is seen for B-Mg, B-Zn and ln(fS-Q10), but the moderate linear positive trend is seen for ln(S-D-25). T bl 2 d Fi 1 t thi b ti I t tl i il t d b d f d lt d hild y y As seen from the Table 1, all correlations were significant due to the large sample size. A weak positive linear trend is seen for B-Mg, B-Zn and ln(fS-Q10), but the moderate linear positive trend is seen for ln(S-D-25). Table 2 and Fig. 1 support this observation. Importantly, similar trends were observed for adults and children. R2=coefficient of determination i trend is seen for B-Mg, B-Zn and ln(fS-Q10), but the moderate linear positive trend is seen for ln(S-D-25). Table 2 and Fig. 1 support this observation. Importantly, similar trends were observed for adults and children. R2 = coefficient of determination.fi fi It can be implicated from this analysis that e.g., for B-Mg the regression coefficient B of 0.004 means that each year the average B-Mg increased by 0.004 units. The coefficient of determination R2 = 0.06 means that 6% of the variance of B-Mg is explained by the year change. All R2 values were below 0.20 indicating that the levels of micronutrients were explained mainly by other factors. https://doi.org/10.1038/s41598-022-09131-x Scientific Reports \| (2022) 12:5008 \| www.nature.com/scientificreports/ Figure 1. Micronutrient levels in adults (● age ≥ 18) and in children (■ age < 18) measured during 1987–202 (A) Means (95% CI) of B-Mg and B-Zn. (B) Geometric means (95% CI) of fS-Q10 and S-D-25. Resultsh The numbers o results varied from year to year: B-Mg (n = 130–2347 for adults and n = 8–165 for children); B-Zn (151–2523 a 8–201); fS-Q10 (725–1889 and 25–132); S-D-25 (800–1854 and 76–143). Adults Children Figure 1. Micronutrient levels in adults (● age ≥ 18) and in children (■ age < 18) measured during 1987–2020. (A) Means (95% CI) of B-Mg and B-Zn. (B) Geometric means (95% CI) of fS-Q10 and S-D-25. The numbers of results varied from year to year: B-Mg (n = 130–2347 for adults and n = 8–165 for children); B-Zn (151–2523 and 8–201); fS-Q10 (725–1889 and 25–132); S-D-25 (800–1854 and 76–143). Figure 1. Micronutrient levels in adults (● age ≥ 18) and in children (■ age < 18) measured during 1987–2020. (A) Means (95% CI) of B-Mg and B-Zn. (B) Geometric means (95% CI) of fS-Q10 and S-D-25. The numbers of results varied from year to year: B-Mg (n = 130–2347 for adults and n = 8–165 for children); B-Zn (151–2523 and 8–201); fS-Q10 (725–1889 and 25–132); S-D-25 (800–1854 and 76–143). Figure 1. Micronutrient levels in adults (● age ≥ 18) and in children (■ age < 18) measured during 1987–2020. (A) Means (95% CI) of B-Mg and B-Zn. (B) Geometric means (95% CI) of fS-Q10 and S-D-25. The numbers of results varied from year to year: B-Mg (n = 130–2347 for adults and n = 8–165 for children); B-Zn (151–2523 and 8–201); fS-Q10 (725–1889 and 25–132); S-D-25 (800–1854 and 76–143). Table 2. The slope and the strength (coefficients of determination R2) of linear trend for several micronutrients. Adults Children Regression coefficient B (P value) R2 Regression coefficient B (P value) R2 B-Mg 0.004 (< 0.001) 0.06 0.002 (< 0.001) 0.03 B-Zn 0.267 (< 0.001) 0.04 0.227 (< 0.001) 0.03 ln (fS-Q10) 0.012 (< 0.001) 0.03 0.007 (< 0.001) 0.02 ln (S-D-25) 0.057 (< 0.001) 0.19 0.047 (< 0.001) 0.17 Table 2. The slope and the strength (coefficients of determination R2) of linear trend for several micronutrients. https://doi.org/10.1038/s41598-022-09131-x Scientific Reports \| (2022) 12:5008 \| www.nature.com/scientificreports/ Table 3. Adjusted means (95% CI) for micronutrients in adults (age ≥ 18 years) measured between 1987 and 2020 (A: Females vs. males B: Age ≥ 50 years vs. 18–49 years). Adjustment was based on the general linear model, where age group (≥ 50 years vs. 18–49 years), gender, and 5-year time periods were included as categorical covariates. Resultsh Adjustment was based on the general linear model, where age group (≥ 50 years vs. 18–49 years), gender, and 5-year time periods were included as categorical covariates. B-Se, fS-BKarot, fS-Q10 and S-D3-25 were logarithmically transformed before analysis. GM geometric mean, RGM ratio of geometric means. Table 3. Adjusted means (95% CI) for micronutrients in adults (age ≥ 18 years) measured between 1987 and 2020 (A: Females vs. males B: Age ≥ 50 years vs. 18–49 years). Adjustment was based on the general linear model, where age group (≥ 50 years vs. 18–49 years), gender, and 5-year time periods were included as categorical covariates. B-Se, fS-BKarot, fS-Q10 and S-D3-25 were logarithmically transformed before analysis. GM geometric mean, RGM ratio of geometric means. and 2020 (A: Females vs. males B: Age ≥ 50 years vs. 18–49 years). Adjustment was based on the general linear model, where age group (≥ 50 years vs. 18–49 years), gender, and 5-year time periods were included a categorical covariates. B-Se, fS-BKarot, fS-Q10 and S-D3-25 were logarithmically transformed before analys GM geometric mean, RGM ratio of geometric means. Influence of gender and age on micronutrients. Next, we wanted to test whether the gender or age have a significant relationship with the level of the micronutrient variable. Table 3 illustrates the adjusted means (95% CI) of micronutrients in adult females and males and in age groups ≥ 50 years and 18–49 years, as well as the differences between those groups. Table 4 illustrates the adjusted means (95% CI) of micronutrients in chil- dren in age groups 10–17 years and < 10 years, as well as the differences between those groups. g g p y yf g p Adjustment was based on the general linear model, where age group (≥ 50 years vs. 18–49 years), gender, and 5-year time periods were included as categorical covariates. B-Se, fS-BKarot, fS-Q10 and S-D3-25 were logarithmically transformed before analysis.h g y y The general linear models showed that: (1) For adults and for children and for all nutrient variables there were significant differences when the adjusted 5-year periods were compared (global P < 0.001), (2) Gender had a significant effect on all nutrient variables except on B-Se (Table 3). (3) Age had a significant effect on nutrient variables except on B-Mn in adults (Tables 3 and 4). Resultsh Importantly, the effects of gender and age were small and were statistically significant due to the large sample size, as seen from the Tables 3 and 4. Correlation between the levels of selected micronutrients. Finally, we wanted to test the possible correlation between the levels of selected micronutrients. We selected E-Cu vs. B-Zn; fS-BKarot vs. fS-A-vit and fS-Q10 vs. B-Se because these pairs of micronutrients have either a synergistic effect (fS-BKarot and fS-A-vit, fS-Q10 and B-Se) or are competitors for the absorption in the intestine (E-Cu and B-Zn).The correlations of selected pairs of nutrient variables were R = 0.07 (P < 0.001) for E-Cu vs. B-Zn, R = 0.08 (P < 0.001) for ln(fS- BKarot) vs. fS-A-Vit and R = 0.36 (P < 0.001) for ln(fS-Q10) vs. ln(B-Se). Even though the P values were signifi- cant, only the latter correlation indicated a weak linear association. Figure 2 illustrates the correlation between the logarithms of blood selenium levels and serum ubiquinone values. Resultsh B-Se, fS-BKarot, fS-Q10 and S-D3-25 were logarithmically transformed before analysis. GM geometric mean, RGM ratio of geometric means. A Females Males Females vs. males N Adj. Mean 95% CI N Adj. Mean 95% CI Adj. Diff 95% CI P B-Mg (mmol/L) 32,821 1.396 1.394–1.397 15,551 1.454 1.452–1.457 − 0.059 − 0.062 to − 0.055 < 0.001 B-Mn (μmol/L) 26,212 0.184 0.183–0.185 13,221 0.170 0.168–0.171 0.015 0.013 to 0.016 < 0.001 B-Zn (μmol/L) 35,472 83.9 83.8–84.0 15,924 91.0 90.8–91.2 − 7.1 − 7.3 to − 6.9 < 0.001 E-Cu (mol/L) 28,932 10.12 10.11–10.14 13,809 10.29 10.27–10.31 − 0.17 − 0.19 to − 0.14 < 0.001 fS-A-vit (μmol/L) 23,960 2.04 2.03–2.05 10,842 2.29 2.27–2.30 − 0.25 − 0.27 to − 0.24 < 0.001 N GM 95% CI N Adj. GM 95% CI Adj. RGM 95% CI P B-Se (μmol/L) 36,815 2.08 2.07–2.09 16,648 2.08 2.07–2.09 1.00 0.99 to 1.01 0.87 fS-BKarot (μmol/L) 21,935 0.85 0.84–0.86 11,008 0.63 0.62–0.64 1.35 1.33 to 1.38 < 0.001 fS-Q10 (μmol/L) 26,633 1.23 1.22–1.24 13,060 1.34 1.33–1.35 0.92 0.91 to 0.93 < 0.001 S-D3-25 (nmol/L) 13,639 68.7 68.2–69.2 6054 66.6 65.9–67.3 1.03 1.02 to 1.04 < 0.001 B Age 18–49 years Age ≥ 50 years Age ≥ 50 years vs. 18–49 years N Adj. Mean 95% CI N Adj. Mean 95% CI Adj. Diff 95% CI P B-Mg (mmol/L) 19,898 1.413 1.411–1.415 28,474 1.437 1.435–1.439 0.024 0.021 to 0.026 < 0.001 B-Mn (μmol/L) 16,059 0.177 0.176–0.178 22,374 0.177 0.176–0.178 − 0.001 − 0.002 to 0.001 0.38 B-Zn (μmol/L) 21,617 86.7 86.5–86.8 29,779 88.3 88.1–88.4 1.6 1.4 to 1.8 < 0.001 E-Cu (mol/L) 17,558 10.29 10.27–10.31 25,183 10.12 10.11–10.14 − 0.16 − 0.19 to − 0.143 < 0.001 fS-A-vit (μmol/L) 15,437 2.11 2.09–2.12 19,365 2.22 2.21–2.23 0.12 0.10 to 0.13 < 0.001 N Adj. GM 95% CI N Adj. GM 95% CI Adj. RGM 95% CI P B-Se (μmol/L) 22,322 2.02 2.01–2.03 31,142 2.14 2.13–2.15 1.06 1.05 to 1.07 < 0.001 fS-BKarot (μmol/L) 14,274 0.70 0.69–0.72 18,669 0.76 0.75–0.77 1.08 1.06 to 1.10 < 0.001 fS-Q10 (μmol/L) 16,795 1.16 1.15–1.17 22,898 1.42 1.41–1.43 1.22 1.21 to 1.23 < 0.001 S-D3-25 (nmol/L) 8503 64.3 63.7–64.9 11,190 71.1 70.5–71.7 1.11 1.09 to 1.12 < 0.001 Table 3. Adjusted means (95% CI) for micronutrients in adults (age ≥ 18 years) measured between 1987 and 2020 (A: Females vs. males B: Age ≥ 50 years vs. 18–49 years). Discussion Adjustment was based on the general linear model, where age group (10–17 years vs. < 10 years) and 5-year time periods were included as categorical covariates. B-Se, fS-BKarot, fS-Q10 and S-D3-25 were logarithmically transformed before analysis. GM geometric mean, RGM ratio of geometric means. Table 4. Adjusted means (95% CI) for micronutrients in children (age < 10 years vs. age 10–17 years) measured between 1987 and 2020. Adjustment was based on the general linear model, where age group (10–17 years vs. < 10 years) and 5-year time periods were included as categorical covariates. B-Se, fS-BKarot, fS-Q10 and S-D3-25 were logarithmically transformed before analysis. GM geometric mean, RGM ratio of geometric means. Figure 2. Scatter plot with regression line to illustrate the relationship between fS-Q10 and B-Se (R = 0.36, p < 0.001), both being logarithmically transformed. The results for children and adults are combined (n = 42,118). Figure 2. Scatter plot with regression line to illustrate the relationship between fS-Q10 and B-Se (R = 0.36, p < 0.001), both being logarithmically transformed. The results for children and adults are combined (n = 42,118). Figure 2. Scatter plot with regression line to illustrate the relationship between fS-Q10 and B-Se (R = 0.36, p < 0.001), both being logarithmically transformed. The results for children and adults are combined (n = 42,118). population of both genders and all ages. The biggest limitation to this extrapolation is that that individuals w lower income might use less private sector services, therefore their results were underrepresented in this st d fl d l d h l l12 Low grade inflammation is very common and poorly reported in the population in general12. In our earlier study25 we found that the effect of inflammation on the measurement of β-carotene was dramatic. In situations of severe inflammation, the decline was 90% and in mild inflammation the levels may be halved. Inflamma- tion interacted with the fS-Q10 levels, and the levels of vitamin A. Inflammation was not considered here as a confounding factor because we used a large sample size collected for 34 years. Rather, we were interested in the long-term trends of several micronutrients. Based on the analysis of this study, we can draw several conclusions: (1) For B-Mg, B-Zn, and ln (fs-Q10) an indication of the positive linear weak trend was detected. (2) A moderate positive trend was observed for ln(S- D-25). Discussion Here, we present the analysis of the trends in several essential micronutrients measured from our out-patients during the years 1987–2020. We believe that the trends observed in this study can be generalized with some limitations to the Finnish population in general because our patients were not hospitalized but represent mixed https://doi.org/10.1038/s41598-022-09131-x Scientific Reports \| (2022) 12:5008 \| www.nature.com/scientificreports/ Table 4. Adjusted means (95% CI) for micronutrients in children (age < 10 years vs. age 10–17 years) measured between 1987 and 2020. Adjustment was based on the general linear model, where age group (10–17 years vs. < 10 years) and 5-year time periods were included as categorical covariates. B-Se, fS-BKarot, fS-Q10 and S-D3-25 were logarithmically transformed before analysis. GM geometric mean, RGM ratio of geometric means. Age < 10 years Age 10–17 years Age 10–17 years vs. Age < 10 years N Adj. Mean 95% CI N Adj. Mean 95% CI Adj. Diff 95% CI P B-Mg (mmol/L) 1179 1.43 1.42–1.44 1552 1.42 1.41–1.42 − 0.02 − 0.03 to − 0.01 0.001 B-Mn (μmol/L) 953 0.201 0.196–0.205 1192 0.194 0.190–0.198 − 0.006 − 0.011 to − 0.001 0.016 B-Zn (μmol/L) 1585 70.6 70.0–71.2 1867 78.0 77.4–78.7 7.5 6.6 to 8.4 < 0.001 E-Cu (mol/L) 1013 11.23 11.14–11.32 1344 10.78 10.68–10.88 − 0.45 − 0.58 to − 0.32 < 0.001 fS-A-vit (μmol/L) 1201 1.36 1.33–1.39 1415 1.70 1.66–1.73 0.34 0.30 to 0.37 < 0.001 N Adj. GM 95% CI N Adj. GM 95% CI Adj. RGM 95% CI P B-Se (μmol/L) 1534 1.74 1.72–1.77 1862 1.82 1.79–1.85 1.05 1.02 to 1.07 < 0.001 fS-BKarot (μmol/L) 916 0.75 0.72–0.79 1274 0.60 0.58–0.63 0.80 0.75 to 0.85 < 0.001 fS-Q10 (μmol/L) 1012 1.03 1.00–1.06 1413 0.94 0.92–0.96 0.91 0.88 to 0.94 < 0.001 S-D3-25 (nmol/L) 684 68.5 66.5–70.6 902 60.2 58.6–61.8 0.88 0.84 to 0.92 < 0.001 Table 4. Adjusted means (95% CI) for micronutrients in children (age < 10 years vs. age 10–17 years) measured between 1987 and 2020. Adjustment was based on the general linear model, where age group (10–17 years vs. < 10 years) and 5-year time periods were included as categorical covariates. B-Se, fS-BKarot, fS-Q10 and S-D3-25 were logarithmically transformed before analysis. GM geometric mean, RGM ratio of geometric means. Table 4. Adjusted means (95% CI) for micronutrients in children (age < 10 years vs. age 10–17 years) measured between 1987 and 2020. www.nature.com/scientificreports/ the routine doctor’s check-up. However, the awareness of the vitamin D immunological and endocrine effects in children and adults13,14 has grown among general population due to good access to medical and popular literature and advertisements by the manufacturers. This awareness has grown at the times of the CoVID-19 -pandemic15. yh g p Vitamin D is probably the best standardized test across variable techniques. In our laboratory vitamin is measured with HPLC but in other laboratories LC–MS/MS (liquid chromatography-tandem mass spectrometry), chemiluminescence, ECLIA, CLIA or immunoassay are used. According to The External Quality Assessment (EQA) scheme for vitamin D metabolites (DEQAS, https://deqas.org)16, some constituents in a sample may have a matrix effect in the ligand binding assays causing inter-sample variability. Matrix does not affect accuracy of the HPLC and LC–MS/MS measurements. It is universally established that the levels < 25, 25–50; 50–75; 75–130 and 375 nmol/L are considered deficient, insufficient, adequate, target values and toxic levels, respectively13. Associa- tion between low levels of serum vitamin D and increased risk of developing several immune-related diseases, including psoriasis, type 1 diabetes, multiple sclerosis, rheumatoid arthritis, tuberculosis, sepsis, respiratory infection, cancer, and CoVID-19, has been observed13,17. However, some individuals might benefit from vitamin D more or less than others as high inter-individual variation in response to vitamin D supplementation has been observed13. It should be noted that the levels of vitamin D are affected by seasonal changes being dependent on the solar UV radiation and the exposure to it in latitudes 60–67.5. Therefore, samples collected during the period from September to May reflect primarily nutritional sufficiency than the sun exposure. The results obtained through our study resulting in the average concentration of vitamin D of 68.2–69.2 and 65.9–67.3 nmol/L in females and males, respectively are higher than those reported in the FINRISK Study 1997 Survey performed from February through March 1998 on healthy volunteers. The mean values of vitamin D were 47 ± 34 nmol/L and 45 ± 35 nmol/L (mean ± SD) for females (n = 202) and males (n = 126), respectively18. Our results are in accordance with the recently published research data from the Northern Finland Birth Cohort that the level of vitamin D has increased in the Finnish middle-aged population during the period 1997–201319. www.nature.com/scientificreports/ g p p g p Determination of vitamin A, β-carotenes and vitamin E (not studied here) from serum, is usually done by HPLC or UPLC (Scottish STEMDRL https://www.trace-elements.co.uk) or by LC–MS/MS (Mayo Clinic, https:// www.mayoclinic.org)). For children, the reference values are usually presented for each age group but for adults the reference values normally are in the range 1.0–3.0 (Scottish STEMDRL) or 1.1–2.7 or 1.1–2.7 (Mayo Clinic). In our laboratory the reference values were calculated as1.4–3.7 and 1.1–3.1 µmol/L for males and females, respectively. Our analysis did not find any trend in the levels of vitamin A, β-carotenes; neither did we found any correlation between the values. This study did not reveal whether there is an insufficiency of these micronutrients in the Finnish population. The importance of vitamin A and β-carotenes is difficult to overestimate. Vitamin A is essential for fetal development and for ocular integrity. The deficiency of vitamin A remains the leading cause of preventable blindness in the world. The deficiency can cause also anemia, and weak resistance to infections. In developed countries, higher consumption of vitamin A, C and E was associated with a significant decreased risk of age-related cataract20. g Usually, determination of ubiquinone (Q10) from serum is done by HPLC with electrochemical detection or using LC–MS/MS methodology. Reduced form of ubiquinone is also measured (Mayo Clinic). There are some variations regarding the reference values being0.42–2.50; 0.43–2.55; 0.50–1.77 and 0.48–1.71 µmol/L. There was a slight positive linear trend over the years in the levels of Q10.This trend can be explained by the aware- ness of the population that the supplementation of Q10 together with selenium may reduce heart complications and improve the overall mental and physical fitness. The four-year prospective placebo-controlled randomized clinical trial found that the overall mortality in elderly has been decreased at the observation time of 12 years21.h y y y There are several methods to measure trace elements from biological samples: the AAS (atomic absorption spectroscopy) and the ICP-MS method. The reference values for E-Zn may vary e.g., 67–132 (Labcorp, https:// www.labcorp.com) to 153–245 µmol/L (Mayo Clinic). The reference values in our laboratory are 65–107 and 68–115 for females and males, respectively. www.nature.com/scientificreports/ The reference values for selenium measured from the whole blood is difficult to compare between the laboratories whereas reference levels for B-Mn are similar being approximately in the range 0.06–0.34 µmol/L (for Labcorp, Mayo Clinic, Scottish STEMDRL).fi g µ p y Due to the extremely low Se intake in the 1970s in Finland, an official decision was made in 1984 to sup- plement multinutrient fertilizers with sodium selenate. Almost all fertilizers used in Finland since 1985 have contained Se22. Despite this public health measure, we could not observe any positive linear trend for blood selenium over the years. Instead, only a mild trend was seen for blood magnesium and zinc.f Trace elements and vitamins exert a combined effect on the function of immune system, and their optimamal and balance concentrations against each other is essential to combat virus infections, and current CoVID-19 pandemics in the first place. sufficiency of vitamin A,C, E, and D and zink for the proper function of the immune cells has been recently reviewed23. y In conclusion, laboratory measurement of micronutrients is a useful tool to assess not only the nutritional status of individuals but to monitor the nutritional policy for the public health.. Only a weak positive trend from 1987 to 2020 was detected for B-Mg, B-Zn and (ln)fS-Q10, both for children and adults whereas no trend was observed for E-Cu, fS-BKarot, fS-A-vit and B-Se despite its use in fertilizers. To our big satisfaction, a moderate positive trend has been detected for the levels of ln(S-D-25) vitamin in adults and children. These data under- line that the usage of laboratory measurement databases is useful for monitoring of nutritional policy among population. Discussion (3) This trend was seen for adults and children. (4) There was a significant positive correlation between micronutrients E-Cu vs. B-Zn, between ln(fS-BKarot) vs. fS-A-vit and between ln(fS-Q10) vs. ln(B-Se). However, the large sample size should be taken into consideration and only the last correlation ln(fS-Q10) vs. ln(B-Se) indicated a meaningful positive correlation.h The large sample size used in this study provides an excellent platform to estimate long-term trends in the general population. But there are some pitfalls. Although statistical analysis may point to a significant change these changes are necessary to estimate from the prisms of its clinical significance. The major observation was that over the years the levels of vitamin D has increased in the Finnish population. This is a very welcome trend since the importance of vitamin D sufficiency cannot be overestimated. Unfortunately, today the importance of vitamin D sufficiency has not yet been clearly emphasized by the Finnish public health sector, and it is rarely measured at Scientific Reports \| (2022) 12:5008 \| https://doi.org/10.1038/s41598-022-09131-x www.nature.com/scientificreports/ www.nature.com/scientificreports/ serum β-carotenes,(fS-BKarot, µmol/L),vitamin A, (fS-A-vit, µmol/L), coenzyme Q10 (Ubiquinone, fS-Q10, µmol/L) and serum vitamin D, (S-D-25, nmol/L). Altogether 67236 results from five out-patient clinics in Fin- land obtained during 1987–2020 were analysed. We collected data also on gender and age.h g y g g The prefaces “fB” refers to a fasting sample, blood; “fS”—the fasting sample, serum; “E” means that the analysis was performed from erythrocytes. The rationale for determination of micronutrients from blood was to accurate estimate the long-term nutrient sufficiency because micronutrients are normally stored in cells, whereas in the serum their concentrations are less stable4,5. The concentration of copper is best measured from washed eryth- rocytes since this minimizes the influence of acute phase reactants24,25. Analytical methods. E-Cu was analysed with atomic absorption spectroscopy (AAS) from 1987 to 2002, thereafter with inductively coupled plasma atomic emission spectroscopy (ICP-OES) and with ICP-MS (induc- tively coupled plasma mass spectrometry) from 2014. B-Se and B-Mn were analysed by AAS until 2007 and 2010, respectively. B-Mg and B-Zn were analyzed by ICP-OES until 2014 and then with ICP-MS. B-Mnwere ana- lysed by the in-house acid digestion and the highly sensitive ICP-MS). The ICP-MS is a type of mass spectrom- etry, where inductively coupled plasma is used to ionize the sample. The sample is atomized, and then atomic and small polyatomic ions are detected (Mineraalilaboratorio Mila Ltd (https://mineraalilaboratoriomila.fi). The values for E-Cu obtained with the new method starting from 2003 were systematically 5% lower than with AAS method, which was acknowledged in the present study. For the other microelements there were no shift in the measured values across the whole observation time.h The fS-BKarot and fS-A-vit are in-house methods (diode array detection (DAD), reverse phase high pressure liquid chromatography (HPLC); fS-Q10 is an in-house method (electrochemical detection, reverse phase HPLC); S-D-25 was detected since 2007 by HPLC. Statistical analysis. First, we wanted to answer the primary question whether there is a linear trend towards an increase or a decrease of micronutrient variables during 1987–2010. The data were plotted as a function of time (Fig. 1), the Pearson correlation coefficient (R) was calculated between calendar year and the micronutrient variable, and the regression analysis was performed for adult and children’s samples (Tables 1,2) to illustrate how steeply the increase or decrease happened per one year. Secondly, the micronutrient variables were analyzed using the general linear model, where age groups (≥ 50 years vs. Received: 13 August 2021; Accepted: 14 March 2022 www.nature.com/scientificreports/ 18–49 years), gender, and 5-year time periods were included as categorical covariates. We grouped the results into the following 5-year periods: 1987–1990; 1991–1995; 1996–2000; 2001–2005; 2006–2010; 2011–2015 and 2016–2020 because general linear trends were not observed. The age and gender distributions were different in different 5-year periods which might influence the results. The purpose of the linear model and the adjustment was to make the different time points more comparable with each other. Adults and children were analyzed separately. Age group (11–17 years vs. < 10 years) and 5-year time periods were included as categorical covariates in children. In the model’s gender, age group and time were forced into the model as predictors, but the interaction terms were included only if the corresponding p < 0.10. The distributions of the B-Se, fS-BKarot, fS-Q10 and S-D-25 were skewed to the right and were logarithmically (ln) transformed before analyses. The results are given as adjusted means (95% CI) or as adjusted geometric means (95% CI) for both genders, age- categories and for each 5-year period. In addition, the adjusted differences (95% CI) are given for both genders and age groups. For the logarithmically transformed micronutrient variable the ratio of geometric means with 95% confidence interval indicates the relative difference.hfi if Thirdly, the Pearson correlation coefficient (R) was calculated to test the possible correlation between the levels of selected micronutrients: between E-Cu and B-Zn; fS-BKarot and fS-A-vit; fS-Q10 and B-Se. These pairs of micronutrients were chosen because they may have either a synergistic effect or are competitors for the absorption in the intestine.f Due to the very large sample size even very weak correlations and small differences between the groups can become statistically significant. Although a weak correlation may be important in an epidemiological study, here we interpreted the magnitude of a correlation coefficient as the following: 0.00–0.19 a very weak, 0.20–0.39 a weak, 0.40–0.59 a moderate, 0.60–0.79 a strong and 0.80–1.00 a very strong positive correlation.i g y g p All statistical tests were two-sided, and p-values < 0.05 were statistically significant. Analyses were performed using IBM SPSS Statistics for Windows (version 27.0, Armonk, NY, USA, IBM Corp.). Materials and methods Samples and nutritional variables. For this study we retrieved from the clinical laboratory database all results on whole blood Magnesium, (B-Mg, mmol/L), Manganese(B-Mn, µmol/L), Zinc(B-Zn, µmol/L), Sele- nium (B-Se, µmol/L) and Copper from erythrocytes (E-Cu, mol/L), the essential microelements, and fasting https://doi.org/10.1038/s41598-022-09131-x Scientific Reports \| (2022) 12:5008 \| www.nature.com/scientificreports/ g We wish to acknowledge Pentti Martikainen MSc. for his inspiring discussions. g We wish to acknowledge Pentti Martikainen MSc. for his inspiring discussions. Statements. 1. All methods were carried out in accordance with Standard Operational Procedures (SOP) of the laboratory which has got accreditation from the Finnish accreditation service FINAS T 309 (EN ISO/IEC17025). All protocols were approved by the ethical board of Minelaililaboratorio Milia, TK1-28.5.2020 1. All methods were carried out in accordance with Standard Operational Procedures (SOP) of the laboratory which has got accreditation from the Finnish accreditation service FINAS T 309 (EN ISO/IEC17025). All protocols were approved by the ethical board of Minelaililaboratorio Milia, TK1-28.5.2020 p pp y 2. Informed consents from all the patients whose blood or sera were analysed during 34 years in Mineraalilabo- ratorio Mila are irrelevant to this study because: (1) No personal information nor clinical data were used to do this data analysis; (2) The data were analysed blind. p pp y 2. Informed consents from all the patients whose blood or sera were analysed during 34 years in Mineraalilabo- ratorio Mila are irrelevant to this study because: (1) No personal information nor clinical data were used to do this data analysis; (2) The data were analysed blind. Received: 13 August 2021; Accepted: 14 March 2022 Scientific Reports \| (2022) 12:5008 \| https://doi.org/10.1038/s41598-022-09131-x www.nature.com/scientificreports/ Competing interests Kaarlo Jaakkola is the founder of Mineraalilaboratorio Mila Oy, all the other authors declare no conflict of interests. References et al. The association between high sensitivity c-reactive protein and micronutrient levels: A cross-sectional analysis based on a laboratory database. Clin. Nutr. ESPEN 33, 283–289 (2019). 25. Tuuminen, T. et al. The association between high sensitivity c-reactive protein and micronutrient levels: A cross-sectional analysis based on a laboratory database. Clin. Nutr. ESPEN 33, 283–289 (2019). 25. Tuuminen, T. et al. The association between high sensitivity c-reactive protein and micronutrient levels: A cross-sectional analysis based on a laboratory database. Clin. Nutr. ESPEN 33, 283–289 (2019). Author contributions T.T. and E.A. designed the study, M.S. retrieved the data, T.P. performed statistical analysis, K.J. was the initiator of the micronutrient measurements in our clinic and together with P.L.L. set the methods in Mineraalilabora- torio Mila Oy. E.M.P. retrieved the data about different techniques and reference methods, M.T. performed the testing and described the techniques used in the laboratory, T.T. wrote the first draft; all the authors contributed equally and approved the final version. References References 1. Arends, J. et al. ESPEN guidelines on nutrition in cancer patients. Clin. Nutr. 36, 11–48. https://doi.org/10.1016/j.clnu.2016.07. 015 (2017). ( ) 2. Barazzoni, R. et al. ESPEN expert statements and practical guidance for nutritional management of individuals with sars-CoV-2 infection. Clin. Nutr. 39, 1631–1638. https://doi.org/10.1016/j.clnu.2020.03.022 (2020). p g j 3. Gröber, U., Holzhauer, P., Kisters, K., Holick, M. F. & Adamietz, I. A. Review micronutrients in oncological intervention. Nutrients 8, 163. https://doi.org/10.3390/nu8030163 (2016). p g 4. Obbagy, J. E. et al. Complementary feeding and micronutrient status: A systematic review. Am. J. Clin. Nutr. 109(Suppl_7) 852S-871S. https://doi.org/10.1093/ajcn/nqy266 (2019). p g j qy ( ) 5. Lorenzo-López, L. et al. Nutritional determinants of frailty in older adults: A systematic review. BMC Geriatr. 17, 108. https://doi org/10.1186/s12877-017-0496-22 (2017).i g 6. Kozeniecki, M., Ludke, R., Kerner, J. & Patterson, B. Micronutrients in liver disease: Roles, risk factors for deficiency, and recom- mendations for supplementation. Nutr. Clin. Pract. 35, 50–62. https://doi.org/10.1002/ncp.10451 (2020). 7. Shi, Y. et al. Trace elements, PPARs, and metabolic syndrome. Int. J. Mol. Sci. 21(7), 2612. https://doi.org/10.3390/ijms21072612 (2020).i 8. Altarelli, M., Ben-Hamouda, N., Schneider, A. & Berger, M. M. Copper deficiency: Causes, manifestations, and treatment. Nutr Clin. Pract. 34, 504–513 (2019).fh 9. Gaffney-Stomberg, E. The impact of trace minerals on bone metabolism. Biol. Trace Elem. Res. 188, 26–34. https://doi.org/10 1007/s12011-018-1583-8 (2019). 10. Gombart, A. F., Pierre, A. & Maggini, S. A review of micronutrients and the immune system-working in harmony to reduce the risk of infection. Nutrients 12, 236. https://doi.org/10.3390/nu12010236 (2020). p g 1. Bae, M. & Kim, H. Mini-review on the roles of vitamin C, vitamin D, and selenium in the immune system against COVID-19 Molecules 25(22), 5346. https://doi.org/10.3390/molecules25225346 (2020).h 11. Bae, M. & Kim, H. Mini-review on the roles of vitamin C, vitamin D, and selenium in t Molecules 25(22), 5346. https://doi.org/10.3390/molecules25225346 (2020).h Molecules 25(22), 5346. https://doi.org/10.3390/molecules25225346 (2020). 12. van den Broek, T. J. et al. The impact of micronutrient status on health: Correlation network analysis to understand the role of micronutrients in metabolic-inflammatory processes regulating homeostasis and phenotypic flexibility. Genet. Nutr. 12, 5. https:// d ( ) ( ) p g ( ) 2. van den Broek, T. J. et al. The impact of micronutrient status on health: Correlation network analysis to understand the role o micronutrients in metabolic-inflammatory processes regulating homeostasis and phenotypic flexibility. Genet. Nutr. 12, 5. https:// doi.org/10.1186/s12263-017-0553-7 (2017).f g ( ) 13. Charoenngam, N. References & Holick, M. F. Immunologic effects of vitamin D on human health and disease. Nutrients 12, 2097 (202 f . & White, J. H. Vitamin D and immunity in infants and children. f 14. Mailhot, G. & White, J. H. Vitamin D and immunity in infant y 5. Mercola, J., William, B., Grant, W. B., Carol, L. & Wagner, C. L. Evidence regarding vitamin D and risk of COVID-19 and its sever- ity. Nutrients 12, 3361 (2020). y 6. Durazo-Arvizu, R. A. et al. Estimating uncertainty of target values for DEQAS serum materials. J. Steroid 188, 90–94. https://doi org/10.1016/j.jsbmb.2018.12.011 (2019). g j j ( ) 17. Trump, D. L. & Johnson, C. S. (Eds). Vitamin D and Cancer. Springer. ISBN 978-1 4419–71876.i g j j 17. Trump, D. L. & Johnson, C. S. (Eds). Vitamin D and Cancer. Springer. ISBN 978-1 4419–71876.i 8. Finick https://keventajat.fi/terveys/suomalaisten-d-vitamiinitilanne-pohjoismaiden-paras-saatko-sina-riittavasti-110893/ (Finn- ish), approached 30.6.2021.h 18. Finick https://keventajat.fi/terveys/suomalaisten-d-vitamiinitilanne-pohjoismaiden-paras-saa ish), approached 30.6.2021.h pp 9. Ikonen, H. et al. The determinants and longitudinal changes in vitamin D status in middle-age: A Northern Finland Birth Cohor 1966 study. Eur. J. Nutr. 60, 4541–4553. https://doi.org/10.1007/s00394-021-02606-z (2021). y p g 0. Jiang, H. et al. Dietary vitamin and carotenoid intake and risk of age-related cataract. Am. J. Clin. Nutr. 109, 43–54. https://doi org/10.1093/ajcn/nqy270 (2019).t 21. Alehagen, U., Aaseth, J., Alexander, J. & Johansson, P. Still reduced cardiovascular mortality 12 years after supplementation with selenium and coenzyme Q10 for four years: A validation of previous 10-year follow-up results of a prospective randomized double- blind placebo-controlled trial in elderly. PLoS ONE 13(4), e0193120. https://doi.org/10.1371/journal.pone.0193120 (2018).tf selenium and coenzyme Q10 for four years: A validation of previous 10-year follow-up results of a prospective randomized double- blind placebo-controlled trial in elderly. PLoS ONE 13(4), e0193120. https://doi.org/10.1371/journal.pone.0193120 (2018).tf p y p g j p 22. Alfthan, G. et al. Selenium Working Group Effects of nationwide addition of selenium to fertilizers on foods, and animal and human health in Finland: From deficiency to optimal selenium status of the population. J. Trace Elem. Med. Biol. 31, 142–147. https://doi.org/10.1016/j.jtemb.2014.04.009 (2015).h p g j j ( ) 3. Pecora, F., Persico, F., Argentiero, A., Neglia, C. & Esposit, S. The role of micronutrients in support of the immune response agains viral infections. Nutrients 12, 3198. https://doi.org/10.3390/nu12103198 (2020). g 4. Jaakkola, K., Grans, L. Amalgam Diseases and Anti-oxidative Treatment. Amalgaanisairaudet ja antioksidatiivinen hoito (Finnish) ISBN 952-90-5309-1 (1994).h 5. Tuuminen, T. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Additional information Correspondence and requests for materials should be addressed to T.T. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. https://doi.org/10.1038/s41598-022-09131-x Scientific Reports \| (2022) 12:5008 \| www.nature.com/scientificreports/ www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2022, corrected publication 2022 https://doi.org/10.1038/s41598-022-09131-x Scientific Reports \| (2022) 12:5008 \|
https://openalex.org/W2776615567	https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0190187&type=printable	English	null	Accounting for tourism benefits in marine reserve design	PloS one	2,017	cc-by	10,656	RESEARCH ARTICLE Accounting for tourism benefits in marine reserve design Daniel F. Viana1, Benjamin S. Halpern1,2,3☯, Steven D. Gaines1☯ 1 Bren School of Environmental Science & Management, University of California, Santa Barbara, Santa Barbara, California, United States of America, 2 National Center for Ecological Analysis and Synthesis, University of California, Santa Barbara, Santa Barbara, California, United States of America, 3 Silwood Park, Imperial College London, Ascot, United Kingdom ☯These authors contributed equally to this work. dviana@bren.ucsb.edu ☯These authors contributed equally to this work. a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Viana DF, Halpern BS, Gaines SD (2017) Accounting for tourism benefits in marine reserve design. PLoS ONE 12(12): e0190187. https://doi. org/10.1371/journal.pone.0190187 Editor: Andrea Belgrano, Sveriges lantbruksuniversitet, SWEDEN Received: September 5, 2017 Accepted: December 8, 2017 Published: December 21, 2017 Copyright: © 2017 Viana et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Editor: Andrea Belgrano, Sveriges lantbruksuniversitet, SWEDEN Received: September 5, 2017 Accepted: December 8, 2017 Published: December 21, 2017 Copyright: © 2017 Viana et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: We cannot publicly provide individual data, because any de-identified data which can be linked to personal data are recognized as personal data according to ethical guidelines in Japan. Additionally, the informed consent we obtained does not include a provision for publicly providing the data. However, a minimal dataset may be available in the form of a collaborative study if approved by the steering committee of the JPHC Study and the Ethics Committee of the National Cancer Center. Please contact the Office of the JPHC Study Group at jphcadmin@ml.res.ncc.go.jp. Abstract Marine reserve design often considers potential benefits to conservation and/or fisheries but typically ignores potential revenues generated through tourism. Since tourism can be the main source of economic benefits for many marine reserves worldwide, ignoring tourism objectives in the design process might lead to sub-optimal outcomes. To incorporate tourism benefits into marine reserve design, we develop a bioeconomic model that tracks tourism and fisheries revenues through time for different management options and location charac- teristics. Results from the model show that accounting for tourism benefits will ultimately motivate greater ocean protection. Our findings demonstrate that marine reserves are part of the optimal economic solution even in situations with optimal fisheries management and low tourism value relative to fisheries. The extent of optimal protection depends on specific location characteristics, such as tourism potential and other local amenities, and the species recreational divers care about. Additionally, as tourism value increases, optimal reserve area also increases. Finally, we demonstrate how tradeoffs between the two services depend on location attributes and management of the fishery outside marine reserve bor- ders. Understanding when unavoidable tradeoffs will arise helps identify those situations where communities must choose between competing interests. Funding: The authors received no specific funding for this work. Funding: The authors received no specific funding for this work. Funding: The authors received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. Several frameworks have now been developed to help capture these joint conservation and economic benefits in effective marine reserve designs [8–10]. One key limitation of the existing work, however, is that by focusing primarily on fisheries economic benefits it has ignored a potentially far larger source of added revenues—tourism. Tourism gains can be obtained through diving operations within the marine reserve [11,12] and the consequent multiplier effects on local businesses related to tourism (e.g. hotels, restaurants). Collectively, these tour- ism benefits can be the main source of economic gains from many marine reserves worldwide [13,14]. To date, there is no clear framework to maximize these potential benefits through effective reserve design. As a result, key questions remain, such as: will the range of conditions where marine reserves are profitable conservation tools grow when tourism is accounted for, and are there inherent economic tradeoffs between reserve benefits to fisheries versus tourism? Competing interests: The authors have declared that no competing interests exist. Marine reserve benefits to conservation, fisheries and tourism all depend on the buildup of biomass and diversity of species within their borders. Thus, many design elements (such as appropriate reserve size relative to scales of fish movement) might align regardless of reserve objectives, while others might be at odds with each other. For example, while fisheries benefits depend on the spillover of adults and/or larval export, tourism and conservation benefits may benefit from higher levels of local retention. This can have important implications in terms of edge location and size of the reserve [3]. Additionally, optimal location of a marine reserve in relation to the coast might differ depending on the objective. Placing a reserve close to port may decrease costs for tourism operators and enforcement agencies while at the same time increase costs for fishers, since they will have to travel longer distances to reach their fishing grounds. Moreover, while conservation objectives require protection of all threatened species and habitats, reserves designed for tourism or fisheries objectives might require only protec- tion of some key species and habitats. This distinction can have important design implications in relation to the location and size of reserves [15]. Introduction Degradation of ocean ecosystems driven by human activities has led to an increased global interest in the establishment of ocean protected areas [1,2]. One type of protected area, where all forms of fishing are prohibited, is known as a “marine reserve” [3]. Much of the interest in marine reserves is driven by their success in recovering important habitats and increasing spe- cies biomass and diversity within the reserve’s boundaries [4]. Although reserves can fail to reach their full potential because of the lack of resources for monitoring and enforcement [5], they are a globally important conservation tool. In addition to these clear conservation bene- fits, the increases in species population size within reserves can also generate important PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 1 / 18 PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Tourism benefits in marine reserve design economic benefits. For example, fisheries benefits can arise through the spillover of adults and/or the export of larvae to surrounding fished areas [6,7]. Funding: The authors received no specific funding for this work. Studies have shown that divers and snorkelers consider ecosystem characteristics and other local amenities when deciding where to visit [16]. Divers are attracted to conservation gains of marine reserves [17] such as increases in the abundance of fish, the diversity of species, iconic species, and coral reef conditions [18–20]. Additionally, since divers are also tourists, other local amenities can also play an important role. Characteristics such as tourism infrastructure, local attractions, proximity to airports, and quality of restaurants and hotels can directly influ- ence a diver’s decision on where to visit [21]. Relative importance of local amenities versus ecosystem health depends on divers’ preferences and availability of different habitats and spe- cies. For example, in the Great Barrier Reef, Australia, whales and dolphins were the preferred draws for divers followed by sharks and rays, overall species richness, turtles and large fish [22]. In the western Caribbean islands, variety of fish, fish abundance and coral variety were the preferred attributes [19]. In contrast, divers from Barbados listed terrestrial characteristics (beaches) and warm and clear water as their main reason for visiting the area followed by coral and fish diversity and abundance [23]. Such differences in preferences show evidence of two categories of divers, one category that is driven by ocean biodiversity and another category that is mainly driven by other local amenities [23]. The former group will likely be attracted by marine reserves, while the latter may be indifferent. Benefits from tourism can in many cases be far greater than the opportunity cost of fore- gone fishing. For example, in the Great Barrier Reef annual revenue from tourism is 36 times greater than income from commercial fishing [14]. In the Medes Islands Marine Reserve (Spain) annual revenue from tourism is about 20 times greater than fishing revenue [24]. 2 / 18 PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Tourism benefits in marine reserve design Potential tourism revenue from marine reserves can be generated directly through user fees [12] or by boosting the tourism economy in the region. Marine Reserves can potentially increase value of all business associated with tourism (e.g. hotels, restaurants), especially those dependent on underwater activities (e.g. dive centers). These benefits depend on the location of the reserve as well as the biomass of fish in the water [16]. Funding: The authors received no specific funding for this work. Reserves located near coastal areas with intense tourism activity and other tourist attractions are likely to have high visita- tion rates quickly after reserve creation [25]. In such situations, the marine reserve may not be the main draw to the area and often does not require high levels of biomass to attract divers. By contrast, locations where there are no other coastal attractions other than the marine reserve may only attract more experienced divers that are drawn by high levels of fish biomass and diversity [26]. These areas may need to be more spectacular and tied with marketing strat- egies to attract large numbers of divers, since the reserves will often be competing with diverse diving options around the globe. Despite growing evidence of economic benefits associated with tourism activities in marine reserves, most spatial planning models only take into account fisheries and/or conservation benefits but ignore tourism gains. To incorporate potential tourism benefits we develop a bioe- conomic model to simulate different marine reserve designs and their predicted impacts on fisheries and tourism revenue. We model the potential benefits for both services under differ- ent tourism and fisheries management scenarios to ask under which conditions are marine reserves part of the optimal solution that maximizes total economic benefits. We then analyze the potential tradeoffs between fisheries and tourism economic benefits to understand the incentives stakeholders face and the situations where conflicts are likely to arise. Material and methods We use a bioeconomic model to simulate different marine reserve designs and the potential economic benefits to fisheries and tourism over time. We divide a hypothetical coastline into 100 homogeneous linear patches where we track the biomass within each patch. Patches are wrapped to eliminate any boundary effect and to make sure all patches are homogeneous. Patches are connected through adult spillover. A fraction of the population emigrates from each patch to nearby patches with a probability that depends on the distance between the patches. A certain fraction of the biomass is also removed through fishing from each patch that is not a marine reserve, with the sum of discounted revenues over time representing the economic gains to fisheries. Larval dispersal is assumed to occur within each patch as popula- tion growth in a patch is only dependent on local population size. Although we acknowledge the important design implications driven by larval dispersal dynamics [27,28], we did not con- sider larval connectivity to simplify the model. Tourism benefits are associated with an increase in the demand for dives inside the marine reserve associated with increased fish den- sity [16]. We did not consider diving activities in fished areas since our source of revenues are the user fees charged to gain access to the marine reserves increase in the demand for dives inside the marine reserve associated with increased fish den- sity [16]. We did not consider diving activities in fished areas since our source of revenues are the user fees charged to gain access to the marine reserves. Tourism benefits in marine reserve design Harvest fraction in each patch, fi, is calculated according to Hilborn et al. 2006, where the intensity of harvest is proportional to the biomass in each patch. We assume that total effort remains constant when a marine reserve is created. This translates to an increased fishing intensity in areas open to fishing as the size of marine reserves grows. The combination of a constant overall fishing effort and a resulting fixed fishing mortality rate in fished patches accounts for the displacement of effort caused by marine reserve placement and creates the fishing the line effect [29] associated with higher catches in patches surrounding marine reserves. For well managed scenarios, total fishing effort is calculated as the amount that gener- ates maximum sustainable yield at equilibrium when the entire area is open to fishing. For overfished scenarios, we assume a fishing effort that would drive fish biomass down to 10% of carrying capacity at equilibrium when all patches are open to fishing. This open access equilib- rium biomass value was assumed according to [30]. Harvest fraction inside patches designated as marine reserves is zero. Initial biomass is assumed to be the equilibrium biomass under the different fisheries management scenarios (50% and 10% of carrying capacity for well managed vs. overfished, respectively). Emigration from patch i (Ei) equals the biomass of fish in the previous year, Bt-1,i, times the movement fraction, represented by μ: Et;i ¼ Bt1;i m ð2Þ ð2Þ Immigration to patch i (Ii) is the sum of the emigration contributions from all other patches j: j: Ii;t ¼ X100 j ¼1 Ej;tpji ð3Þ ð3Þ where the proportion of emigrant fish moving from each patch j to patch i, pji is defined as [16]: pji ¼ expðdjiÞ ð4Þ ð4Þ where di,j is the distance between patch j and patch i. Relative proportions are then normalized so that the proportions moving to all other patches sum to one. where di,j is the distance between patch j and patch i. Relative proportions are then normalized so that the proportions moving to all other patches sum to one. Economic model Fisheries value. Fisheries revenue (Rt) is the sum across all patches of the product of the harvest fraction (fj), resource price (λ) and biomass (Bt,i) in year t. Rt ¼ X100 j¼1 fj Bj l ð5Þ ð5Þ Total net present value of fisheries revenue (FV) is then calculated by summing across all years and applying a discount rate: FV ¼ X50 t¼1 Rt 1 1 þ d t ð6Þ ð6Þ PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Biological model We use a simple logistic model that tracks biomass of a given species in each patch over time: Bt;i ¼ Bt1;i þ g Bt1;i 1 Bt1;i Ki fi Bt1;i Et;i þ It;i ð1Þ ð1Þ ð1Þ Where Bt,i is the biomass in year t and patch i, g is the intrinsic growth rate, Ki is the carrying capacity, fi is the harvest fraction, Et,i is the emigration from patch i and It,i is the immigration to patch i from all other patches. PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 3 / 18 Tourism benefits in marine reserve design congestion effect restricting the total number of divers per marine reserve area per unit time. This reflects the fact that divers prefer less crowded areas, and marine reserves often adopt a cap on the total number of dives per day per area of marine reserve to ensure conservation benefits. Such policy results in a diver carrying capacity inside the marine reserve. The size of the reserve thus limits the potential number of dives per marine reserve area per unit time. We assume that a subset of patches, denoted by M, is designated as a marine reserve. Thus, fjM = 0, and the size of the marine reserve is denoted by x = card(M). Year t biomass in reserve is just SjM Bj,t which is denoted by BM,t. We used a modified version of the equation described by Sala et al. 2013 to model the marginal value of additional dives: Pt ¼ a0 þ f ðBM;tÞ gðxÞDt ð7Þ ð7Þ where Pt is the marginal value of dive Dr,t, α0 is the intercept of the demand function, f(BM,t) is the demand shifter reflecting fish abundance in the marine reserve, and g(x) changes the slope of demand to reflect congestion of divers in the marine reserve (this congestion effect will depend in reserve size, x). The fish abundance effect on demand, f(BM,t), is increasing in fish biomass inside the reserve and the congestion effect, g(x), is decreasing in the size of the reserve (Fig 1). The function forms for f(BM,t) and g(x) are given as follows: g x ð Þ ¼ a1 ðlog100xÞ 1 w ð8Þ ð8Þ where α1 is a location specific price elasticity, x is the reserve size and w controls the slope of the logarithmic function. We assumed a logarithmic function because it allows different slopes to be modeled. The different slopes represent distinct levels of tourism potential, reflecting the where α1 is a location specific price elasticity, x is the reserve size and w controls the slope of the logarithmic function. We assumed a logarithmic function because it allows different slopes to be modeled. The different slopes represent distinct levels of tourism potential, reflecting the fact that when there is a high number of possible divers, small reserves cannot capture all potential tourism revenue because of the congestion effect. Where δ is the discount rate. Tourism value. Tourism value is assumed to be associated with the density of fish inside the marine reserve to reflect the underwater experience of divers. As described by Sala et al. 2013, we assume a dive’s marginal value is directly influenced by the diver’s underwater experi- ence. Increased fish density inside the marine reserve will shift diver’s demand outward, increasing potential revenue generated from the system [16]. Additionally, we assume a 4 / 18 PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 This allows the model to account for crowding issues and diver carrying capacity, which limits the number of divers per area of reserve. We assume that the diver carrying capacity is set to prevent environmental degrada- tion by divers so that tourism activities does not interfere with biomass buildup inside reserves. By setting a cap on the number of dives, reserve area will directly affect the total revenue that can be generated, especially in locations with high tourism potential (S1 Fig, w = 0.25). Under such conditions, tourism value is expected to increase as marine reserve size increases, since more divers will fit in a larger reserve. On the other hand, in locations with low tourism poten- tial, the crowding effect is less important (S1 Fig, w = 6). This is expected to happen, because all potential divers can fit in a relatively small area. Thus increasing reserve size does not imply a significant increase in the number of dives. Although maximum tourism values are scaled to one, revenues generated in locations with high tourism potential can be dramatically higher than locations with low tourism potential. The influence of fish density in the demand curve is represented by f(B), which shifts a dive’s marginal value in a logistic manner: f BM;t ¼ b 1 þ b e c BM;t KM ð9Þ ð9Þ Where BMR,t is the total marine reserve biomass, KMR is the marine reserve carrying capacity, and b and c are the parameters for the logistic curve that regulate the relationship between den- sity and tourism value. Parameter b represents the additional number of dives that can be obtained due to fish density improvements. Parameter c regulates the rate of increase and the minimum density level required for tourism value to begin increasing. We assume that the 5 / 18 PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Tourism benefits in marine reserve design Fig 1. Hypothetical illustration of the effects of congestion, g(x), and fish density, f(BM,t), in divers’ demand (Eq 7). Dotted line illustrates Eq 7 at higher fish density levels. Dashed line illustrates Eq 7 at higher congestion levels. https://doi org/10 1371/journal pone 0190187 g001 Fig 1. Hypothetical illustration of the effects of congestion, g(x), and fish density, f(BM,t), in divers’ demand (Eq 7). Dotted line illustrates Eq 7 at higher fish density levels. PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Tourism benefits in marine reserve design even with relatively low fish densities (S2 Fig, c = 25). An example of this scenario is Barbados, where divers reported that terrestrial characteristics are the main reason for visiting the area [23]. In addition, such a pool of tourists is likely to have a higher fraction of less experienced divers, for whom the underwater experience is not as important. Conversely, in locations where the main tourism draw is the marine reserve itself, diving experience is more important and dive tourism value will likely increase at higher fish densities (S2 Fig, c = 10). An example of this scenario would be Cabo Pulmo, Mexico, an isolated community where the marine reserve is the primary tourism draw and tourism revenues grew rapidly after a 400% increase in the biomass of targeted species [26]. Eq 7 can be used to calculate the number of dives in a given patch for any given price and biomass level. The optimal price (OPt) that maximizes total revenue can also be calculated by taking the derivative of the product of the fee per dive and the number of dives in the reserve and setting the equation equal to zero: OPt ¼ a0 þ a0 þ f ðBM;tÞ 2 þ f BM;tÞ ð10Þ ð10Þ Tourism revenue (TRr,t) is calculated by multiplying the number of dives in the reserve by the optimal price per dive (OPt): Tourism revenue (TRr,t) is calculated by multiplying the number of dives in the reserve by the optimal price per dive (OPt): TRr;t ¼ OPt f ðBÞ þ a0 OPt gðxÞ ð11Þ ð11Þ Equilibrium tourism revenue is calculated as the tourism revenue generated in year 50. Total net present value of tourism revenue (TV) is calculated by summing the predicted reve- nue across all years and applying a discount rate: TV ¼ X50 t TRr;t 1 1 þ d t ð12Þ ð12Þ PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Dashed line illustrates Eq 7 at higher congestion levels. Fig 1. Hypothetical illustration of the effects of congestion, g(x), and fish density, f(BM,t), in divers’ demand (Eq 7). Dotted line illustrates Eq 7 at higher fish density levels. Dashed line illustrates Eq 7 at higher congestion levels. https://doi.org/10.1371/journal.pone.0190187.g001 Fig 1. Hypothetical illustration of the effects of congestion, g(x), and fish density, f(BM,t), in divers’ demand (Eq 7). Dotted line illustrates Eq 7 at higher fish density levels. Dashed line illustrates Eq 7 at higher congestion levels. https://doi.org/10.1371/journal.pone.0190187.g001 Fig 1. Hypothetical illustration of the effects of congestion, g(x), and fish density, f(BM,t), in divers’ demand ( higher fish density levels. Dashed line illustrates Eq 7 at higher congestion levels. https://doi.org/10.1371/journal.pone.0190187.g001 https://doi.org/10.1371/journal.pone.0190187.g001 demand for dives in a marine reserve will shift outward through a logistic relationship with fish density. This assumption is meant to address the fact that marine reserves can achieve a certain threshold of fish density where their attraction to divers will grow far more rapidly (at least more than fish density in areas open to fishing) and after a certain point increasing den- sity will attract few additional divers. This relationship is determined by the c parameter, with actual values representing different location conditions (S2 Fig). In locations where the main draw to the area is not the marine reserve, fish density may not be as important to achieve a given level of tourism revenues. Under such conditions tourism revenues may start growing PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 6 / 18 Tourism benefits in marine reserve design and tourism services. Optimal marine reserve size is defined as the design that maximizes total economic value of the system (tourism + fisheries) for every given relative worth of both ser- vices. The timing component of the model is also explored more explicitly by calculating the number of years required for particular relative tourism values to be realized under different management scenarios. For default values, we assume a movement fraction (μ) of 0.2, an intrinsic growth rate (r) of 0.2 and a 5% discount rate. For the tourism model we assume a moderate dependence of the revenue on fish density (c = 15) and a moderate crowding effect (w = 1). Sensitivity analysis of all model parameters are shown in the supplementary material. https://doi.org/10.1371/journal.pone.0190187.g002 Where δ is the discount rate. Where δ is the discount rate. To obtain general results, we normalize potential tourism and fisheries revenue to each be between 0 and 1, as actual revenue is context dependent. Assuming a 0 to 1 value allows us to test the influence of different relative values from fisheries and tourism on the optimal marine reserve design. Additionally, this assumption does not affect the shape of the tradeoff between these two services, as relative values will only help choose along the tradeoff curve the marine reserve design that provides highest economic returns. We further explore the implication of different relative tourism and fisheries values by demonstrating how actual values can alter optimal marine reserve size. Two metrics are used to determine the value of these services: normalized net present value (NPV) and equilibrium revenue. Net present value of tourism and fisheries services considers the time required for such benefits to be realized. Since future revenues are discounted, timing of benefits becomes a crucial factor. Characteristics such as low initial biomass or slow population growth rates increase the time required for benefits to be realized and therefore negatively affects the NPV. For this metric, a value of one represent the maximum possible NPV that can be achieved for fisheries and tourism services given all possible design and fisheries management options. Equilibrium revenue of fisheries and tour- ism services does not consider the time component. This would be important for stakeholders that have a long-term vision, without time consideration. For this metric, initial biomass or growth rate are not as important. A value of one represent the normalized maximum equilib- rium tourism or fisheries revenue that can be achieved by the system. When considering total revenues, optimal marine reserve design is calculated for different relative values of fisheries PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 7 / 18 Results Additionally, tourism benefits have a maximum value of one in both cases (well managed and overfished), because closing the entire area to fishing does not affect equilibrium values. Despite inherent tradeoffs between tourism and fisheries services, relatively high values of both services can be achieved simultaneously. For example, for all scenarios where maximum tourism can be achieved (Fig 2A, 2C and 2D), both services can simultaneously achieve about 80% of their maximum value. This is the point along the tradeoff curve that maximizes the sum of both normalized values. Interestingly, when considering equilibrium revenues, a reserve of about 40% is desired to maximize the sum of both values (tourism + fisheries), inde- pendent of the management scenario. If for economic or social reasons revenues higher than 80% are desired for one of the two services, it will lead to significant costs to the other. For example, for all scenarios where maximum tourism can be achieved (Fig 2A, 2C and 2D), achieving 90% of tourism benefits will reduce fisheries revenue to about 40% of its maximum value. On the other hand, achieving 90% of fisheries value in well managed scenarios (Fig 2A and 2C) will reduce tourism revenue to about 60% of its maximum value. Sensitivity analysis of crowding (w) and fish density (c) effects on the tradeoffs between tourism and fisheries services show that the shape of the tradeoff is sensitive to these parame- ters (S3 Fig). In locations where diving is not the main driver of tourism benefits (high c value), a small marine reserve might be enough to generate the density of fish needed to attract divers. Locations where the marine reserve is the main tourism driver (low c value) larger areas are necessary to create the density needed for tourism benefits to be realized. Addition- ally, strength of the crowding effect will affect the optimal marine reserve design. Since an area can only fit a certain number of divers at any given time, locations with high tourism potential (low w value) will require more protection to achieve full benefits. Conversely, in locations with low tourism potential (high w value), crowding is not significant. Thus, small marine reserves can accommodate all divers. Simulation of these scenarios shows that although high fish densities can often be achieved with small marine reserves, larger areas may be necessary to capture all potential tourism benefits. Results Expected tradeoffs between fisheries and tourism services vary according to different manage- ment scenarios and metrics (Fig 2). In well managed scenarios, maximum fisheries revenue is achieved with no marine reserves. Fisheries revenues decrease as marine reserve size increases. In such cases, if tourism value is ignored, marine reserves are not part of the optimal economic solution. Thus, with perfect fisheries management, accounting for tourism benefits will be cru- cial for marine reserves to be part of the optimal economic solution. By contrast, in the over- fished scenario higher fisheries value can be achieved with marine reserve implementation. Consequently, even if tourism value is ignored, marine reserves will be part of the optimal solution when resources are overfished. When considering the net present value of fisheries and tourism services (Fig 2A and 2B), overfished areas can only obtain a fraction of the total NPV from well managed systems because of the difference in the initial biomass values and Fig 2. Tradeoffs between fisheries and tourism services for well managed (A and C) and overfished (B and D) scenarios. (A) and (B) demonstrate results in terms of net present value and (C) and (D) demonstrate results in terms of the equilibrium revenue. Colors represent the percent of the area designated as marine reserve. https://doi.org/10.1371/journal.pone.0190187.g002 Fig 2. Tradeoffs between fisheries and tourism services for well managed (A and C) and overfished (B and D) scenarios. (A) and (B) demonstrate results in terms of net present value and (C) and (D) demonstrate results in terms of the equilibrium revenue. Colors represent the percent of the area designated as marine reserve. Fig 2. Tradeoffs between fisheries and tourism services for well managed (A and C) and overfished (B and D) scenarios. (A) and (B) demonstrate results in terms of net present value and (C) and (D) demonstrate results in terms of the equilibrium revenue. Colors represent the percent of the area designated as marine reserve 8 / 18 PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Tourism benefits in marine reserve design harvest levels. In contrast, when considering equilibrium revenues of tourism and fisheries ser- vices, overfished scenarios can achieve much higher values relative to well managed systems. This happens, because equilibrium values do not account for the time required for biomass recovery. Thus, since equilibrium values do not consider discount rate, initial biomass is not as important. PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Results If the planning objective is to maximize overall revenues from both fisheries and tourism services, actual economic values will be crucial to determine the optimal design. Fig 3 shows the influence of relative tourism and fisheries values on optimal marine reserve size under dif- ferent fisheries management scenarios and outcome metrics. Generally, optimal marine reserve size increases as relative tourism value rises, eventually reaching 100% of the area. For overfished scenarios, marine reserves are always part of the optimal solution, even with rela- tively low tourism values. The optimal marine reserve size for overfished scenarios when tour- ism value is extremely low is about 30% of the area. As the relative revenues from tourism and fisheries reach a value close to one, optimal marine reserve size increases rapidly, eventually reaching 100% of the area (Fig 3). In well managed situations, where marine reserves are not part of the optimal solution for fisheries alone, including tourism value changes the outcome even when tourism revenues are well below fisheries values. This happens, because even tiny reserves (1–2%) can bring larger tourism value than the corresponding losses to fisheries value. As tourism value increases relative to fisheries, the optimal marine reserve size grows, eventually reaching 100% of the area. When considering equilibrium revenues (Fig 3B), opti- mal marine reserve area is very similar for both management scenarios when tourism value is about 10 times fisheries value. Sensitivity analysis to crowding and fish density effects show PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 9 / 18 Tourism benefits in marine reserve design Fig 3. Optimal marine reserve size for different relative net present values of tourism and fisheries services. The two figures represent different outcome metrics, where (A) is in terms of net present value and (B) is in terms of equilibrium revenue. https://doi.org/10.1371/journal.pone.0190187.g003 Fig 3. Optimal marine reserve size for different relative net present values of tourism and fisheries services. The two figures represent different outcome metrics, where (A) is in terms of net present value and (B) is in terms of equilibrium revenue. Fig 3. Optimal marine reserve size for different relative net present values of tourism and fisheries services. The two figures represent different outcome metrics, where (A) is in terms of net present value and (B) is in terms of equilibrium revenue. PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Results https://doi.org/10.1371/journal.pone.0190187.g003 https://doi.org/10.1371/journal.pone.0190187.g003 https://doi.org/10.1371/journal.pone.0190187.g003 that optimal marine reserve size for different relative values of tourism and fisheries services can be quite different depending on these parameters values (S4 Fig). Generally, as tourism potential and crowding effect decreases (high w values), higher relative tourism value is needed for high levels of protection. Since higher relative tourism versus fisheries values are harder to achieve when there is low tourism potential, closing big portions of the area becomes less likely. Additionally, with low tourism potential, fish density effect plays an important role. As dependence on fish density increases (locations where the main draw is the marine reserve) greater protection is desired. that optimal marine reserve size for different relative values of tourism and fisheries services can be quite different depending on these parameters values (S4 Fig). Generally, as tourism potential and crowding effect decreases (high w values), higher relative tourism value is needed for high levels of protection. Since higher relative tourism versus fisheries values are harder to achieve when there is low tourism potential, closing big portions of the area becomes less likely. Additionally, with low tourism potential, fish density effect plays an important role. As dependence on fish density increases (locations where the main draw is the marine reserve) greater protection is desired. Timing of benefits is an important factor to consider when creating a marine reserve expecting tourism gains. Fig 4 demonstrates the number of years required for tourism reve- nues to be generated under different marine reserve designs. Because of different starting points and intensities of fishing in open areas, the timing of benefits varies significantly. In well managed scenarios tourism benefits can happen relatively quickly, because stocks inside the marine reserve start at higher values. By contrast, overfished scenarios can take much lon- ger for marine reserve densities to reach peak values. Additionally, the larger the area protected the quicker benefits will be realized, because fewer fish leave the boundaries of the reserve PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 10 / 18 Tourism benefits in marine reserve design Fig 4. Timing of tourism revenue for different marine reserve sizes and fisheries management scenarios. https://doi.org/10.1371/journal.pone.0190187.g004 Fig 4. Timing of tourism revenue for different marine reserve sizes and fisheries management scenarios. Fig 4. Timing of tourism revenue for different marine reserve sizes and fisheries management scenarios. https://doi.org/10.1371/journal.pone.0190187.g004 https://doi.org/10.1371/journal.pone.0190187.g004 where they can be caught. Results For example, it takes 40 years to achieve 0.5 of the maximum tour- ism revenue in overfished scenarios for a marine reserve size of 25%. The more tourism reve- nue is dependent on fish density the longer these benefits take to be realized (S5 and S6 Figs). The benefits of reserves are sensitive to growth and movement characteristics of the target species (S7 Fig). Generally, species that have high movement rates and low growth rates will require larger reserves to achieve tourism benefits. This is consistent with the literature on bio- logical responses of marine reserves, where species that move more require larger areas to be protected [10]. On the other hand, if movement rate is close to zero, relatively small areas will be sufficient to produce fish densities that attract divers, and tourism value will depend mostly on the strength of the crowding effect. Additionally, effects of growth and movement rates on optimal design is greater in overfished relative to well managed scenarios because of the differ- ence in fishing mortality of the fish that spill over from reserves. Tourism benefits in marine reserve design Conflicts are likely to be highest when there is clear preference for one service over the other. Optimal marine reserve design choice will be greatly influenced by the relative social and economic value of tourism and fisheries services. Different stakeholders can have distinct social tradeoffs, which can be related to higher relative profits, social motives such as employ- ment, or cultural reasons such as local traditions and customs. Therefore, some stakeholders might care more about one service than the other, influencing the optimal marine reserve design and inherent tradeoffs between fisheries and tourism services. Stakeholders that depend solely on resource extraction, such as fishers, might only value fisheries and not care about tourism benefits. Consequently, optimal design will be the point along the tradeoff curve that maximizes fisheries value—i.e., the location where a horizontal line reflecting a pure prefer- ence for fishing is tangent to the tradeoff curve. Generally, greater preferences for fisheries services will lead to lower tourism values and less area being protected. In well managed sce- narios, the optimal solution that maximize fisheries services is to open the entire area to fishing and manage the fishery well. This will lead to zero tourism value, since our model assumes that revenue is generated through the collection of user fees that depend on marine reserve estab- lishment. In overfished scenarios, marine reserves are required to maximize fisheries value. This creates a win-win situation with tourism services, where maximizing the value of one ser- vice also generates value to the other service. Stakeholders that rely only on tourism activities (e.g dive operators) might have a high preference for tourism benefits and may not care about fisheries services. When there is a pure preference for tourism services optimal marine reserve design is where tourism value is maximized. This happens where a vertical line is tangent to the tradeoff curve, which for all scenarios is where 100% of the area is set as marine reserve. The greater the preference for tourism services the bigger the compromise to fisheries revenue, leading to a strong tradeoff between the two services. Stakeholders that value both services equally for economic or social reasons (e.g., government managers) might not have a prefer- ence for one or the other service. PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Discussion Accounting for tourism benefits can significantly influence optimal marine reserve design. Results from our model show how considering tourism objectives can be crucial for marine reserves to be part of the optimal economic solution, regardless of the state of the fishery. This result challenges previous findings that marine reserves are not part of the optimal economic solution when the fishery is well managed [31]. Our findings demonstrate how marine reserves should be implemented even in situations with optimal fisheries management and low tourism value relative to fisheries. In such situations, a relatively small reserve can generate more bene- fits from spillover and tourism development than foregone fisheries value. As tourism value increases relative to fisheries revenues, larger areas should be protected to maximize economic outcomes. 11 / 18 PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Thus, following general economic theory, optimal design will be the point along the tradeoff curve that maximizes the sum of both relative values. In this case, optimal design is where a 45 degree line is tangent to the tradeoff curve. For all sce- narios, this equal weighting point has relatively high values of both services and relatively low tradeoffs. Revenues generated through user fees can be used in many ways to offset potential costs associated with the marine reserve. Revenues can be used for direct compensation to fishers, investment in better management of fisheries, creation of alternative livelihoods, community infrastructure, and/or monitoring and enforcement. Direct compensation to fishers can be used to compensate for losses associated with reduced fishing grounds. One example of such a scheme is in the Great Barrier Reef Marine Park, where the Australian government provided compensation for commercial fishers adversely affected by the reserves [32]. Such schemes might be useful to obtain support from key stakeholders, but can create perverse incentives to overharvest areas that are still open to fishing. Utilizing revenues to invest in better manage- ment for adjacent areas open to fisheries can be an alternative to achieve long term sustainabil- ity of the fisheries. Although such investment would not address short term costs, it can help to ensure spillover benefits from marine reserves to affected fishers and achieve better fisheries profits in the future. For example, in the Galapagos National Park, revenue from fees are used by the government to manage the fisheries around the islands [33]. An alternative to investing in the fisheries sector would be to invest in alternative livelihoods such as aquaculture or tour- ism. Such alternatives can have many positive effects by increasing resilience of the system through income diversification. Additionally, it can decrease problems associated with the dis- placement of effort to outside areas by converting some of those fishers into tourism operators or aquaculture farmers. For example, in the Raja Ampat Marine Reserve system located in PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 12 / 18 Tourism benefits in marine reserve design Indonesia, 30% of the user fees are directed to communities in the region for projects related to tourism development [13]. The remaining revenue is used for managing the marine reserves, including costs related with monitoring and enforcement. In many cases, benefits generated through user fees are entirely used by government agencies or NGOs for monitoring and enforcement of the area [34,35]. Using all revenue for reserve management can help enforcement of the area but does not address the root of the problem. In such cases, fishers typically bear all costs and do not have any secure benefits from the reserves, which can lead to strong opposition to any marine reserve creation. Therefore, uncertainty and timing of fisher- ies benefits might lead to increased illegal activities and enforcement costs, which is one reason for many “paper parks” worldwide [5]. Short term costs to fisheries [36] and long term maintenance costs of marine reserves can be a strong deterrent to their success [5]. As fisheries and tourism benefits are related to the density of fish inside closed areas, such benefits can take a long time to be realized depending on reserve size, species characteristics and the fishing pressure before and after reserve creation [37]. Results from our model demonstrate how tourism revenues generated through user fees can take many years to be realized, especially when the fishery is overfished prior to reserve creation. With such benefits occurring in the future, innovative market strategies might be needed to compensate for short term fisheries losses. Such market-based strategies can be a promising solution to use future tourism benefits to offset short-term fisheries losses [36]. For example, in areas with high tourism potential, significant revenues are expected in the future. Thus, agreements between the tourism industry and fishers can be established to ensure fishers are guaranteed a share of future tourism benefits. Although this alternative does not address short term losses it ensures future benefits to fishers, which might be sufficient to gain their support. The level of support might in turn depend on the timing of such benefits and discount rate of the fishers. If their discount rate is high, future benefits can be insignificant compared to short term losses. Another market-based alternative might be to acquire a loan with banks or philanthropic organizations to compensate short-term losses, with payments from future tourism benefits. Philanthropic organizations interested in marine conservation might offer lower discount rates than banks and are usually more willing to take the risks related to an uncertain benefit. PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Tourism benefits in marine reserve design areas might not be viable. On the other hand, for locations with high tourism potential that depend on mobile species for diving activities, benefits from protecting a large area likely out- weighs potential costs to the fisheries sector. Another important assumption of our model is that fisheries target the same species that divers care about. In cases where the main draw for divers are charismatic species not targeted by fisheries (e.g. dolphins, whales, turtles), such an assumption might not hold true. Although such species are not expected to be directly affected by protection as much as species targeted by the fisheries, marine reserves can provide indirect benefits through increased food availability [39]. Additionally, even though there might not be any significant increases in density, marine reserves still create an instrument to collect reve- nue that can be invested in the region. Tourism activities inside marine reserves can have positive and negative effects for marine conservation. Since tourism activities are dependent on marine conservation, high synergies between tourism and conservation services can be expected. Additionally, having a regular presence of divers in the reserves can help with monitoring and enforcement of the area as it can discourage poachers and facilitate detection of illegal fishing activities. On the other hand, inexperienced divers can cause significant habitat degradation and alter important fish behav- iors [40]. Many studies have pointed out the damage caused by divers in sensitive coral reef areas [41,42]. Prevention of damage can be achieved by setting a maximum number of divers for a given area [43] and providing proper training and education to dive masters and recrea- tional divers about best practices and potential harms associated with this activity. Several marine reserves around the world have been using a diving carrying capacity to minimize environmental damage caused by divers. For example, the Mendes Islands Marine Reserve has established a maximum of 450 dives per day [16]. Protecting large portions of the ocean can also help decrease diver density and increase potential conservation benefits. Such methods can significantly decrease adverse tourism effects and increase synergies between conservation and tourism services. In our model, we assume that a maximum number of dives per reserve area is set to prevent environmental degradation by divers. Thus, diving activities does not interfere with biomass buildup inside reserves. PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 The magnitude of uncertainty on future tourism benefits will likely depend on the characteristics of the area related to their tourism potential. For example, in areas where there are no other major attractions other than the marine reserve, marketing campaigns and tourism infrastructure need to be fomented to create a reputation of the area among the diving community and provide minimal conditions for tourists. Otherwise, there is a chance that tourism benefits are going to take too long or will not happen at all, especially if diving experi- ence is not spectacular enough to compete with other marine reserves from around the world. Our model assumes that tourism revenue is associated with the density of fish inside the marine reserve. Although fish density is one of the main ecosystem attributes preferred by divers [19], other characteristics can also be important. For example, diversity and size of fish and corals can be an important factor for divers [20]. Although we don’t explicitly account for these characteristics, such attributes are generally correlated with increases in density inside marine reserves [38]. Additionally, we assumed that divers are driven by only one species of fish, while in reality there will undoubtedly be far more than one important species. Optimal marine reserve design will vary depending on the biological characteristics of the species and focusing on only one may not be sufficient to increase the biomass of the other. One approach would be to focus on the species that have the greatest movement rates to ensure positive growth of all species. Focusing on species with high mobility would mean having to close a rel- atively large area, which might be challenging depending on the context. For example, loca- tions with low tourism potential and high tradeoffs with fisheries services, protecting large PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 13 / 18 Tourism benefits in marine reserve design main source of revenue are transitioning to the tourism sector using user fee revenues to invest in local tourism infrastructure [13]. Additionally, increased tourism activities might influence local consumption of sustainable seafood and increase the price of locally harvested products, allowing reductions in catch without compromising total revenue generated. Conclusion Our model provides the first attempt to incorporate future tourism revenue in the design of marine reserves. Tourism is a way to capture benefits from conservation and turn it to a monetary value, which is crucial when comparing with fisheries value. We provide significant insights on the importance of the specific location characteristics in the prediction of future tourism benefits. Previous tourism infrastructure and other local attractions can play a critical role in determining the expected benefits and their relationship with fish density. Tourism potential of each area can also have significant implications to marine reserve design because of congestion effects. In all scenarios tested, marine reserves were part of the optimal design when considering both tourism and fisheries benefits, even when the fishery is well managed outside. The amount of area to be protected will greatly depend on the value of tourism relative to fisheries. As relative tourism value increases, the percent of the total area to be protected also increases. In areas where tourism value is orders of magnitude greater then fisheries value, it would be optimal to close the entire area to fishing. Therefore, accounting for tourism bene- fits can be crucial to optimally design marine reserves. Additionally, the use of revenues gener- ated through user fees to offset potential costs associated with reserve creation can be crucial to gain support of local stakeholders and increase conservation effectiveness. Supporting information S1 Fig. Relationship between congestion effect, f(x), and marine reserve size for different tourism potential scenarios (represented by w). (TIFF) S1 Fig. Relationship between congestion effect, f(x), and marine reserve size for different tourism potential scenarios (represented by w). (TIFF) S2 Fig. Relationship between the effect of biomass on potential tourism value, f(B), and fish density inside the marine reserve. Different c values represent distinct location character- istics. (TIFF) S3 Fig. Sensitivity analysis of the tradeoffs between fisheries and tourism services for well managed and overfished scenarios under different crowding (w) and fish density (b) effects. (TIFF) S4 Fig. Sensitivity analysis of the optimal marine reserve size for different relative values of tourism and fisheries services under different fisheries management scenarios, and fish density (b) and crowding effects (w) on tourism net present value. (TIFF) S5 Fig. Sensitivity analysis of timing of tourism revenue for different marine reserve sizes to crowding effects (w) and fish density effect (c) for well managed scenarios. (TIFF) S6 Fig. Sensitivity analysis of timing of tourism revenue for different marine reserve sizes to crowding effects (w) and fish density effect (c) for well managed scenarios. (TIFF) S2 Fig. Relationship between the effect of biomass on potential tourism value, f(B), and fish density inside the marine reserve. Different c values represent distinct location character- istics. (TIFF) Future research can relax that assumption and explore how environmental impacts by divers interfere with design outcomes. We use a conservative model in terms of the benefits that can be generated to fisheries. First, our model only considers adult spillover as benefit source. It does not account for potential recruitment increases through larval and egg spillover which in many cases can be the main source of benefit [44,45]. We did not include larval dispersal dynamics in our model in order to obtain simplified but conservative results. In our model, when adult movement rate is zero, there is no possible source of benefit to the fishery. This is not true in many cases where marine reserves can be an important source of eggs and larvae to fished areas thus increasing recruit- ment and growth rate of the fished population. This can have important design implications in terms of reserve location and the expected recruitment benefits to fished areas [27]. Second, we assume that effort is going to remain constant through time, being redistributed into fished areas after reserve creation [46]. This causes an increase in fishing mortality in the outside areas as marine reserve increases and the fishing the line effect [29]. This assumption can be true in many situations with weak management outside reserve boundaries. If the fisheries are opti- mally managed, fishing effort is expected to adjust in order to provide optimal economic returns. As marine reserves increase, overall effort in outside areas should be decreased and concentrated near reserve borders to optimize economic returns [8,46]. Effort reduction can be facilitated with increased tourism activities as it can create alternative livelihoods for the local community. Thus, even though we conservatively assumed that effort is going to remain con- stant, tourism activities in the reserve might in reality decrease fishing effort in outside areas. For example, in Raja Ampat—Indonesia, many locals that used to depend on fishing as their PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 14 / 18 Acknowledgments We are grateful to the Gaines Lab for all helpful feedback and discussions on earlier versions of this manuscript. Author Contributions Conceptualization: Daniel F. Viana, Benjamin S. Halpern, Steven D. Gaines. Formal analysis: Daniel F. Viana. Investigation: Daniel F. Viana. Methodology: Daniel F. Viana, Benjamin S. Halpern, Steven D. Gaines. Project administration: Daniel F. Viana. Supervision: Benjamin S. Halpern, Steven D. Gaines. Validation: Benjamin S. Halpern, Steven D. Gaines. Visualization: Daniel F. Viana. Writing – original draft: Daniel F. Viana, Benjamin S. Halpern, Steven D. Gaines. Writing – review & editing: Daniel F. Viana, Steven D. Gaines. Conceptualization: Daniel F. Viana, Benjamin S. Halpern, Steven D. Gaines. Methodology: Daniel F. Viana, Benjamin S. Halpern, Steven D. Gaines. Supervision: Benjamin S. Halpern, Steven D. Gaines. Validation: Benjamin S. Halpern, Steven D. Gaines. Visualization: Daniel F. Viana. Writing – original draft: Daniel F. Viana, Benjamin S. Halpern, Steven D. Gaines. Writing – review & editing: Daniel F. Viana, Steven D. Gaines. S3 Fig. Sensitivity analysis of the tradeoffs between fisheries and tourism services for well managed and overfished scenarios under different crowding (w) and fish density (b) effects. (TIFF) 15 / 18 PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Tourism benefits in marine reserve design References 1. Halpern BS, Longo C, Hardy D, McLeod KL, Samhouri JF, Katona SK, et al. An index to assess the health and benefits of the global ocean. Nature 2012; 488:615–20. https://doi.org/10.1038/nature11397 PMID: 22895186 2. Halpern BS, Walbridge S, Selkoe K a, Kappel C V, Micheli F, D’Agrosa C, et al. A global map of human impact on marine ecosystems. Science 2008; 319:948–52. https://doi.org/10.1126/science.1149345 PMID: 18276889 3. Gaines SD, White C, Carr MH, Palumbi SR. Designing marine reserve networks for both conservation and fisheries management. Proc Natl Acad Sci U S A 2010; 107:18286–93. https://doi.org/10.1073/ pnas.0906473107 PMID: 20200311 4. Lester SE, Halpern BS. Biological responses in marine no-take reserves versus partially protected areas. Mar Ecol Prog Ser 2008; 367:49–56. https://doi.org/10.3354/meps07599 5. Edgar GJ, Stuart-Smith RD, Willis TJ, Kininmonth S, Baker SC, Banks S, et al. Global conservation out- comes depend on marine protected areas with five key features. Nature 2014; 506:216–20. https://doi. org/10.1038/nature13022 PMID: 24499817 6. Roberts CM, Bohnsack J a, Gell F, Hawkins JP, Goodridge R. Effects of marine reserves on adjacent fisheries. Science 2001; 294:1920–3. https://doi.org/10.1126/science.294.5548.1920 PMID: 11729316 7. Goñi R, Hilborn R, Dı´az D, Mallol S, Adlerstein S. Net contribution of spillover from a marine reserve to fishery catches. Mar Ecol Prog Ser 2010; 400:233–43. https://doi.org/10.3354/meps08419 8. Rassweiler A, Costello C, Hilborn R, Siegel D a. Integrating scientific guidance into marine spatial plan- ning Integrating scientific guidance into marine spatial planning 2014:1–5. 9. Foley MM, Halpern BS, Micheli F, Armsby MH, Caldwell MR, Crain CM, et al. Guiding ecological princi- ples for marine spatial planning. Mar Policy 2010; 34:955–66. https://doi.org/10.1016/j.marpol.2010.02. 001 10. Gaines SD, White C, Carr MH, Palumbi SR. Designing marine reserve networks for both conservation and fisheries management. Proc Natl Acad Sci U S A 2010; 107:18286–93. https://doi.org/10.1073/ pnas.0906473107 PMID: 20200311 16 / 18 PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Tourism benefits in marine reserve design 11. Pascoe S, Doshi A, The´baud O, Thomas CR, Schuttenberg HZ, Heron SF, et al. Estimating the poten- tial impact of entry fees for marine parks on dive tourism in South East Asia. Mar Policy 2014; 47:147– 52. https://doi.org/10.1016/j.marpol.2014.02.017 12. Green E, Donnelly R. Recreational scuba diving in Caribbean marine protected areas: do the users pay? Ambio 2003; 32:140–4. https://doi.org/10.1579/0044-7447-32.2.140 PMID: 12733800 13. Atmodjo E, Lamers M, Mol A. Financing marine conservation tourism: Governing entrance fees in Raja. Mar Policy 2017; 78:181–8. https://doi.org/10.1016/j.marpol.2017.01.023 14. References McCook LJ, Ayling T, Cappo M, Choat JH, Evans RD, De Freitas DM, et al. Adaptive management of the Great Barrier Reef: a globally significant demonstration of the benefits of networks of marine reserves. Proc Natl Acad Sci U S A 2010; 107:18278–85. https://doi.org/10.1073/pnas.0909335107 PMID: 20176947 15. Halpern BS, Warner RR. Matching marine reserve design to reserve objectives. Proc Biol Sci 2003; 270:1871–8. https://doi.org/10.1098/rspb.2003.2405 PMID: 14561299 16. Sala E, Costello C, Dougherty D, Heal G, Kelleher K, Murray JH, et al. A General Business Model for Marine Reserves. PLoS One 2013; 8:1–9. https://doi.org/10.1371/journal.pone.0058799 PMID: 23573192 17. Lester SE, Halpern BS, Grorud-Colvert K, Lubchenco J, Ruttenberg BI, Gaines SD, et al. Biological effects within no-take marine reserves: A global synthesis. Mar Ecol Prog Ser 2009; 384:33–46. https:// doi.org/10.3354/meps08029 18. Alban F, Boncoeur J. ASSESSING THE IMPACT OF MARINE PROTECTED AREAS ON RECREA- TIONAL USES OF A MARINE ECOSYSTEM: THE CASE OF SCUBA DIVING. Portsmouth Proc 2006:1–9. 19. Williams ID, Polunin NVC. Differences between protected and unprotected reefs of the western Carib- bean in attributes preferred by dive tourists. Environ Conserv 2000; 27:382–91. https://doi.org/10.1017/ S0376892900000436 20. Uyarra MC, Watkinson AR, Coˆte´ IM. Managing dive tourism for the sustainable use of coral reefs: Vali- dating diver perceptions of attractive site features. Environ Manage 2009; 43:1–16. https://doi.org/10. 1007/s00267-008-9198-z PMID: 18810525 21. Weaver DB, Lawton LJ. Twenty years on: The state of contemporary ecotourism research. Tour Manag 2007; 28:1168–79. https://doi.org/10.1016/j.tourman.2007.03.004 22. Farr M, Stoeckl N, Alam Beg R. The non-consumptive (tourism) “value” of marine species in the North- ern section of the Great Barrier Reef. Mar Policy 2014; 43:89–103. https://doi.org/10.1016/j.marpol. 2013.05.002 23. Uyarra MC, Coˆte´ IM, Gill J a., Tinch RRT, Viner D, Watkinson AR. Island-specific preferences of tourists for environmental features: implications of climate change for tourism-dependent states. Environ Con- serv 2005; 32:11–9. https://doi.org/10.1017/S0376892904001808 24. Merino G, Maynou F, Boncoeur J. Bioeconomic model for a three-zone marine protected area: A case study of Medes Islands (northwest Mediterranean). ICES J Mar Sci 2009; 66:147–54. https://doi.org/10. 1093/icesjms/fsn200 25. Chae DR, Wattage P, Pascoe S. Recreational benefits from a marine protected area: A travel cost anal- ysis of Lundy. Tour Manag 2012; 33:971–7. https://doi.org/10.1016/j.tourman.2011.10.008 26. Aburto-Oropeza O, Erisman B, Galland GR, Mascareñas-Osorio I, Sala E, Ezcurra E. Large recovery of fish biomass in a no-take marine reserve. PLoS One 2011; 6. https://doi.org/10.1371/journal.pone. 0023601 PMID: 21858183 27. References Siegel D a, Mitarai S, Costello C, Gaines SD, Kendall BE, Warner RR, et al. The stochastic nature of lar- val connectivity among nearshore marine populations. Proc Natl Acad Sci U S A 2008; 105:8974–9. https://doi.org/10.1073/pnas.0802544105 PMID: 18577590 28. Pelc R a, Warner RR, Gaines SD, Paris CB. Detecting larval export from marine reserves. Proc Natl Acad Sci U S A 2010; 107:18266–71. https://doi.org/10.1073/pnas.0907368107 PMID: 20181570 29. Kellner JB, Tetreault I, Gaines SD, Nisbet RM. Fishing the line near marine reserves in single and multi- species fisheries. Ecol Appl 2007; 17:1039–54. https://doi.org/10.1890/05-1845 PMID: 17555217 30. Costello C, Ovando D, Clavelle T, Strauss CK, Hilborn R, Melnychuk MC. Global fishery prospects under contrasting management regimes. Proc Natl Acad Sci 2016:1–5. https://doi.org/10.1073/pnas. 1520420113 PMID: 27035953 31. Buxton CD, Hartmann K, Kearney R, Gardner C. When Is Spillover from Marine Reserves Likely to Ben- efit Fisheries? PLoS One 2014; 9:e107032. https://doi.org/10.1371/journal.pone.0107032 PMID: 25188380 32. Olsson P, Folke C, Hughes TP. Navigating the transition to ecosystem-based management of the Great Barrier Reef, Australia 2008; 105. 17 / 18 PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 Tourism benefits in marine reserve design 33. Castrejo M, Charles A. Improving fisheries co-management through ecosystem-based spatial manage- ment: The Galapagos Marine Reserve 2013; 38:235–45. https://doi.org/10.1016/j.marpol.2012.05.040 33. Castrejo M, Charles A. Improving fisheries co-management through ecosystem-based spatial manage- ment: The Galapagos Marine Reserve 2013; 38:235–45. https://doi.org/10.1016/j.marpol.2012.05.040 34. Thur SM. User fees as sustainable financing mechanisms for marine protected areas: An application to the Bonaire National Marine Park. Mar Policy 2010; 34:63–9. https://doi.org/10.1016/j.marpol.2009.04. 008 35. Reid-Grant K, Bhat MG. Financing marine protected areas in Jamaica: An exploratory study. Mar Policy 2009; 33:128–36. https://doi.org/10.1016/j.marpol.2008.05.004 36. Ovando D, Dougherty D, Wilson JR. Market and design solutions to the short-term economic impacts of marine reserves. Fish Fish 2016:1–16. https://doi.org/10.1111/faf.12153 37. Babcock RC, Shears NT, Alcala AC, Barrett NS, Edgar GJ, Lafferty KD, et al. Decadal trends in marine reserves reveal differential rates of change in direct and indirect effects. Proc Natl Acad Sci U S A 2010; 107:18256–61. https://doi.org/10.1073/pnas.0908012107 PMID: 20176941 38. Lester S, Halpern B, Grorud-Colvert K, Lubchenco J, Ruttenberg B, Gaines S, et al. Biological effects within no-take marine reserves: a global synthesis. Mar Ecol Prog Ser 2009; 384:33–46. https://doi.org/ 10.3354/meps08029 39. Scott R, Hodgson DJ, Witt MJ, Coyne MS, Adnyana W, Blumenthal JM, et al. PLOS ONE \| https://doi.org/10.1371/journal.pone.0190187 December 21, 2017 References Global analysis of satellite tracking data shows that adult green turtles are significantly aggregated in Marine 2012:1053–61. https://doi.org/10.1111/j.1466-8238.2011.00757.x 40. Hawkins JP, Roberts CM, Van’T Hof T, De Meyer K, Tratalos J, Aldam C. Effects of Recreational Scuba Diving on Caribbean Coral and Fish Communities\rEfectos del Buceo Recreacional en Comunidades de Coral y Peces del Caribe. Conserv Biol 1999; 13:888–97. https://doi.org/10.1046/j.1523-1739.1999. 97447.x 41. Tratalos J a., Austin TJ. Impacts of recreational SCUBA diving on coral communities of the Caribbean island of Grand Cayman. Biol Conserv 2001; 102:67–75. https://doi.org/10.1016/S0006-3207(01) 00085-4 42. Milazzo M, Chemello R, Badalamenti F, Riggio RC, Riggio S. The impact of human recreational activi- ties in marine protected areas: What lessons should be learnt in the Mediterranean sea? Mar Ecol 2002; 23:280–90. https://doi.org/10.1111/j.1439-0485.2002.tb00026.x 43. Davis D, Tisdell C. Recreational scuba-diving and carrying capacity in marine protected areas Derrin Davis a & Clem Tisdell b. Ocean Coast Manag 1995; 26:19–40. https://doi.org/10.1016/0964-5691(95) 00004-L 44. Botsford LW, Brumbaugh DR, Grimes C, Kellner JB, Largier J, O’Farrell MR, et al. Connectivity, sus- tainability, and yield: Bridging the gap between conventional fisheries management and marine pro- tected areas. Rev Fish Biol Fish 2009; 19:69–95. https://doi.org/10.1007/s11160-008-9092-z 45. Siegel DA, Kinlan BP, Gaylord B, Gaines SD. Lagrangian descriptions of marine larval dispersion. Mar Ecol Prog Ser 2003; 260:83–96. Ecol Prog Ser 2003; 260:83–96. 46. Hilborn R, Micheli F, De Leo G a. Integrating marine protected areas with catch regulation. Can J Fish A t S i 2006 63 642 9 htt //d i /10 1139/f05 243 46. Hilborn R, Micheli F, De Leo G a. Integrating marine protected areas with catch regulation. Can J Fish Aquat Sci 2006; 63:642–9. https://doi.org/10.1139/f05-243 18 / 18
https://openalex.org/W4372100577	https://zenodo.org/record/7900341/files/Determining%20the%20invariant%20of%20inter-frame%20processing%20for%20constructing%20the%20image%20similarity%20metric.pdf	English	null	Determining the invariant of inter-frame processing for constructing the image similarity metric	Eastern-European journal of enterprise technologies	2,023	cc-by	7,534	Information technology Information technology UDC 004 The relevance of modeling digital images is determined by the need to implement approaches in the study of localization and identification of objects in order to reduce the amount of data. In this paper, the study object is the topology of a discrete two-dimensional image within the framework of the problem of determining the invariants of diffeomorphic transformations. Geographic information objects (GIOs) refer to objects that are on a given surface or objects that locally change the surface. With regard to objects, it is assumed that the change in their geolocation in the process of forming both single images and an extended series of frames obtained in the process of continuous monitoring is insignificant. In the process of scanning the surface, possible changes in the position of the image source are taken into account, for example, such as yawing, rolling, and pitch in the case of unmanned aerial vehicles (UAVs). These maneuvers are represented as a group of diffeomorphisms that are controlled by the internal gyroscopes of the carrier and the external navigation system. Based on the studies reported here, the initial ontology of digital images (ODI) has been determined by using the model of color spaces and functions of a special kind. The presence of an ontology makes it possible to build an adequate topology of color distribution in the image and take into account the specificity of the distribution of different colors in a digital image. The study results indicate that a promising method is to determine the similarity by constructing a color atlas structure graph (CASG) based on ODI and by determining invariants as a fragment of CASG inherited by all images in the sequence. The scope and conditions for the practical use of the result include its application to the analysis of images by methods of artificial intelligence d d l l l h DOI: 10.15587/1729-4061.2023.276650 1. Introduction and representing large amounts of data in real time for the tasks of monitoring GIOs using UAV tools and web cameras is increasingly significant and relevant. The representation of an image in the form of a diagram makes it possible to process large amounts of information. GIOs refer to objects that are on a given surface or ob- jects that change the surface locally. With respect to these objects, it is assumed that the change in their geolocation during the formation of both single digital images of GIO portraits and over a long series of images obtained during continuous monitoring is insignificant. Real-time account- ing, which arises from the need to process the flow of images in the process of inter-frame processing of surface scanning, leads to a sharp increase in the required amount of comput- ing equipment and the time spent on calculations in general. DETERMINING THE INVARIANT OF INTER- FRAME PROCESSING FOR CONSTRUCTING THE IMAGE SIMILARITY METRIC E l e n a G o r d a Corresponding author PhD, Associate Professor Department of Information Technologies* Е-mail: anatlg@ukr.net A n a t o l i i S e r d i u k PhD, Associate Professor Division of Automation and Aeronautical Systems Warsaw University of Technology plac Politechniki, 1, Warszawa, Poland, 00-661 I v a n N a z a r e n k o Doctor of Technical Sciences, Professor, Head of Department Department of Machinery and Equipment of Technological Processes* Kyiv National University of Construction and Architecture Povitroflotskyi ave., 31, Kyiv, Ukraine, 03037 E l e n a G o r d a Corresponding author PhD, Associate Professor Department of Information Technologies Е-mail: anatlg@ukr.net A n a t o l i i S e r d i u k PhD, Associate Professor Division of Automation and Aeronautical Systems Warsaw University of Technology plac Politechniki, 1, Warszawa, Poland, 00-661 I v a n N a z a r e n k o Doctor of Technical Sciences, Professor, Head of Department Department of Machinery and Equipment of Technological Processes* *Kyiv National University of Construction and Architecture Povitroflotskyi ave., 31, Kyiv, Ukraine, 03037 g y f f g Keywords: digital image, color atlas, geographic information object, ontology, measure, similarity, hash function, diffeomorphism, persistent homology How to Cite: Gorda, E., Serdiuk, A., Nazarenko, I. (2023). Determining the invariant of inter-frame processing for constructing the image similarity metric. Eastern-European Journal of Enterprise Technologies, 2 (2 (122)), 19–25. doi: https://doi.org/10.15587/1729-4061.2023.276650 2. Literature review and problem statement Paper [1] discusses the problem-oriented perspective and provides a comprehensive overview of both superficial and deep transfer learning methods for visual recognition between data sets. A comprehensive problem-oriented re- view of advances in transfer learning in relation to this problem revealed not only problems in transfer training for visual recognition but also those that have been almost not studied. Paper [2] discusses the most recent developments in deep learning techniques for semantic segmentation of remote sensing images, including unconventional data such as hyperspectral images and point clouds. Emphasis is on improving pixel-level accuracy. It is noted that the emerging methods of deep learning have demonstrated significantly improved performance on several publicly available data sets. Based on the analysis and classification of recognition problems between data sets and the application of the se- mantic approach [1], it is possible to state the importance of the task of searching for digital image invariants during inter-frame processing [2, 3]. Paper [4] notes that in recent One of the factors in determining the similarity of GIO images is the presence in the image of special points or special areas in terms of brightness, color, or shape, the location of reference points based on the available geodetic references of the object. Such points include, for example, marks remaining after construction, or geodetic marks of the object. They can be determined based on preliminary information about the location of GIO images. The above makes it possible to determine the linear and altitudinal dimensions of GIOs. The synthesis of new models, methods, and devices for monitoring geoinformation objects in this subject area will help successfully meet modern requirements for video infor- mation systems. Research aimed at expanding the application, improve- ment, and development of methods for collecting, processing, 19 2/2 ( 122 ) 2023 2/2 ( 122 ) 2023 Eastern-European Journal of Enterprise Technologies ISSN 1729-3774 methods of the theory of structures (homological algebra) and relations theories, methods of topology, methods of graph theory, methods of geometric hashing, methods of computational and discrete geometry were used. years there has been exponential growth in the processing and transmission of video over the Internet and other ap- plications. However, improving video compression while maintaining the same quality requires further research. 2. Literature review and problem statement The problem of the need to develop approaches and methods that minimize the number of calculations, which makes it possi- ble to increase the efficiency of systems for transmitting and recognizing video information streams, is also drawn atten- tion to in works [5, 6]. Paper [7] discusses modern methods for implementing joint compression of video and other visual streams together with targeted analytics in a wide range us- ing artificial intelligence. To solve this problem, works [8, 9] propose methods of joint compression of the video sequence, taking into account the features of the entire time range (end-to-end technology). The above works also consider the use of methods of intelligent processing of data representing a digital image. The initial information for achieving the goal contains the following: – images contain background (stationary part of the signal), random noise, and GIOs; – the statistical characteristics of the background vary significantly over the field of the frame: the dispersion of the background, the rate of decline of the form of the correlation function. Correlation dependences are anisotropic: the cor- relation in the direction of scanning exceeds the correlation along the scanning ruler; – the random noise accompanying the measurements has the same order of magnitude with one division of the signal quantization scale and is stationary in frame and in time, the nature of the noise distribution is close to normal; Existing intelligent methods of inter-frame image pro- cessing are based on deep learning [10, 11]. Also, for these purposes, steganography methods are used to extract image information [12]. However, these methods require large com- puting power and large amounts of high-quality initial data. nature of the noise distribution is close to normal; Existing intelligent methods of inter-frame image pro- cessing are based on deep learning [10, 11]. Also, for these purposes, steganography methods are used to extract image information [12]. However, these methods require large com- puting power and large amounts of high-quality initial data. A promising method of encoding video in the feature space (FVC), and not in pixel space, is considered in work [13]. The study of the application of new methods based on the consideration of structural features (topology based on color atlas) is reported in [14, 15]. The combination of these methods with methods of rapid processing increases the efficiency of video surveillance systems. 3. The aim and objectives of the study To determine the transformations to the basic orienta- tion of the web camera, the following parameters should be taken into account: yaw, roll, and tonnage of the shooting frame, which are controlled by the internal gyroscopes of the carrier and the external navigation system. It should be especially noted that the area of localization of GIOs is long. Because of this, the accuracy of the positioning of the web camera at the point of image formation relative to the surface of the GIO location becomes an important factor. Camera positioning can be carried out through satellite navigation and binding to the object’s references. The aim of this work is to determine the invariant of inter-frame processing of the time sequence of GIO images obtained from the UAV web camera to determine the metric of image similarity and the necessary initial information. To achieve the set aim, the following tasks have been solved: – to investigate the elements of the initial ontology of the digital image, which determine the initial information received from the web camera, taking into account possible changes in the shooting point, for example, UAV maneuvers, which have recently become widely used as carriers of opti- cal equipment; As a result of converting images into a GrayScale model, it becomes possible to determine the location of light sourc- es or areas with increased luminosity, which, in particular, makes it possible to determine the totality of local bright- ness gradients. A set of local collinear gradients of brightness from areas in the image makes it possible to identify shadows from GIOs or illuminated surfaces. It is noted that when the motion of the web camera changes, collinearity in the field of local brightness gradients is preserved as an invariant. – to build a topological scheme for the representation of a digital image, providing objective characteristics of dy- namically changing in time scenes of geoinformation objects based on the ontology of the initial level of the image (ISO). 2. Literature review and problem statement – GIO and its shape is modified depending on the loca- tion relative to the elements of the scanning ruler; g – frames of the same sequence are not brought to a single coordinate system, along the ruler, the frame is shifted by less than its size; A promising method of encoding video in the feature space (FVC), and not in pixel space, is considered in work [13]. The study of the application of new methods based on the consideration of structural features (topology based on color atlas) is reported in [14, 15]. The combination of these methods with methods of rapid processing increases the efficiency of video surveillance systems. – temporary stationarity of the background is observed only on two or three frames of one scanning direction; – the temporal stationarity of the noise of objects and backgrounds during forward and reverse scanning are sig- nificantly different; – the temporal stationarity of the noise of objects and backgrounds during forward and reverse scanning are sig- nificantly different; – images can contain glare. To study the motion of systems in the frequency range of 50–100 Hz, the method of reducing complex discrete-con- tinuum equations is applicable [16]. Geographic information objects are considered to be localized on a given surface, a horizon line is present, and the light source is localized as a limited elliptical region, or an area extended along the image. As a result of the trans- formation of the image into a halftone, a system of light and dark spots appears on it, corresponding to the distribution of shadows and illuminated sides of the GIOs, as well as internal depression areas. The set of image-fixations of the geographic information object consists of sets obtained as a result of the drift of the web camera vertically, horizontally, in azimuth relative to the GIO area. All this allows us to assert that it is necessary to develop new digital methods and devices that differ not only in effi- ciency and quality but also in the speed of computation by reducing the amount of necessary calculations by encoding video information signals. 4. The study materials and methods Another invariant of converting images to the GrayScale model is the ratio of the sizes of the different regions for each GIO in the luminance channel. During the study, methods of processing digital images, methods of system analysis, methods of cluster analysis, 20 Information technology that appear in the background or, as reflexes, in the color of the adjacent object in the image. The transformation invariant that allows the GIOs to be highlighted against the background of the image is the preservation of the local area containing the anomaly of the brightness gradients in accordance with the distribution of brightness on it. At the same time, not only towering objects above the surface level are highlighted but also holes (de- pressions). The use of color in the analysis of a GIO image makes it possible to identify small structures and poorly structured areas due to the fact that chromaticity is a representable fea- ture of the image, even in the case of low brightness. From the point of view of the color atlas, the image of GIO is the following aggregates: – color spots of different structures (gradients, colors, patterns); – color spots of different structures (gradients, colors, patterns); 5. Results of investigating dynamically changing scenes of geoinformation objects based on the ontology of the image − the ontology of ISO should allow the construction of metrics on a set of ISO in order to apply cluster analysis methods, to develop a computational method for forming an internal catalog of fixed objects on a series of frames. 11. Background on the image media. 12. For areas: colors, gradients, textures. In order to illustrate the ontology of ISO, the ontology model is represented in the form of a graph, where arcs in- dicate the relationships of terms and rules that describe the relationships (is-A connections). In order to illustrate the ontology of ISO, the ontology model is represented in the form of a graph, where arcs in- dicate the relationships of terms and rules that describe the relationships (is-A connections). The practical application of the ontology of the digital image makes it possible, on the one hand, to formalize the de- scription of its structure in the most capacious and complete representation. On the other hand, it is necessary to deter- mine the classes of features, which in turn makes it possible to build adaptive recognition and monitoring algorithms. Localization of GIOs in digital images in the process of mon- itoring using a dynamic image source, for example, UAVs, is implemented as a procedure for determining a measure of similarity based on the hash functions of the color atlas structure graphs (CASG) of images. At the same time, the scheme is CASG, in which all the edges are single. 5. Results of investigating dynamically changing scenes of geoinformation objects based on the ontology of the image p ) – boundaries between color spots; – adjacency relations set on a given set of spots; 5. 1. Investigating elements of the initial ontology of the digital image of the initial information obtained from the webcamera –absolute values given by the functions of the presence of the spot; – mutual proportions. h h f h When considering this issue, the results of work [17] were employed. By analogy with [17], the tasks of creating an image ontology (ISO) include: The synthesis of the initial ontology of the color regions of ISO, containing the terms, objects, processes, and their characteristics that make up the thesaurus (vertices of the graph) of the ontology of ISO includes: − the creation of a holistic knowledge system that pro- vides information support to specialists and experts in vari- ous fields in which ISO is used or researched; − the creation of a holistic knowledge system that pro- vides information support to specialists and experts in vari- ous fields in which ISO is used or researched; g ) gy 1. Universum – direct digital image of the object on the image medium. g 2. Color palettes on the image medium. − the image is a complex, informationally combined field of knowledge. Therefore, its ontology should form a subject area with the ability to provide automation of output within the intellectual knowledge base of ISO. The resulting knowledge base should allow the construction of intelligent (automated) recognition and classification of the corresponding ISO; 3. Image media topology, connectivity, touch, neighbor- hood, area. 4. Structures on the image media. 5. Clusters on the image media. h 6. Features as functions on the image carrier. h d − since the knowledge base in the ontology of ISO is integrated from the database of applied areas, therefore, it must provide a synthesis of inference integration algorithms for each specific formalized application problem; 7. Parameters on the image media. 8. Descriptors on the image media. 9. Figures on the image medium. 10. Noise and noise structures on the image medium. − the ontology of ISO should allow the construction of metrics on a set of ISO in order to apply cluster analysis methods, to develop a computational method for forming an internal catalog of fixed objects on a series of frames. – {dli} – the set of movements of the recorder with respect to fragment l; T T l l T l li l l i n IMG IMG d d IMG ∀= Δ ≠∅∧ ∧Δ ∈   { } 0, : T T l l i n IMG ∀= Δ ≠∅∧ – bypassing the boundary of the area sets the direction of bypassing the ISO as such; ( ) ( ) { }. T l li l l IMG d d IMG ∧Δ ∈   ( ) ( ) { }. T l li l l IMG d d IMG ∧Δ ∈   – alternation of simple edges and vertices and multiple edges. There is a simple edge between the multiples of the edges; The comparison of CASGs for different images of GIOs as graphs gives a measure of the similarity of the structures of different images based on their color regions. – ordering of multiple edges according to the color of the vertices, they connect, cyclic; – the repetition of the same colors in the loop corre- sponds to the same color areas at the point of touch but they are not adjacent in border (do not have a common length boundary greater than unity). – the repetition of the same colors in the loop corre- sponds to the same color areas at the point of touch but they are not adjacent in border (do not have a common length boundary greater than unity). It is noted that due to the color atlas of the image of GIOs, it is possible to display local anisotropic features as indepen- dent changes at each point of the image, forming a heterogene- ity between the locations of the corresponding spots. The construction of the procedure for bringing a set of images of GIOs taken along one trajectory of movement of the web camera to one angle is carried out by comparing the angles of the optical axis of the cone of view to the surface at different points of the trajectory. This procedure is imple- mented through the diffeomorphisms of the color atlas. Fig. 1. Example of plotting the structure graph of a color atlas Frame-by-frame overlap of the GIO image ensures conti- nuity of monitoring while tracking the projection of the tra- jectory of the web camera on the surface of the GIO location. ( ) : jm jl IMG IMG IMG m l = Δ ≠∅ ≠ ⇔  1) { } { } { } { } ∀ ∈Δ ⇒ ∈ ∧ ∈ ; k k i k i l m Ob IMG Ob Ob Ob Ob 1) { } { } { } { } ∀ ∈Δ ⇒ ∈ ∧ ∈ ; k k i k i l m Ob IMG Ob Ob Ob Ob 1) { } { } { } { } ∀ ∈Δ ⇒ ∈ ∧ ∈ ; k k i k i l m Ob IMG Ob Ob Ob Ob { } { } l m 2) ∃ n S – repainting the vertices IMGl or IMGm on the color wheel while maintaining the adjacency of colors and the adjacency of areas; { } { } l m 2) ∃ n S – repainting the vertices IMGl or IMGm on the color wheel while maintaining the adjacency of colors and the adjacency of areas; 3) in the presence of motion (d), the shift of IMGl relative to the recorder to IMGm is defined as ∆IMG=d(IMGl) and GTA from: Based on the color atlas of the GIO image, a structure graph (CASG) is constructed, the vertices of which are above certain points, and the two vertices are connected by an edge if the color spots defined by these vertices are adjacent. 5. 2. Construction of a topological diagram of dynam ically changing in time scenes of geoinformation objects In the case of low light of the object, the color atlas is the main factor since it makes it pos- sible, through coloring, to determine the boundaries of GIOs – the selectable fragment size has a significant impact on the amount of calculations; 21 2/2 ( 122 ) 2023 2/2 ( 122 ) 2023 Eastern-European Journal of Enterprise Technologies ISSN 1729-3774 Each region is represented by the following tuple of values: the area ⇔ def i Ob {number as name; color from a given color wheel; number of touches and their degrees; number of border segments; boundary structure with cyclic permu- tation accuracy – touch segments}. Let’s denote an image of a GIO as IMG, then partitioning ≡⊕ 1 j n j ji i IMG Ob is a direct Each region is represented by the following tuple of values: the area ⇔ def i Ob {number as name; color from a given – real-time accounting arises in connection with the need to process the flow of images to the level of deci- sion-making; color wheel; number of touches and their degrees; number of border segments; boundary structure with cyclic permu- tation accuracy – touch segments}. Let’s denote an image of j n – the continuity of the input data stream imposes restric- tions on processing time, which leads to a sharp increase in the time spent on computations as a whole. In order to reduce computational costs, the raster image is represented in the form of a topological diagram. To build the diagram, topology methods for incident areas and math- ematical graph theory were used. a GIO as IMG, then partitioning = ≡⊕ 1 j j ji i IMG Ob is a direct algebraic sum of regions. g g Intersection of regions For each spot of this GIO ISO, point Oi is defined (Fig. 1) so that together these points are representable to the color spots of this image [14, 15]. It should be noted that in this approach, the background is nothing more than a separate distributed object in the image. At the same time, it may be incoherent and does not require a special definition as a “background”. ∆IMG∈{ГСЦА(IMGl) ˄ГСЦА(IMGm)}. j The CAS graph has the following properties: Then: d – the top of the CASG with one edge corresponds to an isolated region within another region, an isolated edge region; – {dli} – the set of movements of the recorder with respect to fragment l; – {dli} – the set of movements of the recorder with respect to fragment l; – the associated area of color spots is represented by CASG cycles; – 1 , ; li li i d d + ∀  ( ( – the associated area of color spots is represented by CASG cycles; – 1 , ; li li i d d + ∀  ( ( d d – ( ) ( )   1 lT l l d d IMG is the trajectory T of fragment l. For any unclosed trajectory T, the following is true: y – each vertex is assigned a number; – each vertex is assigned a number; – the vertex has a color corresponding to the color con- taining its area; ( ) ( ) : , T l T T l l n n d IMG ∃ = ∅   ( ) ( ) : , T l T T l l n n d IMG ∃ = ∅   – the edges of the graph connect adjacent regions; T ln is the depth of the fragment l relative to the trajectory T. For any closed trajectory T, the following is true: – adjacency – along the boundary; by touch; d l d h l h h b d – edge load – the length of the common border with the adjacent region; ( ) ( ) ∃ =   : . T l T T l l l n n d IMG IMG – the structure of the adjacency of the vertex: the order of adjacency of the edges is given; regions are touched; mul- tiple touches; – the structure of the adjacency of the vertex: the order of adjacency of the edges is given; regions are touched; mul- tiple touches; Then Δ T l IMG is the monitoring invariant relative to trajectory T, where: p – the multiplicity of the edge – the degree of point of contact of the areas by adjacency; – the multiplicity of the edge – the degree of point of contact of the areas by adjacency; { } ( ) ( ) { } 0, : . 5. 2. Construction of a topological diagram of dynam ically changing in time scenes of geoinformation objects The task of processing visual information in real time when scanning a surface is multiparameter as it requires tak- ing into account many different features and factors. In the process of monitoring geoinformation objects, it is necessary to take into account the following features: – the shape of the GIO is significantly modified depend- ing on its location relative to the elements of the scanning ruler, and the brightness changes tenfold; The ontology of the initial level of ISO is a synthesis of knowledge bases of discrete geometry, photogrammetry, taking into account the internal hierarchy of classes of con- cepts and semantic relations in these classes, and is the ini- tial development of ontology. It is built based on the methods of the theory of structures and the theory of relations. – frames of the same sequence are not reduced to a single coordinate system; – temporary stationarity of the background is observed only on two or three frames of one scanning direction; – descriptions of noise, objects, and background during forward and reverse scanning differ significantly from each other; The construction of an ontology is followed by its re- finement based on classification diagrams of compositional schemes, relationship schemes, and diagrams of the state of objects. These steps are a structured methodology for devel- oping, supporting, and studying ontology. – images can contain so-called “glare” – bright, slowly moving areas from frame to frame; – images can contain so-called “glare” – bright, slowly moving areas from frame to frame; – the solution of the problem of classification of GIOs is difficult due to various kinds of distortions, the absence of key points of the elements of group point objects with a shape and internal structure; – the solution of the problem of classification of GIOs is difficult due to various kinds of distortions, the absence of key points of the elements of group point objects with a shape and internal structure; The definition of a measure of ISO similarity provides additional possibilities for identifying GIOs through the use of a color atlas of the image. – {dli} – the set of movements of the recorder with respect to fragment l; Taking into account the timing scale for individual segments of the projection, it is possible to reproduce the relief of color spots of the “lidar” type. Fig. 1. Example of plotting the structure graph of a color atlas 22 Information technology Cross-linking of GIO images is implemented by tracking the shift of the view cone and image change, as a set of related points of areas migrating from one frame to the next (Fig. 2, where T1, T2 are the points of the camera position in time). Also, to determine reference points, it is possible to use information based on the available geodetic references of GIOs left after construction, or geodetic marks of the object. The above (a priori information and local properties of a particular point) makes it possible to determine the linear and elevation dimensions of GIOs. Fig. 2. Tracking the shift of the overview cone and changes in the color atlas structure graph Modern UAV web cameras have rangefinders, altimeters, and gyroscopes, which, together with positioning equip- ment, make it possible, together with positioning equipment, to accurately record the location of the center of gravity and the orientation of the web camera platform. This makes it possible to highlight a cone around the optical axis of the web camera, which scans the GIO on the surface. As a rule, the inner cone of the survey is known, which has the same optical axis and apex. This cone is determined from the condition of exclusion of marginal distortion in the cone of the review. Pitch, yaw, and roll conversions can cause the image of GIOs to scale. This creates image fragmentation by exclud- ing part of the image from the view cone, or by attaching previously unrecorded GIO fragments to the web camera’s view cone. Fig. 2. Tracking the shift of the overview cone and changes in the color atlas structure graph Due to the constructed color atlas of the image of GIOs, it is possible to represent the movement of the web camera as a collinear displacement of a set of representable points of spots-areas. In this case, the symmetric difference in the image of a GIO of the same spot between its two images, as the oriented collinear displacement is the direction of move- ment of the web camera, the totality of all the spots of the image determines the field of gradients. – {dli} – the set of movements of the recorder with respect to fragment l; Along the direction of travel, the initial CASG fragment has disappeared, and the final fragment is moving. The initial fragment of GCAC is reduced frame by frame until it disappears completely. Let’s introduce the following symbols: – T(ti) – conversion of the GIO image due to pitch at time ti; – P(ti) – conversion of the image of GIOs by yaw; – K(ti) – conversion of the GIO image due to the roll of the web camera frame; – { } 0, , i i N t = ti<tj, i<j – moments of fixation of the GIO im- age, and the exposure time is determined from the conditions of clarity of the image for this web camera and the linear speed of its movement in E3, and ensuring the commutativity of the transformation T, P, K for all eight possible combina- tions of 3-termed sequences. – { } 0, , i i N t = ti<tj, i<j – moments of fixation of the GIO im- age, and the exposure time is determined from the conditions of clarity of the image for this web camera and the linear speed of its movement in E3, and ensuring the commutativity of the transformation T, P, K for all eight possible combina- tions of 3-termed sequences. Along the motion path, the depth (the number of consec- utive frames containing the fragment) of the ISO fragment matches is maximum. Let Or(ti) be the orientation of the web camera at time ti. Then its orientation at the next point in time is defined as ( ) ( ) ( ) + ∀∃ = 1 , : , i j i i j Or t Dr Or t where Drj, 1,8, j = Δt are possi- ble three transformations. Using the color atlas method for the image of GIOs, a se- quence of color portraits is constructed taking into account the local gradients of color and the shape of the single-color local area. The resulting color portraits of the GIO image in con- junction with the location and orientation of the web camera make it possible to build a dynamic series of changes in its color atlas. In terms of the color atlas of an image, it makes it possible to construct or define the designs of areas in the image and the shape sets of areas inherent in each particular row. – {dli} – the set of movements of the recorder with respect to fragment l; At the same time, the invariant for determining the similarity of the areas determined by a representative set of points on a given portrait series, taking into account rotations and azimuth removal, is the preservation of proportions between them. Topological relations between points are more resistant to image transformations, do not depend on the coordinate representation of the form under consideration and are in- variant with respect to diffeomorphic transformations. The implementation of the comparison of the similarity of CASG images is based on computational topology algorithms for objects represented by sets of points [18]. For this purpose, for example, the method of geometric hashing and similarity comparison based on Hamming distance (a method widely used in computational geometry to compare images [19]) is used, in conjunction with the pHash perceptual hashing algorithm. A distinctive feature of the use of the pHash comparison method in relation to the methods of algebraic topology is the receipt of invariants of large amounts of data, finding the distance between images by entering non-geo- metric information about the distances between them. When comparing images, hash values are compared using coef- ficients of difference or similarity between two perceptual hash values. One of the factors in determining the similarity of GIO images is the presence of images of special points or special areas in brightness, color, or shape, which can be determined based on preliminary information on the location of GIO images. To detect a feature description of an image, it is neces- sary to bind to its local features – special points. A special point, or feature, is an image point that satisfies a number of properties: – certainty; The development of the application of persistent ho- mologies for CASG is investigating invariants of large data volumes. The use of persistent homologies makes it possible: – robustness; – invariance; – stability; – to increase the amount of information about the shape of the object; – to increase the amount of information about the shape of the object; – interpretability; – to provide the possibility of constructing a diffeomor- phic mapping with the formation of metamorphosis, in which – to provide the possibility of constructing a diffeomor- phic mapping with the formation of metamorphosis, in which – the number of detected special points should provide the required number of them for the detection of objects. – {dli} – the set of movements of the recorder with respect to fragment l; – the number of detected special points should provide the required number of them for the detection of objects. 23 2/2 ( 122 ) 2023 Eastern-European Journal of Enterprise Technologies ISSN 1729-3774 the images are not diffeomorphic and have different topolog- ical characteristics; method is the possibility of including in the search procedure not only additional features that make it possible to clarify the class of CASGs but also various logical or heuristic rules, as well as analytical expressions describing this subject area. The use of tools of logical, heuristic rules, and analytical ex- pressions will make it possible to adjust the assessment of the values of the CASG features in a necessary way and increase the reliability of solving problems. – to find the similarity coefficients of images; – to develop a method that is resistant to various modi- fications of images. 7. Conclusions A feature of the proposed method and the results ob- tained on its basis is the construction of a topological scheme (invariant) of interframe processing of a digital im- age based on topological features. 1. The initial ontology of the digital image is determined based on the initial information obtained from the web camera, taking into account the dynamics of the position of the optical source, for example, UAV. Thus, the objective characteristics, dynamically changing in time scenes of geoinformation objects are taken into account. The results of the research indicate that by constructing a graph of the structure of the color atlas based on ODI, a measure of similarity as a metric of structures is realized. An invariant is defined as a CASG fragment that is inherited by all sequence images (Fig. 2). The search for new invariants of the sequence of images in time is a promising method. 2. A topological scheme representing digital images of inter-frame processing of a temporary sequence of imag- es (topological scheme) taken by a web camera has been constructed, which makes it possible to track the inheritance of primary source information. The constructed scheme is an invariant that is simultaneously resistant to diffeomorphic transformations of images and has an acceptable speed of operation in determining the similarity of the image. q g p g The peculiarity of the proposed method and the results obtained is the application of ODI ontologies in conjunction with the pHash perceptual hashing algorithm. More research is needed on the dependence of the speed of image invariant processing on the factor of detail and col- or space, the height and frequency of image formation, the construction of image processing processors. 1. Zhang, J., Li, W., Ogunbona, P., Xu, D. (2019). Recent Advances in Transfer Learning for Cross-dataset Visual Recognition: A Problem-oriented Perspective. ACM Computing Surveys, 52 (1), 1–38. doi: https://doi.org/10.1145/3291124 2. Yuan, X., Shi, J., Gu, L. (2021). A review of deep learning methods for semantic segmentation of remote sensing imagery. Expert Systems with Applications, 169, 114417. doi: https://doi.org/10.1016/j.eswa.2020.114417 3. Yasin, H. M., Ameen, S. Y. (2021). Review and Evaluation of End-to-End Video Compression with Deep-Learning. 2021 International Conference of Modern Trends in Information and Communication Technology Industry (MTICTI). doi: https:// doi.org/10.1109/MTICTI53925.2021.9664790 4. Fan, L., Zhang, T., Du, W. (2021). Optical-flow-based framework to boost video object detection performance with object enhancement. Expert Systems with Applications, 170, 114544. doi: https://doi.org/10.1016/j.eswa.2020.114544 5. Xue, T., Chen, B., Wu, J., Wei, D., Freeman, W. T. (2019). Video Enhancement with Task-Oriented Flow. International Journal of Computer Vision, 127 (8), 1106–1125. doi: https://doi.org/10.1007/s11263-018-01144-2 6. Que, Z., Lu, G., Xu, D. (2021). VoxelContext-Net: An Octree based Framework for Point Cloud Compression. 2021 IEEE/ CVF Conference on Computer Vision and Pattern Recognition (CVPR). doi: https://doi.org/10.1109/cvpr46437.2021.00598 7. Duan, L., Liu, J., Yang, W., Huang, T., Gao, W. (2020). Video Coding for Machines: A Paradigm of Collaborative Compression and Intelligent Analytics. IEEE Transactions on Image Processing, 29, 8680–8695. doi: https://doi.org/10.1109/tip.2020.3016485 Conf licts of interest The disadvantages of this study include the need for addi- tional studies of UAV recording facilities, which have a limita- tion in terms of considering specific implementations. However, this can be leveled by defining and applying minimum require- ments for the registration means of UAVs of different types. The authors declare that they have no conflicts of in- terest in relation to the current study, including financial, personal, authorship, or any other, that could affect the study and the results reported in this paper. The proposed approach can be successfully developed and applied to the analysis of images by methods of artificial intelligence, learning neural networks, automation of knowl- edge assessment, knowledge engineering, which are widely used in different countries of the world, including Ukraine. The study was conducted without financial support. The study was conducted without financial support. The subject of further research is the experimental confir- mation of the theoretical results obtained in the current work. The conducted research based on the ontology of the initial level of the image makes it possible to provide objec- tive characteristics of dynamically changing in time scenes of geoinformation objects. The advantage of the presented All data are available in the main text of the manuscript. References 1. Zhang, J., Li, W., Ogunbona, P., Xu, D. (2019). Recent Advances in Transfer Learning for Cross-dataset Visual Recognition: A Problem-oriented Perspective. ACM Computing Surveys, 52 (1), 1–38. doi: https://doi.org/10.1145/3291124 Shi, J., Gu, L. (2021). A review of deep learning methods for semantic segmentation of remote sensing imagery. Expert with Applications, 169, 114417. doi: https://doi.org/10.1016/j.eswa.2020.114417 3. Yasin, H. M., Ameen, S. Y. (2021). Review and Evaluation of End-to-End Video Compression with Deep-Learning. 2021 International Conference of Modern Trends in Information and Communication Technology Industry (MTICTI). doi: https:// doi.org/10.1109/MTICTI53925.2021.9664790 4. Fan, L., Zhang, T., Du, W. (2021). Optical-flow-based framework to boost video object detection performance with object enhancement. Expert Systems with Applications, 170, 114544. doi: https://doi.org/10.1016/j.eswa.2020.114544 6. Que, Z., Lu, G., Xu, D. (2021). VoxelContext-Net: An Octree based Framework for Point Cloud Compression. 2021 IEEE/ CVF Conference on Computer Vision and Pattern Recognition (CVPR). doi: https://doi.org/10.1109/cvpr46437.2021.00598 7. Duan, L., Liu, J., Yang, W., Huang, T., Gao, W. (2020). Video Coding for Machines: A Paradigm of Collaborative Compression and Intelligent Analytics. IEEE Transactions on Image Processing, 29, 8680–8695. doi: https://doi.org/10.1109/tip.2020.3016485 24 Information technology 8. Lu, G., Cai, C., Zhang, X., Chen, L., Ouyang, W., Xu, D., Gao, Z. (2020). Content Adaptive and Error Propagation Aware Deep Video Compression. Lecture Notes in Computer Science, 456–472. doi: https://doi.org/10.1007/978-3-030-58536-5_27 9. Ballé, J., Laparra, V., Simoncelli, E. P. (2017). End-to-end Optimized Image Compression. in Proc. Int. C Representations (ICLR). doi: https://doi.org/10.48550/arXiv.1611.01704 parra, V., Simoncelli, E. P. (2017). End-to-end Optimized Image Compression. in Proc. Int. Conf. Learning ons (ICLR). doi: https://doi.org/10.48550/arXiv.1611.01704 10. Liu, D., Li, Y., Lin, J., Li, H., Wu, F. (2020). Deep Learning-Based Video Coding. ACM Computing Surveys, 53 (1), 1–35. doi: https://doi.org/10.1145/3368405 11. Subramanian, N., Elharrouss, O., Al-Maadeed, S., Bouridane, A. (2021). Image Steganography: A Review of the Recent Advances. IEEE Access, 9, 23409–23423. doi: https://doi.org/10.1109/access.2021.3053998 11. Subramanian, N., Elharrouss, O., Al-Maadeed, S., Bouridane, A. (2021). Image Steganography: IEEE Access, 9, 23409–23423. doi: https://doi.org/10.1109/access.2021.3053998 12. Hu, Y., Yang, W., Ma, Z., Liu, J. (2021). Learning End-to-End Lossy Image Compression: A Benchmark. IEEE Transactions on Pattern Analysis and Machine Intelligence. doi: https://doi.org/10.1109/tpami.2021.3065339 Yang, W., Ma, Z., Liu, J. (2021). Learning End-to-End Lossy Image Compression: A Benchmark. IEEE Transactions on ern Analysis and Machine Intelligence. doi: https://doi.org/10.1109/tpami.2021.3065339 Lu, G., Xu, D. (2021). FVC: A New Framework towards Deep Video Compression in Feature Space. 2021 IEEE/CVF 13. Hu, Z., Lu, G., Xu, D. (2021). All data are available in the main text of the manuscript. FVC: A New Framework towards Deep Video Compression in Feature Space. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). doi: https://doi.org/10.1109/cvpr46437.2021.00155 Conference on Computer Vision and Pattern Recognition (CVPR). doi: https://doi.org/10.1109/cvpr46437.2021.00155 14. Horda, O. V. (2016). Construction of the color atlas for digital image of defects such as a “crack.” ScienceRise, 10 (2 (27)), 55–60. doi: https://doi.org/10.15587/2313-8416.2016.80791 E. V., Mykhailenko, V. M. (2016). Struktura tsvetovoho atlasa tsyfrovoho yzobrazhenyia defekta typa "treshchyna ". c discussion, 3, 58–60. 16. Nazarenko, I., Dedov, O., Bernyk, I., Rogovskii, I., Bondarenko, A., Zapryvoda, A. et al. (2020). Determining the regions of stability in the motion regimes and parameters of vibratory machines for different technological purposes. Eastern-European Journal of Enterprise Technologies, 6 (7 (108)), 71–79. doi: https://doi.org/10.15587/1729-4061.2020.217747 17. Gorda, E., Mihaylenko, V. (2017). Ontology of digital image of defects of type "crack" of objects of construction. Upravlinn rozvytkom skladnykh system, 30, 142–145. Available at: http://urss.knuba.edu.ua/files/zbirnyk-30/21.pdf (2017). Ontology of digital image of defects of type "crack" 18. Serdyuk, A. A., Belyavskiy, L. S. (2018). Analiz veroyatnostno-geometricheskogo metoda opredeleniya mes dvizhuschikhsya obektov. Sovremennaya spetsial'naya tekhnika, 1 (52), 6–13. 19. Wolfson, H. J., Rigoutsos, I. (1997). Geometric hashing: an overview. IEEE Computational Science and Engineering, 4 (4), 10–21. doi: https://doi.org/10.1109/99.641604 25
https://openalex.org/W2737369361	https://dash.harvard.edu/bitstream/1/34491858/1/5593637.pdf	English	null	ARD1-mediated aurora kinase A acetylation promotes cell proliferation and migration	Oncotarget	2,017	cc-by	11,191	Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAA Published Version doi:10.18632/oncotarget.19332 doi:10.18632/oncotarget.19332 Permanent link http://nrs.harvard.edu/urn-3:HUL.InstRepos:34491858 ARD1-mediated aurora kinase A acetylation promotes cell proliferation and migration Tam Thuy Lu Vo1, Ji-Hyeon Park1, Ji Hae Seo2, Eun Ji Lee1, Hoon Choi1, Sung-Jin Bae1, Hoang Le1, Sunho An1, Hye Shin Lee1, Hee-Jun Wee1 and Kyu-Won Kim1,3 1SNU-Harvard NeuroVascular Protection Research Center, College of Pharmacy and The Research Institute of Pharmaceutical Sciences, Seoul National University, Seoul 08826, Korea 2Department of Biochemistry, School of Medicine, Keimyung University, Daegu 42601, Korea 3Crop Biotechnology Institute, GreenBio Science and Technology, Seoul National University, Pyeongchang 25354, Korea Correspondence to: Kyu-Won Kim, email: qwonkim@snu.ac.kr Keywords: aurora kinase A, ARD1, lysine acetylation, cell proliferation, cell migration Abbreviations: AuA: Aurora kinase A, ARD1: Arrest defective protein 1, D-box: destruction box, DAD: D-box activating domain Received: May 04, 2017 Accepted: June 30, 2017 Published: July 18, 2017 Copyright: Vo et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License 3.0 (CC BY 3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Tam Thuy Lu Vo1, Ji-Hyeon Park1, Ji Hae Seo2, Eun Ji Lee1, Hoon Choi1, Sung-Jin Bae1, Hoang Le1, Sunho An1, Hye Shin Lee1, Hee-Jun Wee1 and Kyu-Won Kim1,3 1SNU-Harvard NeuroVascular Protection Research Center, College of Pharmacy and The Research Institute of Pharmaceutical Sciences, Seoul National University, Seoul 08826, Korea 2Department of Biochemistry, School of Medicine, Keimyung University, Daegu 42601, Korea 3Crop Biotechnology Institute, GreenBio Science and Technology, Seoul National University, Pyeongchang 25354, Korea Correspondence to: Kyu-Won Kim, email: qwonkim@snu.ac.kr Keywords: aurora kinase A, ARD1, lysine acetylation, cell proliferation, cell migration Abbreviations: AuA: Aurora kinase A, ARD1: Arrest defective protein 1, D-box: destruction box, DAD: D-box activating domain Received: May 04, 2017 Accepted: June 30, 2017 Published: July 18, 2017 Copyright: Vo et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License 3.0 (CC BY 3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. t al. This is an open-access article distributed under the terms of the Creative Commons Attribution License 3.0 ( nrestricted use, distribution, and reproduction in any medium, provided the original author and source are credit ABSTRACT Aurora kinase A (AuA) is a prerequisite for centrosome maturation, separation, and mitotic spindle assembly, thus, it is essential for cell cycle regulation. Overexpression of AuA is implicated in poor prognosis of many types of cancer. However, the regulatory mechanisms underlying the functions of AuA are still not fully understood. Here, we report that AuA colocalizes with arrest defective protein 1 (ARD1) acetyltransferase during cell division and cell migration. Additionally, AuA is acetylated by ARD1 at lysine residues at positions 75 and 125. The double mutations at K75/K125 abolished the kinase activity of AuA. Moreover, the double mutant AuA exhibited diminished ability to promote cell proliferation and cell migration. Mechanistic studies revealed that AuA acetylation at K75/K125 promoted cell proliferation via activation of cyclin E/CDK2 and cyclin B1. In addition, AuA acetylation stimulated cell migration by activating the p38/AKT/MMP-2 pathway. Our findings indicate that ARD1-mediated acetylation of AuA enhances cell proliferation and migration, and probably contributes to cancer development. Share Your Story The Harvard community has made this article openly available. Please share how this access benefits you. Submit a story . Accessibility www.impactjournals.com/oncotarget/ Oncotarget, 2017, Vol. 8, (No. 34), pp: 57216-57230 ARD1-mediated aurora kinase A acetylation promotes cell proliferation and migration Tam Thuy Lu Vo1, Ji-Hyeon Park1, Ji Hae Seo2, Eun Ji Lee1, Hoon Choi1, Sung-Jin Bae1, Hoang Le1, Sunho An1, Hye Shin Lee1, Hee-Jun Wee1 and Kyu-Won Kim1,3 1SNU-Harvard NeuroVascular Protection Research Center, College of Pharmacy and The Research Institute of Pharmaceutical Sciences, Seoul National University, Seoul 08826, Korea 2Department of Biochemistry, School of Medicine, Keimyung University, Daegu 42601, Korea 3Crop Biotechnology Institute, GreenBio Science and Technology, Seoul National University, Pyeongchang 25354, Korea Correspondence to: Kyu-Won Kim, email: qwonkim@snu.ac.kr Keywords: aurora kinase A, ARD1, lysine acetylation, cell proliferation, cell migration Abbreviations: AuA: Aurora kinase A, ARD1: Arrest defective protein 1, D-box: destruction box, DAD: D-box activating domain Received: May 04, 2017 Accepted: June 30, 2017 Published: July 18, 2017 Copyright: Vo et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License 3.0 (CC BY 3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Research Paper www.impactjournals.com/oncotarget/ www.impactjournals.com/oncotarget INTRODUCTION cancers such as rectal, ovarian, prostate, colon, and breast cancers [5–7]. The catalytic domain of AuA shares 70% homology with that of other members of the Aurora kinase family, whereas the N-terminal regulatory domain is distinct. The regulatory domain of AuA is responsible for its binding with protein partners. AuA kinase activity has been extensively studied for its vital role in centrosome functions as well as mitotic regulation [4, 8, 9]. AuA kinase is activated by phosphorylation at threonine 288 by TPX2 [10–12]. AuA then phosphorylates the cell cycle regulator cyclin B1/CDK1, promoting mitotic entry [13–15]. Because of the overexpression of AuA in many types of metastatic cancers and its correlation with poor prognosis [16–20], the non-mitotic functions of AuA, in addition to its role in mitotic regulation, have been elucidated. Recent findings have revealed the contribution of AuA in cell migration and metastasis enhancement. Defective control of cell proliferation is the main characteristic of cancer. In normal cells, cell division is rigorously controlled by complex signaling circuits. However, cancer cells can sustain proliferative signaling themselves and can grow unstoppably. Another hallmark of cancer is cell migration [1]. Cell migration is a highly integrated process, and controlled by dynamic regulatory mechanisms [2]. Hence, understanding the mechanisms of controlling cell proliferation and cell motility could open new horizons in cancer therapy. Aurora kinase A (AuA) is a member of the serine/ threonine kinase family, which has been implicated in controlling the cell cycle and cell division [3, 4]. The AuA-encoding gene is located on chromosome 20q13.2-q13.33, a region frequently amplified in many www.impactjournals.com/oncotarget Oncotarget 57216 lysates of cells overexpressing ARD1 and found that ARD1 was bound to AuA (Figure 2A). As ARD1 is an acetyltransferase that catalyzes the transfer of acetyl groups from acetyl-coA onto lysine residues in its substrates, we hypothesized that AuA could be modulated by ARD1 by acetylation. To examine this hypothesis, we performed an in vitro acetylation assay in which recombinant His-tagged AuA was mixed with recombinant His-tagged ARD1 in the presence of acetyl-CoA. Expectedly, AuA was acetylated by ARD1 (Figure 2B). Consistent with the in vitro experiment, the overexpression of ARD1 significantly upregulated the level of AuA acetylation in cells (Figure 2C). Interestingly, AuA acetylation occurred in a time-dependent manner after autoacetylation of ARD1 (Figure 2D), suggesting that the autoacetylation of ARD1 is essential for regulating AuA acetylation. AuA and ARD1 colocalize during cell division and cell migration AuA comprises 403 amino acids and has two domains, an N-terminal domain spanning residues 1 to 131, and a C-terminal domain spanning residues 132 to 403. The C-terminus includes a catalytic domain that harbors the kinase activity and a destruction box (D-box) that plays a role in ubiquitin-mediated degradation of several mitotic proteins. The N-terminus contains the A-box/D-box activating domain (DAD) that controls AuA degradation (Figure 3A). However, the function of the N-terminal domain is yet unclear [4, 8]. To identify the target sites on AuA that are acetylated by ARD1, we performed in vitro acetylation assays with recombinant AuA. For this, we constructed two truncated fragments of AuA, an N-terminal domain-containing fragment comprising amino acids 1 to 140 and a C-terminal domain-containing fragment comprising residues 126 to 403 (Figure 3A). As shown in Figure 3A, the N-terminal domain of AuA was acetylated, but not the C-terminal domain. To further delineate the residues involved in ARD1-mediated AuA acetylation, a series of N-terminal fragments were generated, in which the lysine residues were substituted with arginine to mimic non-acetylated lysine, and in vitro acetylation assays were performed. Lysines at positions 75 and 125 were identified as preferable sites for AuA acetylation (Figure 3B). Indeed, AuA acetylation was almost negligible when the K75R/K125R double mutant Recent studies showed that ARD1 is essential for cell survival, as knockdown of ARD1 in human cells triggers apoptosis and G1 arrest [27, 29]. We observed the centrosome-like localization of ARD1 (Supplementary Figure 1), and this prompted us to examine whether ARD1 localized to the centrosome. AuA is considered as a centrosome marker owing to its localization to the centrosome and its function in centrosome maturation and separation [30, 31]. Interestingly, ARD1 colocalized with AuA during the cell cycle progression, particularly during the cell division (Figure 1A, 1C). The centrosome not only participates in cell cycle progression, but also contributes to cell migration by undergoing reorientation and supporting the forces generated in cells during cell migration [32]. AuA, as a component of the centrosome, plays a part during cell migration. Here, we also observed the colocalization of ARD1 and AuA in migrating cells (Figure 1B, 1C). These observations suggest that interaction between ARD1 and AuA at the centrosome may play a critical role in cell growth and cell movement. INTRODUCTION Previously, we reported that ARD1, in addition to acetylating a variety of substrates, undergoes self-acetylation and that arginine 82 (R82) and tyrosine 122 (Y122) are required for its acetyltransferase activity [28]. Thus, we examined the levels of AuA acetylation in the presence of functional (wild-type) and R82A/ Y122F mutant ARD1 proteins. It was seen that the AuA acetylation level decreased dramatically when ARD1 was mutated at R82 and Y122 (Figure 2E). Taken together, these data indicate that AuA interacts with ARD1, and AuA acetylation is regulated by functional ARD1. AuA reportedly activates cell migration by modulating MMP-2 expression [21]. However, the mechanisms regulating AuA activity in cell movement are still unclear. Acetylation of lysine residues is one of the major essential post-translational modifications [22], and plays a significant role in modulating histone modification, thus regulating a large number of cellular events. In addition, lysine acetylation modification of non-histone proteins has been shown to be related to several regulatory mechanisms in both healthy and pathological cells [23]. The enzymes responsible for this modification are identified as lysine acetyltransferases. ARD1 is identified as a catalytic subunit of N-acetyltransferase [24], and acts on various substrates related to cell homeostasis, cell movement, and cancer development, such as HSP70 [25], myosin light chain kinase [26], and β-catenin [27]. Furthermore, ARD1 can acetylate itself and this autoacetylation of ARD1 is crucial for its activation [28]. ARD1, therefore, plays important roles in many biological processes.ii In this study, we identified a previously undefined post-translational modification of AuA by ARD1 acetyltransferase. We found that AuA is acetylated at lysines at positions 75 and 125, consequently this contributes to AuA catalytic capacity, then leading to promotion of cell proliferation and stimulation of cell migration. Acetylation of AuA is regulated by ARD1 To determine whether AuA and ARD1 physically interact in cells, we immunoprecipitated ARD1 from www.impactjournals.com/oncotarget Oncotarget 57217 compared to those of other Aurora kinases in regulating cellular processes. AuA was subjected to in vitro acetylation (Figure 3C). Similarly, cells overexpressing AuA double mutant K75R/K125R displayed a dramatically decreased level of acetylated AuA (Figure 3D), suggesting that these sites are critical for the acetylation of AuA by ARD1. These two sites are conserved across species (Figure 3E), indicating that these sites may be essential for regulating AuA activity. However, the sites are not conserved in other members of the Aurora kinase family (Figure 3F), and are probably responsible for the distinct functions of AuA AuA was subjected to in vitro acetylation (Figure 3C). Similarly, cells overexpressing AuA double mutant K75R/K125R displayed a dramatically decreased level of acetylated AuA (Figure 3D), suggesting that these sites are critical for the acetylation of AuA by ARD1. These two sites are conserved across species (Figure 3E), indicating that these sites may be essential for regulating AuA activity. However, the sites are not conserved in other members of the Aurora kinase family (Figure 3F), and are probably responsible for the distinct functions of AuA ARD1-mediated AuA acetylation at K75/K125 enhances AuA kinase activity (A) Left, Aurora A colocalizes with ARD1 at centrosomes during cell division in prophase, metaphase, anaphase and telophase. GFP-ARD1 overexpressing cells (green) were synchronized and stained with anti-AuA antibody (red). DNA was counter-stained by Hoechst. Cells were then viewed under a confocal microscope. Scale bar, 5 µm. Colocalization of ARD1 and AuA is indicated by arrows. Right, Expression of ARD1 and AuA during cell cycle progression. Cyclin B1 was used as the markers of the cell cycle phases. GAPDH was used as loading control. (B) Left, Aurora A colocalizes with ARD1 in migrating cell. Monolayer of GFP-ARD1 overexpressing cells (green) on coverslip was scratched to stimulate migration, then cells were stained with anti-AuA antibody (red). DNA was counter-stained by Hoechst. Scale bar, 5 µm. Colocalization of Figure 1: Aurora kinase A colocalalizes with ARD1 during cell division and cell migration. Figure 1: Aurora kinase A colocalalizes with ARD1 during cell division and cell migration. (A) Left, Aurora A colocalizes with ARD1 at centrosomes during cell division in prophase, metaphase, anaphase and telophase. GFP-ARD1 overexpressing cells (green) were synchronized and stained with anti-AuA antibody (red). DNA was counter-stained by Hoechst. Cells were then viewed under a confocal microscope. Scale bar, 5 µm. Colocalization of ARD1 and AuA is indicated by arrows. Right, Expression of ARD1 and AuA during cell cycle progression. Cyclin B1 was used as the markers of the cell cycle phases. GAPDH was used as loading control. (B) Left, Aurora A colocalizes with ARD1 in migrating cell. Monolayer of GFP-ARD1 overexpressing cells (green) on coverslip was scratched to stimulate migration, then cells were stained with anti-AuA antibody (red). DNA was counter-stained by Hoechst. Scale bar, 5 µm. Colocalization of ARD1 and AuA is indicated by arrows. Right, Expression of ARD1 and AuA during cell migration. Vinculin was used as loading control. (C) The percentage of cells exhibit the colocalization of ARD1 and AuA during the cell cycle progression and cell migration. Figure 1: Aurora kinase A colocalalizes with ARD1 during cell division and cell migration. (A) Left, Aurora A colocalizes with ARD1 at centrosomes during cell division in prophase, metaphase, anaphase and telophase. GFP-ARD1 overexpressing cells (green) were synchronized and stained with anti-AuA antibody (red). DNA was counter-stained by Hoechst. Cells were then viewed under a confocal microscope. Scale bar, 5 µm. Colocalization of ARD1 and AuA is indicated by arrows. ARD1-mediated AuA acetylation at K75/K125 enhances AuA kinase activity The functions of AuA in diverse cellular processes are related to its kinase activity. Several studies have proposed that phosphorylation affects AuA kinase activity. To investigate whether AuA : Aurora kinase A colocalalizes with ARD1 during cell division and cell migration. (A) Left, Aurora A colocalizes 1 at centrosomes during cell division in prophase, metaphase, anaphase and telophase. GFP-ARD1 overexpressing cells (green) hronized and stained with anti-AuA antibody (red). DNA was counter-stained by Hoechst. Cells were then viewed under a microscope. Scale bar, 5 µm. Colocalization of ARD1 and AuA is indicated by arrows. Right, Expression of ARD1 and AuA during progression. Cyclin B1 was used as the markers of the cell cycle phases. GAPDH was used as loading control. (B) Left, Aurora zes with ARD1 in migrating cell. Monolayer of GFP-ARD1 overexpressing cells (green) on coverslip was scratched to stimulate then cells were stained with anti-AuA antibody (red). DNA was counter-stained by Hoechst. Scale bar, 5 µm. Colocalization of d AuA is indicated by arrows. Right, Expression of ARD1 and AuA during cell migration. Vinculin was used as loading control. ercentage of cells exhibit the colocalization of ARD1 and AuA during the cell cycle progression and cell migration. Figure 1: Aurora kinase A colocalalizes with ARD1 during cell division and cell migration. (A) Left, Aurora A colocalizes with ARD1 at centrosomes during cell division in prophase, metaphase, anaphase and telophase. GFP-ARD1 overexpressing cells (green) were synchronized and stained with anti-AuA antibody (red). DNA was counter-stained by Hoechst. Cells were then viewed under a confocal microscope. Scale bar, 5 µm. Colocalization of ARD1 and AuA is indicated by arrows. Right, Expression of ARD1 and AuA during cell cycle progression. Cyclin B1 was used as the markers of the cell cycle phases. GAPDH was used as loading control. (B) Left, Aurora A colocalizes with ARD1 in migrating cell. Monolayer of GFP-ARD1 overexpressing cells (green) on coverslip was scratched to stimulate migration, then cells were stained with anti-AuA antibody (red). DNA was counter-stained by Hoechst. Scale bar, 5 µm. Colocalization of ARD1 and AuA is indicated by arrows. Right, Expression of ARD1 and AuA during cell migration. Vinculin was used as loading control. (C) The percentage of cells exhibit the colocalization of ARD1 and AuA during the cell cycle progression and cell migration. Figure 1: Aurora kinase A colocalalizes with ARD1 during cell division and cell migration. ARD1-mediated AuA acetylation at K75/K125 enhances AuA kinase activity Ponceau S staining shows the quantification of the input proteins. The experiments were performed at least three times independently. (C) Acetylated AuA level increases in GFP-ARD1 overexpressing cells. Lysates from GFP-ARD1 overexpressing MCF7 cells were immuprecipitated with anti- Lys-Ac antibody and analyzed by immunoblotting with anti-AuA antibody or anti-GFP antibody. The experiments were performed at least three times independently. (D) AuA acetylation occurs in a time-dependent manner. His-ARD1 recombinants were subjected to in vitro acetylation assays for series of time, and acetylation levels of recombinants were assessed by western blotting using an anti-Lys-Ac antibody. Quantification of the input proteins were analyzed by Ponceau S staining. The experiments were performed at least three times independently. (E) AuA acetylation is dependent on ARD1 acetyltransferase activity. MCF7 cells were transfected with wild type (WT) GFP-ARD1 or GFP-ARD1 R82F/Y122A mutant. The extracts from the overexpressing cells were immoprecipitated with anti Lys-Ac antibody and acetylated AuA levels were analyzed by immunoblotting with anti-AuA antibody. The experiments were performed at least three times independently. Figure 2: Aurora A is acetylated by ARD1. (A) AuA interacts with ARD1. Lysates from HEK293T cells overexpressing GFP-ARD1 were immunoprecipitated with anti-GFP antibody and immunoblotted with anti-AuA antibody or anti-GFP antibody. The experiments were performed at least three times independently. (B) AuA is acetylated by ARD1 in vitro. His-ARD1, His-AuA recombinants were subjected to in vitro acetylation assays with or without presence of acetyl group donor acetyl- coenzyme A (CoA) for 1 h, and acetylation levels of recombinants were assessed by western blotting using an anti-acetylated lysine antibody (Lys-Ac). Ponceau S staining shows the quantification of the input proteins. The experiments were performed at least three times independently. (C) Acetylated AuA level increases in GFP-ARD1 overexpressing cells. Lysates from GFP-ARD1 overexpressing MCF7 cells were immuprecipitated with anti- Lys-Ac antibody and analyzed by immunoblotting with anti-AuA antibody or anti-GFP antibody. The experiments were performed at least three times independently. (D) AuA acetylation occurs in a time-dependent manner. His-ARD1 recombinants were subjected to in vitro acetylation assays for series of time, and acetylation levels of recombinants were assessed by western blotting using an anti-Lys-Ac antibody. Quantification of the input proteins were analyzed by Ponceau S staining. The experiments were performed at least three times independently. (E) AuA acetylation is dependent on ARD1 acetyltransferase activity. MCF7 cells were transfected with wild type (WT) GFP-ARD1 or GFP-ARD1 R82F/Y122A mutant. ARD1-mediated AuA acetylation at K75/K125 enhances AuA kinase activity Right, Expression of ARD1 and AuA during cell cycle progression. Cyclin B1 was used as the markers of the cell cycle phases. GAPDH was used as loading control. (B) Left, Aurora A colocalizes with ARD1 in migrating cell. Monolayer of GFP-ARD1 overexpressing cells (green) on coverslip was scratched to stimulate migration, then cells were stained with anti-AuA antibody (red). DNA was counter-stained by Hoechst. Scale bar, 5 µm. Colocalization of ARD1 and AuA is indicated by arrows. Right, Expression of ARD1 and AuA during cell migration. Vinculin was used as loading control. (C) The percentage of cells exhibit the colocalization of ARD1 and AuA during the cell cycle progression and cell migration. www.impactjournals.com/oncotarget Oncotarget 57218 re 2: Aurora A is acetylated by ARD1. (A) AuA interacts with ARD1. Lysates from HEK293T cells overexpressing GF immunoprecipitated with anti-GFP antibody and immunoblotted with anti-AuA antibody or anti-GFP antibody. The exp performed at least three times independently. (B) AuA is acetylated by ARD1 in vitro. His-ARD1, His-AuA recombina cted to in vitro acetylation assays with or without presence of acetyl group donor acetyl- coenzyme A (CoA) for 1 h, and ac s of recombinants were assessed by western blotting using an anti-acetylated lysine antibody (Lys-Ac). Ponceau S stainin uantification of the input proteins. The experiments were performed at least three times independently. (C) Acetylated A ases in GFP-ARD1 overexpressing cells. Lysates from GFP-ARD1 overexpressing MCF7 cells were immuprecipitated w Ac antibody and analyzed by immunoblotting with anti-AuA antibody or anti-GFP antibody. The experiments were perf three times independently. (D) AuA acetylation occurs in a time-dependent manner. His-ARD1 recombinants were sub ro acetylation assays for series of time, and acetylation levels of recombinants were assessed by western blotting using an ant ody Quantification of the input proteins were analyzed by Ponceau S staining The experiments were performed at least th Figure 2: Aurora A is acetylated by ARD1. (A) AuA interacts with ARD1. Lysates from HEK293T cells overexpressing GFP-ARD1 were immunoprecipitated with anti-GFP antibody and immunoblotted with anti-AuA antibody or anti-GFP antibody. The experiments were performed at least three times independently. (B) AuA is acetylated by ARD1 in vitro. His-ARD1, His-AuA recombinants were subjected to in vitro acetylation assays with or without presence of acetyl group donor acetyl- coenzyme A (CoA) for 1 h, and acetylation levels of recombinants were assessed by western blotting using an anti-acetylated lysine antibody (Lys-Ac). ARD1-mediated AuA acetylation at K75/K125 enhances AuA kinase activity (B) Lysines at positions 75 and 125 of AuA are acetylated by ARD1. Mutagenesis was performed to substitute lysine residues in N-terminus by arginine. Site-mutated AuA recombinants were then applied to in vitro assays. The experiments were performed at least three times independently. (C) Double mutant AuA K75R/K125R exhibits the negligible acetylation. His-AuA wild type (WT) and K75R/K125R double mutant (DM) were subjected to in vitro acetylation assays with His-ARD1. Acetylation levels were assessed by western blot. Each experiment was repeated at least three times. (D) Double mutant AuA K75R/K125R displays significantly decreased level of acetylated AuA in cells. The lysates from stable cells overexpressing RFP- AuA WT or RFP-AuA K75R/K125R mutant were collected for immunoprecipitation using anti-Lys-Ac antibody. The immunoprecipitates were examined for acetylated AuA by blotting with anti-AuA antibody. Each experiment was repeated three times independently. (E) The lysine residues at positions 75 and 125 are conserved across species. Protein sequence alignment of AuA in various species. (F) The lysine residues at positions 75 and 125 are not conserved among Aurora kinase family proteins. Protein sequence alignment of AuA in Aurora kinase family proteins. Figure 3: K75, K125 of Aurora A are acetylated by ARD1. (A) ARD1 acetylates the N-terminus of AuA in vitro. Top, deletion mutants of His-AuA were subjected to in vitro acetylation assays with ARD1 recombinant protein. Bottom, construction of AuA deletion mutants. AuA-N, AuA N-terminal domain; AuA-C, AuA C-terminal domain. (B) Lysines at positions 75 and 125 of AuA are acetylated by ARD1. Mutagenesis was performed to substitute lysine residues in N-terminus by arginine. Site-mutated AuA recombinants were then applied to in vitro assays. The experiments were performed at least three times independently. (C) Double mutant AuA K75R/K125R exhibits the negligible acetylation. His-AuA wild type (WT) and K75R/K125R double mutant (DM) were subjected to in vitro acetylation assays with His-ARD1. Acetylation levels were assessed by western blot. Each experiment was repeated at least three times. (D) Double mutant AuA K75R/K125R displays significantly decreased level of acetylated AuA in cells. The lysates from stable cells overexpressing RFP- AuA WT or RFP-AuA K75R/K125R mutant were collected for immunoprecipitation using anti-Lys-Ac antibody. The immunoprecipitates were examined for acetylated AuA by blotting with anti-AuA antibody. Each experiment was repeated three times independently. (E) The lysine residues at positions 75 and 125 are conserved across species. Protein sequence alignment of AuA in various species. ARD1-mediated AuA acetylation at K75/K125 enhances AuA kinase activity The extracts from the overexpressing cells were immoprecipitated with anti Lys-Ac antibody and acetylated AuA levels were analyzed by immunoblotting with anti-AuA antibody. The experiments were performed at least three times independently. www.impactjournals.com/oncotarget www.impactjournals.com/oncotarget Oncotarget 57219 Figure 3: K75, K125 of Aurora A are acetylated by ARD1. (A) ARD1 acetylates the N-terminus of AuA in vitro. Top, delet mutants of His-AuA were subjected to in vitro acetylation assays with ARD1 recombinant protein. Bottom, construction of AuA delet mutants. AuA-N, AuA N-terminal domain; AuA-C, AuA C-terminal domain. (B) Lysines at positions 75 and 125 of AuA are acetyla by ARD1. Mutagenesis was performed to substitute lysine residues in N-terminus by arginine. Site-mutated AuA recombinants were th applied to in vitro assays. The experiments were performed at least three times independently. (C) Double mutant AuA K75R/K125R exhib the negligible acetylation. His-AuA wild type (WT) and K75R/K125R double mutant (DM) were subjected to in vitro acetylation assa with His-ARD1. Acetylation levels were assessed by western blot. Each experiment was repeated at least three times. (D) Double mut AuA K75R/K125R displays significantly decreased level of acetylated AuA in cells. The lysates from stable cells overexpressing RF AuA WT or RFP-AuA K75R/K125R mutant were collected for immunoprecipitation using anti-Lys-Ac antibody. The immunoprecipita were examined for acetylated AuA by blotting with anti-AuA antibody. Each experiment was repeated three times independently. (E) T lysine residues at positions 75 and 125 are conserved across species. Protein sequence alignment of AuA in various species. (F) The lys residues at positions 75 and 125 are not conserved among Aurora kinase family proteins. Protein sequence alignment of AuA in Aur kinase family proteins Figure 3: K75, K125 of Aurora A are acetylated by ARD1. (A) ARD1 acetylates the N-terminus of AuA in vitro. Top, delet mutants of His-AuA were subjected to in vitro acetylation assays with ARD1 recombinant protein. Bottom, construction of AuA delet mutants. AuA-N, AuA N-terminal domain; AuA-C, AuA C-terminal domain. (B) Lysines at positions 75 and 125 of AuA are acetyla by ARD1. Mutagenesis was performed to substitute lysine residues in N-terminus by arginine. Site-mutated AuA recombinants were th pplied to in vitro assays. The experiments were performed at least three times independently. (C) Double mutant AuA K75R/K125R exhib he negligible acetylation. His-AuA wild type (WT) and K75R/K125R double mutant (DM) were subjected to in vitro acetylation ass with His-ARD1. Acetylation levels were assessed by western blot. ARD1-mediated AuA acetylation at K75/K125 enhances AuA kinase activity Each experiment was repeated at least three times. (D) Double mut AuA K75R/K125R displays significantly decreased level of acetylated AuA in cells. The lysates from stable cells overexpressing RF AuA WT or RFP-AuA K75R/K125R mutant were collected for immunoprecipitation using anti-Lys-Ac antibody. The immunoprecipita were examined for acetylated AuA by blotting with anti-AuA antibody. Each experiment was repeated three times independently. (E) T ysine residues at positions 75 and 125 are conserved across species. Protein sequence alignment of AuA in various species. (F) The lys esidues at positions 75 and 125 are not conserved among Aurora kinase family proteins. Protein sequence alignment of AuA in Aur kinase family proteins. Figure 3: K75, K125 of Aurora A are acetylated by ARD1. (A) ARD1 acetylates the N-terminus of AuA in vitro. Top, deletion mutants of His-AuA were subjected to in vitro acetylation assays with ARD1 recombinant protein. Bottom, construction of AuA deletion mutants. AuA-N, AuA N-terminal domain; AuA-C, AuA C-terminal domain. (B) Lysines at positions 75 and 125 of AuA are acetylated by ARD1. Mutagenesis was performed to substitute lysine residues in N-terminus by arginine. Site-mutated AuA recombinants were then applied to in vitro assays. The experiments were performed at least three times independently. (C) Double mutant AuA K75R/K125R exhibits the negligible acetylation. His-AuA wild type (WT) and K75R/K125R double mutant (DM) were subjected to in vitro acetylation assays with His-ARD1. Acetylation levels were assessed by western blot. Each experiment was repeated at least three times. (D) Double mutant AuA K75R/K125R displays significantly decreased level of acetylated AuA in cells. The lysates from stable cells overexpressing RFP- AuA WT or RFP-AuA K75R/K125R mutant were collected for immunoprecipitation using anti-Lys-Ac antibody. The immunoprecipitates were examined for acetylated AuA by blotting with anti-AuA antibody. Each experiment was repeated three times independently. (E) The lysine residues at positions 75 and 125 are conserved across species. Protein sequence alignment of AuA in various species. (F) The lysine residues at positions 75 and 125 are not conserved among Aurora kinase family proteins. Protein sequence alignment of AuA in Aurora kinase family proteins. Figure 3: K75, K125 of Aurora A are acetylated by ARD1. (A) ARD1 acetylates the N-terminus of AuA in vitro. Top, deletion mutants of His-AuA were subjected to in vitro acetylation assays with ARD1 recombinant protein. Bottom, construction of AuA deletion mutants. AuA-N, AuA N-terminal domain; AuA-C, AuA C-terminal domain. ARD1-mediated AuA acetylation at K75/K125 enhances AuA kinase activity (F) The lysine residues at positions 75 and 125 are not conserved among Aurora kinase family proteins. Protein sequence alignment of AuA in Aurora kinase family proteins. www.impactjournals.com/oncotarget www.impactjournals.com/oncotarget Oncotarget 57220 acetylation influences its catalytic activity, we examined the phosphorylation level of AuA. Impressively, AuA phosphorylation was elevated in stable cells overexpressing wild-type (WT) AuA, but not in cells expressing the K75R/K125R mutant AuA (Figure 4A). AuA catalytic activity is strictly governed by an ATP cycle comprising ATP binding, ATP hydrolysis, and release of Pi and ADP [33, 34]. This ATP cycle is regulated by the ATP-binding site of AuA, which is located at the interface of the catalytic core [34]. Thus, the ability of AuA to bind ATP is critical for AuA activation. To elucidate the impact of AuA acetylation on its kinase activity, we next probed the ability of WT and K75R/K125R mutant AuA to bind ATP. We found that K75R/K125R mutant AuA exhibited significantly reduced ATP-binding capacity than the WT protein (Figure 4B). Furthermore, the ATPase activity of WT AuA was significantly enhanced in the presence ARD1, which acetylates AuA; however, the mutant AuA, which was unable to be acetylated, did not show significant change in ATPase activity (Figure 4C), indicating that ATP hydrolysis by acetylated AuA is more efficient than that by the non-acetylated mutant. Taken together, these data suggest that ARD1-mediated AuA acetylation at K75/ K125 positively regulates AuA kinase activity. of these processes causes abnormal cell proliferation in various diseases. Cell division is rigorously controlled by checkpoints at cell cycle stage transitions. Because of the important aspect of AuA in cell cycle regulation, we asked whether acetylation modification of AuA could be related to cell cycle regulation via cyclins. We found that cyclin E/CDK2 and cyclin B, which are regulators of G1/S and G2/M checkpoints, respectively, were dramatically upregulated in WT AuA-overexpressing cells, whereas the cells that overexpressed non-acetylated mimic mutation of AuA did not increase the expression levels of cyclin E/CDK2 and cyclin B1, suggesting that the acetylation of AuA at K75/K125 upregulate cyclin E/CDK2 and cyclin B1 expression (Figure 5D). Meanwhile, the expression level of p53 was downregulated in cells overexpressing WT AuA, but elevated in cells overexpressing mutant AuA. This result is consistent with that of a previous study, which showed that AuA-induced p53 phosphorylation triggers p53 degradation [35]. K75/K125 acetylation of AuA promotes cell migration Besides its mitotic functions, AuA was reported to exert non-mitotic effects on cell movement and neurite elongation [36, 37]. The observation that AuA and ARD1 colocalized in migrating cells prompted us to explore the role of AuA acetylation in cell movement. A wound healing assay was implemented, in which a scratch was created in a monolayer of cells, and the closure rate was measured. Defect in AuA acetylation decelerated the wound closure remarkably, demonstrating that AuA acetylation is essential for cell migration (Figure 6A). Furthermore, a transwell invasion assay, wherein cells were seeded on matrigel and stimulated to invade into the lower chamber through the matrigel, was performed. The number of WT AuA-overexpressing cells that invaded the lower chamber was notably higher than that of mutant AuA-expressing cells (Figure 6B), suggesting that AuA acetylation regulated cell movement. Indeed, along with an increase in phosphorylated AuA, acetylated AuA levels were elevated significantly (Figure 6C). To clarify the molecular mechanism of cell migration regulation by K75/ K125 acetylation, we analyzed AKT protein expression levels. Although AKT levels remained unchanged in both WT AuA- and K75R/K125R mutant AuA-overexpressing cells, phosphorylated AKT was augmented in WT AuA- overexpressing cells. AKT can be activated by several signaling pathways. We then tested expression level of p-38, p-ERK and p-FAK which are activators of AKT. In our study, AuA acetylation induced the phosphorylation of AKT via p38, but not via FAK or ERK signaling as phosphorylation level of p38 was elevated in WT AuA overexpressing cells, while the augmented phosphorylation ARD1-mediated AuA acetylation at K75/K125 enhances AuA kinase activity Collectively, these results reveal that AuA acetylation at K75/K125 enhances cell proliferation via cell cycle acceleration. ARD1-mediated AuA acetylation enhances cell proliferation AuA is a key regulator of centrosome maturation and separation, in preparation for cell division. Moreover, various mitotic proteins are regulated by AuA-induced phosphorylation to ensure chromosome segregation and cell cycle progression. Overexpression or downregulation of AuA causes inappropriate cell cycle progression, leading to promotion or inhibition of cell growth. Because acetylation of AuA enhanced AuA kinase activity, we sought to elucidate the role of AuA acetylation in cell proliferation. We seeded MCF7 cells in 96-well plates, incubated the cells for 48 h, treated them with MTT, and measured absorbance as an indicator of cell density. After 48 h, cells overexpressing WT AuA showed significant proliferation compared with cells overexpressing K75R/ K125R mutant AuA (Figure 5A). In confirmation with this data, clonogenic assays displayed the outgrowth of cells expressing acetylated AuA compared with cells expressing nonacetylated mutant AuA (Figure 5B). Then we checked distribution of cells in phases of cell cycle by the flowcytometry. Population of G0/G1 phase of AuA WT overexpressing cells decreased, whereas the number of cells in G2/M phase significantly increased, accompanying with the slight increase in S phase, indicating that acetylated AuA overexpressing cells had higher proliferative proportion than that of non- acetylated AuA mutant overexpressing cells (Figure 5C). Normal cell proliferation requires a balance between cell division and cell death. Dysregulation of either or both www.impactjournals.com/oncotarget Oncotarget 57221 Figure 4: ARD1-mediated AuA acetylation at K75/K125 enhances AuA kinase activity. (A) Phosphorylation level of AuA is decreased in cells overexpressing K75R/K125R mutant. Top, the collected extracts from MCF7 cells overexpressing RFP-AuA WT or DM were immunoprecipitated with anti-RFP antibody and were then immnublotted with anti-phospho- Serine/Threonine antibody (p-Ser/Thr) for examining phosphorylated AuA. Bottom, phosphorylated AuA levels were analyzed using ImageJ program. p < 0.005. Error bar indicates S.E.M (n = 3). (B) ATP-binding ability of AuA mutant is decreased. Lysates of RFP-AuA WT overexpressing cells and RFP-AuA K75R/125R overexpressing cells were incubated with ATP-agarose for 4 h, and ability of AuA binding to ATP was analyzed by western blot using anti- RFP tag antibody. Bottom, ATP-binding capacity was analyzed from western blot density using Image J program. p < 0.005. Error bar indicates S.E.M (n = 3). (C) Acetylation of AuA increases ATPase activity. His-AuA WT and His-AuA DM recombinants were acetylated, and subjected to reaction with ATP. ATPase activity was analyzed by measuring absorbance at 595 nm. p < 0.05; n.s, no significant. ARD1-mediated AuA acetylation enhances cell proliferation Error bar indicates S.E.M (n = 3). Figure 4: ARD1-mediated AuA acetylation at K75/K125 enhances AuA kinase activity. (A) Phosphorylation level of AuA is decreased in cells overexpressing K75R/K125R mutant. Top, the collected extracts from MCF7 cells overexpressing RFP-AuA WT or DM were immunoprecipitated with anti-RFP antibody and were then immnublotted with anti-phospho- Serine/Threonine antibody (p-Ser/Thr) for examining phosphorylated AuA. Bottom, phosphorylated AuA levels were analyzed using ImageJ program. p < 0.005. Error bar indicates S.E.M (n = 3). (B) ATP-binding ability of AuA mutant is decreased. Lysates of RFP-AuA WT overexpressing cells and RFP-AuA K75R/125R overexpressing cells were incubated with ATP-agarose for 4 h, and ability of AuA binding to ATP was analyzed by western blot using anti- RFP tag antibody. Bottom, ATP-binding capacity was analyzed from western blot density using Image J program. p < 0.005. Error bar indicates S.E.M (n = 3). (C) Acetylation of AuA increases ATPase activity. His-AuA WT and His-AuA DM recombinants were acetylated, and subjected to reaction with ATP. ATPase activity was analyzed by measuring absorbance at 595 nm. p < 0.05; n.s, no significant. Error bar indicates S.E.M (n = 3). www.impactjournals.com/oncotarget www.impactjournals.com/oncotarget Oncotarget 57222 Figure 5: ARD1-mediated AuA acetylation enhances cell proliferation. (A) Acetylation of AuA increases cell proliferatio overexpressing RFP-AuA WT and RFP-AuA K75R/K125R MCF7 cells were seeded in 96-well plates and cultured for 48 h. Cell hen treated with MTT and measured absorbance at 490nm. **p < 0.0005; p < 0.05. Error bar indicates S.E.M (n = 3). (B) Acety of AuA promotes colony formation of MCF7 cells. The overexpressing RFP-AuA WT and RFP-AuA K75/125 MCF7 cells were see 6-well plates and cultured for 7 days. Cells were then fixed with formaldehyde and stained with crystal violet to visualize colony form or quantification. Representative image is shown. *p < 0.005, p < 0.05. Error bar indicates S.E.M (n = 3). (C) The overexpressing AuA WT cells shows the decreased population in G0/G1 phase and increased number of cells in S phase and G2/M phase. RFP-Au and RFP-AuA K75R/K125R overexpressing MCF7 cells were stained with propidium iodide and analyzed by BD Calibur flowcyto *p < 0.0005; p < 0.005; p < 0.05. Error bar indicates S.E.M (n = 3). (D) Acetylation of AuA increases expression level of B1, cyclin E and CDK2, whereas decreases expression level of p53. ARD1-mediated AuA acetylation enhances cell proliferation Expression level of target proteins were analyzed by western bl quantified by Image J program. Vinculin was used as loading control. p < 0.05; p < 0.005. Error bar indicates S.E.M (n = 3). Figure 5: ARD1-mediated AuA acetylation enhances cell proliferation. (A) Acetylation of AuA increases cell proliferation. The overexpressing RFP-AuA WT and RFP-AuA K75R/K125R MCF7 cells were seeded in 96-well plates and cultured for 48 h. Cells were then treated with MTT and measured absorbance at 490nm. p < 0.0005; p < 0.05. Error bar indicates S.E.M (n = 3). (B) Acetylation of AuA promotes colony formation of MCF7 cells. The overexpressing RFP-AuA WT and RFP-AuA K75/125 MCF7 cells were seeded in 6-well plates and cultured for 7 days. Cells were then fixed with formaldehyde and stained with crystal violet to visualize colony formation for quantification. Representative image is shown. *p < 0.005, p < 0.05. Error bar indicates S.E.M (n = 3). (C) The overexpressing RFP- AuA WT cells shows the decreased population in G0/G1 phase and increased number of cells in S phase and G2/M phase. RFP-AuA WT and RFP-AuA K75R/K125R overexpressing MCF7 cells were stained with propidium iodide and analyzed by BD Calibur flowcytometry. *p < 0.0005; p < 0.005; p < 0.05. Error bar indicates S.E.M (n = 3). (D) Acetylation of AuA increases expression level of cyclin B1, cyclin E and CDK2, whereas decreases expression level of p53. Expression level of target proteins were analyzed by western blot and quantified by Image J program. Vinculin was used as loading control. p < 0.05; *p < 0.005. Error bar indicates S.E.M (n = 3). www.impactjournals.com/oncotarget www.impactjournals.com/oncotarget Oncotarget 57223 we measured MMP-2 expression levels. We noticed that MMP-2 expression was upregulated when cells exhibited ARD1-mediated AuA acetylation (Figure 6D), of p38 was not observed in the mutant AuA overexpressing cells (Figure 6D). Because MMP-2 is downstream of AKT and crucial for cell migration and cell invasion, Onco 57224 pactjournals.com/oncotarget and crucial for cell migration and cell invasion, exhibited ARD1-mediated AuA acetylation (Figur 6: K75/K125 acetylation of AuA promotes cell migration. (A) Acetylation of AuA promotes cell migration. assay of MCF7 cells was performed, the closure ratios were analyzed after 24 h.p < 0.05. Error bar indicates S.E.M (n = 3 0 μm. (B) Acetylation of AuA promotes cell invasion. Cells were seeded in inserted transwell. ARD1-mediated AuA acetylation enhances cell proliferation After 24 h, transwell was ystal violet and the invasive cells were counted under microscope. p < 0.05. Error bar indicates S.E.M (n = 3). Scale bar, etylation of AuA is upregulated in migrating cell. Cells were scratched to activate cell migration and the lysates were immunoprec ti-Lys-Ac antibody or p-Ser/Thr and immunoblotting using anti-AuA antibody. (D) Acetylation of AuA activates AKT ng but not ERK or FAK. Expression level of target proteins were analyzed by western blot and quantified by Image J p n was used as loading control. p < 0.05; *p < 0.005. Error bar indicates S.E.M (n = 3). Figure 6: K75/K125 acetylation of AuA promotes cell migration. (A) Acetylation of AuA promotes cell migration. Wound healing assay of MCF7 cells was performed, the closure ratios were analyzed after 24 h.p < 0.05. Error bar indicates S.E.M (n = 3). Scale bar, 100 μm. (B) Acetylation of AuA promotes cell invasion. Cells were seeded in inserted transwell. After 24 h, transwell was stained with crystal violet and the invasive cells were counted under microscope. p < 0.05. Error bar indicates S.E.M (n = 3). Scale bar, 50 μm. (C) Acetylation of AuA is upregulated in migrating cell. Cells were scratched to activate cell migration and the lysates were immunoprecipitated with anti-Lys-Ac antibody or p-Ser/Thr and immunoblotting using anti-AuA antibody. (D) Acetylation of AuA activates AKT by p38 signaling but not ERK or FAK. Expression level of target proteins were analyzed by western blot and quantified by Image J program. Vinculin was used as loading control. p < 0.05; **p < 0.005. Error bar indicates S.E.M (n = 3). www.impactjournals.com/oncotarget www.impactjournals.com/oncotarget Oncotarget 57224 suggesting that K75/K125 acetylation regulates cell migration via the p38/AKT/MMP-2 axis. AuA phosphorylation has been well studied for its role in activating AuA kinase activity, thereby inducing signaling cascades in mitotic as well as non-mitotic events [11, 37–39]. In addition to phosphorylation, a novel mechanism for the modulation of AuA function via acetylation by ARD1 was identified in this study (Figures 2B, 5D, 6D). Here, AuA was shown to colocalize with ARD1 at the centrosome (Figure 1A), and the acetylation of AuA at K75 and K125 located at N-terminus was shown to be modulated by ARD1 (Figure 3). The structure of the N-terminal domain of AuA has not been fully characterized yet, as it is difficult to crystalize the full-length AuA protein [40]. Moreover, the role of the N-terminal domain in AuA function is not clearly identified. It is reported that the N-terminus of AuA regulates its binding to partner proteins and orients AuA to the mitotic DISCUSSION To the best of our knowledge, this study is the first to identify post-translational modification of AuA by the acetyltransferase ARD1. We propose a mechanism by which AuA acetylation regulates cell proliferation and migration. AuA is acetylated on lysine residues at positions 75 and 125, resulting in the activation of its kinase activity. This leads to stimulation of cell cycle progression by cyclin E/CDK2 and cyclin B1, promoting cell proliferation on one hand and activation of p38/AKT/ MMP-2 pathway on the other hand, thereby stimulating cell migration (Figure 7). cell migration (Figure 7). binding to partner proteins and orients AuA to the mitotic Figure 7: Schematics for regulation of cell proliferation and migration by ARD1-mediated AuA acetylation. AuA is acetylated by ARD1 at lysine 75 and 125. Then, acetylation of AuA activates cyclin E/CDK2, promoting G1/S transition on one hand. On the other hand, acetylation of AuA increases cyclin B1 level, promoting G2/M transition, therefore, enhancing cell proliferation. In addition, acetylation of AuA by ARD1 stimulates p38 signaling, activating AKT-induced MMP-2, thus, promotes cell migration. Fi 7 S h ti f l ti f ll lif ti d i ti b ARD1 di t d A A t l ti A A i Figure 7: Schematics for regulation of cell proliferation and migration by ARD1-mediated AuA acetylation. AuA is acetylated by ARD1 at lysine 75 and 125. Then, acetylation of AuA activates cyclin E/CDK2, promoting G1/S transition on one hand. On the other hand, acetylation of AuA increases cyclin B1 level, promoting G2/M transition, therefore, enhancing cell proliferation. In addition, acetylation of AuA by ARD1 stimulates p38 signaling, activating AKT-induced MMP-2, thus, promotes cell migration. Figure 7: Schematics for regulation of cell proliferation and migration by ARD1-mediated AuA acetylation. AuA is acetylated by ARD1 at lysine 75 and 125. Then, acetylation of AuA activates cyclin E/CDK2, promoting G1/S transition on one hand. On the other hand, acetylation of AuA increases cyclin B1 level, promoting G2/M transition, therefore, enhancing cell proliferation. In addition, acetylation of AuA by ARD1 stimulates p38 signaling, activating AKT-induced MMP-2, thus, promotes cell migration. www.impactjournals.com/oncotarget www.impactjournals.com/oncotarget Oncotarget 57225 spindle. The members of the Aurora kinase family differ in their N-terminal domains, although sharing 70% homology in their C-terminal domains. The acetylation sites of AuA in the N-terminal domain probably contribute to the regulation of AuA and its localization during mitosis, and distinguish the roles of AuA in mitosis from those of other Aurora kinase family members. Whether AuA N-terminal domain assists the C-terminal domain in catalyzing the phosphorylation of its substrates is still controversial [9, 41]. Nonetheless, the kinase activity of AuA is dependent upon the binding partner and modification sites. In our study, ARD1 was shown to interact with AuA and to acetylate AuA at K75 and K125. Acetylation of K125, which is located near the ATP-binding region of AuA, probably caused conformational changes in the protein, leading to the positive regulation of AuA catalytic activity (Figure 4). It is likely that acetylation of AuA by ARD1 could promote self-phosphorylation of AuA. of AuA in cell migration and tumorigenesis have attracted considerable attention recently [16, 36, 51, 52]. Consistent with previous reports, our study demonstrated that ARD1- mediated AuA acetylation induces the expression of phosphorylated AKT to activate cell migration, and it provides a new regulatory mechanism by which AuA facilitates cell movement. Furthermore, MMP-2, a member of the matrix metalloproteinase family, contributes to cancer owing to its importantrole in degrading structural components of the extracellular matrix, thus paving ways for cells to migrate and invade to further locations [53]. In this study, we showed that the acetylation of K75/K125 of AuA is required for stimulating AKT signaling, resulting in MMP-2 induction to promote cell movement (Figure 6). In conclusion, our study shows that acetylation is a novel modification of AuA and that AuA acetylation stimulates cell proliferation and cell movement. Our findings suggest a possibility that AuA acetylation is a promising therapeutic target for anticancer drug development. Cell proliferation is important for development as it enables organs to attain the appropriate size, and exhibit normal morphology and function. As cells differentiate, their proliferation rate decreases. Uncontrollable cell proliferation results in tumorigenesis and cancer development. The balance between cell division and cell loss maintains proper cell proliferation. Therefore, the sophisticated regulatory mechanism of cell division plays a decisive role in cell proliferation and tumor growth. Cell cycle checkpoints are strictly controlled by multiple complicated mechanisms. Cyclin E/CDK2 is a key regulator of G1/S transition. www.impactjournals.com/oncotarget Upon activation, cyclin E/CDK2 phosphorylates and inactivates retinoblastoma protein, which induced the release of E2F resulting in G1/S transition. The overabundance of cyclin E/CDK2 activity boosts G1/S transition [42]. Cyclin B1 is another switch in cell cycle progression, and the excessive expression of cyclin B1 triggers many subcellular events, resulting in G2/M transition [43]. In this study, we demonstrated that acetylation of AuA at K75 and K125 activates the expression of cyclin E, CDK2, and cyclin B1 (Figure 5D). Thus, activation of cyclin E/CDK2 and cyclin B1 by ARD1- mediated AuA acetylation promotes cell proliferation. Additionally, p53 tumor suppressor involved in multiple cell cycle checkpoints was downregulated (Figure 5D). p53 expression arrests cells at the G1/S and G2/M transitions. In the S phase, p53 expression ensures that cells do not enter mitosis in absence of DNA replication [44]. Hence, the AuA-induced p53 reduction drives cell division, promoting cell proliferation. Antibodies Anti-AuA antibody (ab13824) and Anti-phospho Serine/Threonine (ab17464) antibody were purchased from Abcam. Anti-acetylated lysine (Lys-Ac) antibody (#9941), anti-AKT antibody (#4691S), anti-phospho AKT (S473) antibody (#9271), anti-p38 MAPK antibody (#9212), anti-phospho p38 MAPK antibody (#4511), anti-p44/42 MAPK (Erk1/2) antibody (#9102) were from Cell Signaling Technology. Anti-cyclin B1 antibody (GNS1, sc-245), anti-cyclin E antibody (HE12, sc-247), anti-CDK2 antibody (9E10, sc-40), anti-Vinculin antibody (H300, sc-5573), anti-GFP antibody (B2, sc-9996), anti- MMP2 antibody (H-76, sc-10736), anti-GAPDH antibody (G-9, sc-365062), anti p-ERK antibody (E-4, sc-7383), anti-CDK2 antibody (D-12, sc-6248) were purchased from Santa Cruz. Anti-RFP tag antibody (MA5-15257), anti-phosphor-FAK (S397) antibody (44-624G), and anti- FAK antibody (610087) were obtained from ThermoFisher Scientific, Invitrogen and BD Biolegend respectively. Immunoflourescence Cells were grown on circular glass coverslips plated in 24-well plate. Cells were fixed with formaldehyde in 10 min, permeabilized by incubating in PBS containing 0.25% Triton-X in 10 min and washed 3 times by PBS, then blocked for 1 h in 1% BSA. Primary antibody recognizing AuA (1:500) was diluted in blocking buffer and incubated overnight at 4°C. Alexa Flour®546 goat anti- mouse antibody was used at recommended concentration and incubated in 1h in the dark at room temperature. This was followed by counterstaining with Hoesch and mounting with Gel/mount (Biomeda Corp.). Cells were observed under Axiovert M200 microscope (Zeiss). In vitro acetylation assay BL21 cells transformed with plasmids pET28a- ARD1, pGEX-4T3 or pET28a-AuA were grown to and OD600 of 0.6–0.8. Overall, 1 mM IPTG was added to induce His-ARD1, GST-ARD1 or His-AuA, then cells were grown for overnight at 25°C. For His-tagged protein purification, cells were collected and proteins were extracted with a lysis buffer (pH 7) containing 50 mM NaH2PO4, 300 mM NaCl, 10 mM Imidazole and 1% Triton X-100. His-tagged proteins were purified with Ni-NTA and followed by eluted with an elution buffer containing 50 mM NaH2PO4, 300 mM NaCl, and 250 mM Imidazole. For GST-tagged protein purification, cells were collected and proteins were extracted with lysis buffer containing 250 mM Tris-Cl pH 8.5, 500 mM NaCl, 5 mM EDTA and 1% Triton X-100. Cell lysates were then incubated with Glutathione-Agarose for 4 h, following by washing three times with PBS to obtain purified GST-tagged protein. Acetylation assay was performed as described. Freshly prepared recombinant ARD1 and AuA were incubated in the reaction mixture [50 mmol/L Tris-HCl (pH 8.0), 0.1 mmol/LEDTA, 1mmol/LDTT, 10% glycerol, and 10 mmol/L acetyl- CoA] at 37°C. Each experiment was repeated at least 3 times. Human embryonic kidney HEK293T cell line and human breast cancer MCF7 cell line obtained from ATCC were grown in DMEM medium containing 10% FBS (fetal bovine serum) with penicillin and streptomycin in a 5% CO2 humidified atmosphere at 37oC. Cells were transfected using Polyethylenimine (PEI) reagent following manufacturer’s recommendation. For MCF7 stably overexpressing GFP-ARD1, RFP-AuA, MCF7 cells were transfected with pEFGP-C3-ARD1 and pTagRFP-C- AuA using Lipofectamine 2000 (Lifetechnologies) then selected by using G418 (geneticin). ATP-binding assay Total cell lysates were isolated using cell lysis buffer (Cell signaling). 1mg cell lysates were then subjected to ATP-agarose resins (Innova Biosciences) and incubated at 4°C for 4 h. The ATP-bound protein complexes were washed three times with the same buffer. Proteins bound to ATP were eluted by sample buffer and subjected to electrophoresis and immunoblotting. Each experiment was repeated 3 times. Plasmid construction Oligonucleotide primers were designed to amplified ARD1 gene (GenBank: NM_003491.3) and AuA gene (GenBank: NM_198433.2). Each primer covers additional sequences of restriction enzyme. ARD1 PCR product was digested and cloned into pET28b, pEGFP-C3 and pGEX- 4T3. AuA PCR products were digested and cloned into pTagRFP-C and pET28b. Deletion mutants of His-AuA were constructed from pET28b-AuA plasmid. cDNAs of AuA corresponding to 1–140 aa, and 126–403 aa were amplified by PCR and inserted into pET28b. The AKT pathway is a key signaling pathway in the regulation of physiological processes [45, 46]. Overexpression of AKT has been described in many types of cancer cells [47–49]. AKT is activated by phosphorylation, and it is able to phosphorylate and activate downstream substrates, stimulating cell migration [50]. Besides the conventional function of AuA in the mitotic regulation of the cell cycle, the non-mitotic roles www.impactjournals.com/oncotarget Oncotarget 57226 Cell synchronization MCF7 cells were synchronized by treated with cytosine arabinoside in 16 h for G1/S phase. After cells were released into fresh medium in 4 h, cells were then arrested in early mitosis by using nocodazole at concentration 100 ng/ml for 6 h. For metaphase, cells were then treated with MG132 for 30 min. And cells were in anaphase and telophase after being released from MG132 30 min and 60 min respectively. Immunoblotting and immunoprecipitation Recombinant proteins were applied to in vitro acetylation assay and ATPase activity were measured using High Throughput Colorimetric ATPase Assay Kit (Innova Biosciences) according to the manufacturer’s instructions. Briefly, recombinant proteins were incubated for 30 min at 30°C in a reaction buffer consisting of 50 mM Tris (pH 7.5), 2.5 mM MgCl2 and 0.5 mM ATP. After that PiColorLock Gold reagent and Accelerator were added to the mixture. Then Stabilizer was added and stabilized for 30 min. The absorbance at 595 nm was measured. Each experiment was repeated 3 times. Total cell lysates were isolated using cell lysis buffer (Cell signaling), which contained 20 mM Tris-HCl pH7.5, 150 mM NaCl, 1 mM Na2EDTA, 1 mM EGTA, 1% Triton, 2.5 mM sodium pyrophosphate, 1 mM beta- glycerophosphate, 1 mM Na3VO4, 1 μg/mi leupetin and the protease inhibitor cocktail set (Calbiochem). Lysates were separated by sodium dodecyl sulfate gel electrophoresis, and transferred to nitrocellulose membranes for Western blot analysis. Detection was performed using ECL (enhanced electrochemiluminescence) plus (Amersham Bioscience). For immunoprecipitation, lysates were immunoprecipitated with appropriate primary antibody. The immunoprecipitated complexes were subjected to electrophoresis and immunoblotting. Each experiment was repeated at least 3 times. CONFLICTS OF INTEREST The authors declare no conflicts of interest. Protein sequence alignment Protein sequence alignment was conducted using Clustal Omega. www.impactjournals.com/oncotarget Oncotarget 57227 Author contributions Cells were seeded in 96-well plate at a density of 103 cells per well and cultured for 48 h. The proliferation rates were measured using a Non-Radioactive Cell Proliferation Assay Kit (Promega) according to the manufacturer’s instructions. Briefly, 20 ul of freshly mixed tetrazolium/phenazine methosulfate was added, and the cells were incubated for 1 hour to allow color development. The absorbance at 490 nm was measured to indicate the number of viable cells. Each experiment was repeated 3 times. T.T.L.V designed and performed most of the experiments, assisted by J.H.P., J.H.S., E.J.L., H.C., H.L., S.J.B, S.A. J.H.P, H.-J.W., H.S.L. discussed the results and contributed to the manuscript. T.T.L.V. and K.-W.K. wrote the manuscript. K.-W.K. supervised the project. Cell cycle analysis 2 × 106 cells were harvested and washed twice with PBS, fixed overnight with 70% ethanol. After two washes with cold PBS, cells were incubated with RNAse for 30 min, following up with 30 min incubation with propidium iodine. Cell cycle was analyzed by BD Calibur flowcytometry. Each experiment was repeated 3 times. Transwell invasion assay Invasion assays were conducted using transwell invasion chambers coated with Matrigel matrix (Corning). Cells were seeded in the upper chambers (104 cells/well) in FBS-free DMEM. DMEM containing 10% FBS was added to the lower chambers. After 24 h of incubation, invaded cells on the lower surface were stained with crystal violet stain and counted under a light microscope. Each experiment was repeated 3 times. 6. Zhou H, Kuang J, Zhong L, Kuo Wl, Gray J, Sahin A, Brinkley B, Sen S. Tumour amplified kinase STK15/ BTAK induces centrosome amplification, aneuploidy and transformation. Nat Genet. 1998; 20:189–193. 7. Bischoff JR, Anderson L, Zhu Y, Mossie K, Ng L, Souza B, Schryver B, Flanagan P, Clairvoyant F, Ginther C, Chan CS, Novotny M, Slamon DJ, et al. A homologue of Drosophila aurora kinase is oncogenic and amplified in human colorectal cancers. EMBO J. 1998; 17. REFERENCES 1. Hanahan D, Weinberg Robert A. Hallmarks of Cancer: The Next Generation. Cell. 144:646–674. 1. Hanahan D, Weinberg Robert A. Hallmarks of Cancer: The Next Generation. Cell. 144:646–674. Clonogenic assay Cells were seeded in 6-well plate at a density of 103 cells per well and cultured for 7 days to form colonies. Colonies then were fixed with 1% formaldehyde and stained with crystal violet (0.5%) and counted. Each experiment was repeated 3 times. ACKNOWLEDGMENTS AND FUNDING This work was supported by the Global Research Laboratory Program (2011-0021874), Brain Korea 21 Program, the Global Core Research Center (GCRC) Program (20110030001), Bio & Medical Technology Development Program (2015M3A9E6028949), NRF grant (2015R1C1A2A01054446) through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (MSIP), Basic Science Research Program (2013R1A1A2058956, 2016R1D1A1B03935560) and Basic Science Research Program (NRF-2017R1A6A3A11032239) through the NRF funded by the Ministry of Education. Wound healing assay 2. Vicente-Manzanares M, Webb DJ, Horwitz AR. Cell migration at a glance. J Cell Sci. 2005; 118:4917–4919. 2. Vicente-Manzanares M, Webb DJ, Horwitz AR. Cell migration at a glance. J Cell Sci. 2005; 118:4917–4919. Wounds were created in confluent MCF7 cells by 200 μl tip and washed twice by PBS to remove non- adherent cells. To inhibit the influence of cell proliferation, cells were treated with 10 μM cytosine arabinoside. After 24 h incubation, the wound closure rates were measure under a phase-contrast microscope. Each experiment was repeated 3 times. 3. Nigg EA. Mitotic kinases as regulators of cell division and its checkpoints. Nat Rev Mol Cell Biol. 2001; 2:21–32. 4. Fu J, Bian M, Jiang Q, Zhang C. Roles of Aurora Kinases in Mitosis and Tumorigenesis. Mol Cancer Res. 2007; 5:1–10. 5. Bar-Shira A, Pinthus JH, Rozovsky U, Goldstein M, Sellers WR, Yaron Y, Eshhar Z, Orr-Urtreger A. Multiple Genes in Human 20q13 Chromosomal Region Are Involved in an Advanced Prostate Cancer Xenograft. Cancer Res. 2002; 62:6803–6807. www.impactjournals.com/oncotarget Statistical analysis Data analysis and statistical significance were tested using Graphpad Prism 6.0. P values were calculated by applying two-tailed Student’s t test. Statistical significance was determined at P < 0.05. 8. Marumoto T, Zhang D, Saya H. Aurora-A — A guardian of poles. Nat Rev Cancer. 2005; 5:42–50. 9. Cheetham GMT, Knegtel RMA, Coll JT, Renwick SB, Swenson L, Weber P, Lippke JA, Austen DA. Crystal www.impactjournals.com/oncotarget Oncotarget 57228 22. Farria A, Li W, Dent SYR. KATs in cancer: functions and therapies. Oncogene. 2015; 34:4901–4913. Structure of Aurora-2, an Oncogenic Serine/Threonine Kinase. J Biol Chem. 2002; 277:42419–42422. Structure of Aurora-2, an Oncogenic Serine/Threonine Kinase. J Biol Chem. 2002; 277:42419–42422. 10. Bayliss R, Sardon T, Vernos I, Conti E. Structural Basis of Aurora-A Activation by TPX2 at the Mitotic Spindle. Mol Cell. 2003; 12:851–862. 23. Yang XJ, Seto E. Lysine Acetylation: Codified Crosstalk with Other Posttranslational Modifications. Mol Cell. 31:449–461. 11. Zorba A, Buosi V, Kutter S, Kern N, Pontiggia F, Cho YJ, Kern D. Molecular mechanism of Aurora A kinase autophosphorylation and its allosteric activation by TPX2. eLife. 2014; 3:e02667. 24. Park EC, Szostak JW. ARD1 and NAT1 proteins form a complex that has N-terminal acetyltransferase activity. EMBO J. 1992; 11:2087–2093. 25. Seo JH, Park JH, Lee EJ, Vo TTL, Choi H, Kim JY, Jang JK, Wee HJ, Lee HS, Jang SH, Park ZY, Jeong J, Lee KJ, et al. ARD1-mediated Hsp70 acetylation balances stress-induced protein refolding and degradation. Nat Commun. 2016; 7:12882. 12. Kufer TA, Silljé HHW, Körner R, Gruss OJ, Meraldi P, Nigg EA. Human TPX2 is required for targeting Aurora-A kinase to the spindle. J Cell Biol. 2002; 158:617–623. 26. Shin DH, Chun YS, Lee KH, Shin HW, Park JW. Arrest Defective-1 Controls Tumor Cell Behavior by Acetylating Myosin Light Chain Kinase. PLoS One. 2009; 4:e7451. 13. Hirota T, Kunitoku N, Sasayama T, Marumoto T, Zhang D, Nitta M, Hatakeyama K, Saya H. Aurora-A and an Interacting Activator, the LIM Protein Ajuba, Are Required for Mitotic Commitment in Human Cells. Cell. 2003; 114:585–598. 27. Lim JH, Park JW, Chun YS. Human Arrest Defective 1 Acetylates and Activates β-Catenin, Promoting Lung Cancer Cell Proliferation. Cancer Res. 2006; 66:10677–10682. 14. Satinover DL, Brautigan DL, Stukenberg PT. Aurora-A Kinase and Inhibitor-2 Regulate the Cyclin Threshold for Mitotic Entry in Xenopus Early Embryonic Cell Cycles. Cell Cycle. 2006; 5:2268–2274. 28. Statistical analysis Seo JH, Cha JH, Park JH, Jeong CH, Park ZY, Lee HS, Oh SH, Kang JH, Suh SW, Kim KH, Ha JY, Han SH, Kim SH, et al. Arrest Defective 1 Autoacetylation Is a Critical Step in Its Ability to Stimulate Cancer Cell Proliferation. Cancer Res. 2010; 70:4422–4432. 15. Marumoto T, Hirota T, Morisaki T, Kunitoku N, Zhang D, Ichikawa Y, Sasayama T, Kuninaka S, Mimori T, Tamaki N, Kimura M, Okano Y, Saya H. Roles of aurora-A kinase in mitotic entry and G2 checkpoint in mammalian cells. Genes Cells. 2002; 7:1173–1182. 29. Arnesen T, Gromyko D, Pendino F, Ryningen A, Varhaug JE, Lillehaug JR. Induction of apoptosis in human cells by RNAi-mediated knockdown of hARD1 and NATH, components of the protein N-α-acetyltransferase complex. Oncogene. 2006; 25:4350–4360. 16. Wang Lh, Xiang J, Yan M, Zhang Y, Zhao Y, Yue Cf, Xu J, Zheng Fm, Chen Jn, Kang Z, Chen Ts, Xing D, Liu Q. The Mitotic Kinase Aurora-A Induces Mammary Cell Migration and Breast Cancer Metastasis by Activating the Cofilin-F- actin Pathway. Cancer Res. 2010; 70:9118–9128. 30. Bischoff JR, Plowman GD. The Aurora/Ipl1p kinase family: regulators of chromosome segregation and cytokinesis. Trends Cell Biol. 1999; 9:454–459. 17. D’Assoro AB, Liu T, Quatraro C, Amato A, Opyrchal M, Leontovich A, Ikeda Y, Ohmine S, Lingle W, Suman V, Ecsedy J, Iankov I, Di Leonardo A, et al. The mitotic kinase Aurora-A promotes distant metastases by inducing epithelial-to-mesenchymal transition in ERα+ breast cancer cells. Oncogene. 2014; 33:599–610. 31. Adams RR, Wheatleya SP, Gouldsworthy AM, Kandels- Lewis SE, Carmena M, Smythe C, Gerloff DL, Earnshaw WC. INCENP binds the Aurora-related kinase AIRK2 and is required to target it to chromosomes, the central spindle and cleavage furrow. Curr Biol. 2000; 10:1075–1078. 32. Scliwa M, Höner B. Microtubules, centrosomes and intermediate filaments in directed cell movement. Trends Cell Biol. 1993; 3:377–380. 18. Xu J, Yue Cf, Zhou Wh, Qian Ym, Zhang Y, Wang Sw, Liu AW, Liu Q. Aurora-A contributes to cisplatin resistance and lymphatic metastasis in non-small cell lung cancer and predicts poor prognosis. J Transl Med. 2014; 12:200. 33. Xu X, Wang X, Xiao Z, Li Y, Wang Y. Two TPX2-Dependent Switches Control the Activity of Aurora A. PLoS One. 2011; 6:e16757. 19. Kamran M, Long ZJ, Xu D, Lv SS, Liu B, Wang CL, Xu J, Lam EWF, Liu Q. Statistical analysis Aurora kinase A regulates Survivin stability through targeting FBXL7 in gastric cancer drug resistance and prognosis. Oncogenesis. 2017; 6:e298. 34. Zhao B, Smallwood A, Yang J, Koretke K, Nurse K, Calamari A, Kirkpatrick RB, Lai Z. Modulation of kinase-inhibitor interactions by auxiliary protein binding: Crystallography studies on Aurora A interactions with VX- 680 and with TPX2. Protein Sci. 2008; 17:1791–1797. 20. Goos JACM, Coupe VMH, Diosdado B, Delis-Van Diemen PM, Karga C, Belien JAM, Carvalho B, van den Tol MP, Verheul HMW, Geldof AA, Meijer GA, Hoekstra OS, Fijneman RJA. Aurora kinase A (AURKA) expression in colorectal cancer liver metastasis is associated with poor prognosis. Br J Cancer. 2013; 109:2445–2452. 35. Hobson A, Draney C, Stratford A, Becker TC, Lu D, Arlotto M, Tessem JS. Aurora Kinase A is critical for the Nkx6.1 mediated β-cell proliferation pathway. Islets. 2015; 7:e1027854. 21. Xu X, Wang Y, Chen Z, Sternlicht MD, Hidalgo M, Steffensen B. Matrix Metalloproteinase-2 Contributes to Cancer Cell Migration on Collagen. Cancer Res. 2005; 65:130–136. 36. Wang X, Lu N, Niu B, Chen X, Xie J, Cheng N. Overexpression of Aurora-A Enhances Invasion and Matrix www.impactjournals.com/oncotarget Oncotarget 57229 Metalloproteinase-2 Expression in Esophageal Squamous Cell Carcinoma Cells. Mol Cancer Res. 2012; 10:588–596. 45. Altomare DA, Testa JR. Perturbations of the AKT signaling pathway in human cancer. Oncogene. 2005; 24:7455–7464. Metalloproteinase-2 Expression in Esophageal Squamous Cell Carcinoma Cells. Mol Cancer Res. 2012; 10:588–596. 37. Mori D, Yamada M, Mimori-Kiyosue Y, Shirai Y, Suzuki A, Ohno S, Saya H, Wynshaw-Boris A, Hirotsune S. An essential role of the aPKC-Aurora A-NDEL1 pathway in neurite elongation by modulation of microtubule dynamics. Nat Cell Biol. 2009; 11:1057–1068. 46. Manning BD, Cantley LC. AKT/PKB Signaling: Navigating Downstream. Cell. 2007; 129:1261–1274. 47. David O, Jett J, LeBeau H, Dy G, Hughes J, Friedman M, Brody AR. Phospho-Akt Overexpression in Non–Small Cell Lung Cancer Confers Significant Stage-Independent Survival Disadvantage. Clin Cancer Res. 2004; 10:6865–6871. 38. Ohashi S, Sakashita G, Ban R, Nagasawa M, Matsuzaki H, Murata Y, Taniguchi H, Shima H, Furukawa K, Urano T. Phospho-regulation of human protein kinase Aurora-A: analysis using anti-phospho-Thr288 monoclonal antibodies. Oncogene. 2006; 25:7691–7702. 48. Clark Abbe R, Toker A. Signalling specificity in the Akt pathway in breast cancer. Biochem Soc Trans. 2014; 42:1349–1355. 49. Lee H, Kim M, Baek M, Morales LD, Jang IS, Slaga TJ, DiGiovanni J, Kim DJ. www.impactjournals.com/oncotarget Statistical analysis Targeted disruption of TC-PTP in the proliferative compartment augments STAT3 and AKT signaling and skin tumor development. Sci Rep. 2017; 7:45077. 39. Wike CL, Graves HK, Hawkins R, Gibson MD, Ferdinand MB, Zhang T, Chen Z, Hudson DF, Ottesen JJ, Poirier MG, Schumacher J, Tyler JK. Aurora-A mediated histone H3 phosphorylation of threonine 118 controls condensin I and cohesin occupancy in mitosis. eLife. 2016; 5:e11402. 50. Chin YR, Toker A. The Actin-Bundling Protein Palladin Is an Akt1-Specific Substrate that Regulates Breast Cancer Cell Migration. Mol Cell. 2010; 38:333–344. 40. Bai M, Ni J, Shen S, Wu J, Huang Q, Le Y, Yu L. Two newly identified sites in the N-terminal regulatory domain of Aurora-A are essential for auto-inhibition. Biotechnol Lett. 2014; 36:1595–1604. 51. Wu J, Yang L, Shan Y, Cai C, Wang S, Zhang HAO. AURKA promotes cell migration and invasion of head and neck squamous cell carcinoma through regulation of the AURKA/Akt/FAK signaling pathway. Oncol Lett. 2016; 11:1889–1894. 41. Zhang Y, Ni J, Huang Q, Ren W, Yu L, Zhao S. Identification of the auto-inhibitory domains of Aurora-A kinase. Biochem Biophys Res Commun. 2007; 357:347–352. 52. Do TV, Xiao F, Bickel LE, Klein-Szanto AJ, Pathak HB, Hua X, Howe C, O’Brien SW, Maglaty M, Ecsedy JA, Litwin S, Golemis EA, Schilder RJ, et al. Aurora kinase A mediates epithelial ovarian cancer cell migration and adhesion. Oncogene. 2014; 33:539–549. 42. Hwang HC, Clurman BE. Cyclin E in normal and neoplastic cell cycles. Oncogene. 2005; 24:2776–2786. 43. Pines J. Four-dimensional control of the cell cycle. Nat Cell Biol. 1999; 1:E73–E79. 53. Page-McCaw A, Ewald AJ, Werb Z. Matrix metalloproteinases and the regulation of tissue remodelling. Nat Rev Mol Cell Biol. 2007; 8:221–233. 44. Giono LE, Manfredi JJ. The p53 tumor suppressor participates in multiple cell cycle checkpoints. J Cell Physio. 2006; 209:13–20. www.impactjournals.com/oncotarget Oncotarget 57230

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.